diff options
Diffstat (limited to '')
101 files changed, 20772 insertions, 0 deletions
diff --git a/Documentation/driver-api/80211/cfg80211.rst b/Documentation/driver-api/80211/cfg80211.rst new file mode 100644 index 000000000..eeab91b59 --- /dev/null +++ b/Documentation/driver-api/80211/cfg80211.rst @@ -0,0 +1,354 @@ +================== +cfg80211 subsystem +================== + +.. kernel-doc:: include/net/cfg80211.h + :doc: Introduction + +Device registration +=================== + +.. kernel-doc:: include/net/cfg80211.h + :doc: Device registration + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_channel_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_channel + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_rate_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_rate + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_sta_ht_cap + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_supported_band + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_signal_type + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_params_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy + +.. kernel-doc:: include/net/cfg80211.h + :functions: wireless_dev + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_new + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_read_of_freq_limits + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_register + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_unregister + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_free + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_name + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_dev + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_priv + +.. kernel-doc:: include/net/cfg80211.h + :functions: priv_to_wiphy + +.. kernel-doc:: include/net/cfg80211.h + :functions: set_wiphy_dev + +.. kernel-doc:: include/net/cfg80211.h + :functions: wdev_priv + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_iface_limit + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_iface_combination + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_check_combinations + +Actions and configuration +========================= + +.. kernel-doc:: include/net/cfg80211.h + :doc: Actions and configuration + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_ops + +.. kernel-doc:: include/net/cfg80211.h + :functions: vif_params + +.. kernel-doc:: include/net/cfg80211.h + :functions: key_params + +.. kernel-doc:: include/net/cfg80211.h + :functions: survey_info_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: survey_info + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_beacon_data + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_ap_settings + +.. kernel-doc:: include/net/cfg80211.h + :functions: station_parameters + +.. kernel-doc:: include/net/cfg80211.h + :functions: rate_info_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: rate_info + +.. kernel-doc:: include/net/cfg80211.h + :functions: station_info + +.. kernel-doc:: include/net/cfg80211.h + :functions: monitor_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: mpath_info_flags + +.. kernel-doc:: include/net/cfg80211.h + :functions: mpath_info + +.. kernel-doc:: include/net/cfg80211.h + :functions: bss_parameters + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_txq_params + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_crypto_settings + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_auth_request + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_assoc_request + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_deauth_request + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_disassoc_request + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_ibss_params + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_connect_params + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_pmksa + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_rx_mlme_mgmt + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_auth_timeout + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_rx_assoc_resp + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_assoc_timeout + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_tx_mlme_mgmt + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_ibss_joined + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_connect_resp_params + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_connect_done + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_connect_result + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_connect_bss + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_connect_timeout + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_roamed + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_disconnected + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_ready_on_channel + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_remain_on_channel_expired + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_new_sta + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_rx_mgmt + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_mgmt_tx_status + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_cqm_rssi_notify + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_cqm_pktloss_notify + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_michael_mic_failure + +Scanning and BSS list handling +============================== + +.. kernel-doc:: include/net/cfg80211.h + :doc: Scanning and BSS list handling + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_ssid + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_scan_request + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_scan_done + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_bss + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_inform_bss + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_inform_bss_frame_data + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_inform_bss_data + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_unlink_bss + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_find_ie + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_bss_get_ie + +Utility functions +================= + +.. kernel-doc:: include/net/cfg80211.h + :doc: Utility functions + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_channel_to_frequency + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_frequency_to_channel + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_get_channel + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_get_response_rate + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_hdrlen + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_get_hdrlen_from_skb + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_radiotap_iterator + +Data path helpers +================= + +.. kernel-doc:: include/net/cfg80211.h + :doc: Data path helpers + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_data_to_8023 + +.. kernel-doc:: include/net/cfg80211.h + :functions: ieee80211_amsdu_to_8023s + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_classify8021d + +Regulatory enforcement infrastructure +===================================== + +.. kernel-doc:: include/net/cfg80211.h + :doc: Regulatory enforcement infrastructure + +.. kernel-doc:: include/net/cfg80211.h + :functions: regulatory_hint + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_apply_custom_regulatory + +.. kernel-doc:: include/net/cfg80211.h + :functions: freq_reg_info + +RFkill integration +================== + +.. kernel-doc:: include/net/cfg80211.h + :doc: RFkill integration + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_rfkill_set_hw_state + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_rfkill_start_polling + +.. kernel-doc:: include/net/cfg80211.h + :functions: wiphy_rfkill_stop_polling + +Test mode +========= + +.. kernel-doc:: include/net/cfg80211.h + :doc: Test mode + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_testmode_alloc_reply_skb + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_testmode_reply + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_testmode_alloc_event_skb + +.. kernel-doc:: include/net/cfg80211.h + :functions: cfg80211_testmode_event diff --git a/Documentation/driver-api/80211/conf.py b/Documentation/driver-api/80211/conf.py new file mode 100644 index 000000000..4424b4b0b --- /dev/null +++ b/Documentation/driver-api/80211/conf.py @@ -0,0 +1,10 @@ +# -*- coding: utf-8; mode: python -*- + +project = "Linux 802.11 Driver Developer's Guide" + +tags.add("subproject") + +latex_documents = [ + ('index', '80211.tex', project, + 'The kernel development community', 'manual'), +] diff --git a/Documentation/driver-api/80211/index.rst b/Documentation/driver-api/80211/index.rst new file mode 100644 index 000000000..af210859d --- /dev/null +++ b/Documentation/driver-api/80211/index.rst @@ -0,0 +1,17 @@ +===================================== +Linux 802.11 Driver Developer's Guide +===================================== + +.. toctree:: + + introduction + cfg80211 + mac80211 + mac80211-advanced + +.. only:: subproject and html + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/80211/introduction.rst b/Documentation/driver-api/80211/introduction.rst new file mode 100644 index 000000000..4938fa876 --- /dev/null +++ b/Documentation/driver-api/80211/introduction.rst @@ -0,0 +1,17 @@ +============ +Introduction +============ + +Explaining wireless 802.11 networking in the Linux kernel + +Copyright 2007-2009 Johannes Berg + +These books attempt to give a description of the various subsystems +that play a role in 802.11 wireless networking in Linux. Since these +books are for kernel developers they attempts to document the +structures and functions used in the kernel as well as giving a +higher-level overview. + +The reader is expected to be familiar with the 802.11 standard as +published by the IEEE in 802.11-2007 (or possibly later versions). +References to this standard will be given as "802.11-2007 8.1.5". diff --git a/Documentation/driver-api/80211/mac80211-advanced.rst b/Documentation/driver-api/80211/mac80211-advanced.rst new file mode 100644 index 000000000..70a89b216 --- /dev/null +++ b/Documentation/driver-api/80211/mac80211-advanced.rst @@ -0,0 +1,295 @@ +============================= +mac80211 subsystem (advanced) +============================= + +Information contained within this part of the book is of interest only +for advanced interaction of mac80211 with drivers to exploit more +hardware capabilities and improve performance. + +LED support +=========== + +Mac80211 supports various ways of blinking LEDs. Wherever possible, +device LEDs should be exposed as LED class devices and hooked up to the +appropriate trigger, which will then be triggered appropriately by +mac80211. + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_tx_led_name + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_rx_led_name + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_assoc_led_name + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_radio_led_name + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tpt_blink + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tpt_led_trigger_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_create_tpt_led_trigger + +Hardware crypto acceleration +============================ + +.. kernel-doc:: include/net/mac80211.h + :doc: Hardware crypto acceleration + +.. kernel-doc:: include/net/mac80211.h + :functions: set_key_cmd + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_key_conf + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_key_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_tkip_p1k + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_tkip_p1k_iv + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_tkip_p2k + +Powersave support +================= + +.. kernel-doc:: include/net/mac80211.h + :doc: Powersave support + +Beacon filter support +===================== + +.. kernel-doc:: include/net/mac80211.h + :doc: Beacon filter support + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_beacon_loss + +Multiple queues and QoS support +=============================== + +TBD + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_queue_params + +Access point mode support +========================= + +TBD + +Some parts of the if_conf should be discussed here instead + +Insert notes about VLAN interfaces with hw crypto here or in the hw +crypto chapter. + +support for powersaving clients +------------------------------- + +.. kernel-doc:: include/net/mac80211.h + :doc: AP support for powersaving clients + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_get_buffered_bc + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_beacon_get + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_sta_eosp + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_frame_release_type + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_sta_ps_transition + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_sta_ps_transition_ni + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_sta_set_buffered + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_sta_block_awake + +Supporting multiple virtual interfaces +====================================== + +TBD + +Note: WDS with identical MAC address should almost always be OK + +Insert notes about having multiple virtual interfaces with different MAC +addresses here, note which configurations are supported by mac80211, add +notes about supporting hw crypto with it. + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_iterate_active_interfaces + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_iterate_active_interfaces_atomic + +Station handling +================ + +TODO + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_sta + +.. kernel-doc:: include/net/mac80211.h + :functions: sta_notify_cmd + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_find_sta + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_find_sta_by_ifaddr + +Hardware scan offload +===================== + +TBD + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_scan_completed + +Aggregation +=========== + +TX A-MPDU aggregation +--------------------- + +.. kernel-doc:: net/mac80211/agg-tx.c + :doc: TX A-MPDU aggregation + +.. WARNING: DOCPROC directive not supported: !Cnet/mac80211/agg-tx.c + +RX A-MPDU aggregation +--------------------- + +.. kernel-doc:: net/mac80211/agg-rx.c + :doc: RX A-MPDU aggregation + +.. WARNING: DOCPROC directive not supported: !Cnet/mac80211/agg-rx.c + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_ampdu_mlme_action + +Spatial Multiplexing Powersave (SMPS) +===================================== + +.. kernel-doc:: include/net/mac80211.h + :doc: Spatial multiplexing power save + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_request_smps + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_smps_mode + +TBD + +This part of the book describes the rate control algorithm interface and +how it relates to mac80211 and drivers. + +Rate Control API +================ + +TBD + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_start_tx_ba_session + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_start_tx_ba_cb_irqsafe + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_stop_tx_ba_session + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_stop_tx_ba_cb_irqsafe + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rate_control_changed + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_rate_control + +.. kernel-doc:: include/net/mac80211.h + :functions: rate_control_send_low + +TBD + +This part of the book describes mac80211 internals. + +Key handling +============ + +Key handling basics +------------------- + +.. kernel-doc:: net/mac80211/key.c + :doc: Key handling basics + +MORE TBD +-------- + +TBD + +Receive processing +================== + +TBD + +Transmit processing +=================== + +TBD + +Station info handling +===================== + +Programming information +----------------------- + +.. kernel-doc:: net/mac80211/sta_info.h + :functions: sta_info + +.. kernel-doc:: net/mac80211/sta_info.h + :functions: ieee80211_sta_info_flags + +STA information lifetime rules +------------------------------ + +.. kernel-doc:: net/mac80211/sta_info.c + :doc: STA information lifetime rules + +Aggregation +=========== + +.. kernel-doc:: net/mac80211/sta_info.h + :functions: sta_ampdu_mlme + +.. kernel-doc:: net/mac80211/sta_info.h + :functions: tid_ampdu_tx + +.. kernel-doc:: net/mac80211/sta_info.h + :functions: tid_ampdu_rx + +Synchronisation +=============== + +TBD + +Locking, lots of RCU diff --git a/Documentation/driver-api/80211/mac80211.rst b/Documentation/driver-api/80211/mac80211.rst new file mode 100644 index 000000000..85a8335e8 --- /dev/null +++ b/Documentation/driver-api/80211/mac80211.rst @@ -0,0 +1,216 @@ +=========================== +mac80211 subsystem (basics) +=========================== + +You should read and understand the information contained within this +part of the book while implementing a mac80211 driver. In some chapters, +advanced usage is noted, those may be skipped if this isn't needed. + +This part of the book only covers station and monitor mode +functionality, additional information required to implement the other +modes is covered in the second part of the book. + +Basic hardware handling +======================= + +TBD + +This chapter shall contain information on getting a hw struct allocated +and registered with mac80211. + +Since it is required to allocate rates/modes before registering a hw +struct, this chapter shall also contain information on setting up the +rate/mode structs. + +Additionally, some discussion about the callbacks and the general +programming model should be in here, including the definition of +ieee80211_ops which will be referred to a lot. + +Finally, a discussion of hardware capabilities should be done with +references to other parts of the book. + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_hw + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_hw_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: SET_IEEE80211_DEV + +.. kernel-doc:: include/net/mac80211.h + :functions: SET_IEEE80211_PERM_ADDR + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_ops + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_alloc_hw + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_register_hw + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_unregister_hw + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_free_hw + +PHY configuration +================= + +TBD + +This chapter should describe PHY handling including start/stop callbacks +and the various structures used. + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_conf + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_conf_flags + +Virtual interfaces +================== + +TBD + +This chapter should describe virtual interface basics that are relevant +to the driver (VLANs, MGMT etc are not.) It should explain the use of +the add_iface/remove_iface callbacks as well as the interface +configuration callbacks. + +Things related to AP mode should be discussed there. + +Things related to supporting multiple interfaces should be in the +appropriate chapter, a BIG FAT note should be here about this though and +the recommendation to allow only a single interface in STA mode at +first! + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_vif + +Receive and transmit processing +=============================== + +what should be here +------------------- + +TBD + +This should describe the receive and transmit paths in mac80211/the +drivers as well as transmit status handling. + +Frame format +------------ + +.. kernel-doc:: include/net/mac80211.h + :doc: Frame format + +Packet alignment +---------------- + +.. kernel-doc:: net/mac80211/rx.c + :doc: Packet alignment + +Calling into mac80211 from interrupts +------------------------------------- + +.. kernel-doc:: include/net/mac80211.h + :doc: Calling mac80211 from interrupts + +functions/definitions +--------------------- + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rx_status + +.. kernel-doc:: include/net/mac80211.h + :functions: mac80211_rx_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: mac80211_tx_info_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: mac80211_tx_control_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: mac80211_rate_control_flags + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_rate + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_info + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_info_clear_status + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rx + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rx_ni + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rx_irqsafe + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_status + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_status_ni + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_tx_status_irqsafe + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rts_get + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_rts_duration + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_ctstoself_get + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_ctstoself_duration + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_generic_frame_duration + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_wake_queue + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_stop_queue + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_wake_queues + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_stop_queues + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_queue_stopped + +Frame filtering +=============== + +.. kernel-doc:: include/net/mac80211.h + :doc: Frame filtering + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_filter_flags + +The mac80211 workqueue +====================== + +.. kernel-doc:: include/net/mac80211.h + :doc: mac80211 workqueue + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_queue_work + +.. kernel-doc:: include/net/mac80211.h + :functions: ieee80211_queue_delayed_work diff --git a/Documentation/driver-api/basics.rst b/Documentation/driver-api/basics.rst new file mode 100644 index 000000000..826e85d50 --- /dev/null +++ b/Documentation/driver-api/basics.rst @@ -0,0 +1,129 @@ +Driver Basics +============= + +Driver Entry and Exit points +---------------------------- + +.. kernel-doc:: include/linux/module.h + :internal: + +Driver device table +------------------- + +.. kernel-doc:: include/linux/mod_devicetable.h + :internal: + +Delaying, scheduling, and timer routines +---------------------------------------- + +.. kernel-doc:: include/linux/sched.h + :internal: + +.. kernel-doc:: kernel/sched/core.c + :export: + +.. kernel-doc:: kernel/sched/cpupri.c + :internal: + +.. kernel-doc:: kernel/sched/fair.c + :internal: + +.. kernel-doc:: include/linux/completion.h + :internal: + +.. kernel-doc:: kernel/time/timer.c + :export: + +Wait queues and Wake events +--------------------------- + +.. kernel-doc:: include/linux/wait.h + :internal: + +.. kernel-doc:: kernel/sched/wait.c + :export: + +High-resolution timers +---------------------- + +.. kernel-doc:: include/linux/ktime.h + :internal: + +.. kernel-doc:: include/linux/hrtimer.h + :internal: + +.. kernel-doc:: kernel/time/hrtimer.c + :export: + +Workqueues and Kevents +---------------------- + +.. kernel-doc:: include/linux/workqueue.h + :internal: + +.. kernel-doc:: kernel/workqueue.c + :export: + +Internal Functions +------------------ + +.. kernel-doc:: kernel/exit.c + :internal: + +.. kernel-doc:: kernel/signal.c + :internal: + +.. kernel-doc:: include/linux/kthread.h + :internal: + +.. kernel-doc:: kernel/kthread.c + :export: + +Reference counting +------------------ + +.. kernel-doc:: include/linux/refcount.h + :internal: + +.. kernel-doc:: lib/refcount.c + :export: + +Atomics +------- + +.. kernel-doc:: arch/x86/include/asm/atomic.h + :internal: + +Kernel objects manipulation +--------------------------- + +.. kernel-doc:: lib/kobject.c + :export: + +Kernel utility functions +------------------------ + +.. kernel-doc:: include/linux/kernel.h + :internal: + +.. kernel-doc:: kernel/printk/printk.c + :export: + +.. kernel-doc:: kernel/panic.c + :export: + +.. kernel-doc:: kernel/rcu/tree.c + :export: + +.. kernel-doc:: kernel/rcu/tree_plugin.h + :export: + +.. kernel-doc:: kernel/rcu/update.c + :export: + +Device Resource Management +-------------------------- + +.. kernel-doc:: drivers/base/devres.c + :export: + diff --git a/Documentation/driver-api/clk.rst b/Documentation/driver-api/clk.rst new file mode 100644 index 000000000..593cca505 --- /dev/null +++ b/Documentation/driver-api/clk.rst @@ -0,0 +1,307 @@ +======================== +The Common Clk Framework +======================== + +:Author: Mike Turquette <mturquette@ti.com> + +This document endeavours to explain the common clk framework details, +and how to port a platform over to this framework. It is not yet a +detailed explanation of the clock api in include/linux/clk.h, but +perhaps someday it will include that information. + +Introduction and interface split +================================ + +The common clk framework is an interface to control the clock nodes +available on various devices today. This may come in the form of clock +gating, rate adjustment, muxing or other operations. This framework is +enabled with the CONFIG_COMMON_CLK option. + +The interface itself is divided into two halves, each shielded from the +details of its counterpart. First is the common definition of struct +clk which unifies the framework-level accounting and infrastructure that +has traditionally been duplicated across a variety of platforms. Second +is a common implementation of the clk.h api, defined in +drivers/clk/clk.c. Finally there is struct clk_ops, whose operations +are invoked by the clk api implementation. + +The second half of the interface is comprised of the hardware-specific +callbacks registered with struct clk_ops and the corresponding +hardware-specific structures needed to model a particular clock. For +the remainder of this document any reference to a callback in struct +clk_ops, such as .enable or .set_rate, implies the hardware-specific +implementation of that code. Likewise, references to struct clk_foo +serve as a convenient shorthand for the implementation of the +hardware-specific bits for the hypothetical "foo" hardware. + +Tying the two halves of this interface together is struct clk_hw, which +is defined in struct clk_foo and pointed to within struct clk_core. This +allows for easy navigation between the two discrete halves of the common +clock interface. + +Common data structures and api +============================== + +Below is the common struct clk_core definition from +drivers/clk/clk.c, modified for brevity:: + + struct clk_core { + const char *name; + const struct clk_ops *ops; + struct clk_hw *hw; + struct module *owner; + struct clk_core *parent; + const char **parent_names; + struct clk_core **parents; + u8 num_parents; + u8 new_parent_index; + ... + }; + +The members above make up the core of the clk tree topology. The clk +api itself defines several driver-facing functions which operate on +struct clk. That api is documented in include/linux/clk.h. + +Platforms and devices utilizing the common struct clk_core use the struct +clk_ops pointer in struct clk_core to perform the hardware-specific parts of +the operations defined in clk-provider.h:: + + struct clk_ops { + int (*prepare)(struct clk_hw *hw); + void (*unprepare)(struct clk_hw *hw); + int (*is_prepared)(struct clk_hw *hw); + void (*unprepare_unused)(struct clk_hw *hw); + int (*enable)(struct clk_hw *hw); + void (*disable)(struct clk_hw *hw); + int (*is_enabled)(struct clk_hw *hw); + void (*disable_unused)(struct clk_hw *hw); + unsigned long (*recalc_rate)(struct clk_hw *hw, + unsigned long parent_rate); + long (*round_rate)(struct clk_hw *hw, + unsigned long rate, + unsigned long *parent_rate); + int (*determine_rate)(struct clk_hw *hw, + struct clk_rate_request *req); + int (*set_parent)(struct clk_hw *hw, u8 index); + u8 (*get_parent)(struct clk_hw *hw); + int (*set_rate)(struct clk_hw *hw, + unsigned long rate, + unsigned long parent_rate); + int (*set_rate_and_parent)(struct clk_hw *hw, + unsigned long rate, + unsigned long parent_rate, + u8 index); + unsigned long (*recalc_accuracy)(struct clk_hw *hw, + unsigned long parent_accuracy); + int (*get_phase)(struct clk_hw *hw); + int (*set_phase)(struct clk_hw *hw, int degrees); + void (*init)(struct clk_hw *hw); + void (*debug_init)(struct clk_hw *hw, + struct dentry *dentry); + }; + +Hardware clk implementations +============================ + +The strength of the common struct clk_core comes from its .ops and .hw pointers +which abstract the details of struct clk from the hardware-specific bits, and +vice versa. To illustrate consider the simple gateable clk implementation in +drivers/clk/clk-gate.c:: + + struct clk_gate { + struct clk_hw hw; + void __iomem *reg; + u8 bit_idx; + ... + }; + +struct clk_gate contains struct clk_hw hw as well as hardware-specific +knowledge about which register and bit controls this clk's gating. +Nothing about clock topology or accounting, such as enable_count or +notifier_count, is needed here. That is all handled by the common +framework code and struct clk_core. + +Let's walk through enabling this clk from driver code:: + + struct clk *clk; + clk = clk_get(NULL, "my_gateable_clk"); + + clk_prepare(clk); + clk_enable(clk); + +The call graph for clk_enable is very simple:: + + clk_enable(clk); + clk->ops->enable(clk->hw); + [resolves to...] + clk_gate_enable(hw); + [resolves struct clk gate with to_clk_gate(hw)] + clk_gate_set_bit(gate); + +And the definition of clk_gate_set_bit:: + + static void clk_gate_set_bit(struct clk_gate *gate) + { + u32 reg; + + reg = __raw_readl(gate->reg); + reg |= BIT(gate->bit_idx); + writel(reg, gate->reg); + } + +Note that to_clk_gate is defined as:: + + #define to_clk_gate(_hw) container_of(_hw, struct clk_gate, hw) + +This pattern of abstraction is used for every clock hardware +representation. + +Supporting your own clk hardware +================================ + +When implementing support for a new type of clock it is only necessary to +include the following header:: + + #include <linux/clk-provider.h> + +To construct a clk hardware structure for your platform you must define +the following:: + + struct clk_foo { + struct clk_hw hw; + ... hardware specific data goes here ... + }; + +To take advantage of your data you'll need to support valid operations +for your clk:: + + struct clk_ops clk_foo_ops { + .enable = &clk_foo_enable; + .disable = &clk_foo_disable; + }; + +Implement the above functions using container_of:: + + #define to_clk_foo(_hw) container_of(_hw, struct clk_foo, hw) + + int clk_foo_enable(struct clk_hw *hw) + { + struct clk_foo *foo; + + foo = to_clk_foo(hw); + + ... perform magic on foo ... + + return 0; + }; + +Below is a matrix detailing which clk_ops are mandatory based upon the +hardware capabilities of that clock. A cell marked as "y" means +mandatory, a cell marked as "n" implies that either including that +callback is invalid or otherwise unnecessary. Empty cells are either +optional or must be evaluated on a case-by-case basis. + +.. table:: clock hardware characteristics + + +----------------+------+-------------+---------------+-------------+------+ + | | gate | change rate | single parent | multiplexer | root | + +================+======+=============+===============+=============+======+ + |.prepare | | | | | | + +----------------+------+-------------+---------------+-------------+------+ + |.unprepare | | | | | | + +----------------+------+-------------+---------------+-------------+------+ + +----------------+------+-------------+---------------+-------------+------+ + |.enable | y | | | | | + +----------------+------+-------------+---------------+-------------+------+ + |.disable | y | | | | | + +----------------+------+-------------+---------------+-------------+------+ + |.is_enabled | y | | | | | + +----------------+------+-------------+---------------+-------------+------+ + +----------------+------+-------------+---------------+-------------+------+ + |.recalc_rate | | y | | | | + +----------------+------+-------------+---------------+-------------+------+ + |.round_rate | | y [1]_ | | | | + +----------------+------+-------------+---------------+-------------+------+ + |.determine_rate | | y [1]_ | | | | + +----------------+------+-------------+---------------+-------------+------+ + |.set_rate | | y | | | | + +----------------+------+-------------+---------------+-------------+------+ + +----------------+------+-------------+---------------+-------------+------+ + |.set_parent | | | n | y | n | + +----------------+------+-------------+---------------+-------------+------+ + |.get_parent | | | n | y | n | + +----------------+------+-------------+---------------+-------------+------+ + +----------------+------+-------------+---------------+-------------+------+ + |.recalc_accuracy| | | | | | + +----------------+------+-------------+---------------+-------------+------+ + +----------------+------+-------------+---------------+-------------+------+ + |.init | | | | | | + +----------------+------+-------------+---------------+-------------+------+ + +.. [1] either one of round_rate or determine_rate is required. + +Finally, register your clock at run-time with a hardware-specific +registration function. This function simply populates struct clk_foo's +data and then passes the common struct clk parameters to the framework +with a call to:: + + clk_register(...) + +See the basic clock types in ``drivers/clk/clk-*.c`` for examples. + +Disabling clock gating of unused clocks +======================================= + +Sometimes during development it can be useful to be able to bypass the +default disabling of unused clocks. For example, if drivers aren't enabling +clocks properly but rely on them being on from the bootloader, bypassing +the disabling means that the driver will remain functional while the issues +are sorted out. + +To bypass this disabling, include "clk_ignore_unused" in the bootargs to the +kernel. + +Locking +======= + +The common clock framework uses two global locks, the prepare lock and the +enable lock. + +The enable lock is a spinlock and is held across calls to the .enable, +.disable operations. Those operations are thus not allowed to sleep, +and calls to the clk_enable(), clk_disable() API functions are allowed in +atomic context. + +For clk_is_enabled() API, it is also designed to be allowed to be used in +atomic context. However, it doesn't really make any sense to hold the enable +lock in core, unless you want to do something else with the information of +the enable state with that lock held. Otherwise, seeing if a clk is enabled is +a one-shot read of the enabled state, which could just as easily change after +the function returns because the lock is released. Thus the user of this API +needs to handle synchronizing the read of the state with whatever they're +using it for to make sure that the enable state doesn't change during that +time. + +The prepare lock is a mutex and is held across calls to all other operations. +All those operations are allowed to sleep, and calls to the corresponding API +functions are not allowed in atomic context. + +This effectively divides operations in two groups from a locking perspective. + +Drivers don't need to manually protect resources shared between the operations +of one group, regardless of whether those resources are shared by multiple +clocks or not. However, access to resources that are shared between operations +of the two groups needs to be protected by the drivers. An example of such a +resource would be a register that controls both the clock rate and the clock +enable/disable state. + +The clock framework is reentrant, in that a driver is allowed to call clock +framework functions from within its implementation of clock operations. This +can for instance cause a .set_rate operation of one clock being called from +within the .set_rate operation of another clock. This case must be considered +in the driver implementations, but the code flow is usually controlled by the +driver in that case. + +Note that locking must also be considered when code outside of the common +clock framework needs to access resources used by the clock operations. This +is considered out of scope of this document. diff --git a/Documentation/driver-api/conf.py b/Documentation/driver-api/conf.py new file mode 100644 index 000000000..202726d20 --- /dev/null +++ b/Documentation/driver-api/conf.py @@ -0,0 +1,10 @@ +# -*- coding: utf-8; mode: python -*- + +project = "The Linux driver implementer's API guide" + +tags.add("subproject") + +latex_documents = [ + ('index', 'driver-api.tex', project, + 'The kernel development community', 'manual'), +] diff --git a/Documentation/driver-api/device-io.rst b/Documentation/driver-api/device-io.rst new file mode 100644 index 000000000..b00b23903 --- /dev/null +++ b/Documentation/driver-api/device-io.rst @@ -0,0 +1,201 @@ +.. Copyright 2001 Matthew Wilcox +.. +.. This documentation is free software; you can redistribute +.. it and/or modify it under the terms of the GNU General Public +.. License as published by the Free Software Foundation; either +.. version 2 of the License, or (at your option) any later +.. version. + +=============================== +Bus-Independent Device Accesses +=============================== + +:Author: Matthew Wilcox +:Author: Alan Cox + +Introduction +============ + +Linux provides an API which abstracts performing IO across all busses +and devices, allowing device drivers to be written independently of bus +type. + +Memory Mapped IO +================ + +Getting Access to the Device +---------------------------- + +The most widely supported form of IO is memory mapped IO. That is, a +part of the CPU's address space is interpreted not as accesses to +memory, but as accesses to a device. Some architectures define devices +to be at a fixed address, but most have some method of discovering +devices. The PCI bus walk is a good example of such a scheme. This +document does not cover how to receive such an address, but assumes you +are starting with one. Physical addresses are of type unsigned long. + +This address should not be used directly. Instead, to get an address +suitable for passing to the accessor functions described below, you +should call :c:func:`ioremap()`. An address suitable for accessing +the device will be returned to you. + +After you've finished using the device (say, in your module's exit +routine), call :c:func:`iounmap()` in order to return the address +space to the kernel. Most architectures allocate new address space each +time you call :c:func:`ioremap()`, and they can run out unless you +call :c:func:`iounmap()`. + +Accessing the device +-------------------- + +The part of the interface most used by drivers is reading and writing +memory-mapped registers on the device. Linux provides interfaces to read +and write 8-bit, 16-bit, 32-bit and 64-bit quantities. Due to a +historical accident, these are named byte, word, long and quad accesses. +Both read and write accesses are supported; there is no prefetch support +at this time. + +The functions are named readb(), readw(), readl(), readq(), +readb_relaxed(), readw_relaxed(), readl_relaxed(), readq_relaxed(), +writeb(), writew(), writel() and writeq(). + +Some devices (such as framebuffers) would like to use larger transfers than +8 bytes at a time. For these devices, the :c:func:`memcpy_toio()`, +:c:func:`memcpy_fromio()` and :c:func:`memset_io()` functions are +provided. Do not use memset or memcpy on IO addresses; they are not +guaranteed to copy data in order. + +The read and write functions are defined to be ordered. That is the +compiler is not permitted to reorder the I/O sequence. When the ordering +can be compiler optimised, you can use __readb() and friends to +indicate the relaxed ordering. Use this with care. + +While the basic functions are defined to be synchronous with respect to +each other and ordered with respect to each other the busses the devices +sit on may themselves have asynchronicity. In particular many authors +are burned by the fact that PCI bus writes are posted asynchronously. A +driver author must issue a read from the same device to ensure that +writes have occurred in the specific cases the author cares. This kind +of property cannot be hidden from driver writers in the API. In some +cases, the read used to flush the device may be expected to fail (if the +card is resetting, for example). In that case, the read should be done +from config space, which is guaranteed to soft-fail if the card doesn't +respond. + +The following is an example of flushing a write to a device when the +driver would like to ensure the write's effects are visible prior to +continuing execution:: + + static inline void + qla1280_disable_intrs(struct scsi_qla_host *ha) + { + struct device_reg *reg; + + reg = ha->iobase; + /* disable risc and host interrupts */ + WRT_REG_WORD(®->ictrl, 0); + /* + * The following read will ensure that the above write + * has been received by the device before we return from this + * function. + */ + RD_REG_WORD(®->ictrl); + ha->flags.ints_enabled = 0; + } + +In addition to write posting, on some large multiprocessing systems +(e.g. SGI Challenge, Origin and Altix machines) posted writes won't be +strongly ordered coming from different CPUs. Thus it's important to +properly protect parts of your driver that do memory-mapped writes with +locks and use the :c:func:`mmiowb()` to make sure they arrive in the +order intended. Issuing a regular readX() will also ensure write ordering, +but should only be used when the +driver has to be sure that the write has actually arrived at the device +(not that it's simply ordered with respect to other writes), since a +full readX() is a relatively expensive operation. + +Generally, one should use :c:func:`mmiowb()` prior to releasing a spinlock +that protects regions using :c:func:`writeb()` or similar functions that +aren't surrounded by readb() calls, which will ensure ordering +and flushing. The following pseudocode illustrates what might occur if +write ordering isn't guaranteed via :c:func:`mmiowb()` or one of the +readX() functions:: + + CPU A: spin_lock_irqsave(&dev_lock, flags) + CPU A: ... + CPU A: writel(newval, ring_ptr); + CPU A: spin_unlock_irqrestore(&dev_lock, flags) + ... + CPU B: spin_lock_irqsave(&dev_lock, flags) + CPU B: writel(newval2, ring_ptr); + CPU B: ... + CPU B: spin_unlock_irqrestore(&dev_lock, flags) + +In the case above, newval2 could be written to ring_ptr before newval. +Fixing it is easy though:: + + CPU A: spin_lock_irqsave(&dev_lock, flags) + CPU A: ... + CPU A: writel(newval, ring_ptr); + CPU A: mmiowb(); /* ensure no other writes beat us to the device */ + CPU A: spin_unlock_irqrestore(&dev_lock, flags) + ... + CPU B: spin_lock_irqsave(&dev_lock, flags) + CPU B: writel(newval2, ring_ptr); + CPU B: ... + CPU B: mmiowb(); + CPU B: spin_unlock_irqrestore(&dev_lock, flags) + +See tg3.c for a real world example of how to use :c:func:`mmiowb()` + +PCI ordering rules also guarantee that PIO read responses arrive after any +outstanding DMA writes from that bus, since for some devices the result of +a readb() call may signal to the driver that a DMA transaction is +complete. In many cases, however, the driver may want to indicate that the +next readb() call has no relation to any previous DMA writes +performed by the device. The driver can use readb_relaxed() for +these cases, although only some platforms will honor the relaxed +semantics. Using the relaxed read functions will provide significant +performance benefits on platforms that support it. The qla2xxx driver +provides examples of how to use readX_relaxed(). In many cases, a majority +of the driver's readX() calls can safely be converted to readX_relaxed() +calls, since only a few will indicate or depend on DMA completion. + +Port Space Accesses +=================== + +Port Space Explained +-------------------- + +Another form of IO commonly supported is Port Space. This is a range of +addresses separate to the normal memory address space. Access to these +addresses is generally not as fast as accesses to the memory mapped +addresses, and it also has a potentially smaller address space. + +Unlike memory mapped IO, no preparation is required to access port +space. + +Accessing Port Space +-------------------- + +Accesses to this space are provided through a set of functions which +allow 8-bit, 16-bit and 32-bit accesses; also known as byte, word and +long. These functions are :c:func:`inb()`, :c:func:`inw()`, +:c:func:`inl()`, :c:func:`outb()`, :c:func:`outw()` and +:c:func:`outl()`. + +Some variants are provided for these functions. Some devices require +that accesses to their ports are slowed down. This functionality is +provided by appending a ``_p`` to the end of the function. +There are also equivalents to memcpy. The :c:func:`ins()` and +:c:func:`outs()` functions copy bytes, words or longs to the given +port. + +Public Functions Provided +========================= + +.. kernel-doc:: arch/x86/include/asm/io.h + :internal: + +.. kernel-doc:: lib/pci_iomap.c + :export: diff --git a/Documentation/driver-api/device_connection.rst b/Documentation/driver-api/device_connection.rst new file mode 100644 index 000000000..ba364224c --- /dev/null +++ b/Documentation/driver-api/device_connection.rst @@ -0,0 +1,43 @@ +================== +Device connections +================== + +Introduction +------------ + +Devices often have connections to other devices that are outside of the direct +child/parent relationship. A serial or network communication controller, which +could be a PCI device, may need to be able to get a reference to its PHY +component, which could be attached for example to the I2C bus. Some device +drivers need to be able to control the clocks or the GPIOs for their devices, +and so on. + +Device connections are generic descriptions of any type of connection between +two separate devices. + +Device connections alone do not create a dependency between the two devices. +They are only descriptions which are not tied to either of the devices directly. +A dependency between the two devices exists only if one of the two endpoint +devices requests a reference to the other. The descriptions themselves can be +defined in firmware (not yet supported) or they can be built-in. + +Usage +----- + +Device connections should exist before device ``->probe`` callback is called for +either endpoint device in the description. If the connections are defined in +firmware, this is not a problem. It should be considered if the connection +descriptions are "built-in", and need to be added separately. + +The connection description consists of the names of the two devices with the +connection, i.e. the endpoints, and unique identifier for the connection which +is needed if there are multiple connections between the two devices. + +After a description exists, the devices in it can request reference to the other +endpoint device, or they can request the description itself. + +API +--- + +.. kernel-doc:: drivers/base/devcon.c + :functions: device_connection_find_match device_connection_find device_connection_add device_connection_remove diff --git a/Documentation/driver-api/device_link.rst b/Documentation/driver-api/device_link.rst new file mode 100644 index 000000000..e8b0a8fd1 --- /dev/null +++ b/Documentation/driver-api/device_link.rst @@ -0,0 +1,306 @@ +.. |struct dev_pm_domain| replace:: :c:type:`struct dev_pm_domain <dev_pm_domain>` +.. |struct generic_pm_domain| replace:: :c:type:`struct generic_pm_domain <generic_pm_domain>` + +============ +Device links +============ + +By default, the driver core only enforces dependencies between devices +that are borne out of a parent/child relationship within the device +hierarchy: When suspending, resuming or shutting down the system, devices +are ordered based on this relationship, i.e. children are always suspended +before their parent, and the parent is always resumed before its children. + +Sometimes there is a need to represent device dependencies beyond the +mere parent/child relationship, e.g. between siblings, and have the +driver core automatically take care of them. + +Secondly, the driver core by default does not enforce any driver presence +dependencies, i.e. that one device must be bound to a driver before +another one can probe or function correctly. + +Often these two dependency types come together, so a device depends on +another one both with regards to driver presence *and* with regards to +suspend/resume and shutdown ordering. + +Device links allow representation of such dependencies in the driver core. + +In its standard or *managed* form, a device link combines *both* dependency +types: It guarantees correct suspend/resume and shutdown ordering between a +"supplier" device and its "consumer" devices, and it guarantees driver +presence on the supplier. The consumer devices are not probed before the +supplier is bound to a driver, and they're unbound before the supplier +is unbound. + +When driver presence on the supplier is irrelevant and only correct +suspend/resume and shutdown ordering is needed, the device link may +simply be set up with the ``DL_FLAG_STATELESS`` flag. In other words, +enforcing driver presence on the supplier is optional. + +Another optional feature is runtime PM integration: By setting the +``DL_FLAG_PM_RUNTIME`` flag on addition of the device link, the PM core +is instructed to runtime resume the supplier and keep it active +whenever and for as long as the consumer is runtime resumed. + +Usage +===== + +The earliest point in time when device links can be added is after +:c:func:`device_add()` has been called for the supplier and +:c:func:`device_initialize()` has been called for the consumer. + +It is legal to add them later, but care must be taken that the system +remains in a consistent state: E.g. a device link cannot be added in +the midst of a suspend/resume transition, so either commencement of +such a transition needs to be prevented with :c:func:`lock_system_sleep()`, +or the device link needs to be added from a function which is guaranteed +not to run in parallel to a suspend/resume transition, such as from a +device ``->probe`` callback or a boot-time PCI quirk. + +Another example for an inconsistent state would be a device link that +represents a driver presence dependency, yet is added from the consumer's +``->probe`` callback while the supplier hasn't started to probe yet: Had the +driver core known about the device link earlier, it wouldn't have probed the +consumer in the first place. The onus is thus on the consumer to check +presence of the supplier after adding the link, and defer probing on +non-presence. [Note that it is valid to create a link from the consumer's +``->probe`` callback while the supplier is still probing, but the consumer must +know that the supplier is functional already at the link creation time (that is +the case, for instance, if the consumer has just acquired some resources that +would not have been available had the supplier not been functional then).] + +If a device link with ``DL_FLAG_STATELESS`` set (i.e. a stateless device link) +is added in the ``->probe`` callback of the supplier or consumer driver, it is +typically deleted in its ``->remove`` callback for symmetry. That way, if the +driver is compiled as a module, the device link is added on module load and +orderly deleted on unload. The same restrictions that apply to device link +addition (e.g. exclusion of a parallel suspend/resume transition) apply equally +to deletion. Device links managed by the driver core are deleted automatically +by it. + +Several flags may be specified on device link addition, two of which +have already been mentioned above: ``DL_FLAG_STATELESS`` to express that no +driver presence dependency is needed (but only correct suspend/resume and +shutdown ordering) and ``DL_FLAG_PM_RUNTIME`` to express that runtime PM +integration is desired. + +Two other flags are specifically targeted at use cases where the device +link is added from the consumer's ``->probe`` callback: ``DL_FLAG_RPM_ACTIVE`` +can be specified to runtime resume the supplier upon addition of the +device link. ``DL_FLAG_AUTOREMOVE_CONSUMER`` causes the device link to be +automatically purged when the consumer fails to probe or later unbinds. + +Similarly, when the device link is added from supplier's ``->probe`` callback, +``DL_FLAG_AUTOREMOVE_SUPPLIER`` causes the device link to be automatically +purged when the supplier fails to probe or later unbinds. + +If neither ``DL_FLAG_AUTOREMOVE_CONSUMER`` nor ``DL_FLAG_AUTOREMOVE_SUPPLIER`` +is set, ``DL_FLAG_AUTOPROBE_CONSUMER`` can be used to request the driver core +to probe for a driver for the consumer driver on the link automatically after +a driver has been bound to the supplier device. + +Note, however, that any combinations of ``DL_FLAG_AUTOREMOVE_CONSUMER``, +``DL_FLAG_AUTOREMOVE_SUPPLIER`` or ``DL_FLAG_AUTOPROBE_CONSUMER`` with +``DL_FLAG_STATELESS`` are invalid and cannot be used. + +Limitations +=========== + +Driver authors should be aware that a driver presence dependency for managed +device links (i.e. when ``DL_FLAG_STATELESS`` is not specified on link addition) +may cause probing of the consumer to be deferred indefinitely. This can become +a problem if the consumer is required to probe before a certain initcall level +is reached. Worse, if the supplier driver is blacklisted or missing, the +consumer will never be probed. + +Moreover, managed device links cannot be deleted directly. They are deleted +by the driver core when they are not necessary any more in accordance with the +``DL_FLAG_AUTOREMOVE_CONSUMER`` and ``DL_FLAG_AUTOREMOVE_SUPPLIER`` flags. +However, stateless device links (i.e. device links with ``DL_FLAG_STATELESS`` +set) are expected to be removed by whoever called :c:func:`device_link_add()` +to add them with the help of either :c:func:`device_link_del()` or +:c:func:`device_link_remove()`. + +Sometimes drivers depend on optional resources. They are able to operate +in a degraded mode (reduced feature set or performance) when those resources +are not present. An example is an SPI controller that can use a DMA engine +or work in PIO mode. The controller can determine presence of the optional +resources at probe time but on non-presence there is no way to know whether +they will become available in the near future (due to a supplier driver +probing) or never. Consequently it cannot be determined whether to defer +probing or not. It would be possible to notify drivers when optional +resources become available after probing, but it would come at a high cost +for drivers as switching between modes of operation at runtime based on the +availability of such resources would be much more complex than a mechanism +based on probe deferral. In any case optional resources are beyond the +scope of device links. + +Examples +======== + +* An MMU device exists alongside a busmaster device, both are in the same + power domain. The MMU implements DMA address translation for the busmaster + device and shall be runtime resumed and kept active whenever and as long + as the busmaster device is active. The busmaster device's driver shall + not bind before the MMU is bound. To achieve this, a device link with + runtime PM integration is added from the busmaster device (consumer) + to the MMU device (supplier). The effect with regards to runtime PM + is the same as if the MMU was the parent of the master device. + + The fact that both devices share the same power domain would normally + suggest usage of a |struct dev_pm_domain| or |struct generic_pm_domain|, + however these are not independent devices that happen to share a power + switch, but rather the MMU device serves the busmaster device and is + useless without it. A device link creates a synthetic hierarchical + relationship between the devices and is thus more apt. + +* A Thunderbolt host controller comprises a number of PCIe hotplug ports + and an NHI device to manage the PCIe switch. On resume from system sleep, + the NHI device needs to re-establish PCI tunnels to attached devices + before the hotplug ports can resume. If the hotplug ports were children + of the NHI, this resume order would automatically be enforced by the + PM core, but unfortunately they're aunts. The solution is to add + device links from the hotplug ports (consumers) to the NHI device + (supplier). A driver presence dependency is not necessary for this + use case. + +* Discrete GPUs in hybrid graphics laptops often feature an HDA controller + for HDMI/DP audio. In the device hierarchy the HDA controller is a sibling + of the VGA device, yet both share the same power domain and the HDA + controller is only ever needed when an HDMI/DP display is attached to the + VGA device. A device link from the HDA controller (consumer) to the + VGA device (supplier) aptly represents this relationship. + +* ACPI allows definition of a device start order by way of _DEP objects. + A classical example is when ACPI power management methods on one device + are implemented in terms of I\ :sup:`2`\ C accesses and require a specific + I\ :sup:`2`\ C controller to be present and functional for the power + management of the device in question to work. + +* In some SoCs a functional dependency exists from display, video codec and + video processing IP cores on transparent memory access IP cores that handle + burst access and compression/decompression. + +Alternatives +============ + +* A |struct dev_pm_domain| can be used to override the bus, + class or device type callbacks. It is intended for devices sharing + a single on/off switch, however it does not guarantee a specific + suspend/resume ordering, this needs to be implemented separately. + It also does not by itself track the runtime PM status of the involved + devices and turn off the power switch only when all of them are runtime + suspended. Furthermore it cannot be used to enforce a specific shutdown + ordering or a driver presence dependency. + +* A |struct generic_pm_domain| is a lot more heavyweight than a + device link and does not allow for shutdown ordering or driver presence + dependencies. It also cannot be used on ACPI systems. + +Implementation +============== + +The device hierarchy, which -- as the name implies -- is a tree, +becomes a directed acyclic graph once device links are added. + +Ordering of these devices during suspend/resume is determined by the +dpm_list. During shutdown it is determined by the devices_kset. With +no device links present, the two lists are a flattened, one-dimensional +representations of the device tree such that a device is placed behind +all its ancestors. That is achieved by traversing the ACPI namespace +or OpenFirmware device tree top-down and appending devices to the lists +as they are discovered. + +Once device links are added, the lists need to satisfy the additional +constraint that a device is placed behind all its suppliers, recursively. +To ensure this, upon addition of the device link the consumer and the +entire sub-graph below it (all children and consumers of the consumer) +are moved to the end of the list. (Call to :c:func:`device_reorder_to_tail()` +from :c:func:`device_link_add()`.) + +To prevent introduction of dependency loops into the graph, it is +verified upon device link addition that the supplier is not dependent +on the consumer or any children or consumers of the consumer. +(Call to :c:func:`device_is_dependent()` from :c:func:`device_link_add()`.) +If that constraint is violated, :c:func:`device_link_add()` will return +``NULL`` and a ``WARNING`` will be logged. + +Notably this also prevents the addition of a device link from a parent +device to a child. However the converse is allowed, i.e. a device link +from a child to a parent. Since the driver core already guarantees +correct suspend/resume and shutdown ordering between parent and child, +such a device link only makes sense if a driver presence dependency is +needed on top of that. In this case driver authors should weigh +carefully if a device link is at all the right tool for the purpose. +A more suitable approach might be to simply use deferred probing or +add a device flag causing the parent driver to be probed before the +child one. + +State machine +============= + +.. kernel-doc:: include/linux/device.h + :functions: device_link_state + +:: + + .=============================. + | | + v | + DORMANT <=> AVAILABLE <=> CONSUMER_PROBE => ACTIVE + ^ | + | | + '============ SUPPLIER_UNBIND <============' + +* The initial state of a device link is automatically determined by + :c:func:`device_link_add()` based on the driver presence on the supplier + and consumer. If the link is created before any devices are probed, it + is set to ``DL_STATE_DORMANT``. + +* When a supplier device is bound to a driver, links to its consumers + progress to ``DL_STATE_AVAILABLE``. + (Call to :c:func:`device_links_driver_bound()` from + :c:func:`driver_bound()`.) + +* Before a consumer device is probed, presence of supplier drivers is + verified by checking that links to suppliers are in ``DL_STATE_AVAILABLE`` + state. The state of the links is updated to ``DL_STATE_CONSUMER_PROBE``. + (Call to :c:func:`device_links_check_suppliers()` from + :c:func:`really_probe()`.) + This prevents the supplier from unbinding. + (Call to :c:func:`wait_for_device_probe()` from + :c:func:`device_links_unbind_consumers()`.) + +* If the probe fails, links to suppliers revert back to ``DL_STATE_AVAILABLE``. + (Call to :c:func:`device_links_no_driver()` from :c:func:`really_probe()`.) + +* If the probe succeeds, links to suppliers progress to ``DL_STATE_ACTIVE``. + (Call to :c:func:`device_links_driver_bound()` from :c:func:`driver_bound()`.) + +* When the consumer's driver is later on removed, links to suppliers revert + back to ``DL_STATE_AVAILABLE``. + (Call to :c:func:`__device_links_no_driver()` from + :c:func:`device_links_driver_cleanup()`, which in turn is called from + :c:func:`__device_release_driver()`.) + +* Before a supplier's driver is removed, links to consumers that are not + bound to a driver are updated to ``DL_STATE_SUPPLIER_UNBIND``. + (Call to :c:func:`device_links_busy()` from + :c:func:`__device_release_driver()`.) + This prevents the consumers from binding. + (Call to :c:func:`device_links_check_suppliers()` from + :c:func:`really_probe()`.) + Consumers that are bound are freed from their driver; consumers that are + probing are waited for until they are done. + (Call to :c:func:`device_links_unbind_consumers()` from + :c:func:`__device_release_driver()`.) + Once all links to consumers are in ``DL_STATE_SUPPLIER_UNBIND`` state, + the supplier driver is released and the links revert to ``DL_STATE_DORMANT``. + (Call to :c:func:`device_links_driver_cleanup()` from + :c:func:`__device_release_driver()`.) + +API +=== + +.. kernel-doc:: drivers/base/core.c + :functions: device_link_add device_link_del device_link_remove diff --git a/Documentation/driver-api/dma-buf.rst b/Documentation/driver-api/dma-buf.rst new file mode 100644 index 000000000..b541e97c7 --- /dev/null +++ b/Documentation/driver-api/dma-buf.rst @@ -0,0 +1,168 @@ +Buffer Sharing and Synchronization +================================== + +The dma-buf subsystem provides the framework for sharing buffers for +hardware (DMA) access across multiple device drivers and subsystems, and +for synchronizing asynchronous hardware access. + +This is used, for example, by drm "prime" multi-GPU support, but is of +course not limited to GPU use cases. + +The three main components of this are: (1) dma-buf, representing a +sg_table and exposed to userspace as a file descriptor to allow passing +between devices, (2) fence, which provides a mechanism to signal when +one device as finished access, and (3) reservation, which manages the +shared or exclusive fence(s) associated with the buffer. + +Shared DMA Buffers +------------------ + +This document serves as a guide to device-driver writers on what is the dma-buf +buffer sharing API, how to use it for exporting and using shared buffers. + +Any device driver which wishes to be a part of DMA buffer sharing, can do so as +either the 'exporter' of buffers, or the 'user' or 'importer' of buffers. + +Say a driver A wants to use buffers created by driver B, then we call B as the +exporter, and A as buffer-user/importer. + +The exporter + + - implements and manages operations in :c:type:`struct dma_buf_ops + <dma_buf_ops>` for the buffer, + - allows other users to share the buffer by using dma_buf sharing APIs, + - manages the details of buffer allocation, wrapped int a :c:type:`struct + dma_buf <dma_buf>`, + - decides about the actual backing storage where this allocation happens, + - and takes care of any migration of scatterlist - for all (shared) users of + this buffer. + +The buffer-user + + - is one of (many) sharing users of the buffer. + - doesn't need to worry about how the buffer is allocated, or where. + - and needs a mechanism to get access to the scatterlist that makes up this + buffer in memory, mapped into its own address space, so it can access the + same area of memory. This interface is provided by :c:type:`struct + dma_buf_attachment <dma_buf_attachment>`. + +Any exporters or users of the dma-buf buffer sharing framework must have a +'select DMA_SHARED_BUFFER' in their respective Kconfigs. + +Userspace Interface Notes +~~~~~~~~~~~~~~~~~~~~~~~~~ + +Mostly a DMA buffer file descriptor is simply an opaque object for userspace, +and hence the generic interface exposed is very minimal. There's a few things to +consider though: + +- Since kernel 3.12 the dma-buf FD supports the llseek system call, but only + with offset=0 and whence=SEEK_END|SEEK_SET. SEEK_SET is supported to allow + the usual size discover pattern size = SEEK_END(0); SEEK_SET(0). Every other + llseek operation will report -EINVAL. + + If llseek on dma-buf FDs isn't support the kernel will report -ESPIPE for all + cases. Userspace can use this to detect support for discovering the dma-buf + size using llseek. + +- In order to avoid fd leaks on exec, the FD_CLOEXEC flag must be set + on the file descriptor. This is not just a resource leak, but a + potential security hole. It could give the newly exec'd application + access to buffers, via the leaked fd, to which it should otherwise + not be permitted access. + + The problem with doing this via a separate fcntl() call, versus doing it + atomically when the fd is created, is that this is inherently racy in a + multi-threaded app[3]. The issue is made worse when it is library code + opening/creating the file descriptor, as the application may not even be + aware of the fd's. + + To avoid this problem, userspace must have a way to request O_CLOEXEC + flag be set when the dma-buf fd is created. So any API provided by + the exporting driver to create a dmabuf fd must provide a way to let + userspace control setting of O_CLOEXEC flag passed in to dma_buf_fd(). + +- Memory mapping the contents of the DMA buffer is also supported. See the + discussion below on `CPU Access to DMA Buffer Objects`_ for the full details. + +- The DMA buffer FD is also pollable, see `Fence Poll Support`_ below for + details. + +Basic Operation and Device DMA Access +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/dma-buf.c + :doc: dma buf device access + +CPU Access to DMA Buffer Objects +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/dma-buf.c + :doc: cpu access + +Fence Poll Support +~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/dma-buf.c + :doc: fence polling + +Kernel Functions and Structures Reference +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/dma-buf.c + :export: + +.. kernel-doc:: include/linux/dma-buf.h + :internal: + +Reservation Objects +------------------- + +.. kernel-doc:: drivers/dma-buf/reservation.c + :doc: Reservation Object Overview + +.. kernel-doc:: drivers/dma-buf/reservation.c + :export: + +.. kernel-doc:: include/linux/reservation.h + :internal: + +DMA Fences +---------- + +.. kernel-doc:: drivers/dma-buf/dma-fence.c + :doc: DMA fences overview + +DMA Fences Functions Reference +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/dma-fence.c + :export: + +.. kernel-doc:: include/linux/dma-fence.h + :internal: + +Seqno Hardware Fences +~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: include/linux/seqno-fence.h + :internal: + +DMA Fence Array +~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/dma-fence-array.c + :export: + +.. kernel-doc:: include/linux/dma-fence-array.h + :internal: + +DMA Fence uABI/Sync File +~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/dma-buf/sync_file.c + :export: + +.. kernel-doc:: include/linux/sync_file.h + :internal: + diff --git a/Documentation/driver-api/dmaengine/client.rst b/Documentation/driver-api/dmaengine/client.rst new file mode 100644 index 000000000..fbbb2831f --- /dev/null +++ b/Documentation/driver-api/dmaengine/client.rst @@ -0,0 +1,275 @@ +==================== +DMA Engine API Guide +==================== + +Vinod Koul <vinod dot koul at intel.com> + +.. note:: For DMA Engine usage in async_tx please see: + ``Documentation/crypto/async-tx-api.txt`` + + +Below is a guide to device driver writers on how to use the Slave-DMA API of the +DMA Engine. This is applicable only for slave DMA usage only. + +DMA usage +========= + +The slave DMA usage consists of following steps: + +- Allocate a DMA slave channel + +- Set slave and controller specific parameters + +- Get a descriptor for transaction + +- Submit the transaction + +- Issue pending requests and wait for callback notification + +The details of these operations are: + +1. Allocate a DMA slave channel + + Channel allocation is slightly different in the slave DMA context, + client drivers typically need a channel from a particular DMA + controller only and even in some cases a specific channel is desired. + To request a channel dma_request_chan() API is used. + + Interface: + + .. code-block:: c + + struct dma_chan *dma_request_chan(struct device *dev, const char *name); + + Which will find and return the ``name`` DMA channel associated with the 'dev' + device. The association is done via DT, ACPI or board file based + dma_slave_map matching table. + + A channel allocated via this interface is exclusive to the caller, + until dma_release_channel() is called. + +2. Set slave and controller specific parameters + + Next step is always to pass some specific information to the DMA + driver. Most of the generic information which a slave DMA can use + is in struct dma_slave_config. This allows the clients to specify + DMA direction, DMA addresses, bus widths, DMA burst lengths etc + for the peripheral. + + If some DMA controllers have more parameters to be sent then they + should try to embed struct dma_slave_config in their controller + specific structure. That gives flexibility to client to pass more + parameters, if required. + + Interface: + + .. code-block:: c + + int dmaengine_slave_config(struct dma_chan *chan, + struct dma_slave_config *config) + + Please see the dma_slave_config structure definition in dmaengine.h + for a detailed explanation of the struct members. Please note + that the 'direction' member will be going away as it duplicates the + direction given in the prepare call. + +3. Get a descriptor for transaction + + For slave usage the various modes of slave transfers supported by the + DMA-engine are: + + - slave_sg: DMA a list of scatter gather buffers from/to a peripheral + + - dma_cyclic: Perform a cyclic DMA operation from/to a peripheral till the + operation is explicitly stopped. + + - interleaved_dma: This is common to Slave as well as M2M clients. For slave + address of devices' fifo could be already known to the driver. + Various types of operations could be expressed by setting + appropriate values to the 'dma_interleaved_template' members. + + A non-NULL return of this transfer API represents a "descriptor" for + the given transaction. + + Interface: + + .. code-block:: c + + struct dma_async_tx_descriptor *dmaengine_prep_slave_sg( + struct dma_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, enum dma_data_direction direction, + unsigned long flags); + + struct dma_async_tx_descriptor *dmaengine_prep_dma_cyclic( + struct dma_chan *chan, dma_addr_t buf_addr, size_t buf_len, + size_t period_len, enum dma_data_direction direction); + + struct dma_async_tx_descriptor *dmaengine_prep_interleaved_dma( + struct dma_chan *chan, struct dma_interleaved_template *xt, + unsigned long flags); + + The peripheral driver is expected to have mapped the scatterlist for + the DMA operation prior to calling dmaengine_prep_slave_sg(), and must + keep the scatterlist mapped until the DMA operation has completed. + The scatterlist must be mapped using the DMA struct device. + If a mapping needs to be synchronized later, dma_sync_*_for_*() must be + called using the DMA struct device, too. + So, normal setup should look like this: + + .. code-block:: c + + nr_sg = dma_map_sg(chan->device->dev, sgl, sg_len); + if (nr_sg == 0) + /* error */ + + desc = dmaengine_prep_slave_sg(chan, sgl, nr_sg, direction, flags); + + Once a descriptor has been obtained, the callback information can be + added and the descriptor must then be submitted. Some DMA engine + drivers may hold a spinlock between a successful preparation and + submission so it is important that these two operations are closely + paired. + + .. note:: + + Although the async_tx API specifies that completion callback + routines cannot submit any new operations, this is not the + case for slave/cyclic DMA. + + For slave DMA, the subsequent transaction may not be available + for submission prior to callback function being invoked, so + slave DMA callbacks are permitted to prepare and submit a new + transaction. + + For cyclic DMA, a callback function may wish to terminate the + DMA via dmaengine_terminate_async(). + + Therefore, it is important that DMA engine drivers drop any + locks before calling the callback function which may cause a + deadlock. + + Note that callbacks will always be invoked from the DMA + engines tasklet, never from interrupt context. + +4. Submit the transaction + + Once the descriptor has been prepared and the callback information + added, it must be placed on the DMA engine drivers pending queue. + + Interface: + + .. code-block:: c + + dma_cookie_t dmaengine_submit(struct dma_async_tx_descriptor *desc) + + This returns a cookie can be used to check the progress of DMA engine + activity via other DMA engine calls not covered in this document. + + dmaengine_submit() will not start the DMA operation, it merely adds + it to the pending queue. For this, see step 5, dma_async_issue_pending. + +5. Issue pending DMA requests and wait for callback notification + + The transactions in the pending queue can be activated by calling the + issue_pending API. If channel is idle then the first transaction in + queue is started and subsequent ones queued up. + + On completion of each DMA operation, the next in queue is started and + a tasklet triggered. The tasklet will then call the client driver + completion callback routine for notification, if set. + + Interface: + + .. code-block:: c + + void dma_async_issue_pending(struct dma_chan *chan); + +Further APIs: +------------- + +1. Terminate APIs + + .. code-block:: c + + int dmaengine_terminate_sync(struct dma_chan *chan) + int dmaengine_terminate_async(struct dma_chan *chan) + int dmaengine_terminate_all(struct dma_chan *chan) /* DEPRECATED */ + + This causes all activity for the DMA channel to be stopped, and may + discard data in the DMA FIFO which hasn't been fully transferred. + No callback functions will be called for any incomplete transfers. + + Two variants of this function are available. + + dmaengine_terminate_async() might not wait until the DMA has been fully + stopped or until any running complete callbacks have finished. But it is + possible to call dmaengine_terminate_async() from atomic context or from + within a complete callback. dmaengine_synchronize() must be called before it + is safe to free the memory accessed by the DMA transfer or free resources + accessed from within the complete callback. + + dmaengine_terminate_sync() will wait for the transfer and any running + complete callbacks to finish before it returns. But the function must not be + called from atomic context or from within a complete callback. + + dmaengine_terminate_all() is deprecated and should not be used in new code. + +2. Pause API + + .. code-block:: c + + int dmaengine_pause(struct dma_chan *chan) + + This pauses activity on the DMA channel without data loss. + +3. Resume API + + .. code-block:: c + + int dmaengine_resume(struct dma_chan *chan) + + Resume a previously paused DMA channel. It is invalid to resume a + channel which is not currently paused. + +4. Check Txn complete + + .. code-block:: c + + enum dma_status dma_async_is_tx_complete(struct dma_chan *chan, + dma_cookie_t cookie, dma_cookie_t *last, dma_cookie_t *used) + + This can be used to check the status of the channel. Please see + the documentation in include/linux/dmaengine.h for a more complete + description of this API. + + This can be used in conjunction with dma_async_is_complete() and + the cookie returned from dmaengine_submit() to check for + completion of a specific DMA transaction. + + .. note:: + + Not all DMA engine drivers can return reliable information for + a running DMA channel. It is recommended that DMA engine users + pause or stop (via dmaengine_terminate_all()) the channel before + using this API. + +5. Synchronize termination API + + .. code-block:: c + + void dmaengine_synchronize(struct dma_chan *chan) + + Synchronize the termination of the DMA channel to the current context. + + This function should be used after dmaengine_terminate_async() to synchronize + the termination of the DMA channel to the current context. The function will + wait for the transfer and any running complete callbacks to finish before it + returns. + + If dmaengine_terminate_async() is used to stop the DMA channel this function + must be called before it is safe to free memory accessed by previously + submitted descriptors or to free any resources accessed within the complete + callback of previously submitted descriptors. + + The behavior of this function is undefined if dma_async_issue_pending() has + been called between dmaengine_terminate_async() and this function. diff --git a/Documentation/driver-api/dmaengine/dmatest.rst b/Documentation/driver-api/dmaengine/dmatest.rst new file mode 100644 index 000000000..7ce5e71c3 --- /dev/null +++ b/Documentation/driver-api/dmaengine/dmatest.rst @@ -0,0 +1,114 @@ +============== +DMA Test Guide +============== + +Andy Shevchenko <andriy.shevchenko@linux.intel.com> + +This small document introduces how to test DMA drivers using dmatest module. + +.. note:: + The test suite works only on the channels that have at least one + capability of the following: DMA_MEMCPY (memory-to-memory), DMA_MEMSET + (const-to-memory or memory-to-memory, when emulated), DMA_XOR, DMA_PQ. + +Part 1 - How to build the test module +===================================== + +The menuconfig contains an option that could be found by following path: + + Device Drivers -> DMA Engine support -> DMA Test client + +In the configuration file the option called CONFIG_DMATEST. The dmatest could +be built as module or inside kernel. Let's consider those cases. + +Part 2 - When dmatest is built as a module +========================================== + +Example of usage:: + + % modprobe dmatest channel=dma0chan0 timeout=2000 iterations=1 run=1 + +...or:: + + % modprobe dmatest + % echo dma0chan0 > /sys/module/dmatest/parameters/channel + % echo 2000 > /sys/module/dmatest/parameters/timeout + % echo 1 > /sys/module/dmatest/parameters/iterations + % echo 1 > /sys/module/dmatest/parameters/run + +...or on the kernel command line:: + + dmatest.channel=dma0chan0 dmatest.timeout=2000 dmatest.iterations=1 dmatest.run=1 + +.. hint:: + available channel list could be extracted by running the following command:: + + % ls -1 /sys/class/dma/ + +Once started a message like "dmatest: Started 1 threads using dma0chan0" is +emitted. After that only test failure messages are reported until the test +stops. + +Note that running a new test will not stop any in progress test. + +The following command returns the state of the test. :: + + % cat /sys/module/dmatest/parameters/run + +To wait for test completion userpace can poll 'run' until it is false, or use +the wait parameter. Specifying 'wait=1' when loading the module causes module +initialization to pause until a test run has completed, while reading +/sys/module/dmatest/parameters/wait waits for any running test to complete +before returning. For example, the following scripts wait for 42 tests +to complete before exiting. Note that if 'iterations' is set to 'infinite' then +waiting is disabled. + +Example:: + + % modprobe dmatest run=1 iterations=42 wait=1 + % modprobe -r dmatest + +...or:: + + % modprobe dmatest run=1 iterations=42 + % cat /sys/module/dmatest/parameters/wait + % modprobe -r dmatest + +Part 3 - When built-in in the kernel +==================================== + +The module parameters that is supplied to the kernel command line will be used +for the first performed test. After user gets a control, the test could be +re-run with the same or different parameters. For the details see the above +section `Part 2 - When dmatest is built as a module`_. + +In both cases the module parameters are used as the actual values for the test +case. You always could check them at run-time by running :: + + % grep -H . /sys/module/dmatest/parameters/* + +Part 4 - Gathering the test results +=================================== + +Test results are printed to the kernel log buffer with the format:: + + "dmatest: result <channel>: <test id>: '<error msg>' with src_off=<val> dst_off=<val> len=<val> (<err code>)" + +Example of output:: + + % dmesg | tail -n 1 + dmatest: result dma0chan0-copy0: #1: No errors with src_off=0x7bf dst_off=0x8ad len=0x3fea (0) + +The message format is unified across the different types of errors. A +number in the parentheses represents additional information, e.g. error +code, error counter, or status. A test thread also emits a summary line at +completion listing the number of tests executed, number that failed, and a +result code. + +Example:: + + % dmesg | tail -n 1 + dmatest: dma0chan0-copy0: summary 1 test, 0 failures 1000 iops 100000 KB/s (0) + +The details of a data miscompare error are also emitted, but do not follow the +above format. diff --git a/Documentation/driver-api/dmaengine/index.rst b/Documentation/driver-api/dmaengine/index.rst new file mode 100644 index 000000000..3026fa975 --- /dev/null +++ b/Documentation/driver-api/dmaengine/index.rst @@ -0,0 +1,55 @@ +======================= +DMAEngine documentation +======================= + +DMAEngine documentation provides documents for various aspects of DMAEngine +framework. + +DMAEngine documentation +----------------------- + +This book helps with DMAengine internal APIs and guide for DMAEngine device +driver writers. + +.. toctree:: + :maxdepth: 1 + + provider + +DMAEngine client documentation +------------------------------ + +This book is a guide to device driver writers on how to use the Slave-DMA +API of the DMAEngine. This is applicable only for slave DMA usage only. + +.. toctree:: + :maxdepth: 1 + + client + +DMA Test documentation +---------------------- + +This book introduces how to test DMA drivers using dmatest module. + +.. toctree:: + :maxdepth: 1 + + dmatest + +PXA DMA documentation +---------------------- + +This book adds some notes about PXA DMA + +.. toctree:: + :maxdepth: 1 + + pxa_dma + +.. only:: subproject + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/dmaengine/provider.rst b/Documentation/driver-api/dmaengine/provider.rst new file mode 100644 index 000000000..dfc4486b5 --- /dev/null +++ b/Documentation/driver-api/dmaengine/provider.rst @@ -0,0 +1,504 @@ +================================== +DMAengine controller documentation +================================== + +Hardware Introduction +===================== + +Most of the Slave DMA controllers have the same general principles of +operations. + +They have a given number of channels to use for the DMA transfers, and +a given number of requests lines. + +Requests and channels are pretty much orthogonal. Channels can be used +to serve several to any requests. To simplify, channels are the +entities that will be doing the copy, and requests what endpoints are +involved. + +The request lines actually correspond to physical lines going from the +DMA-eligible devices to the controller itself. Whenever the device +will want to start a transfer, it will assert a DMA request (DRQ) by +asserting that request line. + +A very simple DMA controller would only take into account a single +parameter: the transfer size. At each clock cycle, it would transfer a +byte of data from one buffer to another, until the transfer size has +been reached. + +That wouldn't work well in the real world, since slave devices might +require a specific number of bits to be transferred in a single +cycle. For example, we may want to transfer as much data as the +physical bus allows to maximize performances when doing a simple +memory copy operation, but our audio device could have a narrower FIFO +that requires data to be written exactly 16 or 24 bits at a time. This +is why most if not all of the DMA controllers can adjust this, using a +parameter called the transfer width. + +Moreover, some DMA controllers, whenever the RAM is used as a source +or destination, can group the reads or writes in memory into a buffer, +so instead of having a lot of small memory accesses, which is not +really efficient, you'll get several bigger transfers. This is done +using a parameter called the burst size, that defines how many single +reads/writes it's allowed to do without the controller splitting the +transfer into smaller sub-transfers. + +Our theoretical DMA controller would then only be able to do transfers +that involve a single contiguous block of data. However, some of the +transfers we usually have are not, and want to copy data from +non-contiguous buffers to a contiguous buffer, which is called +scatter-gather. + +DMAEngine, at least for mem2dev transfers, require support for +scatter-gather. So we're left with two cases here: either we have a +quite simple DMA controller that doesn't support it, and we'll have to +implement it in software, or we have a more advanced DMA controller, +that implements in hardware scatter-gather. + +The latter are usually programmed using a collection of chunks to +transfer, and whenever the transfer is started, the controller will go +over that collection, doing whatever we programmed there. + +This collection is usually either a table or a linked list. You will +then push either the address of the table and its number of elements, +or the first item of the list to one channel of the DMA controller, +and whenever a DRQ will be asserted, it will go through the collection +to know where to fetch the data from. + +Either way, the format of this collection is completely dependent on +your hardware. Each DMA controller will require a different structure, +but all of them will require, for every chunk, at least the source and +destination addresses, whether it should increment these addresses or +not and the three parameters we saw earlier: the burst size, the +transfer width and the transfer size. + +The one last thing is that usually, slave devices won't issue DRQ by +default, and you have to enable this in your slave device driver first +whenever you're willing to use DMA. + +These were just the general memory-to-memory (also called mem2mem) or +memory-to-device (mem2dev) kind of transfers. Most devices often +support other kind of transfers or memory operations that dmaengine +support and will be detailed later in this document. + +DMA Support in Linux +==================== + +Historically, DMA controller drivers have been implemented using the +async TX API, to offload operations such as memory copy, XOR, +cryptography, etc., basically any memory to memory operation. + +Over time, the need for memory to device transfers arose, and +dmaengine was extended. Nowadays, the async TX API is written as a +layer on top of dmaengine, and acts as a client. Still, dmaengine +accommodates that API in some cases, and made some design choices to +ensure that it stayed compatible. + +For more information on the Async TX API, please look the relevant +documentation file in Documentation/crypto/async-tx-api.txt. + +DMAEngine APIs +============== + +``struct dma_device`` Initialization +------------------------------------ + +Just like any other kernel framework, the whole DMAEngine registration +relies on the driver filling a structure and registering against the +framework. In our case, that structure is dma_device. + +The first thing you need to do in your driver is to allocate this +structure. Any of the usual memory allocators will do, but you'll also +need to initialize a few fields in there: + +- ``channels``: should be initialized as a list using the + INIT_LIST_HEAD macro for example + +- ``src_addr_widths``: + should contain a bitmask of the supported source transfer width + +- ``dst_addr_widths``: + should contain a bitmask of the supported destination transfer width + +- ``directions``: + should contain a bitmask of the supported slave directions + (i.e. excluding mem2mem transfers) + +- ``residue_granularity``: + granularity of the transfer residue reported to dma_set_residue. + This can be either: + + - Descriptor: + your device doesn't support any kind of residue + reporting. The framework will only know that a particular + transaction descriptor is done. + + - Segment: + your device is able to report which chunks have been transferred + + - Burst: + your device is able to report which burst have been transferred + +- ``dev``: should hold the pointer to the ``struct device`` associated + to your current driver instance. + +Supported transaction types +--------------------------- + +The next thing you need is to set which transaction types your device +(and driver) supports. + +Our ``dma_device structure`` has a field called cap_mask that holds the +various types of transaction supported, and you need to modify this +mask using the dma_cap_set function, with various flags depending on +transaction types you support as an argument. + +All those capabilities are defined in the ``dma_transaction_type enum``, +in ``include/linux/dmaengine.h`` + +Currently, the types available are: + +- DMA_MEMCPY + + - The device is able to do memory to memory copies + +- DMA_XOR + + - The device is able to perform XOR operations on memory areas + + - Used to accelerate XOR intensive tasks, such as RAID5 + +- DMA_XOR_VAL + + - The device is able to perform parity check using the XOR + algorithm against a memory buffer. + +- DMA_PQ + + - The device is able to perform RAID6 P+Q computations, P being a + simple XOR, and Q being a Reed-Solomon algorithm. + +- DMA_PQ_VAL + + - The device is able to perform parity check using RAID6 P+Q + algorithm against a memory buffer. + +- DMA_INTERRUPT + + - The device is able to trigger a dummy transfer that will + generate periodic interrupts + + - Used by the client drivers to register a callback that will be + called on a regular basis through the DMA controller interrupt + +- DMA_PRIVATE + + - The devices only supports slave transfers, and as such isn't + available for async transfers. + +- DMA_ASYNC_TX + + - Must not be set by the device, and will be set by the framework + if needed + + - TODO: What is it about? + +- DMA_SLAVE + + - The device can handle device to memory transfers, including + scatter-gather transfers. + + - While in the mem2mem case we were having two distinct types to + deal with a single chunk to copy or a collection of them, here, + we just have a single transaction type that is supposed to + handle both. + + - If you want to transfer a single contiguous memory buffer, + simply build a scatter list with only one item. + +- DMA_CYCLIC + + - The device can handle cyclic transfers. + + - A cyclic transfer is a transfer where the chunk collection will + loop over itself, with the last item pointing to the first. + + - It's usually used for audio transfers, where you want to operate + on a single ring buffer that you will fill with your audio data. + +- DMA_INTERLEAVE + + - The device supports interleaved transfer. + + - These transfers can transfer data from a non-contiguous buffer + to a non-contiguous buffer, opposed to DMA_SLAVE that can + transfer data from a non-contiguous data set to a continuous + destination buffer. + + - It's usually used for 2d content transfers, in which case you + want to transfer a portion of uncompressed data directly to the + display to print it + +These various types will also affect how the source and destination +addresses change over time. + +Addresses pointing to RAM are typically incremented (or decremented) +after each transfer. In case of a ring buffer, they may loop +(DMA_CYCLIC). Addresses pointing to a device's register (e.g. a FIFO) +are typically fixed. + +Device operations +----------------- + +Our dma_device structure also requires a few function pointers in +order to implement the actual logic, now that we described what +operations we were able to perform. + +The functions that we have to fill in there, and hence have to +implement, obviously depend on the transaction types you reported as +supported. + +- ``device_alloc_chan_resources`` + +- ``device_free_chan_resources`` + + - These functions will be called whenever a driver will call + ``dma_request_channel`` or ``dma_release_channel`` for the first/last + time on the channel associated to that driver. + + - They are in charge of allocating/freeing all the needed + resources in order for that channel to be useful for your driver. + + - These functions can sleep. + +- ``device_prep_dma_*`` + + - These functions are matching the capabilities you registered + previously. + + - These functions all take the buffer or the scatterlist relevant + for the transfer being prepared, and should create a hardware + descriptor or a list of hardware descriptors from it + + - These functions can be called from an interrupt context + + - Any allocation you might do should be using the GFP_NOWAIT + flag, in order not to potentially sleep, but without depleting + the emergency pool either. + + - Drivers should try to pre-allocate any memory they might need + during the transfer setup at probe time to avoid putting to + much pressure on the nowait allocator. + + - It should return a unique instance of the + ``dma_async_tx_descriptor structure``, that further represents this + particular transfer. + + - This structure can be initialized using the function + ``dma_async_tx_descriptor_init``. + + - You'll also need to set two fields in this structure: + + - flags: + TODO: Can it be modified by the driver itself, or + should it be always the flags passed in the arguments + + - tx_submit: A pointer to a function you have to implement, + that is supposed to push the current transaction descriptor to a + pending queue, waiting for issue_pending to be called. + + - In this structure the function pointer callback_result can be + initialized in order for the submitter to be notified that a + transaction has completed. In the earlier code the function pointer + callback has been used. However it does not provide any status to the + transaction and will be deprecated. The result structure defined as + ``dmaengine_result`` that is passed in to callback_result + has two fields: + + - result: This provides the transfer result defined by + ``dmaengine_tx_result``. Either success or some error condition. + + - residue: Provides the residue bytes of the transfer for those that + support residue. + +- ``device_issue_pending`` + + - Takes the first transaction descriptor in the pending queue, + and starts the transfer. Whenever that transfer is done, it + should move to the next transaction in the list. + + - This function can be called in an interrupt context + +- ``device_tx_status`` + + - Should report the bytes left to go over on the given channel + + - Should only care about the transaction descriptor passed as + argument, not the currently active one on a given channel + + - The tx_state argument might be NULL + + - Should use dma_set_residue to report it + + - In the case of a cyclic transfer, it should only take into + account the current period. + + - This function can be called in an interrupt context. + +- device_config + + - Reconfigures the channel with the configuration given as argument + + - This command should NOT perform synchronously, or on any + currently queued transfers, but only on subsequent ones + + - In this case, the function will receive a ``dma_slave_config`` + structure pointer as an argument, that will detail which + configuration to use. + + - Even though that structure contains a direction field, this + field is deprecated in favor of the direction argument given to + the prep_* functions + + - This call is mandatory for slave operations only. This should NOT be + set or expected to be set for memcpy operations. + If a driver support both, it should use this call for slave + operations only and not for memcpy ones. + +- device_pause + + - Pauses a transfer on the channel + + - This command should operate synchronously on the channel, + pausing right away the work of the given channel + +- device_resume + + - Resumes a transfer on the channel + + - This command should operate synchronously on the channel, + resuming right away the work of the given channel + +- device_terminate_all + + - Aborts all the pending and ongoing transfers on the channel + + - For aborted transfers the complete callback should not be called + + - Can be called from atomic context or from within a complete + callback of a descriptor. Must not sleep. Drivers must be able + to handle this correctly. + + - Termination may be asynchronous. The driver does not have to + wait until the currently active transfer has completely stopped. + See device_synchronize. + +- device_synchronize + + - Must synchronize the termination of a channel to the current + context. + + - Must make sure that memory for previously submitted + descriptors is no longer accessed by the DMA controller. + + - Must make sure that all complete callbacks for previously + submitted descriptors have finished running and none are + scheduled to run. + + - May sleep. + + +Misc notes +========== + +(stuff that should be documented, but don't really know +where to put them) + +``dma_run_dependencies`` + +- Should be called at the end of an async TX transfer, and can be + ignored in the slave transfers case. + +- Makes sure that dependent operations are run before marking it + as complete. + +dma_cookie_t + +- it's a DMA transaction ID that will increment over time. + +- Not really relevant any more since the introduction of ``virt-dma`` + that abstracts it away. + +DMA_CTRL_ACK + +- If clear, the descriptor cannot be reused by provider until the + client acknowledges receipt, i.e. has has a chance to establish any + dependency chains + +- This can be acked by invoking async_tx_ack() + +- If set, does not mean descriptor can be reused + +DMA_CTRL_REUSE + +- If set, the descriptor can be reused after being completed. It should + not be freed by provider if this flag is set. + +- The descriptor should be prepared for reuse by invoking + ``dmaengine_desc_set_reuse()`` which will set DMA_CTRL_REUSE. + +- ``dmaengine_desc_set_reuse()`` will succeed only when channel support + reusable descriptor as exhibited by capabilities + +- As a consequence, if a device driver wants to skip the + ``dma_map_sg()`` and ``dma_unmap_sg()`` in between 2 transfers, + because the DMA'd data wasn't used, it can resubmit the transfer right after + its completion. + +- Descriptor can be freed in few ways + + - Clearing DMA_CTRL_REUSE by invoking + ``dmaengine_desc_clear_reuse()`` and submitting for last txn + + - Explicitly invoking ``dmaengine_desc_free()``, this can succeed only + when DMA_CTRL_REUSE is already set + + - Terminating the channel + +- DMA_PREP_CMD + + - If set, the client driver tells DMA controller that passed data in DMA + API is command data. + + - Interpretation of command data is DMA controller specific. It can be + used for issuing commands to other peripherals/register reads/register + writes for which the descriptor should be in different format from + normal data descriptors. + +General Design Notes +==================== + +Most of the DMAEngine drivers you'll see are based on a similar design +that handles the end of transfer interrupts in the handler, but defer +most work to a tasklet, including the start of a new transfer whenever +the previous transfer ended. + +This is a rather inefficient design though, because the inter-transfer +latency will be not only the interrupt latency, but also the +scheduling latency of the tasklet, which will leave the channel idle +in between, which will slow down the global transfer rate. + +You should avoid this kind of practice, and instead of electing a new +transfer in your tasklet, move that part to the interrupt handler in +order to have a shorter idle window (that we can't really avoid +anyway). + +Glossary +======== + +- Burst: A number of consecutive read or write operations that + can be queued to buffers before being flushed to memory. + +- Chunk: A contiguous collection of bursts + +- Transfer: A collection of chunks (be it contiguous or not) diff --git a/Documentation/driver-api/dmaengine/pxa_dma.rst b/Documentation/driver-api/dmaengine/pxa_dma.rst new file mode 100644 index 000000000..442ee691a --- /dev/null +++ b/Documentation/driver-api/dmaengine/pxa_dma.rst @@ -0,0 +1,190 @@ +============================== +PXA/MMP - DMA Slave controller +============================== + +Constraints +=========== + +a) Transfers hot queuing +A driver submitting a transfer and issuing it should be granted the transfer +is queued even on a running DMA channel. +This implies that the queuing doesn't wait for the previous transfer end, +and that the descriptor chaining is not only done in the irq/tasklet code +triggered by the end of the transfer. +A transfer which is submitted and issued on a phy doesn't wait for a phy to +stop and restart, but is submitted on a "running channel". The other +drivers, especially mmp_pdma waited for the phy to stop before relaunching +a new transfer. + +b) All transfers having asked for confirmation should be signaled +Any issued transfer with DMA_PREP_INTERRUPT should trigger a callback call. +This implies that even if an irq/tasklet is triggered by end of tx1, but +at the time of irq/dma tx2 is already finished, tx1->complete() and +tx2->complete() should be called. + +c) Channel running state +A driver should be able to query if a channel is running or not. For the +multimedia case, such as video capture, if a transfer is submitted and then +a check of the DMA channel reports a "stopped channel", the transfer should +not be issued until the next "start of frame interrupt", hence the need to +know if a channel is in running or stopped state. + +d) Bandwidth guarantee +The PXA architecture has 4 levels of DMAs priorities : high, normal, low. +The high priorities get twice as much bandwidth as the normal, which get twice +as much as the low priorities. +A driver should be able to request a priority, especially the real-time +ones such as pxa_camera with (big) throughputs. + +Design +====== +a) Virtual channels +Same concept as in sa11x0 driver, ie. a driver was assigned a "virtual +channel" linked to the requestor line, and the physical DMA channel is +assigned on the fly when the transfer is issued. + +b) Transfer anatomy for a scatter-gather transfer + +:: + + +------------+-----+---------------+----------------+-----------------+ + | desc-sg[0] | ... | desc-sg[last] | status updater | finisher/linker | + +------------+-----+---------------+----------------+-----------------+ + +This structure is pointed by dma->sg_cpu. +The descriptors are used as follows : + + - desc-sg[i]: i-th descriptor, transferring the i-th sg + element to the video buffer scatter gather + + - status updater + Transfers a single u32 to a well known dma coherent memory to leave + a trace that this transfer is done. The "well known" is unique per + physical channel, meaning that a read of this value will tell which + is the last finished transfer at that point in time. + + - finisher: has ddadr=DADDR_STOP, dcmd=ENDIRQEN + + - linker: has ddadr= desc-sg[0] of next transfer, dcmd=0 + +c) Transfers hot-chaining +Suppose the running chain is: + +:: + + Buffer 1 Buffer 2 + +---------+----+---+ +----+----+----+---+ + | d0 | .. | dN | l | | d0 | .. | dN | f | + +---------+----+-|-+ ^----+----+----+---+ + | | + +----+ + +After a call to dmaengine_submit(b3), the chain will look like: + +:: + + Buffer 1 Buffer 2 Buffer 3 + +---------+----+---+ +----+----+----+---+ +----+----+----+---+ + | d0 | .. | dN | l | | d0 | .. | dN | l | | d0 | .. | dN | f | + +---------+----+-|-+ ^----+----+----+-|-+ ^----+----+----+---+ + | | | | + +----+ +----+ + new_link + +If while new_link was created the DMA channel stopped, it is _not_ +restarted. Hot-chaining doesn't break the assumption that +dma_async_issue_pending() is to be used to ensure the transfer is actually started. + +One exception to this rule : + +- if Buffer1 and Buffer2 had all their addresses 8 bytes aligned + +- and if Buffer3 has at least one address not 4 bytes aligned + +- then hot-chaining cannot happen, as the channel must be stopped, the + "align bit" must be set, and the channel restarted As a consequence, + such a transfer tx_submit() will be queued on the submitted queue, and + this specific case if the DMA is already running in aligned mode. + +d) Transfers completion updater +Each time a transfer is completed on a channel, an interrupt might be +generated or not, up to the client's request. But in each case, the last +descriptor of a transfer, the "status updater", will write the latest +transfer being completed into the physical channel's completion mark. + +This will speed up residue calculation, for large transfers such as video +buffers which hold around 6k descriptors or more. This also allows without +any lock to find out what is the latest completed transfer in a running +DMA chain. + +e) Transfers completion, irq and tasklet +When a transfer flagged as "DMA_PREP_INTERRUPT" is finished, the dma irq +is raised. Upon this interrupt, a tasklet is scheduled for the physical +channel. + +The tasklet is responsible for : + +- reading the physical channel last updater mark + +- calling all the transfer callbacks of finished transfers, based on + that mark, and each transfer flags. + +If a transfer is completed while this handling is done, a dma irq will +be raised, and the tasklet will be scheduled once again, having a new +updater mark. + +f) Residue +Residue granularity will be descriptor based. The issued but not completed +transfers will be scanned for all of their descriptors against the +currently running descriptor. + +g) Most complicated case of driver's tx queues +The most tricky situation is when : + + - there are not "acked" transfers (tx0) + + - a driver submitted an aligned tx1, not chained + + - a driver submitted an aligned tx2 => tx2 is cold chained to tx1 + + - a driver issued tx1+tx2 => channel is running in aligned mode + + - a driver submitted an aligned tx3 => tx3 is hot-chained + + - a driver submitted an unaligned tx4 => tx4 is put in submitted queue, + not chained + + - a driver issued tx4 => tx4 is put in issued queue, not chained + + - a driver submitted an aligned tx5 => tx5 is put in submitted queue, not + chained + + - a driver submitted an aligned tx6 => tx6 is put in submitted queue, + cold chained to tx5 + + This translates into (after tx4 is issued) : + + - issued queue + + :: + + +-----+ +-----+ +-----+ +-----+ + | tx1 | | tx2 | | tx3 | | tx4 | + +---|-+ ^---|-+ ^-----+ +-----+ + | | | | + +---+ +---+ + - submitted queue + +-----+ +-----+ + | tx5 | | tx6 | + +---|-+ ^-----+ + | | + +---+ + +- completed queue : empty + +- allocated queue : tx0 + +It should be noted that after tx3 is completed, the channel is stopped, and +restarted in "unaligned mode" to handle tx4. + +Author: Robert Jarzmik <robert.jarzmik@free.fr> diff --git a/Documentation/driver-api/edac.rst b/Documentation/driver-api/edac.rst new file mode 100644 index 000000000..b8c742aa0 --- /dev/null +++ b/Documentation/driver-api/edac.rst @@ -0,0 +1,178 @@ +Error Detection And Correction (EDAC) Devices +============================================= + +Main Concepts used at the EDAC subsystem +---------------------------------------- + +There are several things to be aware of that aren't at all obvious, like +*sockets, *socket sets*, *banks*, *rows*, *chip-select rows*, *channels*, +etc... + +These are some of the many terms that are thrown about that don't always +mean what people think they mean (Inconceivable!). In the interest of +creating a common ground for discussion, terms and their definitions +will be established. + +* Memory devices + +The individual DRAM chips on a memory stick. These devices commonly +output 4 and 8 bits each (x4, x8). Grouping several of these in parallel +provides the number of bits that the memory controller expects: +typically 72 bits, in order to provide 64 bits + 8 bits of ECC data. + +* Memory Stick + +A printed circuit board that aggregates multiple memory devices in +parallel. In general, this is the Field Replaceable Unit (FRU) which +gets replaced, in the case of excessive errors. Most often it is also +called DIMM (Dual Inline Memory Module). + +* Memory Socket + +A physical connector on the motherboard that accepts a single memory +stick. Also called as "slot" on several datasheets. + +* Channel + +A memory controller channel, responsible to communicate with a group of +DIMMs. Each channel has its own independent control (command) and data +bus, and can be used independently or grouped with other channels. + +* Branch + +It is typically the highest hierarchy on a Fully-Buffered DIMM memory +controller. Typically, it contains two channels. Two channels at the +same branch can be used in single mode or in lockstep mode. When +lockstep is enabled, the cacheline is doubled, but it generally brings +some performance penalty. Also, it is generally not possible to point to +just one memory stick when an error occurs, as the error correction code +is calculated using two DIMMs instead of one. Due to that, it is capable +of correcting more errors than on single mode. + +* Single-channel + +The data accessed by the memory controller is contained into one dimm +only. E. g. if the data is 64 bits-wide, the data flows to the CPU using +one 64 bits parallel access. Typically used with SDR, DDR, DDR2 and DDR3 +memories. FB-DIMM and RAMBUS use a different concept for channel, so +this concept doesn't apply there. + +* Double-channel + +The data size accessed by the memory controller is interlaced into two +dimms, accessed at the same time. E. g. if the DIMM is 64 bits-wide (72 +bits with ECC), the data flows to the CPU using a 128 bits parallel +access. + +* Chip-select row + +This is the name of the DRAM signal used to select the DRAM ranks to be +accessed. Common chip-select rows for single channel are 64 bits, for +dual channel 128 bits. It may not be visible by the memory controller, +as some DIMM types have a memory buffer that can hide direct access to +it from the Memory Controller. + +* Single-Ranked stick + +A Single-ranked stick has 1 chip-select row of memory. Motherboards +commonly drive two chip-select pins to a memory stick. A single-ranked +stick, will occupy only one of those rows. The other will be unused. + +.. _doubleranked: + +* Double-Ranked stick + +A double-ranked stick has two chip-select rows which access different +sets of memory devices. The two rows cannot be accessed concurrently. + +* Double-sided stick + +**DEPRECATED TERM**, see :ref:`Double-Ranked stick <doubleranked>`. + +A double-sided stick has two chip-select rows which access different sets +of memory devices. The two rows cannot be accessed concurrently. +"Double-sided" is irrespective of the memory devices being mounted on +both sides of the memory stick. + +* Socket set + +All of the memory sticks that are required for a single memory access or +all of the memory sticks spanned by a chip-select row. A single socket +set has two chip-select rows and if double-sided sticks are used these +will occupy those chip-select rows. + +* Bank + +This term is avoided because it is unclear when needing to distinguish +between chip-select rows and socket sets. + + +Memory Controllers +------------------ + +Most of the EDAC core is focused on doing Memory Controller error detection. +The :c:func:`edac_mc_alloc`. It uses internally the struct ``mem_ctl_info`` +to describe the memory controllers, with is an opaque struct for the EDAC +drivers. Only the EDAC core is allowed to touch it. + +.. kernel-doc:: include/linux/edac.h + +.. kernel-doc:: drivers/edac/edac_mc.h + +PCI Controllers +--------------- + +The EDAC subsystem provides a mechanism to handle PCI controllers by calling +the :c:func:`edac_pci_alloc_ctl_info`. It will use the struct +:c:type:`edac_pci_ctl_info` to describe the PCI controllers. + +.. kernel-doc:: drivers/edac/edac_pci.h + +EDAC Blocks +----------- + +The EDAC subsystem also provides a generic mechanism to report errors on +other parts of the hardware via :c:func:`edac_device_alloc_ctl_info` function. + +The structures :c:type:`edac_dev_sysfs_block_attribute`, +:c:type:`edac_device_block`, :c:type:`edac_device_instance` and +:c:type:`edac_device_ctl_info` provide a generic or abstract 'edac_device' +representation at sysfs. + +This set of structures and the code that implements the APIs for the same, provide for registering EDAC type devices which are NOT standard memory or +PCI, like: + +- CPU caches (L1 and L2) +- DMA engines +- Core CPU switches +- Fabric switch units +- PCIe interface controllers +- other EDAC/ECC type devices that can be monitored for + errors, etc. + +It allows for a 2 level set of hierarchy. + +For example, a cache could be composed of L1, L2 and L3 levels of cache. +Each CPU core would have its own L1 cache, while sharing L2 and maybe L3 +caches. On such case, those can be represented via the following sysfs +nodes:: + + /sys/devices/system/edac/.. + + pci/ <existing pci directory (if available)> + mc/ <existing memory device directory> + cpu/cpu0/.. <L1 and L2 block directory> + /L1-cache/ce_count + /ue_count + /L2-cache/ce_count + /ue_count + cpu/cpu1/.. <L1 and L2 block directory> + /L1-cache/ce_count + /ue_count + /L2-cache/ce_count + /ue_count + ... + + the L1 and L2 directories would be "edac_device_block's" + +.. kernel-doc:: drivers/edac/edac_device.h diff --git a/Documentation/driver-api/firmware/built-in-fw.rst b/Documentation/driver-api/firmware/built-in-fw.rst new file mode 100644 index 000000000..396cdf591 --- /dev/null +++ b/Documentation/driver-api/firmware/built-in-fw.rst @@ -0,0 +1,33 @@ +================= +Built-in firmware +================= + +Firmware can be built-in to the kernel, this means building the firmware +into vmlinux directly, to enable avoiding having to look for firmware from +the filesystem. Instead, firmware can be looked for inside the kernel +directly. You can enable built-in firmware using the kernel configuration +options: + + * CONFIG_EXTRA_FIRMWARE + * CONFIG_EXTRA_FIRMWARE_DIR + +There are a few reasons why you might want to consider building your firmware +into the kernel with CONFIG_EXTRA_FIRMWARE: + +* Speed +* Firmware is needed for accessing the boot device, and the user doesn't + want to stuff the firmware into the boot initramfs. + +Even if you have these needs there are a few reasons why you may not be +able to make use of built-in firmware: + +* Legalese - firmware is non-GPL compatible +* Some firmware may be optional +* Firmware upgrades are possible, therefore a new firmware would implicate + a complete kernel rebuild. +* Some firmware files may be really large in size. The remote-proc subsystem + is an example subsystem which deals with these sorts of firmware +* The firmware may need to be scraped out from some device specific location + dynamically, an example is calibration data for for some WiFi chipsets. This + calibration data can be unique per sold device. + diff --git a/Documentation/driver-api/firmware/core.rst b/Documentation/driver-api/firmware/core.rst new file mode 100644 index 000000000..1d1688cbc --- /dev/null +++ b/Documentation/driver-api/firmware/core.rst @@ -0,0 +1,16 @@ +========================== +Firmware API core features +========================== + +The firmware API has a rich set of core features available. This section +documents these features. + +.. toctree:: + + fw_search_path + built-in-fw + firmware_cache + direct-fs-lookup + fallback-mechanisms + lookup-order + diff --git a/Documentation/driver-api/firmware/direct-fs-lookup.rst b/Documentation/driver-api/firmware/direct-fs-lookup.rst new file mode 100644 index 000000000..82b4d585a --- /dev/null +++ b/Documentation/driver-api/firmware/direct-fs-lookup.rst @@ -0,0 +1,30 @@ +======================== +Direct filesystem lookup +======================== + +Direct filesystem lookup is the most common form of firmware lookup performed +by the kernel. The kernel looks for the firmware directly on the root +filesystem in the paths documented in the section 'Firmware search paths'. +The filesystem lookup is implemented in fw_get_filesystem_firmware(), it +uses common core kernel file loader facility kernel_read_file_from_path(). +The max path allowed is PATH_MAX -- currently this is 4096 characters. + +It is recommended you keep /lib/firmware paths on your root filesystem, +avoid having a separate partition for them in order to avoid possible +races with lookups and avoid uses of the custom fallback mechanisms +documented below. + +Firmware and initramfs +---------------------- + +Drivers which are built-in to the kernel should have the firmware integrated +also as part of the initramfs used to boot the kernel given that otherwise +a race is possible with loading the driver and the real rootfs not yet being +available. Stuffing the firmware into initramfs resolves this race issue, +however note that using initrd does not suffice to address the same race. + +There are circumstances that justify not wanting to include firmware into +initramfs, such as dealing with large firmware firmware files for the +remote-proc subsystem. For such cases using a userspace fallback mechanism +is currently the only viable solution as only userspace can know for sure +when the real rootfs is ready and mounted. diff --git a/Documentation/driver-api/firmware/fallback-mechanisms.rst b/Documentation/driver-api/firmware/fallback-mechanisms.rst new file mode 100644 index 000000000..8b041d0ab --- /dev/null +++ b/Documentation/driver-api/firmware/fallback-mechanisms.rst @@ -0,0 +1,204 @@ +=================== +Fallback mechanisms +=================== + +A fallback mechanism is supported to allow to overcome failures to do a direct +filesystem lookup on the root filesystem or when the firmware simply cannot be +installed for practical reasons on the root filesystem. The kernel +configuration options related to supporting the firmware fallback mechanism are: + + * CONFIG_FW_LOADER_USER_HELPER: enables building the firmware fallback + mechanism. Most distributions enable this option today. If enabled but + CONFIG_FW_LOADER_USER_HELPER_FALLBACK is disabled, only the custom fallback + mechanism is available and for the request_firmware_nowait() call. + * CONFIG_FW_LOADER_USER_HELPER_FALLBACK: force enables each request to + enable the kobject uevent fallback mechanism on all firmware API calls + except request_firmware_direct(). Most distributions disable this option + today. The call request_firmware_nowait() allows for one alternative + fallback mechanism: if this kconfig option is enabled and your second + argument to request_firmware_nowait(), uevent, is set to false you are + informing the kernel that you have a custom fallback mechanism and it will + manually load the firmware. Read below for more details. + +Note that this means when having this configuration: + +CONFIG_FW_LOADER_USER_HELPER=y +CONFIG_FW_LOADER_USER_HELPER_FALLBACK=n + +the kobject uevent fallback mechanism will never take effect even +for request_firmware_nowait() when uevent is set to true. + +Justifying the firmware fallback mechanism +========================================== + +Direct filesystem lookups may fail for a variety of reasons. Known reasons for +this are worth itemizing and documenting as it justifies the need for the +fallback mechanism: + +* Race against access with the root filesystem upon bootup. + +* Races upon resume from suspend. This is resolved by the firmware cache, but + the firmware cache is only supported if you use uevents, and its not + supported for request_firmware_into_buf(). + +* Firmware is not accessible through typical means: + * It cannot be installed into the root filesystem + * The firmware provides very unique device specific data tailored for + the unit gathered with local information. An example is calibration + data for WiFi chipsets for mobile devices. This calibration data is + not common to all units, but tailored per unit. Such information may + be installed on a separate flash partition other than where the root + filesystem is provided. + +Types of fallback mechanisms +============================ + +There are really two fallback mechanisms available using one shared sysfs +interface as a loading facility: + +* Kobject uevent fallback mechanism +* Custom fallback mechanism + +First lets document the shared sysfs loading facility. + +Firmware sysfs loading facility +=============================== + +In order to help device drivers upload firmware using a fallback mechanism +the firmware infrastructure creates a sysfs interface to enable userspace +to load and indicate when firmware is ready. The sysfs directory is created +via fw_create_instance(). This call creates a new struct device named after +the firmware requested, and establishes it in the device hierarchy by +associating the device used to make the request as the device's parent. +The sysfs directory's file attributes are defined and controlled through +the new device's class (firmware_class) and group (fw_dev_attr_groups). +This is actually where the original firmware_class module name came from, +given that originally the only firmware loading mechanism available was the +mechanism we now use as a fallback mechanism, which registers a struct class +firmware_class. Because the attributes exposed are part of the module name, the +module name firmware_class cannot be renamed in the future, to ensure backward +compatibility with old userspace. + +To load firmware using the sysfs interface we expose a loading indicator, +and a file upload firmware into: + + * /sys/$DEVPATH/loading + * /sys/$DEVPATH/data + +To upload firmware you will echo 1 onto the loading file to indicate +you are loading firmware. You then write the firmware into the data file, +and you notify the kernel the firmware is ready by echo'ing 0 onto +the loading file. + +The firmware device used to help load firmware using sysfs is only created if +direct firmware loading fails and if the fallback mechanism is enabled for your +firmware request, this is set up with :c:func:`firmware_fallback_sysfs`. It is +important to re-iterate that no device is created if a direct filesystem lookup +succeeded. + +Using:: + + echo 1 > /sys/$DEVPATH/loading + +Will clean any previous partial load at once and make the firmware API +return an error. When loading firmware the firmware_class grows a buffer +for the firmware in PAGE_SIZE increments to hold the image as it comes in. + +firmware_data_read() and firmware_loading_show() are just provided for the +test_firmware driver for testing, they are not called in normal use or +expected to be used regularly by userspace. + +firmware_fallback_sysfs +----------------------- +.. kernel-doc:: drivers/base/firmware_loader/fallback.c + :functions: firmware_fallback_sysfs + +Firmware kobject uevent fallback mechanism +========================================== + +Since a device is created for the sysfs interface to help load firmware as a +fallback mechanism userspace can be informed of the addition of the device by +relying on kobject uevents. The addition of the device into the device +hierarchy means the fallback mechanism for firmware loading has been initiated. +For details of implementation refer to fw_load_sysfs_fallback(), in particular +on the use of dev_set_uevent_suppress() and kobject_uevent(). + +The kernel's kobject uevent mechanism is implemented in lib/kobject_uevent.c, +it issues uevents to userspace. As a supplement to kobject uevents Linux +distributions could also enable CONFIG_UEVENT_HELPER_PATH, which makes use of +core kernel's usermode helper (UMH) functionality to call out to a userspace +helper for kobject uevents. In practice though no standard distribution has +ever used the CONFIG_UEVENT_HELPER_PATH. If CONFIG_UEVENT_HELPER_PATH is +enabled this binary would be called each time kobject_uevent_env() gets called +in the kernel for each kobject uevent triggered. + +Different implementations have been supported in userspace to take advantage of +this fallback mechanism. When firmware loading was only possible using the +sysfs mechanism the userspace component "hotplug" provided the functionality of +monitoring for kobject events. Historically this was superseded be systemd's +udev, however firmware loading support was removed from udev as of systemd +commit be2ea723b1d0 ("udev: remove userspace firmware loading support") +as of v217 on August, 2014. This means most Linux distributions today are +not using or taking advantage of the firmware fallback mechanism provided +by kobject uevents. This is specially exacerbated due to the fact that most +distributions today disable CONFIG_FW_LOADER_USER_HELPER_FALLBACK. + +Refer to do_firmware_uevent() for details of the kobject event variables +setup. The variables currently passed to userspace with a "kobject add" +event are: + +* FIRMWARE=firmware name +* TIMEOUT=timeout value +* ASYNC=whether or not the API request was asynchronous + +By default DEVPATH is set by the internal kernel kobject infrastructure. +Below is an example simple kobject uevent script:: + + # Both $DEVPATH and $FIRMWARE are already provided in the environment. + MY_FW_DIR=/lib/firmware/ + echo 1 > /sys/$DEVPATH/loading + cat $MY_FW_DIR/$FIRMWARE > /sys/$DEVPATH/data + echo 0 > /sys/$DEVPATH/loading + +Firmware custom fallback mechanism +================================== + +Users of the request_firmware_nowait() call have yet another option available +at their disposal: rely on the sysfs fallback mechanism but request that no +kobject uevents be issued to userspace. The original logic behind this +was that utilities other than udev might be required to lookup firmware +in non-traditional paths -- paths outside of the listing documented in the +section 'Direct filesystem lookup'. This option is not available to any of +the other API calls as uevents are always forced for them. + +Since uevents are only meaningful if the fallback mechanism is enabled +in your kernel it would seem odd to enable uevents with kernels that do not +have the fallback mechanism enabled in their kernels. Unfortunately we also +rely on the uevent flag which can be disabled by request_firmware_nowait() to +also setup the firmware cache for firmware requests. As documented above, +the firmware cache is only set up if uevent is enabled for an API call. +Although this can disable the firmware cache for request_firmware_nowait() +calls, users of this API should not use it for the purposes of disabling +the cache as that was not the original purpose of the flag. Not setting +the uevent flag means you want to opt-in for the firmware fallback mechanism +but you want to suppress kobject uevents, as you have a custom solution which +will monitor for your device addition into the device hierarchy somehow and +load firmware for you through a custom path. + +Firmware fallback timeout +========================= + +The firmware fallback mechanism has a timeout. If firmware is not loaded +onto the sysfs interface by the timeout value an error is sent to the +driver. By default the timeout is set to 60 seconds if uevents are +desirable, otherwise MAX_JIFFY_OFFSET is used (max timeout possible). +The logic behind using MAX_JIFFY_OFFSET for non-uevents is that a custom +solution will have as much time as it needs to load firmware. + +You can customize the firmware timeout by echo'ing your desired timeout into +the following file: + +* /sys/class/firmware/timeout + +If you echo 0 into it means MAX_JIFFY_OFFSET will be used. The data type +for the timeout is an int. diff --git a/Documentation/driver-api/firmware/firmware_cache.rst b/Documentation/driver-api/firmware/firmware_cache.rst new file mode 100644 index 000000000..c2e69d9c6 --- /dev/null +++ b/Documentation/driver-api/firmware/firmware_cache.rst @@ -0,0 +1,51 @@ +============== +Firmware cache +============== + +When Linux resumes from suspend some device drivers require firmware lookups to +re-initialize devices. During resume there may be a period of time during which +firmware lookups are not possible, during this short period of time firmware +requests will fail. Time is of essence though, and delaying drivers to wait for +the root filesystem for firmware delays user experience with device +functionality. In order to support these requirements the firmware +infrastructure implements a firmware cache for device drivers for most API +calls, automatically behind the scenes. + +The firmware cache makes using certain firmware API calls safe during a device +driver's suspend and resume callback. Users of these API calls needn't cache +the firmware by themselves for dealing with firmware loss during system resume. + +The firmware cache works by requesting for firmware prior to suspend and +caching it in memory. Upon resume device drivers using the firmware API will +have access to the firmware immediately, without having to wait for the root +filesystem to mount or dealing with possible race issues with lookups as the +root filesystem mounts. + +Some implementation details about the firmware cache setup: + +* The firmware cache is setup by adding a devres entry for each device that + uses all synchronous call except :c:func:`request_firmware_into_buf`. + +* If an asynchronous call is used the firmware cache is only set up for a + device if if the second argument (uevent) to request_firmware_nowait() is + true. When uevent is true it requests that a kobject uevent be sent to + userspace for the firmware request through the sysfs fallback mechanism + if the firmware file is not found. + +* If the firmware cache is determined to be needed as per the above two + criteria the firmware cache is setup by adding a devres entry for the + device making the firmware request. + +* The firmware devres entry is maintained throughout the lifetime of the + device. This means that even if you release_firmware() the firmware cache + will still be used on resume from suspend. + +* The timeout for the fallback mechanism is temporarily reduced to 10 seconds + as the firmware cache is set up during suspend, the timeout is set back to + the old value you had configured after the cache is set up. + +* Upon suspend any pending non-uevent firmware requests are killed to avoid + stalling the kernel, this is done with kill_requests_without_uevent(). Kernel + calls requiring the non-uevent therefore need to implement their own firmware + cache mechanism but must not use the firmware API on suspend. + diff --git a/Documentation/driver-api/firmware/fw_search_path.rst b/Documentation/driver-api/firmware/fw_search_path.rst new file mode 100644 index 000000000..a360f1009 --- /dev/null +++ b/Documentation/driver-api/firmware/fw_search_path.rst @@ -0,0 +1,26 @@ +===================== +Firmware search paths +===================== + +The following search paths are used to look for firmware on your +root filesystem. + +* fw_path_para - module parameter - default is empty so this is ignored +* /lib/firmware/updates/UTS_RELEASE/ +* /lib/firmware/updates/ +* /lib/firmware/UTS_RELEASE/ +* /lib/firmware/ + +The module parameter ''path'' can be passed to the firmware_class module +to activate the first optional custom fw_path_para. The custom path can +only be up to 256 characters long. The kernel parameter passed would be: + +* 'firmware_class.path=$CUSTOMIZED_PATH' + +There is an alternative to customize the path at run time after bootup, you +can use the file: + +* /sys/module/firmware_class/parameters/path + +You would echo into it your custom path and firmware requested will be +searched for there first. diff --git a/Documentation/driver-api/firmware/index.rst b/Documentation/driver-api/firmware/index.rst new file mode 100644 index 000000000..29da39ec4 --- /dev/null +++ b/Documentation/driver-api/firmware/index.rst @@ -0,0 +1,17 @@ +================== +Linux Firmware API +================== + +.. toctree:: + + introduction + core + request_firmware + other_interfaces + +.. only:: subproject and html + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/firmware/introduction.rst b/Documentation/driver-api/firmware/introduction.rst new file mode 100644 index 000000000..211cb44eb --- /dev/null +++ b/Documentation/driver-api/firmware/introduction.rst @@ -0,0 +1,27 @@ +============ +Introduction +============ + +The firmware API enables kernel code to request files required +for functionality from userspace, the uses vary: + +* Microcode for CPU errata +* Device driver firmware, required to be loaded onto device + microcontrollers +* Device driver information data (calibration data, EEPROM overrides), + some of which can be completely optional. + +Types of firmware requests +========================== + +There are two types of calls: + +* Synchronous +* Asynchronous + +Which one you use vary depending on your requirements, the rule of thumb +however is you should strive to use the asynchronous APIs unless you also +are already using asynchronous initialization mechanisms which will not +stall or delay boot. Even if loading firmware does not take a lot of time +processing firmware might, and this can still delay boot or initialization, +as such mechanisms such as asynchronous probe can help supplement drivers. diff --git a/Documentation/driver-api/firmware/lookup-order.rst b/Documentation/driver-api/firmware/lookup-order.rst new file mode 100644 index 000000000..88c817396 --- /dev/null +++ b/Documentation/driver-api/firmware/lookup-order.rst @@ -0,0 +1,18 @@ +===================== +Firmware lookup order +===================== + +Different functionality is available to enable firmware to be found. +Below is chronological order of how firmware will be looked for once +a driver issues a firmware API call. + +* The ''Built-in firmware'' is checked first, if the firmware is present we + return it immediately +* The ''Firmware cache'' is looked at next. If the firmware is found we + return it immediately +* The ''Direct filesystem lookup'' is performed next, if found we + return it immediately +* If no firmware has been found and the fallback mechanism was enabled + the sysfs interface is created. After this either a kobject uevent + is issued or the custom firmware loading is relied upon for firmware + loading up to the timeout value. diff --git a/Documentation/driver-api/firmware/other_interfaces.rst b/Documentation/driver-api/firmware/other_interfaces.rst new file mode 100644 index 000000000..36c47b1e9 --- /dev/null +++ b/Documentation/driver-api/firmware/other_interfaces.rst @@ -0,0 +1,15 @@ +Other Firmware Interfaces +========================= + +DMI Interfaces +-------------- + +.. kernel-doc:: drivers/firmware/dmi_scan.c + :export: + +EDD Interfaces +-------------- + +.. kernel-doc:: drivers/firmware/edd.c + :internal: + diff --git a/Documentation/driver-api/firmware/request_firmware.rst b/Documentation/driver-api/firmware/request_firmware.rst new file mode 100644 index 000000000..f62bdcbfe --- /dev/null +++ b/Documentation/driver-api/firmware/request_firmware.rst @@ -0,0 +1,75 @@ +==================== +request_firmware API +==================== + +You would typically load firmware and then load it into your device somehow. +The typical firmware work flow is reflected below:: + + if(request_firmware(&fw_entry, $FIRMWARE, device) == 0) + copy_fw_to_device(fw_entry->data, fw_entry->size); + release_firmware(fw_entry); + +Synchronous firmware requests +============================= + +Synchronous firmware requests will wait until the firmware is found or until +an error is returned. + +request_firmware +---------------- +.. kernel-doc:: drivers/base/firmware_loader/main.c + :functions: request_firmware + +firmware_request_nowarn +----------------------- +.. kernel-doc:: drivers/base/firmware_loader/main.c + :functions: firmware_request_nowarn + +request_firmware_direct +----------------------- +.. kernel-doc:: drivers/base/firmware_loader/main.c + :functions: request_firmware_direct + +request_firmware_into_buf +------------------------- +.. kernel-doc:: drivers/base/firmware_loader/main.c + :functions: request_firmware_into_buf + +Asynchronous firmware requests +============================== + +Asynchronous firmware requests allow driver code to not have to wait +until the firmware or an error is returned. Function callbacks are +provided so that when the firmware or an error is found the driver is +informed through the callback. request_firmware_nowait() cannot be called +in atomic contexts. + +request_firmware_nowait +----------------------- +.. kernel-doc:: drivers/base/firmware_loader/main.c + :functions: request_firmware_nowait + +Special optimizations on reboot +=============================== + +Some devices have an optimization in place to enable the firmware to be +retained during system reboot. When such optimizations are used the driver +author must ensure the firmware is still available on resume from suspend, +this can be done with firmware_request_cache() instead of requesting for the +firmware to be loaded. + +firmware_request_cache() +------------------------ +.. kernel-doc:: drivers/base/firmware_loader/main.c + :functions: firmware_request_cache + +request firmware API expected driver use +======================================== + +Once an API call returns you process the firmware and then release the +firmware. For example if you used request_firmware() and it returns, +the driver has the firmware image accessible in fw_entry->{data,size}. +If something went wrong request_firmware() returns non-zero and fw_entry +is set to NULL. Once your driver is done with processing the firmware it +can call call release_firmware(fw_entry) to release the firmware image +and any related resource. diff --git a/Documentation/driver-api/fpga/fpga-bridge.rst b/Documentation/driver-api/fpga/fpga-bridge.rst new file mode 100644 index 000000000..2c2aaca89 --- /dev/null +++ b/Documentation/driver-api/fpga/fpga-bridge.rst @@ -0,0 +1,49 @@ +FPGA Bridge +=========== + +API to implement a new FPGA bridge +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: include/linux/fpga/fpga-bridge.h + :functions: fpga_bridge + +.. kernel-doc:: include/linux/fpga/fpga-bridge.h + :functions: fpga_bridge_ops + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_create + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_free + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_register + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_unregister + +API to control an FPGA bridge +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +You probably won't need these directly. FPGA regions should handle this. + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: of_fpga_bridge_get + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_get + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_put + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_get_to_list + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: of_fpga_bridge_get_to_list + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_enable + +.. kernel-doc:: drivers/fpga/fpga-bridge.c + :functions: fpga_bridge_disable diff --git a/Documentation/driver-api/fpga/fpga-mgr.rst b/Documentation/driver-api/fpga/fpga-mgr.rst new file mode 100644 index 000000000..82b6dbbd3 --- /dev/null +++ b/Documentation/driver-api/fpga/fpga-mgr.rst @@ -0,0 +1,225 @@ +FPGA Manager +============ + +Overview +-------- + +The FPGA manager core exports a set of functions for programming an FPGA with +an image. The API is manufacturer agnostic. All manufacturer specifics are +hidden away in a low level driver which registers a set of ops with the core. +The FPGA image data itself is very manufacturer specific, but for our purposes +it's just binary data. The FPGA manager core won't parse it. + +The FPGA image to be programmed can be in a scatter gather list, a single +contiguous buffer, or a firmware file. Because allocating contiguous kernel +memory for the buffer should be avoided, users are encouraged to use a scatter +gather list instead if possible. + +The particulars for programming the image are presented in a structure (struct +fpga_image_info). This struct contains parameters such as pointers to the +FPGA image as well as image-specific particulars such as whether the image was +built for full or partial reconfiguration. + +How to support a new FPGA device +-------------------------------- + +To add another FPGA manager, write a driver that implements a set of ops. The +probe function calls fpga_mgr_register(), such as:: + + static const struct fpga_manager_ops socfpga_fpga_ops = { + .write_init = socfpga_fpga_ops_configure_init, + .write = socfpga_fpga_ops_configure_write, + .write_complete = socfpga_fpga_ops_configure_complete, + .state = socfpga_fpga_ops_state, + }; + + static int socfpga_fpga_probe(struct platform_device *pdev) + { + struct device *dev = &pdev->dev; + struct socfpga_fpga_priv *priv; + struct fpga_manager *mgr; + int ret; + + priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + /* + * do ioremaps, get interrupts, etc. and save + * them in priv + */ + + mgr = fpga_mgr_create(dev, "Altera SOCFPGA FPGA Manager", + &socfpga_fpga_ops, priv); + if (!mgr) + return -ENOMEM; + + platform_set_drvdata(pdev, mgr); + + ret = fpga_mgr_register(mgr); + if (ret) + fpga_mgr_free(mgr); + + return ret; + } + + static int socfpga_fpga_remove(struct platform_device *pdev) + { + struct fpga_manager *mgr = platform_get_drvdata(pdev); + + fpga_mgr_unregister(mgr); + + return 0; + } + + +The ops will implement whatever device specific register writes are needed to +do the programming sequence for this particular FPGA. These ops return 0 for +success or negative error codes otherwise. + +The programming sequence is:: + 1. .write_init + 2. .write or .write_sg (may be called once or multiple times) + 3. .write_complete + +The .write_init function will prepare the FPGA to receive the image data. The +buffer passed into .write_init will be at most .initial_header_size bytes long; +if the whole bitstream is not immediately available then the core code will +buffer up at least this much before starting. + +The .write function writes a buffer to the FPGA. The buffer may be contain the +whole FPGA image or may be a smaller chunk of an FPGA image. In the latter +case, this function is called multiple times for successive chunks. This interface +is suitable for drivers which use PIO. + +The .write_sg version behaves the same as .write except the input is a sg_table +scatter list. This interface is suitable for drivers which use DMA. + +The .write_complete function is called after all the image has been written +to put the FPGA into operating mode. + +The ops include a .state function which will determine the state the FPGA is in +and return a code of type enum fpga_mgr_states. It doesn't result in a change +in state. + +How to write an image buffer to a supported FPGA +------------------------------------------------ + +Some sample code:: + + #include <linux/fpga/fpga-mgr.h> + + struct fpga_manager *mgr; + struct fpga_image_info *info; + int ret; + + /* + * Get a reference to FPGA manager. The manager is not locked, so you can + * hold onto this reference without it preventing programming. + * + * This example uses the device node of the manager. Alternatively, use + * fpga_mgr_get(dev) instead if you have the device. + */ + mgr = of_fpga_mgr_get(mgr_node); + + /* struct with information about the FPGA image to program. */ + info = fpga_image_info_alloc(dev); + + /* flags indicates whether to do full or partial reconfiguration */ + info->flags = FPGA_MGR_PARTIAL_RECONFIG; + + /* + * At this point, indicate where the image is. This is pseudo-code; you're + * going to use one of these three. + */ + if (image is in a scatter gather table) { + + info->sgt = [your scatter gather table] + + } else if (image is in a buffer) { + + info->buf = [your image buffer] + info->count = [image buffer size] + + } else if (image is in a firmware file) { + + info->firmware_name = devm_kstrdup(dev, firmware_name, GFP_KERNEL); + + } + + /* Get exclusive control of FPGA manager */ + ret = fpga_mgr_lock(mgr); + + /* Load the buffer to the FPGA */ + ret = fpga_mgr_buf_load(mgr, &info, buf, count); + + /* Release the FPGA manager */ + fpga_mgr_unlock(mgr); + fpga_mgr_put(mgr); + + /* Deallocate the image info if you're done with it */ + fpga_image_info_free(info); + +API for implementing a new FPGA Manager driver +---------------------------------------------- + +.. kernel-doc:: include/linux/fpga/fpga-mgr.h + :functions: fpga_manager + +.. kernel-doc:: include/linux/fpga/fpga-mgr.h + :functions: fpga_manager_ops + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_create + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_free + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_register + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_unregister + +API for programming an FPGA +--------------------------- + +FPGA Manager flags + +.. kernel-doc:: include/linux/fpga/fpga-mgr.h + :doc: FPGA Manager flags + +.. kernel-doc:: include/linux/fpga/fpga-mgr.h + :functions: fpga_image_info + +.. kernel-doc:: include/linux/fpga/fpga-mgr.h + :functions: fpga_mgr_states + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_image_info_alloc + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_image_info_free + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: of_fpga_mgr_get + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_get + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_put + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_lock + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_unlock + +.. kernel-doc:: include/linux/fpga/fpga-mgr.h + :functions: fpga_mgr_states + +Note - use :c:func:`fpga_region_program_fpga()` instead of :c:func:`fpga_mgr_load()` + +.. kernel-doc:: drivers/fpga/fpga-mgr.c + :functions: fpga_mgr_load diff --git a/Documentation/driver-api/fpga/fpga-region.rst b/Documentation/driver-api/fpga/fpga-region.rst new file mode 100644 index 000000000..f30333ce8 --- /dev/null +++ b/Documentation/driver-api/fpga/fpga-region.rst @@ -0,0 +1,102 @@ +FPGA Region +=========== + +Overview +-------- + +This document is meant to be a brief overview of the FPGA region API usage. A +more conceptual look at regions can be found in the Device Tree binding +document [#f1]_. + +For the purposes of this API document, let's just say that a region associates +an FPGA Manager and a bridge (or bridges) with a reprogrammable region of an +FPGA or the whole FPGA. The API provides a way to register a region and to +program a region. + +Currently the only layer above fpga-region.c in the kernel is the Device Tree +support (of-fpga-region.c) described in [#f1]_. The DT support layer uses regions +to program the FPGA and then DT to handle enumeration. The common region code +is intended to be used by other schemes that have other ways of accomplishing +enumeration after programming. + +An fpga-region can be set up to know the following things: + + * which FPGA manager to use to do the programming + + * which bridges to disable before programming and enable afterwards. + +Additional info needed to program the FPGA image is passed in the struct +fpga_image_info including: + + * pointers to the image as either a scatter-gather buffer, a contiguous + buffer, or the name of firmware file + + * flags indicating specifics such as whether the image is for partial + reconfiguration. + +How to program an FPGA using a region +------------------------------------- + +First, allocate the info struct:: + + info = fpga_image_info_alloc(dev); + if (!info) + return -ENOMEM; + +Set flags as needed, i.e.:: + + info->flags |= FPGA_MGR_PARTIAL_RECONFIG; + +Point to your FPGA image, such as:: + + info->sgt = &sgt; + +Add info to region and do the programming:: + + region->info = info; + ret = fpga_region_program_fpga(region); + +:c:func:`fpga_region_program_fpga()` operates on info passed in the +fpga_image_info (region->info). This function will attempt to: + + * lock the region's mutex + * lock the region's FPGA manager + * build a list of FPGA bridges if a method has been specified to do so + * disable the bridges + * program the FPGA + * re-enable the bridges + * release the locks + +Then you will want to enumerate whatever hardware has appeared in the FPGA. + +How to add a new FPGA region +---------------------------- + +An example of usage can be seen in the probe function of [#f2]_. + +.. [#f1] ../devicetree/bindings/fpga/fpga-region.txt +.. [#f2] ../../drivers/fpga/of-fpga-region.c + +API to program an FPGA +---------------------- + +.. kernel-doc:: drivers/fpga/fpga-region.c + :functions: fpga_region_program_fpga + +API to add a new FPGA region +---------------------------- + +.. kernel-doc:: include/linux/fpga/fpga-region.h + :functions: fpga_region + +.. kernel-doc:: drivers/fpga/fpga-region.c + :functions: fpga_region_create + +.. kernel-doc:: drivers/fpga/fpga-region.c + :functions: fpga_region_free + +.. kernel-doc:: drivers/fpga/fpga-region.c + :functions: fpga_region_register + +.. kernel-doc:: drivers/fpga/fpga-region.c + :functions: fpga_region_unregister diff --git a/Documentation/driver-api/fpga/index.rst b/Documentation/driver-api/fpga/index.rst new file mode 100644 index 000000000..c51e5ebd5 --- /dev/null +++ b/Documentation/driver-api/fpga/index.rst @@ -0,0 +1,13 @@ +============== +FPGA Subsystem +============== + +:Author: Alan Tull + +.. toctree:: + :maxdepth: 2 + + intro + fpga-mgr + fpga-bridge + fpga-region diff --git a/Documentation/driver-api/fpga/intro.rst b/Documentation/driver-api/fpga/intro.rst new file mode 100644 index 000000000..50d1cab84 --- /dev/null +++ b/Documentation/driver-api/fpga/intro.rst @@ -0,0 +1,54 @@ +Introduction +============ + +The FPGA subsystem supports reprogramming FPGAs dynamically under +Linux. Some of the core intentions of the FPGA subsystems are: + +* The FPGA subsystem is vendor agnostic. + +* The FPGA subsystem separates upper layers (userspace interfaces and + enumeration) from lower layers that know how to program a specific + FPGA. + +* Code should not be shared between upper and lower layers. This + should go without saying. If that seems necessary, there's probably + framework functionality that can be added that will benefit + other users. Write the linux-fpga mailing list and maintainers and + seek out a solution that expands the framework for broad reuse. + +* Generally, when adding code, think of the future. Plan for reuse. + +The framework in the kernel is divided into: + +FPGA Manager +------------ + +If you are adding a new FPGA or a new method of programming an FPGA, +this is the subsystem for you. Low level FPGA manager drivers contain +the knowledge of how to program a specific device. This subsystem +includes the framework in fpga-mgr.c and the low level drivers that +are registered with it. + +FPGA Bridge +----------- + +FPGA Bridges prevent spurious signals from going out of an FPGA or a +region of an FPGA during programming. They are disabled before +programming begins and re-enabled afterwards. An FPGA bridge may be +actual hard hardware that gates a bus to a CPU or a soft ("freeze") +bridge in FPGA fabric that surrounds a partial reconfiguration region +of an FPGA. This subsystem includes fpga-bridge.c and the low level +drivers that are registered with it. + +FPGA Region +----------- + +If you are adding a new interface to the FPGA framework, add it on top +of an FPGA region to allow the most reuse of your interface. + +The FPGA Region framework (fpga-region.c) associates managers and +bridges as reconfigurable regions. A region may refer to the whole +FPGA in full reconfiguration or to a partial reconfiguration region. + +The Device Tree FPGA Region support (of-fpga-region.c) handles +reprogramming FPGAs when device tree overlays are applied. diff --git a/Documentation/driver-api/frame-buffer.rst b/Documentation/driver-api/frame-buffer.rst new file mode 100644 index 000000000..9dd3060f0 --- /dev/null +++ b/Documentation/driver-api/frame-buffer.rst @@ -0,0 +1,62 @@ +Frame Buffer Library +==================== + +The frame buffer drivers depend heavily on four data structures. These +structures are declared in include/linux/fb.h. They are fb_info, +fb_var_screeninfo, fb_fix_screeninfo and fb_monospecs. The last +three can be made available to and from userland. + +fb_info defines the current state of a particular video card. Inside +fb_info, there exists a fb_ops structure which is a collection of +needed functions to make fbdev and fbcon work. fb_info is only visible +to the kernel. + +fb_var_screeninfo is used to describe the features of a video card +that are user defined. With fb_var_screeninfo, things such as depth +and the resolution may be defined. + +The next structure is fb_fix_screeninfo. This defines the properties +of a card that are created when a mode is set and can't be changed +otherwise. A good example of this is the start of the frame buffer +memory. This "locks" the address of the frame buffer memory, so that it +cannot be changed or moved. + +The last structure is fb_monospecs. In the old API, there was little +importance for fb_monospecs. This allowed for forbidden things such as +setting a mode of 800x600 on a fix frequency monitor. With the new API, +fb_monospecs prevents such things, and if used correctly, can prevent a +monitor from being cooked. fb_monospecs will not be useful until +kernels 2.5.x. + +Frame Buffer Memory +------------------- + +.. kernel-doc:: drivers/video/fbdev/core/fbmem.c + :export: + +Frame Buffer Colormap +--------------------- + +.. kernel-doc:: drivers/video/fbdev/core/fbcmap.c + :export: + +Frame Buffer Video Mode Database +-------------------------------- + +.. kernel-doc:: drivers/video/fbdev/core/modedb.c + :internal: + +.. kernel-doc:: drivers/video/fbdev/core/modedb.c + :export: + +Frame Buffer Macintosh Video Mode Database +------------------------------------------ + +.. kernel-doc:: drivers/video/fbdev/macmodes.c + :export: + +Frame Buffer Fonts +------------------ + +Refer to the file lib/fonts/fonts.c for more information. + diff --git a/Documentation/driver-api/gpio/board.rst b/Documentation/driver-api/gpio/board.rst new file mode 100644 index 000000000..2c112553d --- /dev/null +++ b/Documentation/driver-api/gpio/board.rst @@ -0,0 +1,195 @@ +============= +GPIO Mappings +============= + +This document explains how GPIOs can be assigned to given devices and functions. + +Note that it only applies to the new descriptor-based interface. For a +description of the deprecated integer-based GPIO interface please refer to +gpio-legacy.txt (actually, there is no real mapping possible with the old +interface; you just fetch an integer from somewhere and request the +corresponding GPIO). + +All platforms can enable the GPIO library, but if the platform strictly +requires GPIO functionality to be present, it needs to select GPIOLIB from its +Kconfig. Then, how GPIOs are mapped depends on what the platform uses to +describe its hardware layout. Currently, mappings can be defined through device +tree, ACPI, and platform data. + +Device Tree +----------- +GPIOs can easily be mapped to devices and functions in the device tree. The +exact way to do it depends on the GPIO controller providing the GPIOs, see the +device tree bindings for your controller. + +GPIOs mappings are defined in the consumer device's node, in a property named +<function>-gpios, where <function> is the function the driver will request +through gpiod_get(). For example:: + + foo_device { + compatible = "acme,foo"; + ... + led-gpios = <&gpio 15 GPIO_ACTIVE_HIGH>, /* red */ + <&gpio 16 GPIO_ACTIVE_HIGH>, /* green */ + <&gpio 17 GPIO_ACTIVE_HIGH>; /* blue */ + + power-gpios = <&gpio 1 GPIO_ACTIVE_LOW>; + }; + +Properties named <function>-gpio are also considered valid and old bindings use +it but are only supported for compatibility reasons and should not be used for +newer bindings since it has been deprecated. + +This property will make GPIOs 15, 16 and 17 available to the driver under the +"led" function, and GPIO 1 as the "power" GPIO:: + + struct gpio_desc *red, *green, *blue, *power; + + red = gpiod_get_index(dev, "led", 0, GPIOD_OUT_HIGH); + green = gpiod_get_index(dev, "led", 1, GPIOD_OUT_HIGH); + blue = gpiod_get_index(dev, "led", 2, GPIOD_OUT_HIGH); + + power = gpiod_get(dev, "power", GPIOD_OUT_HIGH); + +The led GPIOs will be active high, while the power GPIO will be active low (i.e. +gpiod_is_active_low(power) will be true). + +The second parameter of the gpiod_get() functions, the con_id string, has to be +the <function>-prefix of the GPIO suffixes ("gpios" or "gpio", automatically +looked up by the gpiod functions internally) used in the device tree. With above +"led-gpios" example, use the prefix without the "-" as con_id parameter: "led". + +Internally, the GPIO subsystem prefixes the GPIO suffix ("gpios" or "gpio") +with the string passed in con_id to get the resulting string +(``snprintf(... "%s-%s", con_id, gpio_suffixes[]``). + +ACPI +---- +ACPI also supports function names for GPIOs in a similar fashion to DT. +The above DT example can be converted to an equivalent ACPI description +with the help of _DSD (Device Specific Data), introduced in ACPI 5.1:: + + Device (FOO) { + Name (_CRS, ResourceTemplate () { + GpioIo (Exclusive, ..., IoRestrictionOutputOnly, + "\\_SB.GPI0") {15} // red + GpioIo (Exclusive, ..., IoRestrictionOutputOnly, + "\\_SB.GPI0") {16} // green + GpioIo (Exclusive, ..., IoRestrictionOutputOnly, + "\\_SB.GPI0") {17} // blue + GpioIo (Exclusive, ..., IoRestrictionOutputOnly, + "\\_SB.GPI0") {1} // power + }) + + Name (_DSD, Package () { + ToUUID("daffd814-6eba-4d8c-8a91-bc9bbf4aa301"), + Package () { + Package () { + "led-gpios", + Package () { + ^FOO, 0, 0, 1, + ^FOO, 1, 0, 1, + ^FOO, 2, 0, 1, + } + }, + Package () { + "power-gpios", + Package () {^FOO, 3, 0, 0}, + }, + } + }) + } + +For more information about the ACPI GPIO bindings see +Documentation/acpi/gpio-properties.txt. + +Platform Data +------------- +Finally, GPIOs can be bound to devices and functions using platform data. Board +files that desire to do so need to include the following header:: + + #include <linux/gpio/machine.h> + +GPIOs are mapped by the means of tables of lookups, containing instances of the +gpiod_lookup structure. Two macros are defined to help declaring such mappings:: + + GPIO_LOOKUP(chip_label, chip_hwnum, con_id, flags) + GPIO_LOOKUP_IDX(chip_label, chip_hwnum, con_id, idx, flags) + +where + + - chip_label is the label of the gpiod_chip instance providing the GPIO + - chip_hwnum is the hardware number of the GPIO within the chip + - con_id is the name of the GPIO function from the device point of view. It + can be NULL, in which case it will match any function. + - idx is the index of the GPIO within the function. + - flags is defined to specify the following properties: + * GPIO_ACTIVE_HIGH - GPIO line is active high + * GPIO_ACTIVE_LOW - GPIO line is active low + * GPIO_OPEN_DRAIN - GPIO line is set up as open drain + * GPIO_OPEN_SOURCE - GPIO line is set up as open source + * GPIO_PERSISTENT - GPIO line is persistent during + suspend/resume and maintains its value + * GPIO_TRANSITORY - GPIO line is transitory and may loose its + electrical state during suspend/resume + +In the future, these flags might be extended to support more properties. + +Note that GPIO_LOOKUP() is just a shortcut to GPIO_LOOKUP_IDX() where idx = 0. + +A lookup table can then be defined as follows, with an empty entry defining its +end. The 'dev_id' field of the table is the identifier of the device that will +make use of these GPIOs. It can be NULL, in which case it will be matched for +calls to gpiod_get() with a NULL device. + +.. code-block:: c + + struct gpiod_lookup_table gpios_table = { + .dev_id = "foo.0", + .table = { + GPIO_LOOKUP_IDX("gpio.0", 15, "led", 0, GPIO_ACTIVE_HIGH), + GPIO_LOOKUP_IDX("gpio.0", 16, "led", 1, GPIO_ACTIVE_HIGH), + GPIO_LOOKUP_IDX("gpio.0", 17, "led", 2, GPIO_ACTIVE_HIGH), + GPIO_LOOKUP("gpio.0", 1, "power", GPIO_ACTIVE_LOW), + { }, + }, + }; + +And the table can be added by the board code as follows:: + + gpiod_add_lookup_table(&gpios_table); + +The driver controlling "foo.0" will then be able to obtain its GPIOs as follows:: + + struct gpio_desc *red, *green, *blue, *power; + + red = gpiod_get_index(dev, "led", 0, GPIOD_OUT_HIGH); + green = gpiod_get_index(dev, "led", 1, GPIOD_OUT_HIGH); + blue = gpiod_get_index(dev, "led", 2, GPIOD_OUT_HIGH); + + power = gpiod_get(dev, "power", GPIOD_OUT_HIGH); + +Since the "led" GPIOs are mapped as active-high, this example will switch their +signals to 1, i.e. enabling the LEDs. And for the "power" GPIO, which is mapped +as active-low, its actual signal will be 0 after this code. Contrary to the +legacy integer GPIO interface, the active-low property is handled during +mapping and is thus transparent to GPIO consumers. + +A set of functions such as gpiod_set_value() is available to work with +the new descriptor-oriented interface. + +Boards using platform data can also hog GPIO lines by defining GPIO hog tables. + +.. code-block:: c + + struct gpiod_hog gpio_hog_table[] = { + GPIO_HOG("gpio.0", 10, "foo", GPIO_ACTIVE_LOW, GPIOD_OUT_HIGH), + { } + }; + +And the table can be added to the board code as follows:: + + gpiod_add_hogs(gpio_hog_table); + +The line will be hogged as soon as the gpiochip is created or - in case the +chip was created earlier - when the hog table is registered. diff --git a/Documentation/driver-api/gpio/consumer.rst b/Documentation/driver-api/gpio/consumer.rst new file mode 100644 index 000000000..aa03f389d --- /dev/null +++ b/Documentation/driver-api/gpio/consumer.rst @@ -0,0 +1,439 @@ +================================== +GPIO Descriptor Consumer Interface +================================== + +This document describes the consumer interface of the GPIO framework. Note that +it describes the new descriptor-based interface. For a description of the +deprecated integer-based GPIO interface please refer to gpio-legacy.txt. + + +Guidelines for GPIOs consumers +============================== + +Drivers that can't work without standard GPIO calls should have Kconfig entries +that depend on GPIOLIB or select GPIOLIB. The functions that allow a driver to +obtain and use GPIOs are available by including the following file: + + #include <linux/gpio/consumer.h> + +There are static inline stubs for all functions in the header file in the case +where GPIOLIB is disabled. When these stubs are called they will emit +warnings. These stubs are used for two use cases: + +- Simple compile coverage with e.g. COMPILE_TEST - it does not matter that + the current platform does not enable or select GPIOLIB because we are not + going to execute the system anyway. + +- Truly optional GPIOLIB support - where the driver does not really make use + of the GPIOs on certain compile-time configurations for certain systems, but + will use it under other compile-time configurations. In this case the + consumer must make sure not to call into these functions, or the user will + be met with console warnings that may be perceived as intimidating. + +All the functions that work with the descriptor-based GPIO interface are +prefixed with ``gpiod_``. The ``gpio_`` prefix is used for the legacy +interface. No other function in the kernel should use these prefixes. The use +of the legacy functions is strongly discouraged, new code should use +<linux/gpio/consumer.h> and descriptors exclusively. + + +Obtaining and Disposing GPIOs +============================= + +With the descriptor-based interface, GPIOs are identified with an opaque, +non-forgeable handler that must be obtained through a call to one of the +gpiod_get() functions. Like many other kernel subsystems, gpiod_get() takes the +device that will use the GPIO and the function the requested GPIO is supposed to +fulfill:: + + struct gpio_desc *gpiod_get(struct device *dev, const char *con_id, + enum gpiod_flags flags) + +If a function is implemented by using several GPIOs together (e.g. a simple LED +device that displays digits), an additional index argument can be specified:: + + struct gpio_desc *gpiod_get_index(struct device *dev, + const char *con_id, unsigned int idx, + enum gpiod_flags flags) + +For a more detailed description of the con_id parameter in the DeviceTree case +see Documentation/driver-api/gpio/board.rst + +The flags parameter is used to optionally specify a direction and initial value +for the GPIO. Values can be: + +* GPIOD_ASIS or 0 to not initialize the GPIO at all. The direction must be set + later with one of the dedicated functions. +* GPIOD_IN to initialize the GPIO as input. +* GPIOD_OUT_LOW to initialize the GPIO as output with a value of 0. +* GPIOD_OUT_HIGH to initialize the GPIO as output with a value of 1. +* GPIOD_OUT_LOW_OPEN_DRAIN same as GPIOD_OUT_LOW but also enforce the line + to be electrically used with open drain. +* GPIOD_OUT_HIGH_OPEN_DRAIN same as GPIOD_OUT_HIGH but also enforce the line + to be electrically used with open drain. + +The two last flags are used for use cases where open drain is mandatory, such +as I2C: if the line is not already configured as open drain in the mappings +(see board.txt), then open drain will be enforced anyway and a warning will be +printed that the board configuration needs to be updated to match the use case. + +Both functions return either a valid GPIO descriptor, or an error code checkable +with IS_ERR() (they will never return a NULL pointer). -ENOENT will be returned +if and only if no GPIO has been assigned to the device/function/index triplet, +other error codes are used for cases where a GPIO has been assigned but an error +occurred while trying to acquire it. This is useful to discriminate between mere +errors and an absence of GPIO for optional GPIO parameters. For the common +pattern where a GPIO is optional, the gpiod_get_optional() and +gpiod_get_index_optional() functions can be used. These functions return NULL +instead of -ENOENT if no GPIO has been assigned to the requested function:: + + struct gpio_desc *gpiod_get_optional(struct device *dev, + const char *con_id, + enum gpiod_flags flags) + + struct gpio_desc *gpiod_get_index_optional(struct device *dev, + const char *con_id, + unsigned int index, + enum gpiod_flags flags) + +Note that gpio_get*_optional() functions (and their managed variants), unlike +the rest of gpiolib API, also return NULL when gpiolib support is disabled. +This is helpful to driver authors, since they do not need to special case +-ENOSYS return codes. System integrators should however be careful to enable +gpiolib on systems that need it. + +For a function using multiple GPIOs all of those can be obtained with one call:: + + struct gpio_descs *gpiod_get_array(struct device *dev, + const char *con_id, + enum gpiod_flags flags) + +This function returns a struct gpio_descs which contains an array of +descriptors:: + + struct gpio_descs { + unsigned int ndescs; + struct gpio_desc *desc[]; + } + +The following function returns NULL instead of -ENOENT if no GPIOs have been +assigned to the requested function:: + + struct gpio_descs *gpiod_get_array_optional(struct device *dev, + const char *con_id, + enum gpiod_flags flags) + +Device-managed variants of these functions are also defined:: + + struct gpio_desc *devm_gpiod_get(struct device *dev, const char *con_id, + enum gpiod_flags flags) + + struct gpio_desc *devm_gpiod_get_index(struct device *dev, + const char *con_id, + unsigned int idx, + enum gpiod_flags flags) + + struct gpio_desc *devm_gpiod_get_optional(struct device *dev, + const char *con_id, + enum gpiod_flags flags) + + struct gpio_desc *devm_gpiod_get_index_optional(struct device *dev, + const char *con_id, + unsigned int index, + enum gpiod_flags flags) + + struct gpio_descs *devm_gpiod_get_array(struct device *dev, + const char *con_id, + enum gpiod_flags flags) + + struct gpio_descs *devm_gpiod_get_array_optional(struct device *dev, + const char *con_id, + enum gpiod_flags flags) + +A GPIO descriptor can be disposed of using the gpiod_put() function:: + + void gpiod_put(struct gpio_desc *desc) + +For an array of GPIOs this function can be used:: + + void gpiod_put_array(struct gpio_descs *descs) + +It is strictly forbidden to use a descriptor after calling these functions. +It is also not allowed to individually release descriptors (using gpiod_put()) +from an array acquired with gpiod_get_array(). + +The device-managed variants are, unsurprisingly:: + + void devm_gpiod_put(struct device *dev, struct gpio_desc *desc) + + void devm_gpiod_put_array(struct device *dev, struct gpio_descs *descs) + + +Using GPIOs +=========== + +Setting Direction +----------------- +The first thing a driver must do with a GPIO is setting its direction. If no +direction-setting flags have been given to gpiod_get*(), this is done by +invoking one of the gpiod_direction_*() functions:: + + int gpiod_direction_input(struct gpio_desc *desc) + int gpiod_direction_output(struct gpio_desc *desc, int value) + +The return value is zero for success, else a negative errno. It should be +checked, since the get/set calls don't return errors and since misconfiguration +is possible. You should normally issue these calls from a task context. However, +for spinlock-safe GPIOs it is OK to use them before tasking is enabled, as part +of early board setup. + +For output GPIOs, the value provided becomes the initial output value. This +helps avoid signal glitching during system startup. + +A driver can also query the current direction of a GPIO:: + + int gpiod_get_direction(const struct gpio_desc *desc) + +This function returns 0 for output, 1 for input, or an error code in case of error. + +Be aware that there is no default direction for GPIOs. Therefore, **using a GPIO +without setting its direction first is illegal and will result in undefined +behavior!** + + +Spinlock-Safe GPIO Access +------------------------- +Most GPIO controllers can be accessed with memory read/write instructions. Those +don't need to sleep, and can safely be done from inside hard (non-threaded) IRQ +handlers and similar contexts. + +Use the following calls to access GPIOs from an atomic context:: + + int gpiod_get_value(const struct gpio_desc *desc); + void gpiod_set_value(struct gpio_desc *desc, int value); + +The values are boolean, zero for low, nonzero for high. When reading the value +of an output pin, the value returned should be what's seen on the pin. That +won't always match the specified output value, because of issues including +open-drain signaling and output latencies. + +The get/set calls do not return errors because "invalid GPIO" should have been +reported earlier from gpiod_direction_*(). However, note that not all platforms +can read the value of output pins; those that can't should always return zero. +Also, using these calls for GPIOs that can't safely be accessed without sleeping +(see below) is an error. + + +GPIO Access That May Sleep +-------------------------- +Some GPIO controllers must be accessed using message based buses like I2C or +SPI. Commands to read or write those GPIO values require waiting to get to the +head of a queue to transmit a command and get its response. This requires +sleeping, which can't be done from inside IRQ handlers. + +Platforms that support this type of GPIO distinguish them from other GPIOs by +returning nonzero from this call:: + + int gpiod_cansleep(const struct gpio_desc *desc) + +To access such GPIOs, a different set of accessors is defined:: + + int gpiod_get_value_cansleep(const struct gpio_desc *desc) + void gpiod_set_value_cansleep(struct gpio_desc *desc, int value) + +Accessing such GPIOs requires a context which may sleep, for example a threaded +IRQ handler, and those accessors must be used instead of spinlock-safe +accessors without the cansleep() name suffix. + +Other than the fact that these accessors might sleep, and will work on GPIOs +that can't be accessed from hardIRQ handlers, these calls act the same as the +spinlock-safe calls. + + +The active low and open drain semantics +--------------------------------------- +As a consumer should not have to care about the physical line level, all of the +gpiod_set_value_xxx() or gpiod_set_array_value_xxx() functions operate with +the *logical* value. With this they take the active low property into account. +This means that they check whether the GPIO is configured to be active low, +and if so, they manipulate the passed value before the physical line level is +driven. + +The same is applicable for open drain or open source output lines: those do not +actively drive their output high (open drain) or low (open source), they just +switch their output to a high impedance value. The consumer should not need to +care. (For details read about open drain in driver.txt.) + +With this, all the gpiod_set_(array)_value_xxx() functions interpret the +parameter "value" as "asserted" ("1") or "de-asserted" ("0"). The physical line +level will be driven accordingly. + +As an example, if the active low property for a dedicated GPIO is set, and the +gpiod_set_(array)_value_xxx() passes "asserted" ("1"), the physical line level +will be driven low. + +To summarize:: + + Function (example) line property physical line + gpiod_set_raw_value(desc, 0); don't care low + gpiod_set_raw_value(desc, 1); don't care high + gpiod_set_value(desc, 0); default (active high) low + gpiod_set_value(desc, 1); default (active high) high + gpiod_set_value(desc, 0); active low high + gpiod_set_value(desc, 1); active low low + gpiod_set_value(desc, 0); default (active high) low + gpiod_set_value(desc, 1); default (active high) high + gpiod_set_value(desc, 0); open drain low + gpiod_set_value(desc, 1); open drain high impedance + gpiod_set_value(desc, 0); open source high impedance + gpiod_set_value(desc, 1); open source high + +It is possible to override these semantics using the set_raw/get_raw functions +but it should be avoided as much as possible, especially by system-agnostic drivers +which should not need to care about the actual physical line level and worry about +the logical value instead. + + +Accessing raw GPIO values +------------------------- +Consumers exist that need to manage the logical state of a GPIO line, i.e. the value +their device will actually receive, no matter what lies between it and the GPIO +line. + +The following set of calls ignore the active-low or open drain property of a GPIO and +work on the raw line value:: + + int gpiod_get_raw_value(const struct gpio_desc *desc) + void gpiod_set_raw_value(struct gpio_desc *desc, int value) + int gpiod_get_raw_value_cansleep(const struct gpio_desc *desc) + void gpiod_set_raw_value_cansleep(struct gpio_desc *desc, int value) + int gpiod_direction_output_raw(struct gpio_desc *desc, int value) + +The active low state of a GPIO can also be queried using the following call:: + + int gpiod_is_active_low(const struct gpio_desc *desc) + +Note that these functions should only be used with great moderation; a driver +should not have to care about the physical line level or open drain semantics. + + +Access multiple GPIOs with a single function call +------------------------------------------------- +The following functions get or set the values of an array of GPIOs:: + + int gpiod_get_array_value(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array); + int gpiod_get_raw_array_value(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array); + int gpiod_get_array_value_cansleep(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array); + int gpiod_get_raw_array_value_cansleep(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array); + + void gpiod_set_array_value(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array) + void gpiod_set_raw_array_value(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array) + void gpiod_set_array_value_cansleep(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array) + void gpiod_set_raw_array_value_cansleep(unsigned int array_size, + struct gpio_desc **desc_array, + int *value_array) + +The array can be an arbitrary set of GPIOs. The functions will try to access +GPIOs belonging to the same bank or chip simultaneously if supported by the +corresponding chip driver. In that case a significantly improved performance +can be expected. If simultaneous access is not possible the GPIOs will be +accessed sequentially. + +The functions take three arguments: + * array_size - the number of array elements + * desc_array - an array of GPIO descriptors + * value_array - an array to store the GPIOs' values (get) or + an array of values to assign to the GPIOs (set) + +The descriptor array can be obtained using the gpiod_get_array() function +or one of its variants. If the group of descriptors returned by that function +matches the desired group of GPIOs, those GPIOs can be accessed by simply using +the struct gpio_descs returned by gpiod_get_array():: + + struct gpio_descs *my_gpio_descs = gpiod_get_array(...); + gpiod_set_array_value(my_gpio_descs->ndescs, my_gpio_descs->desc, + my_gpio_values); + +It is also possible to access a completely arbitrary array of descriptors. The +descriptors may be obtained using any combination of gpiod_get() and +gpiod_get_array(). Afterwards the array of descriptors has to be setup +manually before it can be passed to one of the above functions. + +Note that for optimal performance GPIOs belonging to the same chip should be +contiguous within the array of descriptors. + +The return value of gpiod_get_array_value() and its variants is 0 on success +or negative on error. Note the difference to gpiod_get_value(), which returns +0 or 1 on success to convey the GPIO value. With the array functions, the GPIO +values are stored in value_array rather than passed back as return value. + + +GPIOs mapped to IRQs +-------------------- +GPIO lines can quite often be used as IRQs. You can get the IRQ number +corresponding to a given GPIO using the following call:: + + int gpiod_to_irq(const struct gpio_desc *desc) + +It will return an IRQ number, or a negative errno code if the mapping can't be +done (most likely because that particular GPIO cannot be used as IRQ). It is an +unchecked error to use a GPIO that wasn't set up as an input using +gpiod_direction_input(), or to use an IRQ number that didn't originally come +from gpiod_to_irq(). gpiod_to_irq() is not allowed to sleep. + +Non-error values returned from gpiod_to_irq() can be passed to request_irq() or +free_irq(). They will often be stored into IRQ resources for platform devices, +by the board-specific initialization code. Note that IRQ trigger options are +part of the IRQ interface, e.g. IRQF_TRIGGER_FALLING, as are system wakeup +capabilities. + + +GPIOs and ACPI +============== + +On ACPI systems, GPIOs are described by GpioIo()/GpioInt() resources listed by +the _CRS configuration objects of devices. Those resources do not provide +connection IDs (names) for GPIOs, so it is necessary to use an additional +mechanism for this purpose. + +Systems compliant with ACPI 5.1 or newer may provide a _DSD configuration object +which, among other things, may be used to provide connection IDs for specific +GPIOs described by the GpioIo()/GpioInt() resources in _CRS. If that is the +case, it will be handled by the GPIO subsystem automatically. However, if the +_DSD is not present, the mappings between GpioIo()/GpioInt() resources and GPIO +connection IDs need to be provided by device drivers. + +For details refer to Documentation/acpi/gpio-properties.txt + + +Interacting With the Legacy GPIO Subsystem +========================================== +Many kernel subsystems still handle GPIOs using the legacy integer-based +interface. Although it is strongly encouraged to upgrade them to the safer +descriptor-based API, the following two functions allow you to convert a GPIO +descriptor into the GPIO integer namespace and vice-versa:: + + int desc_to_gpio(const struct gpio_desc *desc) + struct gpio_desc *gpio_to_desc(unsigned gpio) + +The GPIO number returned by desc_to_gpio() can be safely used as long as the +GPIO descriptor has not been freed. All the same, a GPIO number passed to +gpio_to_desc() must have been properly acquired, and usage of the returned GPIO +descriptor is only possible after the GPIO number has been released. + +Freeing a GPIO obtained by one API with the other API is forbidden and an +unchecked error. diff --git a/Documentation/driver-api/gpio/driver.rst b/Documentation/driver-api/gpio/driver.rst new file mode 100644 index 000000000..cbe024284 --- /dev/null +++ b/Documentation/driver-api/gpio/driver.rst @@ -0,0 +1,429 @@ +================================ +GPIO Descriptor Driver Interface +================================ + +This document serves as a guide for GPIO chip drivers writers. Note that it +describes the new descriptor-based interface. For a description of the +deprecated integer-based GPIO interface please refer to gpio-legacy.txt. + +Each GPIO controller driver needs to include the following header, which defines +the structures used to define a GPIO driver: + + #include <linux/gpio/driver.h> + + +Internal Representation of GPIOs +================================ + +Inside a GPIO driver, individual GPIOs are identified by their hardware number, +which is a unique number between 0 and n, n being the number of GPIOs managed by +the chip. This number is purely internal: the hardware number of a particular +GPIO descriptor is never made visible outside of the driver. + +On top of this internal number, each GPIO also need to have a global number in +the integer GPIO namespace so that it can be used with the legacy GPIO +interface. Each chip must thus have a "base" number (which can be automatically +assigned), and for each GPIO the global number will be (base + hardware number). +Although the integer representation is considered deprecated, it still has many +users and thus needs to be maintained. + +So for example one platform could use numbers 32-159 for GPIOs, with a +controller defining 128 GPIOs at a "base" of 32 ; while another platform uses +numbers 0..63 with one set of GPIO controllers, 64-79 with another type of GPIO +controller, and on one particular board 80-95 with an FPGA. The numbers need not +be contiguous; either of those platforms could also use numbers 2000-2063 to +identify GPIOs in a bank of I2C GPIO expanders. + + +Controller Drivers: gpio_chip +============================= + +In the gpiolib framework each GPIO controller is packaged as a "struct +gpio_chip" (see linux/gpio/driver.h for its complete definition) with members +common to each controller of that type: + + - methods to establish GPIO line direction + - methods used to access GPIO line values + - method to set electrical configuration for a given GPIO line + - method to return the IRQ number associated to a given GPIO line + - flag saying whether calls to its methods may sleep + - optional line names array to identify lines + - optional debugfs dump method (showing extra state like pullup config) + - optional base number (will be automatically assigned if omitted) + - optional label for diagnostics and GPIO chip mapping using platform data + +The code implementing a gpio_chip should support multiple instances of the +controller, possibly using the driver model. That code will configure each +gpio_chip and issue ``gpiochip_add[_data]()`` or ``devm_gpiochip_add_data()``. +Removing a GPIO controller should be rare; use ``[devm_]gpiochip_remove()`` +when it is unavoidable. + +Often a gpio_chip is part of an instance-specific structure with states not +exposed by the GPIO interfaces, such as addressing, power management, and more. +Chips such as audio codecs will have complex non-GPIO states. + +Any debugfs dump method should normally ignore signals which haven't been +requested as GPIOs. They can use gpiochip_is_requested(), which returns either +NULL or the label associated with that GPIO when it was requested. + +RT_FULL: the GPIO driver should not use spinlock_t or any sleepable APIs +(like PM runtime) in its gpio_chip implementation (.get/.set and direction +control callbacks) if it is expected to call GPIO APIs from atomic context +on -RT (inside hard IRQ handlers and similar contexts). Normally this should +not be required. + + +GPIO electrical configuration +----------------------------- + +GPIOs can be configured for several electrical modes of operation by using the +.set_config() callback. Currently this API supports setting debouncing and +single-ended modes (open drain/open source). These settings are described +below. + +The .set_config() callback uses the same enumerators and configuration +semantics as the generic pin control drivers. This is not a coincidence: it is +possible to assign the .set_config() to the function gpiochip_generic_config() +which will result in pinctrl_gpio_set_config() being called and eventually +ending up in the pin control back-end "behind" the GPIO controller, usually +closer to the actual pins. This way the pin controller can manage the below +listed GPIO configurations. + +If a pin controller back-end is used, the GPIO controller or hardware +description needs to provide "GPIO ranges" mapping the GPIO line offsets to pin +numbers on the pin controller so they can properly cross-reference each other. + + +GPIOs with debounce support +--------------------------- + +Debouncing is a configuration set to a pin indicating that it is connected to +a mechanical switch or button, or similar that may bounce. Bouncing means the +line is pulled high/low quickly at very short intervals for mechanical +reasons. This can result in the value being unstable or irqs fireing repeatedly +unless the line is debounced. + +Debouncing in practice involves setting up a timer when something happens on +the line, wait a little while and then sample the line again, so see if it +still has the same value (low or high). This could also be repeated by a clever +state machine, waiting for a line to become stable. In either case, it sets +a certain number of milliseconds for debouncing, or just "on/off" if that time +is not configurable. + + +GPIOs with open drain/source support +------------------------------------ + +Open drain (CMOS) or open collector (TTL) means the line is not actively driven +high: instead you provide the drain/collector as output, so when the transistor +is not open, it will present a high-impedance (tristate) to the external rail:: + + + CMOS CONFIGURATION TTL CONFIGURATION + + ||--- out +--- out + in ----|| |/ + ||--+ in ----| + | |\ + GND GND + +This configuration is normally used as a way to achieve one of two things: + +- Level-shifting: to reach a logical level higher than that of the silicon + where the output resides. + +- inverse wire-OR on an I/O line, for example a GPIO line, making it possible + for any driving stage on the line to drive it low even if any other output + to the same line is simultaneously driving it high. A special case of this + is driving the SCL and SCA lines of an I2C bus, which is by definition a + wire-OR bus. + +Both usecases require that the line be equipped with a pull-up resistor. This +resistor will make the line tend to high level unless one of the transistors on +the rail actively pulls it down. + +The level on the line will go as high as the VDD on the pull-up resistor, which +may be higher than the level supported by the transistor, achieving a +level-shift to the higher VDD. + +Integrated electronics often have an output driver stage in the form of a CMOS +"totem-pole" with one N-MOS and one P-MOS transistor where one of them drives +the line high and one of them drives the line low. This is called a push-pull +output. The "totem-pole" looks like so:: + + VDD + | + OD ||--+ + +--/ ---o|| P-MOS-FET + | ||--+ + IN --+ +----- out + | ||--+ + +--/ ----|| N-MOS-FET + OS ||--+ + | + GND + +The desired output signal (e.g. coming directly from some GPIO output register) +arrives at IN. The switches named "OD" and "OS" are normally closed, creating +a push-pull circuit. + +Consider the little "switches" named "OD" and "OS" that enable/disable the +P-MOS or N-MOS transistor right after the split of the input. As you can see, +either transistor will go totally numb if this switch is open. The totem-pole +is then halved and give high impedance instead of actively driving the line +high or low respectively. That is usually how software-controlled open +drain/source works. + +Some GPIO hardware come in open drain / open source configuration. Some are +hard-wired lines that will only support open drain or open source no matter +what: there is only one transistor there. Some are software-configurable: +by flipping a bit in a register the output can be configured as open drain +or open source, in practice by flicking open the switches labeled "OD" and "OS" +in the drawing above. + +By disabling the P-MOS transistor, the output can be driven between GND and +high impedance (open drain), and by disabling the N-MOS transistor, the output +can be driven between VDD and high impedance (open source). In the first case, +a pull-up resistor is needed on the outgoing rail to complete the circuit, and +in the second case, a pull-down resistor is needed on the rail. + +Hardware that supports open drain or open source or both, can implement a +special callback in the gpio_chip: .set_config() that takes a generic +pinconf packed value telling whether to configure the line as open drain, +open source or push-pull. This will happen in response to the +GPIO_OPEN_DRAIN or GPIO_OPEN_SOURCE flag set in the machine file, or coming +from other hardware descriptions. + +If this state can not be configured in hardware, i.e. if the GPIO hardware does +not support open drain/open source in hardware, the GPIO library will instead +use a trick: when a line is set as output, if the line is flagged as open +drain, and the IN output value is low, it will be driven low as usual. But +if the IN output value is set to high, it will instead *NOT* be driven high, +instead it will be switched to input, as input mode is high impedance, thus +achieveing an "open drain emulation" of sorts: electrically the behaviour will +be identical, with the exception of possible hardware glitches when switching +the mode of the line. + +For open source configuration the same principle is used, just that instead +of actively driving the line low, it is set to input. + + +GPIO drivers providing IRQs +--------------------------- +It is custom that GPIO drivers (GPIO chips) are also providing interrupts, +most often cascaded off a parent interrupt controller, and in some special +cases the GPIO logic is melded with a SoC's primary interrupt controller. + +The IRQ portions of the GPIO block are implemented using an irqchip, using +the header <linux/irq.h>. So basically such a driver is utilizing two sub- +systems simultaneously: gpio and irq. + +RT_FULL: a realtime compliant GPIO driver should not use spinlock_t or any +sleepable APIs (like PM runtime) as part of its irq_chip implementation. + +* spinlock_t should be replaced with raw_spinlock_t [1]. +* If sleepable APIs have to be used, these can be done from the .irq_bus_lock() + and .irq_bus_unlock() callbacks, as these are the only slowpath callbacks + on an irqchip. Create the callbacks if needed [2]. + +GPIO irqchips usually fall in one of two categories: + +* CHAINED GPIO irqchips: these are usually the type that is embedded on + an SoC. This means that there is a fast IRQ flow handler for the GPIOs that + gets called in a chain from the parent IRQ handler, most typically the + system interrupt controller. This means that the GPIO irqchip handler will + be called immediately from the parent irqchip, while holding the IRQs + disabled. The GPIO irqchip will then end up calling something like this + sequence in its interrupt handler:: + + static irqreturn_t foo_gpio_irq(int irq, void *data) + chained_irq_enter(...); + generic_handle_irq(...); + chained_irq_exit(...); + + Chained GPIO irqchips typically can NOT set the .can_sleep flag on + struct gpio_chip, as everything happens directly in the callbacks: no + slow bus traffic like I2C can be used. + + RT_FULL: Note, chained IRQ handlers will not be forced threaded on -RT. + As result, spinlock_t or any sleepable APIs (like PM runtime) can't be used + in chained IRQ handler. + If required (and if it can't be converted to the nested threaded GPIO irqchip) + a chained IRQ handler can be converted to generic irq handler and this way + it will be a threaded IRQ handler on -RT and a hard IRQ handler on non-RT + (for example, see [3]). + Know W/A: The generic_handle_irq() is expected to be called with IRQ disabled, + so the IRQ core will complain if it is called from an IRQ handler which is + forced to a thread. The "fake?" raw lock can be used to W/A this problem:: + + raw_spinlock_t wa_lock; + static irqreturn_t omap_gpio_irq_handler(int irq, void *gpiobank) + unsigned long wa_lock_flags; + raw_spin_lock_irqsave(&bank->wa_lock, wa_lock_flags); + generic_handle_irq(irq_find_mapping(bank->chip.irq.domain, bit)); + raw_spin_unlock_irqrestore(&bank->wa_lock, wa_lock_flags); + +* GENERIC CHAINED GPIO irqchips: these are the same as "CHAINED GPIO irqchips", + but chained IRQ handlers are not used. Instead GPIO IRQs dispatching is + performed by generic IRQ handler which is configured using request_irq(). + The GPIO irqchip will then end up calling something like this sequence in + its interrupt handler:: + + static irqreturn_t gpio_rcar_irq_handler(int irq, void *dev_id) + for each detected GPIO IRQ + generic_handle_irq(...); + + RT_FULL: Such kind of handlers will be forced threaded on -RT, as result IRQ + core will complain that generic_handle_irq() is called with IRQ enabled and + the same W/A as for "CHAINED GPIO irqchips" can be applied. + +* NESTED THREADED GPIO irqchips: these are off-chip GPIO expanders and any + other GPIO irqchip residing on the other side of a sleeping bus. Of course + such drivers that need slow bus traffic to read out IRQ status and similar, + traffic which may in turn incur other IRQs to happen, cannot be handled + in a quick IRQ handler with IRQs disabled. Instead they need to spawn a + thread and then mask the parent IRQ line until the interrupt is handled + by the driver. The hallmark of this driver is to call something like + this in its interrupt handler:: + + static irqreturn_t foo_gpio_irq(int irq, void *data) + ... + handle_nested_irq(irq); + + The hallmark of threaded GPIO irqchips is that they set the .can_sleep + flag on struct gpio_chip to true, indicating that this chip may sleep + when accessing the GPIOs. + +To help out in handling the set-up and management of GPIO irqchips and the +associated irqdomain and resource allocation callbacks, the gpiolib has +some helpers that can be enabled by selecting the GPIOLIB_IRQCHIP Kconfig +symbol: + +* gpiochip_irqchip_add(): adds a chained irqchip to a gpiochip. It will pass + the struct gpio_chip* for the chip to all IRQ callbacks, so the callbacks + need to embed the gpio_chip in its state container and obtain a pointer + to the container using container_of(). + (See Documentation/driver-model/design-patterns.txt) + +* gpiochip_irqchip_add_nested(): adds a nested irqchip to a gpiochip. + Apart from that it works exactly like the chained irqchip. + +* gpiochip_set_chained_irqchip(): sets up a chained irq handler for a + gpio_chip from a parent IRQ and passes the struct gpio_chip* as handler + data. (Notice handler data, since the irqchip data is likely used by the + parent irqchip!). + +* gpiochip_set_nested_irqchip(): sets up a nested irq handler for a + gpio_chip from a parent IRQ. As the parent IRQ has usually been + explicitly requested by the driver, this does very little more than + mark all the child IRQs as having the other IRQ as parent. + +If there is a need to exclude certain GPIOs from the IRQ domain, you can +set .irq.need_valid_mask of the gpiochip before gpiochip_add_data() is +called. This allocates an .irq.valid_mask with as many bits set as there +are GPIOs in the chip. Drivers can exclude GPIOs by clearing bits from this +mask. The mask must be filled in before gpiochip_irqchip_add() or +gpiochip_irqchip_add_nested() is called. + +To use the helpers please keep the following in mind: + +- Make sure to assign all relevant members of the struct gpio_chip so that + the irqchip can initialize. E.g. .dev and .can_sleep shall be set up + properly. + +- Nominally set all handlers to handle_bad_irq() in the setup call and pass + handle_bad_irq() as flow handler parameter in gpiochip_irqchip_add() if it is + expected for GPIO driver that irqchip .set_type() callback have to be called + before using/enabling GPIO IRQ. Then set the handler to handle_level_irq() + and/or handle_edge_irq() in the irqchip .set_type() callback depending on + what your controller supports. + +It is legal for any IRQ consumer to request an IRQ from any irqchip no matter +if that is a combined GPIO+IRQ driver. The basic premise is that gpio_chip and +irq_chip are orthogonal, and offering their services independent of each +other. + +gpiod_to_irq() is just a convenience function to figure out the IRQ for a +certain GPIO line and should not be relied upon to have been called before +the IRQ is used. + +So always prepare the hardware and make it ready for action in respective +callbacks from the GPIO and irqchip APIs. Do not rely on gpiod_to_irq() having +been called first. + +This orthogonality leads to ambiguities that we need to solve: if there is +competition inside the subsystem which side is using the resource (a certain +GPIO line and register for example) it needs to deny certain operations and +keep track of usage inside of the gpiolib subsystem. This is why the API +below exists. + + +Locking IRQ usage +----------------- +Input GPIOs can be used as IRQ signals. When this happens, a driver is requested +to mark the GPIO as being used as an IRQ:: + + int gpiochip_lock_as_irq(struct gpio_chip *chip, unsigned int offset) + +This will prevent the use of non-irq related GPIO APIs until the GPIO IRQ lock +is released:: + + void gpiochip_unlock_as_irq(struct gpio_chip *chip, unsigned int offset) + +When implementing an irqchip inside a GPIO driver, these two functions should +typically be called in the .startup() and .shutdown() callbacks from the +irqchip. + +When using the gpiolib irqchip helpers, these callback are automatically +assigned. + +Real-Time compliance for GPIO IRQ chips +--------------------------------------- + +Any provider of irqchips needs to be carefully tailored to support Real Time +preemption. It is desirable that all irqchips in the GPIO subsystem keep this +in mind and do the proper testing to assure they are real time-enabled. +So, pay attention on above " RT_FULL:" notes, please. +The following is a checklist to follow when preparing a driver for real +time-compliance: + +- ensure spinlock_t is not used as part irq_chip implementation; +- ensure that sleepable APIs are not used as part irq_chip implementation. + If sleepable APIs have to be used, these can be done from the .irq_bus_lock() + and .irq_bus_unlock() callbacks; +- Chained GPIO irqchips: ensure spinlock_t or any sleepable APIs are not used + from chained IRQ handler; +- Generic chained GPIO irqchips: take care about generic_handle_irq() calls and + apply corresponding W/A; +- Chained GPIO irqchips: get rid of chained IRQ handler and use generic irq + handler if possible :) +- regmap_mmio: Sry, but you are in trouble :( if MMIO regmap is used as for + GPIO IRQ chip implementation; +- Test your driver with the appropriate in-kernel real time test cases for both + level and edge IRQs. + + +Requesting self-owned GPIO pins +------------------------------- + +Sometimes it is useful to allow a GPIO chip driver to request its own GPIO +descriptors through the gpiolib API. Using gpio_request() for this purpose +does not help since it pins the module to the kernel forever (it calls +try_module_get()). A GPIO driver can use the following functions instead +to request and free descriptors without being pinned to the kernel forever:: + + struct gpio_desc *gpiochip_request_own_desc(struct gpio_desc *desc, + const char *label) + + void gpiochip_free_own_desc(struct gpio_desc *desc) + +Descriptors requested with gpiochip_request_own_desc() must be released with +gpiochip_free_own_desc(). + +These functions must be used with care since they do not affect module use +count. Do not use the functions to request gpio descriptors not owned by the +calling driver. + +* [1] http://www.spinics.net/lists/linux-omap/msg120425.html +* [2] https://lkml.org/lkml/2015/9/25/494 +* [3] https://lkml.org/lkml/2015/9/25/495 diff --git a/Documentation/driver-api/gpio/drivers-on-gpio.rst b/Documentation/driver-api/gpio/drivers-on-gpio.rst new file mode 100644 index 000000000..f3a189320 --- /dev/null +++ b/Documentation/driver-api/gpio/drivers-on-gpio.rst @@ -0,0 +1,101 @@ +============================ +Subsystem drivers using GPIO +============================ + +Note that standard kernel drivers exist for common GPIO tasks and will provide +the right in-kernel and userspace APIs/ABIs for the job, and that these +drivers can quite easily interconnect with other kernel subsystems using +hardware descriptions such as device tree or ACPI: + +- leds-gpio: drivers/leds/leds-gpio.c will handle LEDs connected to GPIO + lines, giving you the LED sysfs interface + +- ledtrig-gpio: drivers/leds/trigger/ledtrig-gpio.c will provide a LED trigger, + i.e. a LED will turn on/off in response to a GPIO line going high or low + (and that LED may in turn use the leds-gpio as per above). + +- gpio-keys: drivers/input/keyboard/gpio_keys.c is used when your GPIO line + can generate interrupts in response to a key press. Also supports debounce. + +- gpio-keys-polled: drivers/input/keyboard/gpio_keys_polled.c is used when your + GPIO line cannot generate interrupts, so it needs to be periodically polled + by a timer. + +- gpio_mouse: drivers/input/mouse/gpio_mouse.c is used to provide a mouse with + up to three buttons by simply using GPIOs and no mouse port. You can cut the + mouse cable and connect the wires to GPIO lines or solder a mouse connector + to the lines for a more permanent solution of this type. + +- gpio-beeper: drivers/input/misc/gpio-beeper.c is used to provide a beep from + an external speaker connected to a GPIO line. + +- extcon-gpio: drivers/extcon/extcon-gpio.c is used when you need to read an + external connector status, such as a headset line for an audio driver or an + HDMI connector. It will provide a better userspace sysfs interface than GPIO. + +- restart-gpio: drivers/power/reset/gpio-restart.c is used to restart/reboot + the system by pulling a GPIO line and will register a restart handler so + userspace can issue the right system call to restart the system. + +- poweroff-gpio: drivers/power/reset/gpio-poweroff.c is used to power the + system down by pulling a GPIO line and will register a pm_power_off() + callback so that userspace can issue the right system call to power down the + system. + +- gpio-gate-clock: drivers/clk/clk-gpio.c is used to control a gated clock + (off/on) that uses a GPIO, and integrated with the clock subsystem. + +- i2c-gpio: drivers/i2c/busses/i2c-gpio.c is used to drive an I2C bus + (two wires, SDA and SCL lines) by hammering (bitbang) two GPIO lines. It will + appear as any other I2C bus to the system and makes it possible to connect + drivers for the I2C devices on the bus like any other I2C bus driver. + +- spi_gpio: drivers/spi/spi-gpio.c is used to drive an SPI bus (variable number + of wires, at least SCK and optionally MISO, MOSI and chip select lines) using + GPIO hammering (bitbang). It will appear as any other SPI bus on the system + and makes it possible to connect drivers for SPI devices on the bus like + any other SPI bus driver. For example any MMC/SD card can then be connected + to this SPI by using the mmc_spi host from the MMC/SD card subsystem. + +- w1-gpio: drivers/w1/masters/w1-gpio.c is used to drive a one-wire bus using + a GPIO line, integrating with the W1 subsystem and handling devices on + the bus like any other W1 device. + +- gpio-fan: drivers/hwmon/gpio-fan.c is used to control a fan for cooling the + system, connected to a GPIO line (and optionally a GPIO alarm line), + presenting all the right in-kernel and sysfs interfaces to make your system + not overheat. + +- gpio-regulator: drivers/regulator/gpio-regulator.c is used to control a + regulator providing a certain voltage by pulling a GPIO line, integrating + with the regulator subsystem and giving you all the right interfaces. + +- gpio-wdt: drivers/watchdog/gpio_wdt.c is used to provide a watchdog timer + that will periodically "ping" a hardware connected to a GPIO line by toggling + it from 1-to-0-to-1. If that hardware does not receive its "ping" + periodically, it will reset the system. + +- gpio-nand: drivers/mtd/nand/raw/gpio.c is used to connect a NAND flash chip + to a set of simple GPIO lines: RDY, NCE, ALE, CLE, NWP. It interacts with the + NAND flash MTD subsystem and provides chip access and partition parsing like + any other NAND driving hardware. + +- ps2-gpio: drivers/input/serio/ps2-gpio.c is used to drive a PS/2 (IBM) serio + bus, data and clock line, by bit banging two GPIO lines. It will appear as + any other serio bus to the system and makes it possible to connect drivers + for e.g. keyboards and other PS/2 protocol based devices. + +- cec-gpio: drivers/media/platform/cec-gpio/ is used to interact with a CEC + Consumer Electronics Control bus using only GPIO. It is used to communicate + with devices on the HDMI bus. + +Apart from this there are special GPIO drivers in subsystems like MMC/SD to +read card detect and write protect GPIO lines, and in the TTY serial subsystem +to emulate MCTRL (modem control) signals CTS/RTS by using two GPIO lines. The +MTD NOR flash has add-ons for extra GPIO lines too, though the address bus is +usually connected directly to the flash. + +Use those instead of talking directly to the GPIOs using sysfs; they integrate +with kernel frameworks better than your userspace code could. Needless to say, +just using the appropriate kernel drivers will simplify and speed up your +embedded hacking in particular by providing ready-made components. diff --git a/Documentation/driver-api/gpio/index.rst b/Documentation/driver-api/gpio/index.rst new file mode 100644 index 000000000..6a374ded1 --- /dev/null +++ b/Documentation/driver-api/gpio/index.rst @@ -0,0 +1,48 @@ +=================================== +General Purpose Input/Output (GPIO) +=================================== + +Contents: + +.. toctree:: + :maxdepth: 2 + + intro + driver + consumer + board + drivers-on-gpio + legacy + +Core +==== + +.. kernel-doc:: include/linux/gpio/driver.h + :internal: + +.. kernel-doc:: drivers/gpio/gpiolib.c + :export: + +ACPI support +============ + +.. kernel-doc:: drivers/gpio/gpiolib-acpi.c + :export: + +Device tree support +=================== + +.. kernel-doc:: drivers/gpio/gpiolib-of.c + :export: + +Device-managed API +================== + +.. kernel-doc:: drivers/gpio/devres.c + :export: + +sysfs helpers +============= + +.. kernel-doc:: drivers/gpio/gpiolib-sysfs.c + :export: diff --git a/Documentation/driver-api/gpio/intro.rst b/Documentation/driver-api/gpio/intro.rst new file mode 100644 index 000000000..74591489d --- /dev/null +++ b/Documentation/driver-api/gpio/intro.rst @@ -0,0 +1,124 @@ +============ +Introduction +============ + + +GPIO Interfaces +=============== + +The documents in this directory give detailed instructions on how to access +GPIOs in drivers, and how to write a driver for a device that provides GPIOs +itself. + +Due to the history of GPIO interfaces in the kernel, there are two different +ways to obtain and use GPIOs: + + - The descriptor-based interface is the preferred way to manipulate GPIOs, + and is described by all the files in this directory excepted gpio-legacy.txt. + - The legacy integer-based interface which is considered deprecated (but still + usable for compatibility reasons) is documented in gpio-legacy.txt. + +The remainder of this document applies to the new descriptor-based interface. +gpio-legacy.txt contains the same information applied to the legacy +integer-based interface. + + +What is a GPIO? +=============== + +A "General Purpose Input/Output" (GPIO) is a flexible software-controlled +digital signal. They are provided from many kinds of chip, and are familiar +to Linux developers working with embedded and custom hardware. Each GPIO +represents a bit connected to a particular pin, or "ball" on Ball Grid Array +(BGA) packages. Board schematics show which external hardware connects to +which GPIOs. Drivers can be written generically, so that board setup code +passes such pin configuration data to drivers. + +System-on-Chip (SOC) processors heavily rely on GPIOs. In some cases, every +non-dedicated pin can be configured as a GPIO; and most chips have at least +several dozen of them. Programmable logic devices (like FPGAs) can easily +provide GPIOs; multifunction chips like power managers, and audio codecs +often have a few such pins to help with pin scarcity on SOCs; and there are +also "GPIO Expander" chips that connect using the I2C or SPI serial buses. +Most PC southbridges have a few dozen GPIO-capable pins (with only the BIOS +firmware knowing how they're used). + +The exact capabilities of GPIOs vary between systems. Common options: + + - Output values are writable (high=1, low=0). Some chips also have + options about how that value is driven, so that for example only one + value might be driven, supporting "wire-OR" and similar schemes for the + other value (notably, "open drain" signaling). + + - Input values are likewise readable (1, 0). Some chips support readback + of pins configured as "output", which is very useful in such "wire-OR" + cases (to support bidirectional signaling). GPIO controllers may have + input de-glitch/debounce logic, sometimes with software controls. + + - Inputs can often be used as IRQ signals, often edge triggered but + sometimes level triggered. Such IRQs may be configurable as system + wakeup events, to wake the system from a low power state. + + - Usually a GPIO will be configurable as either input or output, as needed + by different product boards; single direction ones exist too. + + - Most GPIOs can be accessed while holding spinlocks, but those accessed + through a serial bus normally can't. Some systems support both types. + +On a given board each GPIO is used for one specific purpose like monitoring +MMC/SD card insertion/removal, detecting card write-protect status, driving +a LED, configuring a transceiver, bit-banging a serial bus, poking a hardware +watchdog, sensing a switch, and so on. + + +Common GPIO Properties +====================== + +These properties are met through all the other documents of the GPIO interface +and it is useful to understand them, especially if you need to define GPIO +mappings. + +Active-High and Active-Low +-------------------------- +It is natural to assume that a GPIO is "active" when its output signal is 1 +("high"), and inactive when it is 0 ("low"). However in practice the signal of a +GPIO may be inverted before is reaches its destination, or a device could decide +to have different conventions about what "active" means. Such decisions should +be transparent to device drivers, therefore it is possible to define a GPIO as +being either active-high ("1" means "active", the default) or active-low ("0" +means "active") so that drivers only need to worry about the logical signal and +not about what happens at the line level. + +Open Drain and Open Source +-------------------------- +Sometimes shared signals need to use "open drain" (where only the low signal +level is actually driven), or "open source" (where only the high signal level is +driven) signaling. That term applies to CMOS transistors; "open collector" is +used for TTL. A pullup or pulldown resistor causes the high or low signal level. +This is sometimes called a "wire-AND"; or more practically, from the negative +logic (low=true) perspective this is a "wire-OR". + +One common example of an open drain signal is a shared active-low IRQ line. +Also, bidirectional data bus signals sometimes use open drain signals. + +Some GPIO controllers directly support open drain and open source outputs; many +don't. When you need open drain signaling but your hardware doesn't directly +support it, there's a common idiom you can use to emulate it with any GPIO pin +that can be used as either an input or an output: + + LOW: gpiod_direction_output(gpio, 0) ... this drives the signal and overrides + the pullup. + + HIGH: gpiod_direction_input(gpio) ... this turns off the output, so the pullup + (or some other device) controls the signal. + +The same logic can be applied to emulate open source signaling, by driving the +high signal and configuring the GPIO as input for low. This open drain/open +source emulation can be handled transparently by the GPIO framework. + +If you are "driving" the signal high but gpiod_get_value(gpio) reports a low +value (after the appropriate rise time passes), you know some other component is +driving the shared signal low. That's not necessarily an error. As one common +example, that's how I2C clocks are stretched: a slave that needs a slower clock +delays the rising edge of SCK, and the I2C master adjusts its signaling rate +accordingly. diff --git a/Documentation/driver-api/gpio/legacy.rst b/Documentation/driver-api/gpio/legacy.rst new file mode 100644 index 000000000..5e9421e05 --- /dev/null +++ b/Documentation/driver-api/gpio/legacy.rst @@ -0,0 +1,770 @@ +====================== +Legacy GPIO Interfaces +====================== + +This provides an overview of GPIO access conventions on Linux. + +These calls use the gpio_* naming prefix. No other calls should use that +prefix, or the related __gpio_* prefix. + + +What is a GPIO? +=============== +A "General Purpose Input/Output" (GPIO) is a flexible software-controlled +digital signal. They are provided from many kinds of chip, and are familiar +to Linux developers working with embedded and custom hardware. Each GPIO +represents a bit connected to a particular pin, or "ball" on Ball Grid Array +(BGA) packages. Board schematics show which external hardware connects to +which GPIOs. Drivers can be written generically, so that board setup code +passes such pin configuration data to drivers. + +System-on-Chip (SOC) processors heavily rely on GPIOs. In some cases, every +non-dedicated pin can be configured as a GPIO; and most chips have at least +several dozen of them. Programmable logic devices (like FPGAs) can easily +provide GPIOs; multifunction chips like power managers, and audio codecs +often have a few such pins to help with pin scarcity on SOCs; and there are +also "GPIO Expander" chips that connect using the I2C or SPI serial busses. +Most PC southbridges have a few dozen GPIO-capable pins (with only the BIOS +firmware knowing how they're used). + +The exact capabilities of GPIOs vary between systems. Common options: + + - Output values are writable (high=1, low=0). Some chips also have + options about how that value is driven, so that for example only one + value might be driven ... supporting "wire-OR" and similar schemes + for the other value (notably, "open drain" signaling). + + - Input values are likewise readable (1, 0). Some chips support readback + of pins configured as "output", which is very useful in such "wire-OR" + cases (to support bidirectional signaling). GPIO controllers may have + input de-glitch/debounce logic, sometimes with software controls. + + - Inputs can often be used as IRQ signals, often edge triggered but + sometimes level triggered. Such IRQs may be configurable as system + wakeup events, to wake the system from a low power state. + + - Usually a GPIO will be configurable as either input or output, as needed + by different product boards; single direction ones exist too. + + - Most GPIOs can be accessed while holding spinlocks, but those accessed + through a serial bus normally can't. Some systems support both types. + +On a given board each GPIO is used for one specific purpose like monitoring +MMC/SD card insertion/removal, detecting card writeprotect status, driving +a LED, configuring a transceiver, bitbanging a serial bus, poking a hardware +watchdog, sensing a switch, and so on. + + +GPIO conventions +================ +Note that this is called a "convention" because you don't need to do it this +way, and it's no crime if you don't. There **are** cases where portability +is not the main issue; GPIOs are often used for the kind of board-specific +glue logic that may even change between board revisions, and can't ever be +used on a board that's wired differently. Only least-common-denominator +functionality can be very portable. Other features are platform-specific, +and that can be critical for glue logic. + +Plus, this doesn't require any implementation framework, just an interface. +One platform might implement it as simple inline functions accessing chip +registers; another might implement it by delegating through abstractions +used for several very different kinds of GPIO controller. (There is some +optional code supporting such an implementation strategy, described later +in this document, but drivers acting as clients to the GPIO interface must +not care how it's implemented.) + +That said, if the convention is supported on their platform, drivers should +use it when possible. Platforms must select GPIOLIB if GPIO functionality +is strictly required. Drivers that can't work without +standard GPIO calls should have Kconfig entries which depend on GPIOLIB. The +GPIO calls are available, either as "real code" or as optimized-away stubs, +when drivers use the include file: + + #include <linux/gpio.h> + +If you stick to this convention then it'll be easier for other developers to +see what your code is doing, and help maintain it. + +Note that these operations include I/O barriers on platforms which need to +use them; drivers don't need to add them explicitly. + + +Identifying GPIOs +----------------- +GPIOs are identified by unsigned integers in the range 0..MAX_INT. That +reserves "negative" numbers for other purposes like marking signals as +"not available on this board", or indicating faults. Code that doesn't +touch the underlying hardware treats these integers as opaque cookies. + +Platforms define how they use those integers, and usually #define symbols +for the GPIO lines so that board-specific setup code directly corresponds +to the relevant schematics. In contrast, drivers should only use GPIO +numbers passed to them from that setup code, using platform_data to hold +board-specific pin configuration data (along with other board specific +data they need). That avoids portability problems. + +So for example one platform uses numbers 32-159 for GPIOs; while another +uses numbers 0..63 with one set of GPIO controllers, 64-79 with another +type of GPIO controller, and on one particular board 80-95 with an FPGA. +The numbers need not be contiguous; either of those platforms could also +use numbers 2000-2063 to identify GPIOs in a bank of I2C GPIO expanders. + +If you want to initialize a structure with an invalid GPIO number, use +some negative number (perhaps "-EINVAL"); that will never be valid. To +test if such number from such a structure could reference a GPIO, you +may use this predicate: + + int gpio_is_valid(int number); + +A number that's not valid will be rejected by calls which may request +or free GPIOs (see below). Other numbers may also be rejected; for +example, a number might be valid but temporarily unused on a given board. + +Whether a platform supports multiple GPIO controllers is a platform-specific +implementation issue, as are whether that support can leave "holes" in the space +of GPIO numbers, and whether new controllers can be added at runtime. Such issues +can affect things including whether adjacent GPIO numbers are both valid. + +Using GPIOs +----------- +The first thing a system should do with a GPIO is allocate it, using +the gpio_request() call; see later. + +One of the next things to do with a GPIO, often in board setup code when +setting up a platform_device using the GPIO, is mark its direction:: + + /* set as input or output, returning 0 or negative errno */ + int gpio_direction_input(unsigned gpio); + int gpio_direction_output(unsigned gpio, int value); + +The return value is zero for success, else a negative errno. It should +be checked, since the get/set calls don't have error returns and since +misconfiguration is possible. You should normally issue these calls from +a task context. However, for spinlock-safe GPIOs it's OK to use them +before tasking is enabled, as part of early board setup. + +For output GPIOs, the value provided becomes the initial output value. +This helps avoid signal glitching during system startup. + +For compatibility with legacy interfaces to GPIOs, setting the direction +of a GPIO implicitly requests that GPIO (see below) if it has not been +requested already. That compatibility is being removed from the optional +gpiolib framework. + +Setting the direction can fail if the GPIO number is invalid, or when +that particular GPIO can't be used in that mode. It's generally a bad +idea to rely on boot firmware to have set the direction correctly, since +it probably wasn't validated to do more than boot Linux. (Similarly, +that board setup code probably needs to multiplex that pin as a GPIO, +and configure pullups/pulldowns appropriately.) + + +Spinlock-Safe GPIO access +------------------------- +Most GPIO controllers can be accessed with memory read/write instructions. +Those don't need to sleep, and can safely be done from inside hard +(nonthreaded) IRQ handlers and similar contexts. + +Use the following calls to access such GPIOs, +for which gpio_cansleep() will always return false (see below):: + + /* GPIO INPUT: return zero or nonzero */ + int gpio_get_value(unsigned gpio); + + /* GPIO OUTPUT */ + void gpio_set_value(unsigned gpio, int value); + +The values are boolean, zero for low, nonzero for high. When reading the +value of an output pin, the value returned should be what's seen on the +pin ... that won't always match the specified output value, because of +issues including open-drain signaling and output latencies. + +The get/set calls have no error returns because "invalid GPIO" should have +been reported earlier from gpio_direction_*(). However, note that not all +platforms can read the value of output pins; those that can't should always +return zero. Also, using these calls for GPIOs that can't safely be accessed +without sleeping (see below) is an error. + +Platform-specific implementations are encouraged to optimize the two +calls to access the GPIO value in cases where the GPIO number (and for +output, value) are constant. It's normal for them to need only a couple +of instructions in such cases (reading or writing a hardware register), +and not to need spinlocks. Such optimized calls can make bitbanging +applications a lot more efficient (in both space and time) than spending +dozens of instructions on subroutine calls. + + +GPIO access that may sleep +-------------------------- +Some GPIO controllers must be accessed using message based busses like I2C +or SPI. Commands to read or write those GPIO values require waiting to +get to the head of a queue to transmit a command and get its response. +This requires sleeping, which can't be done from inside IRQ handlers. + +Platforms that support this type of GPIO distinguish them from other GPIOs +by returning nonzero from this call (which requires a valid GPIO number, +which should have been previously allocated with gpio_request):: + + int gpio_cansleep(unsigned gpio); + +To access such GPIOs, a different set of accessors is defined:: + + /* GPIO INPUT: return zero or nonzero, might sleep */ + int gpio_get_value_cansleep(unsigned gpio); + + /* GPIO OUTPUT, might sleep */ + void gpio_set_value_cansleep(unsigned gpio, int value); + + +Accessing such GPIOs requires a context which may sleep, for example +a threaded IRQ handler, and those accessors must be used instead of +spinlock-safe accessors without the cansleep() name suffix. + +Other than the fact that these accessors might sleep, and will work +on GPIOs that can't be accessed from hardIRQ handlers, these calls act +the same as the spinlock-safe calls. + +**IN ADDITION** calls to setup and configure such GPIOs must be made +from contexts which may sleep, since they may need to access the GPIO +controller chip too (These setup calls are usually made from board +setup or driver probe/teardown code, so this is an easy constraint.):: + + gpio_direction_input() + gpio_direction_output() + gpio_request() + + ## gpio_request_one() + ## gpio_request_array() + ## gpio_free_array() + + gpio_free() + gpio_set_debounce() + + + +Claiming and Releasing GPIOs +---------------------------- +To help catch system configuration errors, two calls are defined:: + + /* request GPIO, returning 0 or negative errno. + * non-null labels may be useful for diagnostics. + */ + int gpio_request(unsigned gpio, const char *label); + + /* release previously-claimed GPIO */ + void gpio_free(unsigned gpio); + +Passing invalid GPIO numbers to gpio_request() will fail, as will requesting +GPIOs that have already been claimed with that call. The return value of +gpio_request() must be checked. You should normally issue these calls from +a task context. However, for spinlock-safe GPIOs it's OK to request GPIOs +before tasking is enabled, as part of early board setup. + +These calls serve two basic purposes. One is marking the signals which +are actually in use as GPIOs, for better diagnostics; systems may have +several hundred potential GPIOs, but often only a dozen are used on any +given board. Another is to catch conflicts, identifying errors when +(a) two or more drivers wrongly think they have exclusive use of that +signal, or (b) something wrongly believes it's safe to remove drivers +needed to manage a signal that's in active use. That is, requesting a +GPIO can serve as a kind of lock. + +Some platforms may also use knowledge about what GPIOs are active for +power management, such as by powering down unused chip sectors and, more +easily, gating off unused clocks. + +For GPIOs that use pins known to the pinctrl subsystem, that subsystem should +be informed of their use; a gpiolib driver's .request() operation may call +pinctrl_gpio_request(), and a gpiolib driver's .free() operation may call +pinctrl_gpio_free(). The pinctrl subsystem allows a pinctrl_gpio_request() +to succeed concurrently with a pin or pingroup being "owned" by a device for +pin multiplexing. + +Any programming of pin multiplexing hardware that is needed to route the +GPIO signal to the appropriate pin should occur within a GPIO driver's +.direction_input() or .direction_output() operations, and occur after any +setup of an output GPIO's value. This allows a glitch-free migration from a +pin's special function to GPIO. This is sometimes required when using a GPIO +to implement a workaround on signals typically driven by a non-GPIO HW block. + +Some platforms allow some or all GPIO signals to be routed to different pins. +Similarly, other aspects of the GPIO or pin may need to be configured, such as +pullup/pulldown. Platform software should arrange that any such details are +configured prior to gpio_request() being called for those GPIOs, e.g. using +the pinctrl subsystem's mapping table, so that GPIO users need not be aware +of these details. + +Also note that it's your responsibility to have stopped using a GPIO +before you free it. + +Considering in most cases GPIOs are actually configured right after they +are claimed, three additional calls are defined:: + + /* request a single GPIO, with initial configuration specified by + * 'flags', identical to gpio_request() wrt other arguments and + * return value + */ + int gpio_request_one(unsigned gpio, unsigned long flags, const char *label); + + /* request multiple GPIOs in a single call + */ + int gpio_request_array(struct gpio *array, size_t num); + + /* release multiple GPIOs in a single call + */ + void gpio_free_array(struct gpio *array, size_t num); + +where 'flags' is currently defined to specify the following properties: + + * GPIOF_DIR_IN - to configure direction as input + * GPIOF_DIR_OUT - to configure direction as output + + * GPIOF_INIT_LOW - as output, set initial level to LOW + * GPIOF_INIT_HIGH - as output, set initial level to HIGH + * GPIOF_OPEN_DRAIN - gpio pin is open drain type. + * GPIOF_OPEN_SOURCE - gpio pin is open source type. + + * GPIOF_EXPORT_DIR_FIXED - export gpio to sysfs, keep direction + * GPIOF_EXPORT_DIR_CHANGEABLE - also export, allow changing direction + +since GPIOF_INIT_* are only valid when configured as output, so group valid +combinations as: + + * GPIOF_IN - configure as input + * GPIOF_OUT_INIT_LOW - configured as output, initial level LOW + * GPIOF_OUT_INIT_HIGH - configured as output, initial level HIGH + +When setting the flag as GPIOF_OPEN_DRAIN then it will assume that pins is +open drain type. Such pins will not be driven to 1 in output mode. It is +require to connect pull-up on such pins. By enabling this flag, gpio lib will +make the direction to input when it is asked to set value of 1 in output mode +to make the pin HIGH. The pin is make to LOW by driving value 0 in output mode. + +When setting the flag as GPIOF_OPEN_SOURCE then it will assume that pins is +open source type. Such pins will not be driven to 0 in output mode. It is +require to connect pull-down on such pin. By enabling this flag, gpio lib will +make the direction to input when it is asked to set value of 0 in output mode +to make the pin LOW. The pin is make to HIGH by driving value 1 in output mode. + +In the future, these flags can be extended to support more properties. + +Further more, to ease the claim/release of multiple GPIOs, 'struct gpio' is +introduced to encapsulate all three fields as:: + + struct gpio { + unsigned gpio; + unsigned long flags; + const char *label; + }; + +A typical example of usage:: + + static struct gpio leds_gpios[] = { + { 32, GPIOF_OUT_INIT_HIGH, "Power LED" }, /* default to ON */ + { 33, GPIOF_OUT_INIT_LOW, "Green LED" }, /* default to OFF */ + { 34, GPIOF_OUT_INIT_LOW, "Red LED" }, /* default to OFF */ + { 35, GPIOF_OUT_INIT_LOW, "Blue LED" }, /* default to OFF */ + { ... }, + }; + + err = gpio_request_one(31, GPIOF_IN, "Reset Button"); + if (err) + ... + + err = gpio_request_array(leds_gpios, ARRAY_SIZE(leds_gpios)); + if (err) + ... + + gpio_free_array(leds_gpios, ARRAY_SIZE(leds_gpios)); + + +GPIOs mapped to IRQs +-------------------- +GPIO numbers are unsigned integers; so are IRQ numbers. These make up +two logically distinct namespaces (GPIO 0 need not use IRQ 0). You can +map between them using calls like:: + + /* map GPIO numbers to IRQ numbers */ + int gpio_to_irq(unsigned gpio); + + /* map IRQ numbers to GPIO numbers (avoid using this) */ + int irq_to_gpio(unsigned irq); + +Those return either the corresponding number in the other namespace, or +else a negative errno code if the mapping can't be done. (For example, +some GPIOs can't be used as IRQs.) It is an unchecked error to use a GPIO +number that wasn't set up as an input using gpio_direction_input(), or +to use an IRQ number that didn't originally come from gpio_to_irq(). + +These two mapping calls are expected to cost on the order of a single +addition or subtraction. They're not allowed to sleep. + +Non-error values returned from gpio_to_irq() can be passed to request_irq() +or free_irq(). They will often be stored into IRQ resources for platform +devices, by the board-specific initialization code. Note that IRQ trigger +options are part of the IRQ interface, e.g. IRQF_TRIGGER_FALLING, as are +system wakeup capabilities. + +Non-error values returned from irq_to_gpio() would most commonly be used +with gpio_get_value(), for example to initialize or update driver state +when the IRQ is edge-triggered. Note that some platforms don't support +this reverse mapping, so you should avoid using it. + + +Emulating Open Drain Signals +---------------------------- +Sometimes shared signals need to use "open drain" signaling, where only the +low signal level is actually driven. (That term applies to CMOS transistors; +"open collector" is used for TTL.) A pullup resistor causes the high signal +level. This is sometimes called a "wire-AND"; or more practically, from the +negative logic (low=true) perspective this is a "wire-OR". + +One common example of an open drain signal is a shared active-low IRQ line. +Also, bidirectional data bus signals sometimes use open drain signals. + +Some GPIO controllers directly support open drain outputs; many don't. When +you need open drain signaling but your hardware doesn't directly support it, +there's a common idiom you can use to emulate it with any GPIO pin that can +be used as either an input or an output: + + LOW: gpio_direction_output(gpio, 0) ... this drives the signal + and overrides the pullup. + + HIGH: gpio_direction_input(gpio) ... this turns off the output, + so the pullup (or some other device) controls the signal. + +If you are "driving" the signal high but gpio_get_value(gpio) reports a low +value (after the appropriate rise time passes), you know some other component +is driving the shared signal low. That's not necessarily an error. As one +common example, that's how I2C clocks are stretched: a slave that needs a +slower clock delays the rising edge of SCK, and the I2C master adjusts its +signaling rate accordingly. + + +GPIO controllers and the pinctrl subsystem +------------------------------------------ + +A GPIO controller on a SOC might be tightly coupled with the pinctrl +subsystem, in the sense that the pins can be used by other functions +together with an optional gpio feature. We have already covered the +case where e.g. a GPIO controller need to reserve a pin or set the +direction of a pin by calling any of:: + + pinctrl_gpio_request() + pinctrl_gpio_free() + pinctrl_gpio_direction_input() + pinctrl_gpio_direction_output() + +But how does the pin control subsystem cross-correlate the GPIO +numbers (which are a global business) to a certain pin on a certain +pin controller? + +This is done by registering "ranges" of pins, which are essentially +cross-reference tables. These are described in +Documentation/driver-api/pinctl.rst + +While the pin allocation is totally managed by the pinctrl subsystem, +gpio (under gpiolib) is still maintained by gpio drivers. It may happen +that different pin ranges in a SoC is managed by different gpio drivers. + +This makes it logical to let gpio drivers announce their pin ranges to +the pin ctrl subsystem before it will call 'pinctrl_gpio_request' in order +to request the corresponding pin to be prepared by the pinctrl subsystem +before any gpio usage. + +For this, the gpio controller can register its pin range with pinctrl +subsystem. There are two ways of doing it currently: with or without DT. + +For with DT support refer to Documentation/devicetree/bindings/gpio/gpio.txt. + +For non-DT support, user can call gpiochip_add_pin_range() with appropriate +parameters to register a range of gpio pins with a pinctrl driver. For this +exact name string of pinctrl device has to be passed as one of the +argument to this routine. + + +What do these conventions omit? +=============================== +One of the biggest things these conventions omit is pin multiplexing, since +this is highly chip-specific and nonportable. One platform might not need +explicit multiplexing; another might have just two options for use of any +given pin; another might have eight options per pin; another might be able +to route a given GPIO to any one of several pins. (Yes, those examples all +come from systems that run Linux today.) + +Related to multiplexing is configuration and enabling of the pullups or +pulldowns integrated on some platforms. Not all platforms support them, +or support them in the same way; and any given board might use external +pullups (or pulldowns) so that the on-chip ones should not be used. +(When a circuit needs 5 kOhm, on-chip 100 kOhm resistors won't do.) +Likewise drive strength (2 mA vs 20 mA) and voltage (1.8V vs 3.3V) is a +platform-specific issue, as are models like (not) having a one-to-one +correspondence between configurable pins and GPIOs. + +There are other system-specific mechanisms that are not specified here, +like the aforementioned options for input de-glitching and wire-OR output. +Hardware may support reading or writing GPIOs in gangs, but that's usually +configuration dependent: for GPIOs sharing the same bank. (GPIOs are +commonly grouped in banks of 16 or 32, with a given SOC having several such +banks.) Some systems can trigger IRQs from output GPIOs, or read values +from pins not managed as GPIOs. Code relying on such mechanisms will +necessarily be nonportable. + +Dynamic definition of GPIOs is not currently standard; for example, as +a side effect of configuring an add-on board with some GPIO expanders. + + +GPIO implementor's framework (OPTIONAL) +======================================= +As noted earlier, there is an optional implementation framework making it +easier for platforms to support different kinds of GPIO controller using +the same programming interface. This framework is called "gpiolib". + +As a debugging aid, if debugfs is available a /sys/kernel/debug/gpio file +will be found there. That will list all the controllers registered through +this framework, and the state of the GPIOs currently in use. + + +Controller Drivers: gpio_chip +----------------------------- +In this framework each GPIO controller is packaged as a "struct gpio_chip" +with information common to each controller of that type: + + - methods to establish GPIO direction + - methods used to access GPIO values + - flag saying whether calls to its methods may sleep + - optional debugfs dump method (showing extra state like pullup config) + - label for diagnostics + +There is also per-instance data, which may come from device.platform_data: +the number of its first GPIO, and how many GPIOs it exposes. + +The code implementing a gpio_chip should support multiple instances of the +controller, possibly using the driver model. That code will configure each +gpio_chip and issue gpiochip_add(). Removing a GPIO controller should be +rare; use gpiochip_remove() when it is unavoidable. + +Most often a gpio_chip is part of an instance-specific structure with state +not exposed by the GPIO interfaces, such as addressing, power management, +and more. Chips such as codecs will have complex non-GPIO state. + +Any debugfs dump method should normally ignore signals which haven't been +requested as GPIOs. They can use gpiochip_is_requested(), which returns +either NULL or the label associated with that GPIO when it was requested. + + +Platform Support +---------------- +To force-enable this framework, a platform's Kconfig will "select" GPIOLIB, +else it is up to the user to configure support for GPIO. + +It may also provide a custom value for ARCH_NR_GPIOS, so that it better +reflects the number of GPIOs in actual use on that platform, without +wasting static table space. (It should count both built-in/SoC GPIOs and +also ones on GPIO expanders. + +If neither of these options are selected, the platform does not support +GPIOs through GPIO-lib and the code cannot be enabled by the user. + +Trivial implementations of those functions can directly use framework +code, which always dispatches through the gpio_chip:: + + #define gpio_get_value __gpio_get_value + #define gpio_set_value __gpio_set_value + #define gpio_cansleep __gpio_cansleep + +Fancier implementations could instead define those as inline functions with +logic optimizing access to specific SOC-based GPIOs. For example, if the +referenced GPIO is the constant "12", getting or setting its value could +cost as little as two or three instructions, never sleeping. When such an +optimization is not possible those calls must delegate to the framework +code, costing at least a few dozen instructions. For bitbanged I/O, such +instruction savings can be significant. + +For SOCs, platform-specific code defines and registers gpio_chip instances +for each bank of on-chip GPIOs. Those GPIOs should be numbered/labeled to +match chip vendor documentation, and directly match board schematics. They +may well start at zero and go up to a platform-specific limit. Such GPIOs +are normally integrated into platform initialization to make them always be +available, from arch_initcall() or earlier; they can often serve as IRQs. + + +Board Support +------------- +For external GPIO controllers -- such as I2C or SPI expanders, ASICs, multi +function devices, FPGAs or CPLDs -- most often board-specific code handles +registering controller devices and ensures that their drivers know what GPIO +numbers to use with gpiochip_add(). Their numbers often start right after +platform-specific GPIOs. + +For example, board setup code could create structures identifying the range +of GPIOs that chip will expose, and passes them to each GPIO expander chip +using platform_data. Then the chip driver's probe() routine could pass that +data to gpiochip_add(). + +Initialization order can be important. For example, when a device relies on +an I2C-based GPIO, its probe() routine should only be called after that GPIO +becomes available. That may mean the device should not be registered until +calls for that GPIO can work. One way to address such dependencies is for +such gpio_chip controllers to provide setup() and teardown() callbacks to +board specific code; those board specific callbacks would register devices +once all the necessary resources are available, and remove them later when +the GPIO controller device becomes unavailable. + + +Sysfs Interface for Userspace (OPTIONAL) +======================================== +Platforms which use the "gpiolib" implementors framework may choose to +configure a sysfs user interface to GPIOs. This is different from the +debugfs interface, since it provides control over GPIO direction and +value instead of just showing a gpio state summary. Plus, it could be +present on production systems without debugging support. + +Given appropriate hardware documentation for the system, userspace could +know for example that GPIO #23 controls the write protect line used to +protect boot loader segments in flash memory. System upgrade procedures +may need to temporarily remove that protection, first importing a GPIO, +then changing its output state, then updating the code before re-enabling +the write protection. In normal use, GPIO #23 would never be touched, +and the kernel would have no need to know about it. + +Again depending on appropriate hardware documentation, on some systems +userspace GPIO can be used to determine system configuration data that +standard kernels won't know about. And for some tasks, simple userspace +GPIO drivers could be all that the system really needs. + +Note that standard kernel drivers exist for common "LEDs and Buttons" +GPIO tasks: "leds-gpio" and "gpio_keys", respectively. Use those +instead of talking directly to the GPIOs; they integrate with kernel +frameworks better than your userspace code could. + + +Paths in Sysfs +-------------- +There are three kinds of entry in /sys/class/gpio: + + - Control interfaces used to get userspace control over GPIOs; + + - GPIOs themselves; and + + - GPIO controllers ("gpio_chip" instances). + +That's in addition to standard files including the "device" symlink. + +The control interfaces are write-only: + + /sys/class/gpio/ + + "export" ... Userspace may ask the kernel to export control of + a GPIO to userspace by writing its number to this file. + + Example: "echo 19 > export" will create a "gpio19" node + for GPIO #19, if that's not requested by kernel code. + + "unexport" ... Reverses the effect of exporting to userspace. + + Example: "echo 19 > unexport" will remove a "gpio19" + node exported using the "export" file. + +GPIO signals have paths like /sys/class/gpio/gpio42/ (for GPIO #42) +and have the following read/write attributes: + + /sys/class/gpio/gpioN/ + + "direction" ... reads as either "in" or "out". This value may + normally be written. Writing as "out" defaults to + initializing the value as low. To ensure glitch free + operation, values "low" and "high" may be written to + configure the GPIO as an output with that initial value. + + Note that this attribute *will not exist* if the kernel + doesn't support changing the direction of a GPIO, or + it was exported by kernel code that didn't explicitly + allow userspace to reconfigure this GPIO's direction. + + "value" ... reads as either 0 (low) or 1 (high). If the GPIO + is configured as an output, this value may be written; + any nonzero value is treated as high. + + If the pin can be configured as interrupt-generating interrupt + and if it has been configured to generate interrupts (see the + description of "edge"), you can poll(2) on that file and + poll(2) will return whenever the interrupt was triggered. If + you use poll(2), set the events POLLPRI and POLLERR. If you + use select(2), set the file descriptor in exceptfds. After + poll(2) returns, either lseek(2) to the beginning of the sysfs + file and read the new value or close the file and re-open it + to read the value. + + "edge" ... reads as either "none", "rising", "falling", or + "both". Write these strings to select the signal edge(s) + that will make poll(2) on the "value" file return. + + This file exists only if the pin can be configured as an + interrupt generating input pin. + + "active_low" ... reads as either 0 (false) or 1 (true). Write + any nonzero value to invert the value attribute both + for reading and writing. Existing and subsequent + poll(2) support configuration via the edge attribute + for "rising" and "falling" edges will follow this + setting. + +GPIO controllers have paths like /sys/class/gpio/gpiochip42/ (for the +controller implementing GPIOs starting at #42) and have the following +read-only attributes: + + /sys/class/gpio/gpiochipN/ + + "base" ... same as N, the first GPIO managed by this chip + + "label" ... provided for diagnostics (not always unique) + + "ngpio" ... how many GPIOs this manges (N to N + ngpio - 1) + +Board documentation should in most cases cover what GPIOs are used for +what purposes. However, those numbers are not always stable; GPIOs on +a daughtercard might be different depending on the base board being used, +or other cards in the stack. In such cases, you may need to use the +gpiochip nodes (possibly in conjunction with schematics) to determine +the correct GPIO number to use for a given signal. + + +Exporting from Kernel code +-------------------------- +Kernel code can explicitly manage exports of GPIOs which have already been +requested using gpio_request():: + + /* export the GPIO to userspace */ + int gpio_export(unsigned gpio, bool direction_may_change); + + /* reverse gpio_export() */ + void gpio_unexport(); + + /* create a sysfs link to an exported GPIO node */ + int gpio_export_link(struct device *dev, const char *name, + unsigned gpio) + +After a kernel driver requests a GPIO, it may only be made available in +the sysfs interface by gpio_export(). The driver can control whether the +signal direction may change. This helps drivers prevent userspace code +from accidentally clobbering important system state. + +This explicit exporting can help with debugging (by making some kinds +of experiments easier), or can provide an always-there interface that's +suitable for documenting as part of a board support package. + +After the GPIO has been exported, gpio_export_link() allows creating +symlinks from elsewhere in sysfs to the GPIO sysfs node. Drivers can +use this to provide the interface under their own device in sysfs with +a descriptive name. + + +API Reference +============= + +The functions listed in this section are deprecated. The GPIO descriptor based +API should be used in new code. + +.. kernel-doc:: drivers/gpio/gpiolib-legacy.c + :export: diff --git a/Documentation/driver-api/hsi.rst b/Documentation/driver-api/hsi.rst new file mode 100644 index 000000000..f9cec02b7 --- /dev/null +++ b/Documentation/driver-api/hsi.rst @@ -0,0 +1,88 @@ +High Speed Synchronous Serial Interface (HSI) +============================================= + +Introduction +--------------- + +High Speed Syncronous Interface (HSI) is a fullduplex, low latency protocol, +that is optimized for die-level interconnect between an Application Processor +and a Baseband chipset. It has been specified by the MIPI alliance in 2003 and +implemented by multiple vendors since then. + +The HSI interface supports full duplex communication over multiple channels +(typically 8) and is capable of reaching speeds up to 200 Mbit/s. + +The serial protocol uses two signals, DATA and FLAG as combined data and clock +signals and an additional READY signal for flow control. An additional WAKE +signal can be used to wakeup the chips from standby modes. The signals are +commonly prefixed by AC for signals going from the application die to the +cellular die and CA for signals going the other way around. + +:: + + +------------+ +---------------+ + | Cellular | | Application | + | Die | | Die | + | | - - - - - - CAWAKE - - - - - - >| | + | T|------------ CADATA ------------>|R | + | X|------------ CAFLAG ------------>|X | + | |<----------- ACREADY ------------| | + | | | | + | | | | + | |< - - - - - ACWAKE - - - - - - -| | + | R|<----------- ACDATA -------------|T | + | X|<----------- ACFLAG -------------|X | + | |------------ CAREADY ----------->| | + | | | | + | | | | + +------------+ +---------------+ + +HSI Subsystem in Linux +------------------------- + +In the Linux kernel the hsi subsystem is supposed to be used for HSI devices. +The hsi subsystem contains drivers for hsi controllers including support for +multi-port controllers and provides a generic API for using the HSI ports. + +It also contains HSI client drivers, which make use of the generic API to +implement a protocol used on the HSI interface. These client drivers can +use an arbitrary number of channels. + +hsi-char Device +------------------ + +Each port automatically registers a generic client driver called hsi_char, +which provides a charecter device for userspace representing the HSI port. +It can be used to communicate via HSI from userspace. Userspace may +configure the hsi_char device using the following ioctl commands: + +HSC_RESET + flush the HSI port + +HSC_SET_PM + enable or disable the client. + +HSC_SEND_BREAK + send break + +HSC_SET_RX + set RX configuration + +HSC_GET_RX + get RX configuration + +HSC_SET_TX + set TX configuration + +HSC_GET_TX + get TX configuration + +The kernel HSI API +------------------ + +.. kernel-doc:: include/linux/hsi/hsi.h + :internal: + +.. kernel-doc:: drivers/hsi/hsi_core.c + :export: + diff --git a/Documentation/driver-api/i2c.rst b/Documentation/driver-api/i2c.rst new file mode 100644 index 000000000..7582c079d --- /dev/null +++ b/Documentation/driver-api/i2c.rst @@ -0,0 +1,48 @@ +I\ :sup:`2`\ C and SMBus Subsystem +================================== + +I\ :sup:`2`\ C (or without fancy typography, "I2C") is an acronym for +the "Inter-IC" bus, a simple bus protocol which is widely used where low +data rate communications suffice. Since it's also a licensed trademark, +some vendors use another name (such as "Two-Wire Interface", TWI) for +the same bus. I2C only needs two signals (SCL for clock, SDA for data), +conserving board real estate and minimizing signal quality issues. Most +I2C devices use seven bit addresses, and bus speeds of up to 400 kHz; +there's a high speed extension (3.4 MHz) that's not yet found wide use. +I2C is a multi-master bus; open drain signaling is used to arbitrate +between masters, as well as to handshake and to synchronize clocks from +slower clients. + +The Linux I2C programming interfaces support the master side of bus +interactions and the slave side. The programming interface is +structured around two kinds of driver, and two kinds of device. An I2C +"Adapter Driver" abstracts the controller hardware; it binds to a +physical device (perhaps a PCI device or platform_device) and exposes a +:c:type:`struct i2c_adapter <i2c_adapter>` representing each +I2C bus segment it manages. On each I2C bus segment will be I2C devices +represented by a :c:type:`struct i2c_client <i2c_client>`. +Those devices will be bound to a :c:type:`struct i2c_driver +<i2c_driver>`, which should follow the standard Linux driver model. There +are functions to perform various I2C protocol operations; at this writing +all such functions are usable only from task context. + +The System Management Bus (SMBus) is a sibling protocol. Most SMBus +systems are also I2C conformant. The electrical constraints are tighter +for SMBus, and it standardizes particular protocol messages and idioms. +Controllers that support I2C can also support most SMBus operations, but +SMBus controllers don't support all the protocol options that an I2C +controller will. There are functions to perform various SMBus protocol +operations, either using I2C primitives or by issuing SMBus commands to +i2c_adapter devices which don't support those I2C operations. + +.. kernel-doc:: include/linux/i2c.h + :internal: + +.. kernel-doc:: drivers/i2c/i2c-boardinfo.c + :functions: i2c_register_board_info + +.. kernel-doc:: drivers/i2c/i2c-core-base.c + :export: + +.. kernel-doc:: drivers/i2c/i2c-core-smbus.c + :export: diff --git a/Documentation/driver-api/iio/buffers.rst b/Documentation/driver-api/iio/buffers.rst new file mode 100644 index 000000000..02c99a6be --- /dev/null +++ b/Documentation/driver-api/iio/buffers.rst @@ -0,0 +1,125 @@ +======= +Buffers +======= + +* struct :c:type:`iio_buffer` — general buffer structure +* :c:func:`iio_validate_scan_mask_onehot` — Validates that exactly one channel + is selected +* :c:func:`iio_buffer_get` — Grab a reference to the buffer +* :c:func:`iio_buffer_put` — Release the reference to the buffer + +The Industrial I/O core offers a way for continuous data capture based on a +trigger source. Multiple data channels can be read at once from +:file:`/dev/iio:device{X}` character device node, thus reducing the CPU load. + +IIO buffer sysfs interface +========================== +An IIO buffer has an associated attributes directory under +:file:`/sys/bus/iio/iio:device{X}/buffer/*`. Here are some of the existing +attributes: + +* :file:`length`, the total number of data samples (capacity) that can be + stored by the buffer. +* :file:`enable`, activate buffer capture. + +IIO buffer setup +================ + +The meta information associated with a channel reading placed in a buffer is +called a scan element . The important bits configuring scan elements are +exposed to userspace applications via the +:file:`/sys/bus/iio/iio:device{X}/scan_elements/*` directory. This file contains +attributes of the following form: + +* :file:`enable`, used for enabling a channel. If and only if its attribute + is non *zero*, then a triggered capture will contain data samples for this + channel. +* :file:`type`, description of the scan element data storage within the buffer + and hence the form in which it is read from user space. + Format is [be|le]:[s|u]bits/storagebitsXrepeat[>>shift] . + * *be* or *le*, specifies big or little endian. + * *s* or *u*, specifies if signed (2's complement) or unsigned. + * *bits*, is the number of valid data bits. + * *storagebits*, is the number of bits (after padding) that it occupies in the + buffer. + * *shift*, if specified, is the shift that needs to be applied prior to + masking out unused bits. + * *repeat*, specifies the number of bits/storagebits repetitions. When the + repeat element is 0 or 1, then the repeat value is omitted. + +For example, a driver for a 3-axis accelerometer with 12 bit resolution where +data is stored in two 8-bits registers as follows:: + + 7 6 5 4 3 2 1 0 + +---+---+---+---+---+---+---+---+ + |D3 |D2 |D1 |D0 | X | X | X | X | (LOW byte, address 0x06) + +---+---+---+---+---+---+---+---+ + + 7 6 5 4 3 2 1 0 + +---+---+---+---+---+---+---+---+ + |D11|D10|D9 |D8 |D7 |D6 |D5 |D4 | (HIGH byte, address 0x07) + +---+---+---+---+---+---+---+---+ + +will have the following scan element type for each axis:: + + $ cat /sys/bus/iio/devices/iio:device0/scan_elements/in_accel_y_type + le:s12/16>>4 + +A user space application will interpret data samples read from the buffer as +two byte little endian signed data, that needs a 4 bits right shift before +masking out the 12 valid bits of data. + +For implementing buffer support a driver should initialize the following +fields in iio_chan_spec definition:: + + struct iio_chan_spec { + /* other members */ + int scan_index + struct { + char sign; + u8 realbits; + u8 storagebits; + u8 shift; + u8 repeat; + enum iio_endian endianness; + } scan_type; + }; + +The driver implementing the accelerometer described above will have the +following channel definition:: + + struct struct iio_chan_spec accel_channels[] = { + { + .type = IIO_ACCEL, + .modified = 1, + .channel2 = IIO_MOD_X, + /* other stuff here */ + .scan_index = 0, + .scan_type = { + .sign = 's', + .realbits = 12, + .storagebits = 16, + .shift = 4, + .endianness = IIO_LE, + }, + } + /* similar for Y (with channel2 = IIO_MOD_Y, scan_index = 1) + * and Z (with channel2 = IIO_MOD_Z, scan_index = 2) axis + */ + } + +Here **scan_index** defines the order in which the enabled channels are placed +inside the buffer. Channels with a lower **scan_index** will be placed before +channels with a higher index. Each channel needs to have a unique +**scan_index**. + +Setting **scan_index** to -1 can be used to indicate that the specific channel +does not support buffered capture. In this case no entries will be created for +the channel in the scan_elements directory. + +More details +============ +.. kernel-doc:: include/linux/iio/buffer.h +.. kernel-doc:: drivers/iio/industrialio-buffer.c + :export: + diff --git a/Documentation/driver-api/iio/core.rst b/Documentation/driver-api/iio/core.rst new file mode 100644 index 000000000..9a34ae03b --- /dev/null +++ b/Documentation/driver-api/iio/core.rst @@ -0,0 +1,182 @@ +============= +Core elements +============= + +The Industrial I/O core offers a unified framework for writing drivers for +many different types of embedded sensors. a standard interface to user space +applications manipulating sensors. The implementation can be found under +:file:`drivers/iio/industrialio-*` + +Industrial I/O Devices +---------------------- + +* struct :c:type:`iio_dev` - industrial I/O device +* :c:func:`iio_device_alloc()` - alocate an :c:type:`iio_dev` from a driver +* :c:func:`iio_device_free()` - free an :c:type:`iio_dev` from a driver +* :c:func:`iio_device_register()` - register a device with the IIO subsystem +* :c:func:`iio_device_unregister()` - unregister a device from the IIO + subsystem + +An IIO device usually corresponds to a single hardware sensor and it +provides all the information needed by a driver handling a device. +Let's first have a look at the functionality embedded in an IIO device +then we will show how a device driver makes use of an IIO device. + +There are two ways for a user space application to interact with an IIO driver. + +1. :file:`/sys/bus/iio/iio:device{X}/`, this represents a hardware sensor + and groups together the data channels of the same chip. +2. :file:`/dev/iio:device{X}`, character device node interface used for + buffered data transfer and for events information retrieval. + +A typical IIO driver will register itself as an :doc:`I2C <../i2c>` or +:doc:`SPI <../spi>` driver and will create two routines, probe and remove. + +At probe: + +1. Call :c:func:`iio_device_alloc()`, which allocates memory for an IIO device. +2. Initialize IIO device fields with driver specific information (e.g. + device name, device channels). +3. Call :c:func:`iio_device_register()`, this registers the device with the + IIO core. After this call the device is ready to accept requests from user + space applications. + +At remove, we free the resources allocated in probe in reverse order: + +1. :c:func:`iio_device_unregister()`, unregister the device from the IIO core. +2. :c:func:`iio_device_free()`, free the memory allocated for the IIO device. + +IIO device sysfs interface +========================== + +Attributes are sysfs files used to expose chip info and also allowing +applications to set various configuration parameters. For device with +index X, attributes can be found under /sys/bus/iio/iio:deviceX/ directory. +Common attributes are: + +* :file:`name`, description of the physical chip. +* :file:`dev`, shows the major:minor pair associated with + :file:`/dev/iio:deviceX` node. +* :file:`sampling_frequency_available`, available discrete set of sampling + frequency values for device. +* Available standard attributes for IIO devices are described in the + :file:`Documentation/ABI/testing/sysfs-bus-iio` file in the Linux kernel + sources. + +IIO device channels +=================== + +struct :c:type:`iio_chan_spec` - specification of a single channel + +An IIO device channel is a representation of a data channel. An IIO device can +have one or multiple channels. For example: + +* a thermometer sensor has one channel representing the temperature measurement. +* a light sensor with two channels indicating the measurements in the visible + and infrared spectrum. +* an accelerometer can have up to 3 channels representing acceleration on X, Y + and Z axes. + +An IIO channel is described by the struct :c:type:`iio_chan_spec`. +A thermometer driver for the temperature sensor in the example above would +have to describe its channel as follows:: + + static const struct iio_chan_spec temp_channel[] = { + { + .type = IIO_TEMP, + .info_mask_separate = BIT(IIO_CHAN_INFO_PROCESSED), + }, + }; + +Channel sysfs attributes exposed to userspace are specified in the form of +bitmasks. Depending on their shared info, attributes can be set in one of the +following masks: + +* **info_mask_separate**, attributes will be specific to + this channel +* **info_mask_shared_by_type**, attributes are shared by all channels of the + same type +* **info_mask_shared_by_dir**, attributes are shared by all channels of the same + direction +* **info_mask_shared_by_all**, attributes are shared by all channels + +When there are multiple data channels per channel type we have two ways to +distinguish between them: + +* set **.modified** field of :c:type:`iio_chan_spec` to 1. Modifiers are + specified using **.channel2** field of the same :c:type:`iio_chan_spec` + structure and are used to indicate a physically unique characteristic of the + channel such as its direction or spectral response. For example, a light + sensor can have two channels, one for infrared light and one for both + infrared and visible light. +* set **.indexed** field of :c:type:`iio_chan_spec` to 1. In this case the + channel is simply another instance with an index specified by the **.channel** + field. + +Here is how we can make use of the channel's modifiers:: + + static const struct iio_chan_spec light_channels[] = { + { + .type = IIO_INTENSITY, + .modified = 1, + .channel2 = IIO_MOD_LIGHT_IR, + .info_mask_separate = BIT(IIO_CHAN_INFO_RAW), + .info_mask_shared = BIT(IIO_CHAN_INFO_SAMP_FREQ), + }, + { + .type = IIO_INTENSITY, + .modified = 1, + .channel2 = IIO_MOD_LIGHT_BOTH, + .info_mask_separate = BIT(IIO_CHAN_INFO_RAW), + .info_mask_shared = BIT(IIO_CHAN_INFO_SAMP_FREQ), + }, + { + .type = IIO_LIGHT, + .info_mask_separate = BIT(IIO_CHAN_INFO_PROCESSED), + .info_mask_shared = BIT(IIO_CHAN_INFO_SAMP_FREQ), + }, + } + +This channel's definition will generate two separate sysfs files for raw data +retrieval: + +* :file:`/sys/bus/iio/iio:device{X}/in_intensity_ir_raw` +* :file:`/sys/bus/iio/iio:device{X}/in_intensity_both_raw` + +one file for processed data: + +* :file:`/sys/bus/iio/iio:device{X}/in_illuminance_input` + +and one shared sysfs file for sampling frequency: + +* :file:`/sys/bus/iio/iio:device{X}/sampling_frequency`. + +Here is how we can make use of the channel's indexing:: + + static const struct iio_chan_spec light_channels[] = { + { + .type = IIO_VOLTAGE, + .indexed = 1, + .channel = 0, + .info_mask_separate = BIT(IIO_CHAN_INFO_RAW), + }, + { + .type = IIO_VOLTAGE, + .indexed = 1, + .channel = 1, + .info_mask_separate = BIT(IIO_CHAN_INFO_RAW), + }, + } + +This will generate two separate attributes files for raw data retrieval: + +* :file:`/sys/bus/iio/devices/iio:device{X}/in_voltage0_raw`, representing + voltage measurement for channel 0. +* :file:`/sys/bus/iio/devices/iio:device{X}/in_voltage1_raw`, representing + voltage measurement for channel 1. + +More details +============ +.. kernel-doc:: include/linux/iio/iio.h +.. kernel-doc:: drivers/iio/industrialio-core.c + :export: diff --git a/Documentation/driver-api/iio/hw-consumer.rst b/Documentation/driver-api/iio/hw-consumer.rst new file mode 100644 index 000000000..8facce6a6 --- /dev/null +++ b/Documentation/driver-api/iio/hw-consumer.rst @@ -0,0 +1,51 @@ +=========== +HW consumer +=========== +An IIO device can be directly connected to another device in hardware. in this +case the buffers between IIO provider and IIO consumer are handled by hardware. +The Industrial I/O HW consumer offers a way to bond these IIO devices without +software buffer for data. The implementation can be found under +:file:`drivers/iio/buffer/hw-consumer.c` + + +* struct :c:type:`iio_hw_consumer` — Hardware consumer structure +* :c:func:`iio_hw_consumer_alloc` — Allocate IIO hardware consumer +* :c:func:`iio_hw_consumer_free` — Free IIO hardware consumer +* :c:func:`iio_hw_consumer_enable` — Enable IIO hardware consumer +* :c:func:`iio_hw_consumer_disable` — Disable IIO hardware consumer + + +HW consumer setup +================= + +As standard IIO device the implementation is based on IIO provider/consumer. +A typical IIO HW consumer setup looks like this:: + + static struct iio_hw_consumer *hwc; + + static const struct iio_info adc_info = { + .read_raw = adc_read_raw, + }; + + static int adc_read_raw(struct iio_dev *indio_dev, + struct iio_chan_spec const *chan, int *val, + int *val2, long mask) + { + ret = iio_hw_consumer_enable(hwc); + + /* Acquire data */ + + ret = iio_hw_consumer_disable(hwc); + } + + static int adc_probe(struct platform_device *pdev) + { + hwc = devm_iio_hw_consumer_alloc(&iio->dev); + } + +More details +============ +.. kernel-doc:: include/linux/iio/hw-consumer.h +.. kernel-doc:: drivers/iio/buffer/industrialio-hw-consumer.c + :export: + diff --git a/Documentation/driver-api/iio/index.rst b/Documentation/driver-api/iio/index.rst new file mode 100644 index 000000000..7fba341bd --- /dev/null +++ b/Documentation/driver-api/iio/index.rst @@ -0,0 +1,18 @@ +.. include:: <isonum.txt> + +Industrial I/O +============== + +**Copyright** |copy| 2015 Intel Corporation + +Contents: + +.. toctree:: + :maxdepth: 2 + + intro + core + buffers + triggers + triggered-buffers + hw-consumer diff --git a/Documentation/driver-api/iio/intro.rst b/Documentation/driver-api/iio/intro.rst new file mode 100644 index 000000000..3653fbd57 --- /dev/null +++ b/Documentation/driver-api/iio/intro.rst @@ -0,0 +1,33 @@ +.. include:: <isonum.txt> + +============ +Introduction +============ + +The main purpose of the Industrial I/O subsystem (IIO) is to provide support +for devices that in some sense perform either +analog-to-digital conversion (ADC) or digital-to-analog conversion (DAC) +or both. The aim is to fill the gap between the somewhat similar hwmon and +:doc:`input <../input>` subsystems. Hwmon is directed at low sample rate +sensors used to monitor and control the system itself, like fan speed control +or temperature measurement. :doc:`Input <../input>` is, as its name suggests, +focused on human interaction input devices (keyboard, mouse, touchscreen). +In some cases there is considerable overlap between these and IIO. + +Devices that fall into this category include: + +* analog to digital converters (ADCs) +* accelerometers +* capacitance to digital converters (CDCs) +* digital to analog converters (DACs) +* gyroscopes +* inertial measurement units (IMUs) +* color and light sensors +* magnetometers +* pressure sensors +* proximity sensors +* temperature sensors + +Usually these sensors are connected via :doc:`SPI <../spi>` or +:doc:`I2C <../i2c>`. A common use case of the sensors devices is to have +combined functionality (e.g. light plus proximity sensor). diff --git a/Documentation/driver-api/iio/triggered-buffers.rst b/Documentation/driver-api/iio/triggered-buffers.rst new file mode 100644 index 000000000..0db12660c --- /dev/null +++ b/Documentation/driver-api/iio/triggered-buffers.rst @@ -0,0 +1,69 @@ +================= +Triggered Buffers +================= + +Now that we know what buffers and triggers are let's see how they work together. + +IIO triggered buffer setup +========================== + +* :c:func:`iio_triggered_buffer_setup` — Setup triggered buffer and pollfunc +* :c:func:`iio_triggered_buffer_cleanup` — Free resources allocated by + :c:func:`iio_triggered_buffer_setup` +* struct :c:type:`iio_buffer_setup_ops` — buffer setup related callbacks + +A typical triggered buffer setup looks like this:: + + const struct iio_buffer_setup_ops sensor_buffer_setup_ops = { + .preenable = sensor_buffer_preenable, + .postenable = sensor_buffer_postenable, + .postdisable = sensor_buffer_postdisable, + .predisable = sensor_buffer_predisable, + }; + + irqreturn_t sensor_iio_pollfunc(int irq, void *p) + { + pf->timestamp = iio_get_time_ns((struct indio_dev *)p); + return IRQ_WAKE_THREAD; + } + + irqreturn_t sensor_trigger_handler(int irq, void *p) + { + u16 buf[8]; + int i = 0; + + /* read data for each active channel */ + for_each_set_bit(bit, active_scan_mask, masklength) + buf[i++] = sensor_get_data(bit) + + iio_push_to_buffers_with_timestamp(indio_dev, buf, timestamp); + + iio_trigger_notify_done(trigger); + return IRQ_HANDLED; + } + + /* setup triggered buffer, usually in probe function */ + iio_triggered_buffer_setup(indio_dev, sensor_iio_polfunc, + sensor_trigger_handler, + sensor_buffer_setup_ops); + +The important things to notice here are: + +* :c:type:`iio_buffer_setup_ops`, the buffer setup functions to be called at + predefined points in the buffer configuration sequence (e.g. before enable, + after disable). If not specified, the IIO core uses the default + iio_triggered_buffer_setup_ops. +* **sensor_iio_pollfunc**, the function that will be used as top half of poll + function. It should do as little processing as possible, because it runs in + interrupt context. The most common operation is recording of the current + timestamp and for this reason one can use the IIO core defined + :c:func:`iio_pollfunc_store_time` function. +* **sensor_trigger_handler**, the function that will be used as bottom half of + the poll function. This runs in the context of a kernel thread and all the + processing takes place here. It usually reads data from the device and + stores it in the internal buffer together with the timestamp recorded in the + top half. + +More details +============ +.. kernel-doc:: drivers/iio/buffer/industrialio-triggered-buffer.c diff --git a/Documentation/driver-api/iio/triggers.rst b/Documentation/driver-api/iio/triggers.rst new file mode 100644 index 000000000..f89d37e7d --- /dev/null +++ b/Documentation/driver-api/iio/triggers.rst @@ -0,0 +1,80 @@ +======== +Triggers +======== + +* struct :c:type:`iio_trigger` — industrial I/O trigger device +* :c:func:`devm_iio_trigger_alloc` — Resource-managed iio_trigger_alloc +* :c:func:`devm_iio_trigger_free` — Resource-managed iio_trigger_free +* :c:func:`devm_iio_trigger_register` — Resource-managed iio_trigger_register +* :c:func:`devm_iio_trigger_unregister` — Resource-managed + iio_trigger_unregister +* :c:func:`iio_trigger_validate_own_device` — Check if a trigger and IIO + device belong to the same device + +In many situations it is useful for a driver to be able to capture data based +on some external event (trigger) as opposed to periodically polling for data. +An IIO trigger can be provided by a device driver that also has an IIO device +based on hardware generated events (e.g. data ready or threshold exceeded) or +provided by a separate driver from an independent interrupt source (e.g. GPIO +line connected to some external system, timer interrupt or user space writing +a specific file in sysfs). A trigger may initiate data capture for a number of +sensors and also it may be completely unrelated to the sensor itself. + +IIO trigger sysfs interface +=========================== + +There are two locations in sysfs related to triggers: + +* :file:`/sys/bus/iio/devices/trigger{Y}/*`, this file is created once an + IIO trigger is registered with the IIO core and corresponds to trigger + with index Y. + Because triggers can be very different depending on type there are few + standard attributes that we can describe here: + + * :file:`name`, trigger name that can be later used for association with a + device. + * :file:`sampling_frequency`, some timer based triggers use this attribute to + specify the frequency for trigger calls. + +* :file:`/sys/bus/iio/devices/iio:device{X}/trigger/*`, this directory is + created once the device supports a triggered buffer. We can associate a + trigger with our device by writing the trigger's name in the + :file:`current_trigger` file. + +IIO trigger setup +================= + +Let's see a simple example of how to setup a trigger to be used by a driver:: + + struct iio_trigger_ops trigger_ops = { + .set_trigger_state = sample_trigger_state, + .validate_device = sample_validate_device, + } + + struct iio_trigger *trig; + + /* first, allocate memory for our trigger */ + trig = iio_trigger_alloc(dev, "trig-%s-%d", name, idx); + + /* setup trigger operations field */ + trig->ops = &trigger_ops; + + /* now register the trigger with the IIO core */ + iio_trigger_register(trig); + +IIO trigger ops +=============== + +* struct :c:type:`iio_trigger_ops` — operations structure for an iio_trigger. + +Notice that a trigger has a set of operations attached: + +* :file:`set_trigger_state`, switch the trigger on/off on demand. +* :file:`validate_device`, function to validate the device when the current + trigger gets changed. + +More details +============ +.. kernel-doc:: include/linux/iio/trigger.h +.. kernel-doc:: drivers/iio/industrialio-trigger.c + :export: diff --git a/Documentation/driver-api/index.rst b/Documentation/driver-api/index.rst new file mode 100644 index 000000000..6d9f2f9fe --- /dev/null +++ b/Documentation/driver-api/index.rst @@ -0,0 +1,62 @@ +======================================== +The Linux driver implementer's API guide +======================================== + +The kernel offers a wide variety of interfaces to support the development +of device drivers. This document is an only somewhat organized collection +of some of those interfaces — it will hopefully get better over time! The +available subsections can be seen below. + +.. class:: toc-title + + Table of contents + +.. toctree:: + :maxdepth: 2 + + basics + infrastructure + pm/index + clk + device-io + device_connection + dma-buf + device_link + message-based + sound + frame-buffer + regulator + iio/index + input + usb/index + pci + spi + i2c + hsi + edac + scsi + libata + target + mtdnand + miscellaneous + w1 + rapidio + s390-drivers + vme + 80211/index + uio-howto + firmware/index + pinctl + gpio/index + misc_devices + dmaengine/index + slimbus + soundwire/index + fpga/index + +.. only:: subproject and html + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/infrastructure.rst b/Documentation/driver-api/infrastructure.rst new file mode 100644 index 000000000..6172f3cc3 --- /dev/null +++ b/Documentation/driver-api/infrastructure.rst @@ -0,0 +1,84 @@ +Device drivers infrastructure +============================= + +The Basic Device Driver-Model Structures +---------------------------------------- + +.. kernel-doc:: include/linux/device.h + :internal: + +Device Drivers Base +------------------- + +.. kernel-doc:: drivers/base/init.c + :internal: + +.. kernel-doc:: drivers/base/driver.c + :export: + +.. kernel-doc:: drivers/base/core.c + :export: + +.. kernel-doc:: drivers/base/syscore.c + :export: + +.. kernel-doc:: drivers/base/class.c + :export: + +.. kernel-doc:: drivers/base/node.c + :internal: + +.. kernel-doc:: drivers/base/firmware_loader/main.c + :export: + +.. kernel-doc:: drivers/base/transport_class.c + :export: + +.. kernel-doc:: drivers/base/dd.c + :export: + +.. kernel-doc:: include/linux/platform_device.h + :internal: + +.. kernel-doc:: drivers/base/platform.c + :export: + +.. kernel-doc:: drivers/base/bus.c + :export: + +Device Drivers DMA Management +----------------------------- + +.. kernel-doc:: kernel/dma/coherent.c + :export: + +.. kernel-doc:: kernel/dma/mapping.c + :export: + +Device drivers PnP support +-------------------------- + +.. kernel-doc:: drivers/pnp/core.c + :internal: + +.. kernel-doc:: drivers/pnp/card.c + :export: + +.. kernel-doc:: drivers/pnp/driver.c + :internal: + +.. kernel-doc:: drivers/pnp/manager.c + :export: + +.. kernel-doc:: drivers/pnp/support.c + :export: + +Userspace IO devices +-------------------- + +.. kernel-doc:: drivers/uio/uio.c + :export: + +.. kernel-doc:: include/linux/uio_driver.h + :internal: + diff --git a/Documentation/driver-api/input.rst b/Documentation/driver-api/input.rst new file mode 100644 index 000000000..d05bf58fa --- /dev/null +++ b/Documentation/driver-api/input.rst @@ -0,0 +1,51 @@ +Input Subsystem +=============== + +Input core +---------- + +.. kernel-doc:: include/linux/input.h + :internal: + +.. kernel-doc:: drivers/input/input.c + :export: + +.. kernel-doc:: drivers/input/ff-core.c + :export: + +.. kernel-doc:: drivers/input/ff-memless.c + :export: + +Multitouch Library +------------------ + +.. kernel-doc:: include/linux/input/mt.h + :internal: + +.. kernel-doc:: drivers/input/input-mt.c + :export: + +Polled input devices +-------------------- + +.. kernel-doc:: include/linux/input-polldev.h + :internal: + +.. kernel-doc:: drivers/input/input-polldev.c + :export: + +Matrix keyboards/keypads +------------------------ + +.. kernel-doc:: include/linux/input/matrix_keypad.h + :internal: + +Sparse keymap support +--------------------- + +.. kernel-doc:: include/linux/input/sparse-keymap.h + :internal: + +.. kernel-doc:: drivers/input/sparse-keymap.c + :export: + diff --git a/Documentation/driver-api/libata.rst b/Documentation/driver-api/libata.rst new file mode 100644 index 000000000..9f3e5dc31 --- /dev/null +++ b/Documentation/driver-api/libata.rst @@ -0,0 +1,1030 @@ +======================== +libATA Developer's Guide +======================== + +:Author: Jeff Garzik + +Introduction +============ + +libATA is a library used inside the Linux kernel to support ATA host +controllers and devices. libATA provides an ATA driver API, class +transports for ATA and ATAPI devices, and SCSI<->ATA translation for ATA +devices according to the T10 SAT specification. + +This Guide documents the libATA driver API, library functions, library +internals, and a couple sample ATA low-level drivers. + +libata Driver API +================= + +:c:type:`struct ata_port_operations <ata_port_operations>` +is defined for every low-level libata +hardware driver, and it controls how the low-level driver interfaces +with the ATA and SCSI layers. + +FIS-based drivers will hook into the system with ``->qc_prep()`` and +``->qc_issue()`` high-level hooks. Hardware which behaves in a manner +similar to PCI IDE hardware may utilize several generic helpers, +defining at a bare minimum the bus I/O addresses of the ATA shadow +register blocks. + +:c:type:`struct ata_port_operations <ata_port_operations>` +---------------------------------------------------------- + +Disable ATA port +~~~~~~~~~~~~~~~~ + +:: + + void (*port_disable) (struct ata_port *); + + +Called from :c:func:`ata_bus_probe` error path, as well as when unregistering +from the SCSI module (rmmod, hot unplug). This function should do +whatever needs to be done to take the port out of use. In most cases, +:c:func:`ata_port_disable` can be used as this hook. + +Called from :c:func:`ata_bus_probe` on a failed probe. Called from +:c:func:`ata_scsi_release`. + +Post-IDENTIFY device configuration +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + void (*dev_config) (struct ata_port *, struct ata_device *); + + +Called after IDENTIFY [PACKET] DEVICE is issued to each device found. +Typically used to apply device-specific fixups prior to issue of SET +FEATURES - XFER MODE, and prior to operation. + +This entry may be specified as NULL in ata_port_operations. + +Set PIO/DMA mode +~~~~~~~~~~~~~~~~ + +:: + + void (*set_piomode) (struct ata_port *, struct ata_device *); + void (*set_dmamode) (struct ata_port *, struct ata_device *); + void (*post_set_mode) (struct ata_port *); + unsigned int (*mode_filter) (struct ata_port *, struct ata_device *, unsigned int); + + +Hooks called prior to the issue of SET FEATURES - XFER MODE command. The +optional ``->mode_filter()`` hook is called when libata has built a mask of +the possible modes. This is passed to the ``->mode_filter()`` function +which should return a mask of valid modes after filtering those +unsuitable due to hardware limits. It is not valid to use this interface +to add modes. + +``dev->pio_mode`` and ``dev->dma_mode`` are guaranteed to be valid when +``->set_piomode()`` and when ``->set_dmamode()`` is called. The timings for +any other drive sharing the cable will also be valid at this point. That +is the library records the decisions for the modes of each drive on a +channel before it attempts to set any of them. + +``->post_set_mode()`` is called unconditionally, after the SET FEATURES - +XFER MODE command completes successfully. + +``->set_piomode()`` is always called (if present), but ``->set_dma_mode()`` +is only called if DMA is possible. + +Taskfile read/write +~~~~~~~~~~~~~~~~~~~ + +:: + + void (*sff_tf_load) (struct ata_port *ap, struct ata_taskfile *tf); + void (*sff_tf_read) (struct ata_port *ap, struct ata_taskfile *tf); + + +``->tf_load()`` is called to load the given taskfile into hardware +registers / DMA buffers. ``->tf_read()`` is called to read the hardware +registers / DMA buffers, to obtain the current set of taskfile register +values. Most drivers for taskfile-based hardware (PIO or MMIO) use +:c:func:`ata_sff_tf_load` and :c:func:`ata_sff_tf_read` for these hooks. + +PIO data read/write +~~~~~~~~~~~~~~~~~~~ + +:: + + void (*sff_data_xfer) (struct ata_device *, unsigned char *, unsigned int, int); + + +All bmdma-style drivers must implement this hook. This is the low-level +operation that actually copies the data bytes during a PIO data +transfer. Typically the driver will choose one of +:c:func:`ata_sff_data_xfer`, or :c:func:`ata_sff_data_xfer32`. + +ATA command execute +~~~~~~~~~~~~~~~~~~~ + +:: + + void (*sff_exec_command)(struct ata_port *ap, struct ata_taskfile *tf); + + +causes an ATA command, previously loaded with ``->tf_load()``, to be +initiated in hardware. Most drivers for taskfile-based hardware use +:c:func:`ata_sff_exec_command` for this hook. + +Per-cmd ATAPI DMA capabilities filter +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + int (*check_atapi_dma) (struct ata_queued_cmd *qc); + + +Allow low-level driver to filter ATA PACKET commands, returning a status +indicating whether or not it is OK to use DMA for the supplied PACKET +command. + +This hook may be specified as NULL, in which case libata will assume +that atapi dma can be supported. + +Read specific ATA shadow registers +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + u8 (*sff_check_status)(struct ata_port *ap); + u8 (*sff_check_altstatus)(struct ata_port *ap); + + +Reads the Status/AltStatus ATA shadow register from hardware. On some +hardware, reading the Status register has the side effect of clearing +the interrupt condition. Most drivers for taskfile-based hardware use +:c:func:`ata_sff_check_status` for this hook. + +Write specific ATA shadow register +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + void (*sff_set_devctl)(struct ata_port *ap, u8 ctl); + + +Write the device control ATA shadow register to the hardware. Most +drivers don't need to define this. + +Select ATA device on bus +~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + void (*sff_dev_select)(struct ata_port *ap, unsigned int device); + + +Issues the low-level hardware command(s) that causes one of N hardware +devices to be considered 'selected' (active and available for use) on +the ATA bus. This generally has no meaning on FIS-based devices. + +Most drivers for taskfile-based hardware use :c:func:`ata_sff_dev_select` for +this hook. + +Private tuning method +~~~~~~~~~~~~~~~~~~~~~ + +:: + + void (*set_mode) (struct ata_port *ap); + + +By default libata performs drive and controller tuning in accordance +with the ATA timing rules and also applies blacklists and cable limits. +Some controllers need special handling and have custom tuning rules, +typically raid controllers that use ATA commands but do not actually do +drive timing. + + **Warning** + + This hook should not be used to replace the standard controller + tuning logic when a controller has quirks. Replacing the default + tuning logic in that case would bypass handling for drive and bridge + quirks that may be important to data reliability. If a controller + needs to filter the mode selection it should use the mode_filter + hook instead. + +Control PCI IDE BMDMA engine +~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + void (*bmdma_setup) (struct ata_queued_cmd *qc); + void (*bmdma_start) (struct ata_queued_cmd *qc); + void (*bmdma_stop) (struct ata_port *ap); + u8 (*bmdma_status) (struct ata_port *ap); + + +When setting up an IDE BMDMA transaction, these hooks arm +(``->bmdma_setup``), fire (``->bmdma_start``), and halt (``->bmdma_stop``) the +hardware's DMA engine. ``->bmdma_status`` is used to read the standard PCI +IDE DMA Status register. + +These hooks are typically either no-ops, or simply not implemented, in +FIS-based drivers. + +Most legacy IDE drivers use :c:func:`ata_bmdma_setup` for the +:c:func:`bmdma_setup` hook. :c:func:`ata_bmdma_setup` will write the pointer +to the PRD table to the IDE PRD Table Address register, enable DMA in the DMA +Command register, and call :c:func:`exec_command` to begin the transfer. + +Most legacy IDE drivers use :c:func:`ata_bmdma_start` for the +:c:func:`bmdma_start` hook. :c:func:`ata_bmdma_start` will write the +ATA_DMA_START flag to the DMA Command register. + +Many legacy IDE drivers use :c:func:`ata_bmdma_stop` for the +:c:func:`bmdma_stop` hook. :c:func:`ata_bmdma_stop` clears the ATA_DMA_START +flag in the DMA command register. + +Many legacy IDE drivers use :c:func:`ata_bmdma_status` as the +:c:func:`bmdma_status` hook. + +High-level taskfile hooks +~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + enum ata_completion_errors (*qc_prep) (struct ata_queued_cmd *qc); + int (*qc_issue) (struct ata_queued_cmd *qc); + + +Higher-level hooks, these two hooks can potentially supercede several of +the above taskfile/DMA engine hooks. ``->qc_prep`` is called after the +buffers have been DMA-mapped, and is typically used to populate the +hardware's DMA scatter-gather table. Most drivers use the standard +:c:func:`ata_qc_prep` helper function, but more advanced drivers roll their +own. + +``->qc_issue`` is used to make a command active, once the hardware and S/G +tables have been prepared. IDE BMDMA drivers use the helper function +:c:func:`ata_qc_issue_prot` for taskfile protocol-based dispatch. More +advanced drivers implement their own ``->qc_issue``. + +:c:func:`ata_qc_issue_prot` calls ``->tf_load()``, ``->bmdma_setup()``, and +``->bmdma_start()`` as necessary to initiate a transfer. + +Exception and probe handling (EH) +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + void (*eng_timeout) (struct ata_port *ap); + void (*phy_reset) (struct ata_port *ap); + + +Deprecated. Use ``->error_handler()`` instead. + +:: + + void (*freeze) (struct ata_port *ap); + void (*thaw) (struct ata_port *ap); + + +:c:func:`ata_port_freeze` is called when HSM violations or some other +condition disrupts normal operation of the port. A frozen port is not +allowed to perform any operation until the port is thawed, which usually +follows a successful reset. + +The optional ``->freeze()`` callback can be used for freezing the port +hardware-wise (e.g. mask interrupt and stop DMA engine). If a port +cannot be frozen hardware-wise, the interrupt handler must ack and clear +interrupts unconditionally while the port is frozen. + +The optional ``->thaw()`` callback is called to perform the opposite of +``->freeze()``: prepare the port for normal operation once again. Unmask +interrupts, start DMA engine, etc. + +:: + + void (*error_handler) (struct ata_port *ap); + + +``->error_handler()`` is a driver's hook into probe, hotplug, and recovery +and other exceptional conditions. The primary responsibility of an +implementation is to call :c:func:`ata_do_eh` or :c:func:`ata_bmdma_drive_eh` +with a set of EH hooks as arguments: + +'prereset' hook (may be NULL) is called during an EH reset, before any +other actions are taken. + +'postreset' hook (may be NULL) is called after the EH reset is +performed. Based on existing conditions, severity of the problem, and +hardware capabilities, + +Either 'softreset' (may be NULL) or 'hardreset' (may be NULL) will be +called to perform the low-level EH reset. + +:: + + void (*post_internal_cmd) (struct ata_queued_cmd *qc); + + +Perform any hardware-specific actions necessary to finish processing +after executing a probe-time or EH-time command via +:c:func:`ata_exec_internal`. + +Hardware interrupt handling +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + irqreturn_t (*irq_handler)(int, void *, struct pt_regs *); + void (*irq_clear) (struct ata_port *); + + +``->irq_handler`` is the interrupt handling routine registered with the +system, by libata. ``->irq_clear`` is called during probe just before the +interrupt handler is registered, to be sure hardware is quiet. + +The second argument, dev_instance, should be cast to a pointer to +:c:type:`struct ata_host_set <ata_host_set>`. + +Most legacy IDE drivers use :c:func:`ata_sff_interrupt` for the irq_handler +hook, which scans all ports in the host_set, determines which queued +command was active (if any), and calls ata_sff_host_intr(ap,qc). + +Most legacy IDE drivers use :c:func:`ata_sff_irq_clear` for the +:c:func:`irq_clear` hook, which simply clears the interrupt and error flags +in the DMA status register. + +SATA phy read/write +~~~~~~~~~~~~~~~~~~~ + +:: + + int (*scr_read) (struct ata_port *ap, unsigned int sc_reg, + u32 *val); + int (*scr_write) (struct ata_port *ap, unsigned int sc_reg, + u32 val); + + +Read and write standard SATA phy registers. Currently only used if +``->phy_reset`` hook called the :c:func:`sata_phy_reset` helper function. +sc_reg is one of SCR_STATUS, SCR_CONTROL, SCR_ERROR, or SCR_ACTIVE. + +Init and shutdown +~~~~~~~~~~~~~~~~~ + +:: + + int (*port_start) (struct ata_port *ap); + void (*port_stop) (struct ata_port *ap); + void (*host_stop) (struct ata_host_set *host_set); + + +``->port_start()`` is called just after the data structures for each port +are initialized. Typically this is used to alloc per-port DMA buffers / +tables / rings, enable DMA engines, and similar tasks. Some drivers also +use this entry point as a chance to allocate driver-private memory for +``ap->private_data``. + +Many drivers use :c:func:`ata_port_start` as this hook or call it from their +own :c:func:`port_start` hooks. :c:func:`ata_port_start` allocates space for +a legacy IDE PRD table and returns. + +``->port_stop()`` is called after ``->host_stop()``. Its sole function is to +release DMA/memory resources, now that they are no longer actively being +used. Many drivers also free driver-private data from port at this time. + +``->host_stop()`` is called after all ``->port_stop()`` calls have completed. +The hook must finalize hardware shutdown, release DMA and other +resources, etc. This hook may be specified as NULL, in which case it is +not called. + +Error handling +============== + +This chapter describes how errors are handled under libata. Readers are +advised to read SCSI EH (Documentation/scsi/scsi_eh.txt) and ATA +exceptions doc first. + +Origins of commands +------------------- + +In libata, a command is represented with +:c:type:`struct ata_queued_cmd <ata_queued_cmd>` or qc. +qc's are preallocated during port initialization and repetitively used +for command executions. Currently only one qc is allocated per port but +yet-to-be-merged NCQ branch allocates one for each tag and maps each qc +to NCQ tag 1-to-1. + +libata commands can originate from two sources - libata itself and SCSI +midlayer. libata internal commands are used for initialization and error +handling. All normal blk requests and commands for SCSI emulation are +passed as SCSI commands through queuecommand callback of SCSI host +template. + +How commands are issued +----------------------- + +Internal commands + First, qc is allocated and initialized using :c:func:`ata_qc_new_init`. + Although :c:func:`ata_qc_new_init` doesn't implement any wait or retry + mechanism when qc is not available, internal commands are currently + issued only during initialization and error recovery, so no other + command is active and allocation is guaranteed to succeed. + + Once allocated qc's taskfile is initialized for the command to be + executed. qc currently has two mechanisms to notify completion. One + is via ``qc->complete_fn()`` callback and the other is completion + ``qc->waiting``. ``qc->complete_fn()`` callback is the asynchronous path + used by normal SCSI translated commands and ``qc->waiting`` is the + synchronous (issuer sleeps in process context) path used by internal + commands. + + Once initialization is complete, host_set lock is acquired and the + qc is issued. + +SCSI commands + All libata drivers use :c:func:`ata_scsi_queuecmd` as + ``hostt->queuecommand`` callback. scmds can either be simulated or + translated. No qc is involved in processing a simulated scmd. The + result is computed right away and the scmd is completed. + + For a translated scmd, :c:func:`ata_qc_new_init` is invoked to allocate a + qc and the scmd is translated into the qc. SCSI midlayer's + completion notification function pointer is stored into + ``qc->scsidone``. + + ``qc->complete_fn()`` callback is used for completion notification. ATA + commands use :c:func:`ata_scsi_qc_complete` while ATAPI commands use + :c:func:`atapi_qc_complete`. Both functions end up calling ``qc->scsidone`` + to notify upper layer when the qc is finished. After translation is + completed, the qc is issued with :c:func:`ata_qc_issue`. + + Note that SCSI midlayer invokes hostt->queuecommand while holding + host_set lock, so all above occur while holding host_set lock. + +How commands are processed +-------------------------- + +Depending on which protocol and which controller are used, commands are +processed differently. For the purpose of discussion, a controller which +uses taskfile interface and all standard callbacks is assumed. + +Currently 6 ATA command protocols are used. They can be sorted into the +following four categories according to how they are processed. + +ATA NO DATA or DMA + ATA_PROT_NODATA and ATA_PROT_DMA fall into this category. These + types of commands don't require any software intervention once + issued. Device will raise interrupt on completion. + +ATA PIO + ATA_PROT_PIO is in this category. libata currently implements PIO + with polling. ATA_NIEN bit is set to turn off interrupt and + pio_task on ata_wq performs polling and IO. + +ATAPI NODATA or DMA + ATA_PROT_ATAPI_NODATA and ATA_PROT_ATAPI_DMA are in this + category. packet_task is used to poll BSY bit after issuing PACKET + command. Once BSY is turned off by the device, packet_task + transfers CDB and hands off processing to interrupt handler. + +ATAPI PIO + ATA_PROT_ATAPI is in this category. ATA_NIEN bit is set and, as + in ATAPI NODATA or DMA, packet_task submits cdb. However, after + submitting cdb, further processing (data transfer) is handed off to + pio_task. + +How commands are completed +-------------------------- + +Once issued, all qc's are either completed with :c:func:`ata_qc_complete` or +time out. For commands which are handled by interrupts, +:c:func:`ata_host_intr` invokes :c:func:`ata_qc_complete`, and, for PIO tasks, +pio_task invokes :c:func:`ata_qc_complete`. In error cases, packet_task may +also complete commands. + +:c:func:`ata_qc_complete` does the following. + +1. DMA memory is unmapped. + +2. ATA_QCFLAG_ACTIVE is cleared from qc->flags. + +3. :c:func:`qc->complete_fn` callback is invoked. If the return value of the + callback is not zero. Completion is short circuited and + :c:func:`ata_qc_complete` returns. + +4. :c:func:`__ata_qc_complete` is called, which does + + 1. ``qc->flags`` is cleared to zero. + + 2. ``ap->active_tag`` and ``qc->tag`` are poisoned. + + 3. ``qc->waiting`` is cleared & completed (in that order). + + 4. qc is deallocated by clearing appropriate bit in ``ap->qactive``. + +So, it basically notifies upper layer and deallocates qc. One exception +is short-circuit path in #3 which is used by :c:func:`atapi_qc_complete`. + +For all non-ATAPI commands, whether it fails or not, almost the same +code path is taken and very little error handling takes place. A qc is +completed with success status if it succeeded, with failed status +otherwise. + +However, failed ATAPI commands require more handling as REQUEST SENSE is +needed to acquire sense data. If an ATAPI command fails, +:c:func:`ata_qc_complete` is invoked with error status, which in turn invokes +:c:func:`atapi_qc_complete` via ``qc->complete_fn()`` callback. + +This makes :c:func:`atapi_qc_complete` set ``scmd->result`` to +SAM_STAT_CHECK_CONDITION, complete the scmd and return 1. As the +sense data is empty but ``scmd->result`` is CHECK CONDITION, SCSI midlayer +will invoke EH for the scmd, and returning 1 makes :c:func:`ata_qc_complete` +to return without deallocating the qc. This leads us to +:c:func:`ata_scsi_error` with partially completed qc. + +:c:func:`ata_scsi_error` +------------------------ + +:c:func:`ata_scsi_error` is the current ``transportt->eh_strategy_handler()`` +for libata. As discussed above, this will be entered in two cases - +timeout and ATAPI error completion. This function calls low level libata +driver's :c:func:`eng_timeout` callback, the standard callback for which is +:c:func:`ata_eng_timeout`. It checks if a qc is active and calls +:c:func:`ata_qc_timeout` on the qc if so. Actual error handling occurs in +:c:func:`ata_qc_timeout`. + +If EH is invoked for timeout, :c:func:`ata_qc_timeout` stops BMDMA and +completes the qc. Note that as we're currently in EH, we cannot call +scsi_done. As described in SCSI EH doc, a recovered scmd should be +either retried with :c:func:`scsi_queue_insert` or finished with +:c:func:`scsi_finish_command`. Here, we override ``qc->scsidone`` with +:c:func:`scsi_finish_command` and calls :c:func:`ata_qc_complete`. + +If EH is invoked due to a failed ATAPI qc, the qc here is completed but +not deallocated. The purpose of this half-completion is to use the qc as +place holder to make EH code reach this place. This is a bit hackish, +but it works. + +Once control reaches here, the qc is deallocated by invoking +:c:func:`__ata_qc_complete` explicitly. Then, internal qc for REQUEST SENSE +is issued. Once sense data is acquired, scmd is finished by directly +invoking :c:func:`scsi_finish_command` on the scmd. Note that as we already +have completed and deallocated the qc which was associated with the +scmd, we don't need to/cannot call :c:func:`ata_qc_complete` again. + +Problems with the current EH +---------------------------- + +- Error representation is too crude. Currently any and all error + conditions are represented with ATA STATUS and ERROR registers. + Errors which aren't ATA device errors are treated as ATA device + errors by setting ATA_ERR bit. Better error descriptor which can + properly represent ATA and other errors/exceptions is needed. + +- When handling timeouts, no action is taken to make device forget + about the timed out command and ready for new commands. + +- EH handling via :c:func:`ata_scsi_error` is not properly protected from + usual command processing. On EH entrance, the device is not in + quiescent state. Timed out commands may succeed or fail any time. + pio_task and atapi_task may still be running. + +- Too weak error recovery. Devices / controllers causing HSM mismatch + errors and other errors quite often require reset to return to known + state. Also, advanced error handling is necessary to support features + like NCQ and hotplug. + +- ATA errors are directly handled in the interrupt handler and PIO + errors in pio_task. This is problematic for advanced error handling + for the following reasons. + + First, advanced error handling often requires context and internal qc + execution. + + Second, even a simple failure (say, CRC error) needs information + gathering and could trigger complex error handling (say, resetting & + reconfiguring). Having multiple code paths to gather information, + enter EH and trigger actions makes life painful. + + Third, scattered EH code makes implementing low level drivers + difficult. Low level drivers override libata callbacks. If EH is + scattered over several places, each affected callbacks should perform + its part of error handling. This can be error prone and painful. + +libata Library +============== + +.. kernel-doc:: drivers/ata/libata-core.c + :export: + +libata Core Internals +===================== + +.. kernel-doc:: drivers/ata/libata-core.c + :internal: + +.. kernel-doc:: drivers/ata/libata-eh.c + +libata SCSI translation/emulation +================================= + +.. kernel-doc:: drivers/ata/libata-scsi.c + :export: + +.. kernel-doc:: drivers/ata/libata-scsi.c + :internal: + +ATA errors and exceptions +========================= + +This chapter tries to identify what error/exception conditions exist for +ATA/ATAPI devices and describe how they should be handled in +implementation-neutral way. + +The term 'error' is used to describe conditions where either an explicit +error condition is reported from device or a command has timed out. + +The term 'exception' is either used to describe exceptional conditions +which are not errors (say, power or hotplug events), or to describe both +errors and non-error exceptional conditions. Where explicit distinction +between error and exception is necessary, the term 'non-error exception' +is used. + +Exception categories +-------------------- + +Exceptions are described primarily with respect to legacy taskfile + bus +master IDE interface. If a controller provides other better mechanism +for error reporting, mapping those into categories described below +shouldn't be difficult. + +In the following sections, two recovery actions - reset and +reconfiguring transport - are mentioned. These are described further in +`EH recovery actions <#exrec>`__. + +HSM violation +~~~~~~~~~~~~~ + +This error is indicated when STATUS value doesn't match HSM requirement +during issuing or execution any ATA/ATAPI command. + +- ATA_STATUS doesn't contain !BSY && DRDY && !DRQ while trying to + issue a command. + +- !BSY && !DRQ during PIO data transfer. + +- DRQ on command completion. + +- !BSY && ERR after CDB transfer starts but before the last byte of CDB + is transferred. ATA/ATAPI standard states that "The device shall not + terminate the PACKET command with an error before the last byte of + the command packet has been written" in the error outputs description + of PACKET command and the state diagram doesn't include such + transitions. + +In these cases, HSM is violated and not much information regarding the +error can be acquired from STATUS or ERROR register. IOW, this error can +be anything - driver bug, faulty device, controller and/or cable. + +As HSM is violated, reset is necessary to restore known state. +Reconfiguring transport for lower speed might be helpful too as +transmission errors sometimes cause this kind of errors. + +ATA/ATAPI device error (non-NCQ / non-CHECK CONDITION) +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +These are errors detected and reported by ATA/ATAPI devices indicating +device problems. For this type of errors, STATUS and ERROR register +values are valid and describe error condition. Note that some of ATA bus +errors are detected by ATA/ATAPI devices and reported using the same +mechanism as device errors. Those cases are described later in this +section. + +For ATA commands, this type of errors are indicated by !BSY && ERR +during command execution and on completion. + +For ATAPI commands, + +- !BSY && ERR && ABRT right after issuing PACKET indicates that PACKET + command is not supported and falls in this category. + +- !BSY && ERR(==CHK) && !ABRT after the last byte of CDB is transferred + indicates CHECK CONDITION and doesn't fall in this category. + +- !BSY && ERR(==CHK) && ABRT after the last byte of CDB is transferred + \*probably\* indicates CHECK CONDITION and doesn't fall in this + category. + +Of errors detected as above, the following are not ATA/ATAPI device +errors but ATA bus errors and should be handled according to +`ATA bus error <#excatATAbusErr>`__. + +CRC error during data transfer + This is indicated by ICRC bit in the ERROR register and means that + corruption occurred during data transfer. Up to ATA/ATAPI-7, the + standard specifies that this bit is only applicable to UDMA + transfers but ATA/ATAPI-8 draft revision 1f says that the bit may be + applicable to multiword DMA and PIO. + +ABRT error during data transfer or on completion + Up to ATA/ATAPI-7, the standard specifies that ABRT could be set on + ICRC errors and on cases where a device is not able to complete a + command. Combined with the fact that MWDMA and PIO transfer errors + aren't allowed to use ICRC bit up to ATA/ATAPI-7, it seems to imply + that ABRT bit alone could indicate transfer errors. + + However, ATA/ATAPI-8 draft revision 1f removes the part that ICRC + errors can turn on ABRT. So, this is kind of gray area. Some + heuristics are needed here. + +ATA/ATAPI device errors can be further categorized as follows. + +Media errors + This is indicated by UNC bit in the ERROR register. ATA devices + reports UNC error only after certain number of retries cannot + recover the data, so there's nothing much else to do other than + notifying upper layer. + + READ and WRITE commands report CHS or LBA of the first failed sector + but ATA/ATAPI standard specifies that the amount of transferred data + on error completion is indeterminate, so we cannot assume that + sectors preceding the failed sector have been transferred and thus + cannot complete those sectors successfully as SCSI does. + +Media changed / media change requested error + <<TODO: fill here>> + +Address error + This is indicated by IDNF bit in the ERROR register. Report to upper + layer. + +Other errors + This can be invalid command or parameter indicated by ABRT ERROR bit + or some other error condition. Note that ABRT bit can indicate a lot + of things including ICRC and Address errors. Heuristics needed. + +Depending on commands, not all STATUS/ERROR bits are applicable. These +non-applicable bits are marked with "na" in the output descriptions but +up to ATA/ATAPI-7 no definition of "na" can be found. However, +ATA/ATAPI-8 draft revision 1f describes "N/A" as follows. + + 3.2.3.3a N/A + A keyword the indicates a field has no defined value in this + standard and should not be checked by the host or device. N/A + fields should be cleared to zero. + +So, it seems reasonable to assume that "na" bits are cleared to zero by +devices and thus need no explicit masking. + +ATAPI device CHECK CONDITION +~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +ATAPI device CHECK CONDITION error is indicated by set CHK bit (ERR bit) +in the STATUS register after the last byte of CDB is transferred for a +PACKET command. For this kind of errors, sense data should be acquired +to gather information regarding the errors. REQUEST SENSE packet command +should be used to acquire sense data. + +Once sense data is acquired, this type of errors can be handled +similarly to other SCSI errors. Note that sense data may indicate ATA +bus error (e.g. Sense Key 04h HARDWARE ERROR && ASC/ASCQ 47h/00h SCSI +PARITY ERROR). In such cases, the error should be considered as an ATA +bus error and handled according to `ATA bus error <#excatATAbusErr>`__. + +ATA device error (NCQ) +~~~~~~~~~~~~~~~~~~~~~~ + +NCQ command error is indicated by cleared BSY and set ERR bit during NCQ +command phase (one or more NCQ commands outstanding). Although STATUS +and ERROR registers will contain valid values describing the error, READ +LOG EXT is required to clear the error condition, determine which +command has failed and acquire more information. + +READ LOG EXT Log Page 10h reports which tag has failed and taskfile +register values describing the error. With this information the failed +command can be handled as a normal ATA command error as in +`ATA/ATAPI device error (non-NCQ / non-CHECK CONDITION) <#excatDevErr>`__ +and all other in-flight commands must be retried. Note that this retry +should not be counted - it's likely that commands retried this way would +have completed normally if it were not for the failed command. + +Note that ATA bus errors can be reported as ATA device NCQ errors. This +should be handled as described in `ATA bus error <#excatATAbusErr>`__. + +If READ LOG EXT Log Page 10h fails or reports NQ, we're thoroughly +screwed. This condition should be treated according to +`HSM violation <#excatHSMviolation>`__. + +ATA bus error +~~~~~~~~~~~~~ + +ATA bus error means that data corruption occurred during transmission +over ATA bus (SATA or PATA). This type of errors can be indicated by + +- ICRC or ABRT error as described in + `ATA/ATAPI device error (non-NCQ / non-CHECK CONDITION) <#excatDevErr>`__. + +- Controller-specific error completion with error information + indicating transmission error. + +- On some controllers, command timeout. In this case, there may be a + mechanism to determine that the timeout is due to transmission error. + +- Unknown/random errors, timeouts and all sorts of weirdities. + +As described above, transmission errors can cause wide variety of +symptoms ranging from device ICRC error to random device lockup, and, +for many cases, there is no way to tell if an error condition is due to +transmission error or not; therefore, it's necessary to employ some kind +of heuristic when dealing with errors and timeouts. For example, +encountering repetitive ABRT errors for known supported command is +likely to indicate ATA bus error. + +Once it's determined that ATA bus errors have possibly occurred, +lowering ATA bus transmission speed is one of actions which may +alleviate the problem. See `Reconfigure transport <#exrecReconf>`__ for +more information. + +PCI bus error +~~~~~~~~~~~~~ + +Data corruption or other failures during transmission over PCI (or other +system bus). For standard BMDMA, this is indicated by Error bit in the +BMDMA Status register. This type of errors must be logged as it +indicates something is very wrong with the system. Resetting host +controller is recommended. + +Late completion +~~~~~~~~~~~~~~~ + +This occurs when timeout occurs and the timeout handler finds out that +the timed out command has completed successfully or with error. This is +usually caused by lost interrupts. This type of errors must be logged. +Resetting host controller is recommended. + +Unknown error (timeout) +~~~~~~~~~~~~~~~~~~~~~~~ + +This is when timeout occurs and the command is still processing or the +host and device are in unknown state. When this occurs, HSM could be in +any valid or invalid state. To bring the device to known state and make +it forget about the timed out command, resetting is necessary. The timed +out command may be retried. + +Timeouts can also be caused by transmission errors. Refer to +`ATA bus error <#excatATAbusErr>`__ for more details. + +Hotplug and power management exceptions +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +<<TODO: fill here>> + +EH recovery actions +------------------- + +This section discusses several important recovery actions. + +Clearing error condition +~~~~~~~~~~~~~~~~~~~~~~~~ + +Many controllers require its error registers to be cleared by error +handler. Different controllers may have different requirements. + +For SATA, it's strongly recommended to clear at least SError register +during error handling. + +Reset +~~~~~ + +During EH, resetting is necessary in the following cases. + +- HSM is in unknown or invalid state + +- HBA is in unknown or invalid state + +- EH needs to make HBA/device forget about in-flight commands + +- HBA/device behaves weirdly + +Resetting during EH might be a good idea regardless of error condition +to improve EH robustness. Whether to reset both or either one of HBA and +device depends on situation but the following scheme is recommended. + +- When it's known that HBA is in ready state but ATA/ATAPI device is in + unknown state, reset only device. + +- If HBA is in unknown state, reset both HBA and device. + +HBA resetting is implementation specific. For a controller complying to +taskfile/BMDMA PCI IDE, stopping active DMA transaction may be +sufficient iff BMDMA state is the only HBA context. But even mostly +taskfile/BMDMA PCI IDE complying controllers may have implementation +specific requirements and mechanism to reset themselves. This must be +addressed by specific drivers. + +OTOH, ATA/ATAPI standard describes in detail ways to reset ATA/ATAPI +devices. + +PATA hardware reset + This is hardware initiated device reset signalled with asserted PATA + RESET- signal. There is no standard way to initiate hardware reset + from software although some hardware provides registers that allow + driver to directly tweak the RESET- signal. + +Software reset + This is achieved by turning CONTROL SRST bit on for at least 5us. + Both PATA and SATA support it but, in case of SATA, this may require + controller-specific support as the second Register FIS to clear SRST + should be transmitted while BSY bit is still set. Note that on PATA, + this resets both master and slave devices on a channel. + +EXECUTE DEVICE DIAGNOSTIC command + Although ATA/ATAPI standard doesn't describe exactly, EDD implies + some level of resetting, possibly similar level with software reset. + Host-side EDD protocol can be handled with normal command processing + and most SATA controllers should be able to handle EDD's just like + other commands. As in software reset, EDD affects both devices on a + PATA bus. + + Although EDD does reset devices, this doesn't suit error handling as + EDD cannot be issued while BSY is set and it's unclear how it will + act when device is in unknown/weird state. + +ATAPI DEVICE RESET command + This is very similar to software reset except that reset can be + restricted to the selected device without affecting the other device + sharing the cable. + +SATA phy reset + This is the preferred way of resetting a SATA device. In effect, + it's identical to PATA hardware reset. Note that this can be done + with the standard SCR Control register. As such, it's usually easier + to implement than software reset. + +One more thing to consider when resetting devices is that resetting +clears certain configuration parameters and they need to be set to their +previous or newly adjusted values after reset. + +Parameters affected are. + +- CHS set up with INITIALIZE DEVICE PARAMETERS (seldom used) + +- Parameters set with SET FEATURES including transfer mode setting + +- Block count set with SET MULTIPLE MODE + +- Other parameters (SET MAX, MEDIA LOCK...) + +ATA/ATAPI standard specifies that some parameters must be maintained +across hardware or software reset, but doesn't strictly specify all of +them. Always reconfiguring needed parameters after reset is required for +robustness. Note that this also applies when resuming from deep sleep +(power-off). + +Also, ATA/ATAPI standard requires that IDENTIFY DEVICE / IDENTIFY PACKET +DEVICE is issued after any configuration parameter is updated or a +hardware reset and the result used for further operation. OS driver is +required to implement revalidation mechanism to support this. + +Reconfigure transport +~~~~~~~~~~~~~~~~~~~~~ + +For both PATA and SATA, a lot of corners are cut for cheap connectors, +cables or controllers and it's quite common to see high transmission +error rate. This can be mitigated by lowering transmission speed. + +The following is a possible scheme Jeff Garzik suggested. + + If more than $N (3?) transmission errors happen in 15 minutes, + + - if SATA, decrease SATA PHY speed. if speed cannot be decreased, + + - decrease UDMA xfer speed. if at UDMA0, switch to PIO4, + + - decrease PIO xfer speed. if at PIO3, complain, but continue + +ata_piix Internals +=================== + +.. kernel-doc:: drivers/ata/ata_piix.c + :internal: + +sata_sil Internals +=================== + +.. kernel-doc:: drivers/ata/sata_sil.c + :internal: + +Thanks +====== + +The bulk of the ATA knowledge comes thanks to long conversations with +Andre Hedrick (www.linux-ide.org), and long hours pondering the ATA and +SCSI specifications. + +Thanks to Alan Cox for pointing out similarities between SATA and SCSI, +and in general for motivation to hack on libata. + +libata's device detection method, ata_pio_devchk, and in general all +the early probing was based on extensive study of Hale Landis's +probe/reset code in his ATADRVR driver (www.ata-atapi.com). diff --git a/Documentation/driver-api/message-based.rst b/Documentation/driver-api/message-based.rst new file mode 100644 index 000000000..18ff94ef6 --- /dev/null +++ b/Documentation/driver-api/message-based.rst @@ -0,0 +1,12 @@ +Message-based devices +===================== + +Fusion message devices +---------------------- + +.. kernel-doc:: drivers/message/fusion/mptbase.c + :export: + +.. kernel-doc:: drivers/message/fusion/mptscsih.c + :export: + diff --git a/Documentation/driver-api/misc_devices.rst b/Documentation/driver-api/misc_devices.rst new file mode 100644 index 000000000..c7ee7b02b --- /dev/null +++ b/Documentation/driver-api/misc_devices.rst @@ -0,0 +1,5 @@ +Miscellaneous Devices +===================== + +.. kernel-doc:: drivers/char/misc.c + :export: diff --git a/Documentation/driver-api/miscellaneous.rst b/Documentation/driver-api/miscellaneous.rst new file mode 100644 index 000000000..304ffb146 --- /dev/null +++ b/Documentation/driver-api/miscellaneous.rst @@ -0,0 +1,49 @@ +Parallel Port Devices +===================== + +.. kernel-doc:: include/linux/parport.h + :internal: + +.. kernel-doc:: drivers/parport/ieee1284.c + :export: + +.. kernel-doc:: drivers/parport/share.c + :export: + +.. kernel-doc:: drivers/parport/daisy.c + :internal: + +16x50 UART Driver +================= + +.. kernel-doc:: drivers/tty/serial/serial_core.c + :export: + +.. kernel-doc:: drivers/tty/serial/8250/8250_core.c + :export: + +Pulse-Width Modulation (PWM) +============================ + +Pulse-width modulation is a modulation technique primarily used to +control power supplied to electrical devices. + +The PWM framework provides an abstraction for providers and consumers of +PWM signals. A controller that provides one or more PWM signals is +registered as :c:type:`struct pwm_chip <pwm_chip>`. Providers +are expected to embed this structure in a driver-specific structure. +This structure contains fields that describe a particular chip. + +A chip exposes one or more PWM signal sources, each of which exposed as +a :c:type:`struct pwm_device <pwm_device>`. Operations can be +performed on PWM devices to control the period, duty cycle, polarity and +active state of the signal. + +Note that PWM devices are exclusive resources: they can always only be +used by one consumer at a time. + +.. kernel-doc:: include/linux/pwm.h + :internal: + +.. kernel-doc:: drivers/pwm/core.c + :export: diff --git a/Documentation/driver-api/mtdnand.rst b/Documentation/driver-api/mtdnand.rst new file mode 100644 index 000000000..5470a3d6b --- /dev/null +++ b/Documentation/driver-api/mtdnand.rst @@ -0,0 +1,1007 @@ +===================================== +MTD NAND Driver Programming Interface +===================================== + +:Author: Thomas Gleixner + +Introduction +============ + +The generic NAND driver supports almost all NAND and AG-AND based chips +and connects them to the Memory Technology Devices (MTD) subsystem of +the Linux Kernel. + +This documentation is provided for developers who want to implement +board drivers or filesystem drivers suitable for NAND devices. + +Known Bugs And Assumptions +========================== + +None. + +Documentation hints +=================== + +The function and structure docs are autogenerated. Each function and +struct member has a short description which is marked with an [XXX] +identifier. The following chapters explain the meaning of those +identifiers. + +Function identifiers [XXX] +-------------------------- + +The functions are marked with [XXX] identifiers in the short comment. +The identifiers explain the usage and scope of the functions. Following +identifiers are used: + +- [MTD Interface] + + These functions provide the interface to the MTD kernel API. They are + not replaceable and provide functionality which is complete hardware + independent. + +- [NAND Interface] + + These functions are exported and provide the interface to the NAND + kernel API. + +- [GENERIC] + + Generic functions are not replaceable and provide functionality which + is complete hardware independent. + +- [DEFAULT] + + Default functions provide hardware related functionality which is + suitable for most of the implementations. These functions can be + replaced by the board driver if necessary. Those functions are called + via pointers in the NAND chip description structure. The board driver + can set the functions which should be replaced by board dependent + functions before calling nand_scan(). If the function pointer is + NULL on entry to nand_scan() then the pointer is set to the default + function which is suitable for the detected chip type. + +Struct member identifiers [XXX] +------------------------------- + +The struct members are marked with [XXX] identifiers in the comment. The +identifiers explain the usage and scope of the members. Following +identifiers are used: + +- [INTERN] + + These members are for NAND driver internal use only and must not be + modified. Most of these values are calculated from the chip geometry + information which is evaluated during nand_scan(). + +- [REPLACEABLE] + + Replaceable members hold hardware related functions which can be + provided by the board driver. The board driver can set the functions + which should be replaced by board dependent functions before calling + nand_scan(). If the function pointer is NULL on entry to + nand_scan() then the pointer is set to the default function which is + suitable for the detected chip type. + +- [BOARDSPECIFIC] + + Board specific members hold hardware related information which must + be provided by the board driver. The board driver must set the + function pointers and datafields before calling nand_scan(). + +- [OPTIONAL] + + Optional members can hold information relevant for the board driver. + The generic NAND driver code does not use this information. + +Basic board driver +================== + +For most boards it will be sufficient to provide just the basic +functions and fill out some really board dependent members in the nand +chip description structure. + +Basic defines +------------- + +At least you have to provide a nand_chip structure and a storage for +the ioremap'ed chip address. You can allocate the nand_chip structure +using kmalloc or you can allocate it statically. The NAND chip structure +embeds an mtd structure which will be registered to the MTD subsystem. +You can extract a pointer to the mtd structure from a nand_chip pointer +using the nand_to_mtd() helper. + +Kmalloc based example + +:: + + static struct mtd_info *board_mtd; + static void __iomem *baseaddr; + + +Static example + +:: + + static struct nand_chip board_chip; + static void __iomem *baseaddr; + + +Partition defines +----------------- + +If you want to divide your device into partitions, then define a +partitioning scheme suitable to your board. + +:: + + #define NUM_PARTITIONS 2 + static struct mtd_partition partition_info[] = { + { .name = "Flash partition 1", + .offset = 0, + .size = 8 * 1024 * 1024 }, + { .name = "Flash partition 2", + .offset = MTDPART_OFS_NEXT, + .size = MTDPART_SIZ_FULL }, + }; + + +Hardware control function +------------------------- + +The hardware control function provides access to the control pins of the +NAND chip(s). The access can be done by GPIO pins or by address lines. +If you use address lines, make sure that the timing requirements are +met. + +*GPIO based example* + +:: + + static void board_hwcontrol(struct mtd_info *mtd, int cmd) + { + switch(cmd){ + case NAND_CTL_SETCLE: /* Set CLE pin high */ break; + case NAND_CTL_CLRCLE: /* Set CLE pin low */ break; + case NAND_CTL_SETALE: /* Set ALE pin high */ break; + case NAND_CTL_CLRALE: /* Set ALE pin low */ break; + case NAND_CTL_SETNCE: /* Set nCE pin low */ break; + case NAND_CTL_CLRNCE: /* Set nCE pin high */ break; + } + } + + +*Address lines based example.* It's assumed that the nCE pin is driven +by a chip select decoder. + +:: + + static void board_hwcontrol(struct mtd_info *mtd, int cmd) + { + struct nand_chip *this = mtd_to_nand(mtd); + switch(cmd){ + case NAND_CTL_SETCLE: this->IO_ADDR_W |= CLE_ADRR_BIT; break; + case NAND_CTL_CLRCLE: this->IO_ADDR_W &= ~CLE_ADRR_BIT; break; + case NAND_CTL_SETALE: this->IO_ADDR_W |= ALE_ADRR_BIT; break; + case NAND_CTL_CLRALE: this->IO_ADDR_W &= ~ALE_ADRR_BIT; break; + } + } + + +Device ready function +--------------------- + +If the hardware interface has the ready busy pin of the NAND chip +connected to a GPIO or other accessible I/O pin, this function is used +to read back the state of the pin. The function has no arguments and +should return 0, if the device is busy (R/B pin is low) and 1, if the +device is ready (R/B pin is high). If the hardware interface does not +give access to the ready busy pin, then the function must not be defined +and the function pointer this->dev_ready is set to NULL. + +Init function +------------- + +The init function allocates memory and sets up all the board specific +parameters and function pointers. When everything is set up nand_scan() +is called. This function tries to detect and identify then chip. If a +chip is found all the internal data fields are initialized accordingly. +The structure(s) have to be zeroed out first and then filled with the +necessary information about the device. + +:: + + static int __init board_init (void) + { + struct nand_chip *this; + int err = 0; + + /* Allocate memory for MTD device structure and private data */ + this = kzalloc(sizeof(struct nand_chip), GFP_KERNEL); + if (!this) { + printk ("Unable to allocate NAND MTD device structure.\n"); + err = -ENOMEM; + goto out; + } + + board_mtd = nand_to_mtd(this); + + /* map physical address */ + baseaddr = ioremap(CHIP_PHYSICAL_ADDRESS, 1024); + if (!baseaddr) { + printk("Ioremap to access NAND chip failed\n"); + err = -EIO; + goto out_mtd; + } + + /* Set address of NAND IO lines */ + this->IO_ADDR_R = baseaddr; + this->IO_ADDR_W = baseaddr; + /* Reference hardware control function */ + this->hwcontrol = board_hwcontrol; + /* Set command delay time, see datasheet for correct value */ + this->chip_delay = CHIP_DEPENDEND_COMMAND_DELAY; + /* Assign the device ready function, if available */ + this->dev_ready = board_dev_ready; + this->eccmode = NAND_ECC_SOFT; + + /* Scan to find existence of the device */ + if (nand_scan (this, 1)) { + err = -ENXIO; + goto out_ior; + } + + add_mtd_partitions(board_mtd, partition_info, NUM_PARTITIONS); + goto out; + + out_ior: + iounmap(baseaddr); + out_mtd: + kfree (this); + out: + return err; + } + module_init(board_init); + + +Exit function +------------- + +The exit function is only necessary if the driver is compiled as a +module. It releases all resources which are held by the chip driver and +unregisters the partitions in the MTD layer. + +:: + + #ifdef MODULE + static void __exit board_cleanup (void) + { + /* Release resources, unregister device */ + nand_release (mtd_to_nand(board_mtd)); + + /* unmap physical address */ + iounmap(baseaddr); + + /* Free the MTD device structure */ + kfree (mtd_to_nand(board_mtd)); + } + module_exit(board_cleanup); + #endif + + +Advanced board driver functions +=============================== + +This chapter describes the advanced functionality of the NAND driver. +For a list of functions which can be overridden by the board driver see +the documentation of the nand_chip structure. + +Multiple chip control +--------------------- + +The nand driver can control chip arrays. Therefore the board driver must +provide an own select_chip function. This function must (de)select the +requested chip. The function pointer in the nand_chip structure must be +set before calling nand_scan(). The maxchip parameter of nand_scan() +defines the maximum number of chips to scan for. Make sure that the +select_chip function can handle the requested number of chips. + +The nand driver concatenates the chips to one virtual chip and provides +this virtual chip to the MTD layer. + +*Note: The driver can only handle linear chip arrays of equally sized +chips. There is no support for parallel arrays which extend the +buswidth.* + +*GPIO based example* + +:: + + static void board_select_chip (struct mtd_info *mtd, int chip) + { + /* Deselect all chips, set all nCE pins high */ + GPIO(BOARD_NAND_NCE) |= 0xff; + if (chip >= 0) + GPIO(BOARD_NAND_NCE) &= ~ (1 << chip); + } + + +*Address lines based example.* Its assumed that the nCE pins are +connected to an address decoder. + +:: + + static void board_select_chip (struct mtd_info *mtd, int chip) + { + struct nand_chip *this = mtd_to_nand(mtd); + + /* Deselect all chips */ + this->IO_ADDR_R &= ~BOARD_NAND_ADDR_MASK; + this->IO_ADDR_W &= ~BOARD_NAND_ADDR_MASK; + switch (chip) { + case 0: + this->IO_ADDR_R |= BOARD_NAND_ADDR_CHIP0; + this->IO_ADDR_W |= BOARD_NAND_ADDR_CHIP0; + break; + .... + case n: + this->IO_ADDR_R |= BOARD_NAND_ADDR_CHIPn; + this->IO_ADDR_W |= BOARD_NAND_ADDR_CHIPn; + break; + } + } + + +Hardware ECC support +-------------------- + +Functions and constants +~~~~~~~~~~~~~~~~~~~~~~~ + +The nand driver supports three different types of hardware ECC. + +- NAND_ECC_HW3_256 + + Hardware ECC generator providing 3 bytes ECC per 256 byte. + +- NAND_ECC_HW3_512 + + Hardware ECC generator providing 3 bytes ECC per 512 byte. + +- NAND_ECC_HW6_512 + + Hardware ECC generator providing 6 bytes ECC per 512 byte. + +- NAND_ECC_HW8_512 + + Hardware ECC generator providing 8 bytes ECC per 512 byte. + +If your hardware generator has a different functionality add it at the +appropriate place in nand_base.c + +The board driver must provide following functions: + +- enable_hwecc + + This function is called before reading / writing to the chip. Reset + or initialize the hardware generator in this function. The function + is called with an argument which let you distinguish between read and + write operations. + +- calculate_ecc + + This function is called after read / write from / to the chip. + Transfer the ECC from the hardware to the buffer. If the option + NAND_HWECC_SYNDROME is set then the function is only called on + write. See below. + +- correct_data + + In case of an ECC error this function is called for error detection + and correction. Return 1 respectively 2 in case the error can be + corrected. If the error is not correctable return -1. If your + hardware generator matches the default algorithm of the nand_ecc + software generator then use the correction function provided by + nand_ecc instead of implementing duplicated code. + +Hardware ECC with syndrome calculation +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Many hardware ECC implementations provide Reed-Solomon codes and +calculate an error syndrome on read. The syndrome must be converted to a +standard Reed-Solomon syndrome before calling the error correction code +in the generic Reed-Solomon library. + +The ECC bytes must be placed immediately after the data bytes in order +to make the syndrome generator work. This is contrary to the usual +layout used by software ECC. The separation of data and out of band area +is not longer possible. The nand driver code handles this layout and the +remaining free bytes in the oob area are managed by the autoplacement +code. Provide a matching oob-layout in this case. See rts_from4.c and +diskonchip.c for implementation reference. In those cases we must also +use bad block tables on FLASH, because the ECC layout is interfering +with the bad block marker positions. See bad block table support for +details. + +Bad block table support +----------------------- + +Most NAND chips mark the bad blocks at a defined position in the spare +area. Those blocks must not be erased under any circumstances as the bad +block information would be lost. It is possible to check the bad block +mark each time when the blocks are accessed by reading the spare area of +the first page in the block. This is time consuming so a bad block table +is used. + +The nand driver supports various types of bad block tables. + +- Per device + + The bad block table contains all bad block information of the device + which can consist of multiple chips. + +- Per chip + + A bad block table is used per chip and contains the bad block + information for this particular chip. + +- Fixed offset + + The bad block table is located at a fixed offset in the chip + (device). This applies to various DiskOnChip devices. + +- Automatic placed + + The bad block table is automatically placed and detected either at + the end or at the beginning of a chip (device) + +- Mirrored tables + + The bad block table is mirrored on the chip (device) to allow updates + of the bad block table without data loss. + +nand_scan() calls the function nand_default_bbt(). +nand_default_bbt() selects appropriate default bad block table +descriptors depending on the chip information which was retrieved by +nand_scan(). + +The standard policy is scanning the device for bad blocks and build a +ram based bad block table which allows faster access than always +checking the bad block information on the flash chip itself. + +Flash based tables +~~~~~~~~~~~~~~~~~~ + +It may be desired or necessary to keep a bad block table in FLASH. For +AG-AND chips this is mandatory, as they have no factory marked bad +blocks. They have factory marked good blocks. The marker pattern is +erased when the block is erased to be reused. So in case of powerloss +before writing the pattern back to the chip this block would be lost and +added to the bad blocks. Therefore we scan the chip(s) when we detect +them the first time for good blocks and store this information in a bad +block table before erasing any of the blocks. + +The blocks in which the tables are stored are protected against +accidental access by marking them bad in the memory bad block table. The +bad block table management functions are allowed to circumvent this +protection. + +The simplest way to activate the FLASH based bad block table support is +to set the option NAND_BBT_USE_FLASH in the bbt_option field of the +nand chip structure before calling nand_scan(). For AG-AND chips is +this done by default. This activates the default FLASH based bad block +table functionality of the NAND driver. The default bad block table +options are + +- Store bad block table per chip + +- Use 2 bits per block + +- Automatic placement at the end of the chip + +- Use mirrored tables with version numbers + +- Reserve 4 blocks at the end of the chip + +User defined tables +~~~~~~~~~~~~~~~~~~~ + +User defined tables are created by filling out a nand_bbt_descr +structure and storing the pointer in the nand_chip structure member +bbt_td before calling nand_scan(). If a mirror table is necessary a +second structure must be created and a pointer to this structure must be +stored in bbt_md inside the nand_chip structure. If the bbt_md member +is set to NULL then only the main table is used and no scan for the +mirrored table is performed. + +The most important field in the nand_bbt_descr structure is the +options field. The options define most of the table properties. Use the +predefined constants from rawnand.h to define the options. + +- Number of bits per block + + The supported number of bits is 1, 2, 4, 8. + +- Table per chip + + Setting the constant NAND_BBT_PERCHIP selects that a bad block + table is managed for each chip in a chip array. If this option is not + set then a per device bad block table is used. + +- Table location is absolute + + Use the option constant NAND_BBT_ABSPAGE and define the absolute + page number where the bad block table starts in the field pages. If + you have selected bad block tables per chip and you have a multi chip + array then the start page must be given for each chip in the chip + array. Note: there is no scan for a table ident pattern performed, so + the fields pattern, veroffs, offs, len can be left uninitialized + +- Table location is automatically detected + + The table can either be located in the first or the last good blocks + of the chip (device). Set NAND_BBT_LASTBLOCK to place the bad block + table at the end of the chip (device). The bad block tables are + marked and identified by a pattern which is stored in the spare area + of the first page in the block which holds the bad block table. Store + a pointer to the pattern in the pattern field. Further the length of + the pattern has to be stored in len and the offset in the spare area + must be given in the offs member of the nand_bbt_descr structure. + For mirrored bad block tables different patterns are mandatory. + +- Table creation + + Set the option NAND_BBT_CREATE to enable the table creation if no + table can be found during the scan. Usually this is done only once if + a new chip is found. + +- Table write support + + Set the option NAND_BBT_WRITE to enable the table write support. + This allows the update of the bad block table(s) in case a block has + to be marked bad due to wear. The MTD interface function + block_markbad is calling the update function of the bad block table. + If the write support is enabled then the table is updated on FLASH. + + Note: Write support should only be enabled for mirrored tables with + version control. + +- Table version control + + Set the option NAND_BBT_VERSION to enable the table version + control. It's highly recommended to enable this for mirrored tables + with write support. It makes sure that the risk of losing the bad + block table information is reduced to the loss of the information + about the one worn out block which should be marked bad. The version + is stored in 4 consecutive bytes in the spare area of the device. The + position of the version number is defined by the member veroffs in + the bad block table descriptor. + +- Save block contents on write + + In case that the block which holds the bad block table does contain + other useful information, set the option NAND_BBT_SAVECONTENT. When + the bad block table is written then the whole block is read the bad + block table is updated and the block is erased and everything is + written back. If this option is not set only the bad block table is + written and everything else in the block is ignored and erased. + +- Number of reserved blocks + + For automatic placement some blocks must be reserved for bad block + table storage. The number of reserved blocks is defined in the + maxblocks member of the bad block table description structure. + Reserving 4 blocks for mirrored tables should be a reasonable number. + This also limits the number of blocks which are scanned for the bad + block table ident pattern. + +Spare area (auto)placement +-------------------------- + +The nand driver implements different possibilities for placement of +filesystem data in the spare area, + +- Placement defined by fs driver + +- Automatic placement + +The default placement function is automatic placement. The nand driver +has built in default placement schemes for the various chiptypes. If due +to hardware ECC functionality the default placement does not fit then +the board driver can provide a own placement scheme. + +File system drivers can provide a own placement scheme which is used +instead of the default placement scheme. + +Placement schemes are defined by a nand_oobinfo structure + +:: + + struct nand_oobinfo { + int useecc; + int eccbytes; + int eccpos[24]; + int oobfree[8][2]; + }; + + +- useecc + + The useecc member controls the ecc and placement function. The header + file include/mtd/mtd-abi.h contains constants to select ecc and + placement. MTD_NANDECC_OFF switches off the ecc complete. This is + not recommended and available for testing and diagnosis only. + MTD_NANDECC_PLACE selects caller defined placement, + MTD_NANDECC_AUTOPLACE selects automatic placement. + +- eccbytes + + The eccbytes member defines the number of ecc bytes per page. + +- eccpos + + The eccpos array holds the byte offsets in the spare area where the + ecc codes are placed. + +- oobfree + + The oobfree array defines the areas in the spare area which can be + used for automatic placement. The information is given in the format + {offset, size}. offset defines the start of the usable area, size the + length in bytes. More than one area can be defined. The list is + terminated by an {0, 0} entry. + +Placement defined by fs driver +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The calling function provides a pointer to a nand_oobinfo structure +which defines the ecc placement. For writes the caller must provide a +spare area buffer along with the data buffer. The spare area buffer size +is (number of pages) \* (size of spare area). For reads the buffer size +is (number of pages) \* ((size of spare area) + (number of ecc steps per +page) \* sizeof (int)). The driver stores the result of the ecc check +for each tuple in the spare buffer. The storage sequence is:: + + <spare data page 0><ecc result 0>...<ecc result n> + + ... + + <spare data page n><ecc result 0>...<ecc result n> + +This is a legacy mode used by YAFFS1. + +If the spare area buffer is NULL then only the ECC placement is done +according to the given scheme in the nand_oobinfo structure. + +Automatic placement +~~~~~~~~~~~~~~~~~~~ + +Automatic placement uses the built in defaults to place the ecc bytes in +the spare area. If filesystem data have to be stored / read into the +spare area then the calling function must provide a buffer. The buffer +size per page is determined by the oobfree array in the nand_oobinfo +structure. + +If the spare area buffer is NULL then only the ECC placement is done +according to the default builtin scheme. + +Spare area autoplacement default schemes +---------------------------------------- + +256 byte pagesize +~~~~~~~~~~~~~~~~~ + +======== ================== =================================================== +Offset Content Comment +======== ================== =================================================== +0x00 ECC byte 0 Error correction code byte 0 +0x01 ECC byte 1 Error correction code byte 1 +0x02 ECC byte 2 Error correction code byte 2 +0x03 Autoplace 0 +0x04 Autoplace 1 +0x05 Bad block marker If any bit in this byte is zero, then this + block is bad. This applies only to the first + page in a block. In the remaining pages this + byte is reserved +0x06 Autoplace 2 +0x07 Autoplace 3 +======== ================== =================================================== + +512 byte pagesize +~~~~~~~~~~~~~~~~~ + + +============= ================== ============================================== +Offset Content Comment +============= ================== ============================================== +0x00 ECC byte 0 Error correction code byte 0 of the lower + 256 Byte data in this page +0x01 ECC byte 1 Error correction code byte 1 of the lower + 256 Bytes of data in this page +0x02 ECC byte 2 Error correction code byte 2 of the lower + 256 Bytes of data in this page +0x03 ECC byte 3 Error correction code byte 0 of the upper + 256 Bytes of data in this page +0x04 reserved reserved +0x05 Bad block marker If any bit in this byte is zero, then this + block is bad. This applies only to the first + page in a block. In the remaining pages this + byte is reserved +0x06 ECC byte 4 Error correction code byte 1 of the upper + 256 Bytes of data in this page +0x07 ECC byte 5 Error correction code byte 2 of the upper + 256 Bytes of data in this page +0x08 - 0x0F Autoplace 0 - 7 +============= ================== ============================================== + +2048 byte pagesize +~~~~~~~~~~~~~~~~~~ + +=========== ================== ================================================ +Offset Content Comment +=========== ================== ================================================ +0x00 Bad block marker If any bit in this byte is zero, then this block + is bad. This applies only to the first page in a + block. In the remaining pages this byte is + reserved +0x01 Reserved Reserved +0x02-0x27 Autoplace 0 - 37 +0x28 ECC byte 0 Error correction code byte 0 of the first + 256 Byte data in this page +0x29 ECC byte 1 Error correction code byte 1 of the first + 256 Bytes of data in this page +0x2A ECC byte 2 Error correction code byte 2 of the first + 256 Bytes data in this page +0x2B ECC byte 3 Error correction code byte 0 of the second + 256 Bytes of data in this page +0x2C ECC byte 4 Error correction code byte 1 of the second + 256 Bytes of data in this page +0x2D ECC byte 5 Error correction code byte 2 of the second + 256 Bytes of data in this page +0x2E ECC byte 6 Error correction code byte 0 of the third + 256 Bytes of data in this page +0x2F ECC byte 7 Error correction code byte 1 of the third + 256 Bytes of data in this page +0x30 ECC byte 8 Error correction code byte 2 of the third + 256 Bytes of data in this page +0x31 ECC byte 9 Error correction code byte 0 of the fourth + 256 Bytes of data in this page +0x32 ECC byte 10 Error correction code byte 1 of the fourth + 256 Bytes of data in this page +0x33 ECC byte 11 Error correction code byte 2 of the fourth + 256 Bytes of data in this page +0x34 ECC byte 12 Error correction code byte 0 of the fifth + 256 Bytes of data in this page +0x35 ECC byte 13 Error correction code byte 1 of the fifth + 256 Bytes of data in this page +0x36 ECC byte 14 Error correction code byte 2 of the fifth + 256 Bytes of data in this page +0x37 ECC byte 15 Error correction code byte 0 of the sixth + 256 Bytes of data in this page +0x38 ECC byte 16 Error correction code byte 1 of the sixth + 256 Bytes of data in this page +0x39 ECC byte 17 Error correction code byte 2 of the sixth + 256 Bytes of data in this page +0x3A ECC byte 18 Error correction code byte 0 of the seventh + 256 Bytes of data in this page +0x3B ECC byte 19 Error correction code byte 1 of the seventh + 256 Bytes of data in this page +0x3C ECC byte 20 Error correction code byte 2 of the seventh + 256 Bytes of data in this page +0x3D ECC byte 21 Error correction code byte 0 of the eighth + 256 Bytes of data in this page +0x3E ECC byte 22 Error correction code byte 1 of the eighth + 256 Bytes of data in this page +0x3F ECC byte 23 Error correction code byte 2 of the eighth + 256 Bytes of data in this page +=========== ================== ================================================ + +Filesystem support +================== + +The NAND driver provides all necessary functions for a filesystem via +the MTD interface. + +Filesystems must be aware of the NAND peculiarities and restrictions. +One major restrictions of NAND Flash is, that you cannot write as often +as you want to a page. The consecutive writes to a page, before erasing +it again, are restricted to 1-3 writes, depending on the manufacturers +specifications. This applies similar to the spare area. + +Therefore NAND aware filesystems must either write in page size chunks +or hold a writebuffer to collect smaller writes until they sum up to +pagesize. Available NAND aware filesystems: JFFS2, YAFFS. + +The spare area usage to store filesystem data is controlled by the spare +area placement functionality which is described in one of the earlier +chapters. + +Tools +===== + +The MTD project provides a couple of helpful tools to handle NAND Flash. + +- flasherase, flasheraseall: Erase and format FLASH partitions + +- nandwrite: write filesystem images to NAND FLASH + +- nanddump: dump the contents of a NAND FLASH partitions + +These tools are aware of the NAND restrictions. Please use those tools +instead of complaining about errors which are caused by non NAND aware +access methods. + +Constants +========= + +This chapter describes the constants which might be relevant for a +driver developer. + +Chip option constants +--------------------- + +Constants for chip id table +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +These constants are defined in rawnand.h. They are OR-ed together to +describe the chip functionality:: + + /* Buswitdh is 16 bit */ + #define NAND_BUSWIDTH_16 0x00000002 + /* Device supports partial programming without padding */ + #define NAND_NO_PADDING 0x00000004 + /* Chip has cache program function */ + #define NAND_CACHEPRG 0x00000008 + /* Chip has copy back function */ + #define NAND_COPYBACK 0x00000010 + /* AND Chip which has 4 banks and a confusing page / block + * assignment. See Renesas datasheet for further information */ + #define NAND_IS_AND 0x00000020 + /* Chip has a array of 4 pages which can be read without + * additional ready /busy waits */ + #define NAND_4PAGE_ARRAY 0x00000040 + + +Constants for runtime options +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +These constants are defined in rawnand.h. They are OR-ed together to +describe the functionality:: + + /* The hw ecc generator provides a syndrome instead a ecc value on read + * This can only work if we have the ecc bytes directly behind the + * data bytes. Applies for DOC and AG-AND Renesas HW Reed Solomon generators */ + #define NAND_HWECC_SYNDROME 0x00020000 + + +ECC selection constants +----------------------- + +Use these constants to select the ECC algorithm:: + + /* No ECC. Usage is not recommended ! */ + #define NAND_ECC_NONE 0 + /* Software ECC 3 byte ECC per 256 Byte data */ + #define NAND_ECC_SOFT 1 + /* Hardware ECC 3 byte ECC per 256 Byte data */ + #define NAND_ECC_HW3_256 2 + /* Hardware ECC 3 byte ECC per 512 Byte data */ + #define NAND_ECC_HW3_512 3 + /* Hardware ECC 6 byte ECC per 512 Byte data */ + #define NAND_ECC_HW6_512 4 + /* Hardware ECC 8 byte ECC per 512 Byte data */ + #define NAND_ECC_HW8_512 6 + + +Hardware control related constants +---------------------------------- + +These constants describe the requested hardware access function when the +boardspecific hardware control function is called:: + + /* Select the chip by setting nCE to low */ + #define NAND_CTL_SETNCE 1 + /* Deselect the chip by setting nCE to high */ + #define NAND_CTL_CLRNCE 2 + /* Select the command latch by setting CLE to high */ + #define NAND_CTL_SETCLE 3 + /* Deselect the command latch by setting CLE to low */ + #define NAND_CTL_CLRCLE 4 + /* Select the address latch by setting ALE to high */ + #define NAND_CTL_SETALE 5 + /* Deselect the address latch by setting ALE to low */ + #define NAND_CTL_CLRALE 6 + /* Set write protection by setting WP to high. Not used! */ + #define NAND_CTL_SETWP 7 + /* Clear write protection by setting WP to low. Not used! */ + #define NAND_CTL_CLRWP 8 + + +Bad block table related constants +--------------------------------- + +These constants describe the options used for bad block table +descriptors:: + + /* Options for the bad block table descriptors */ + + /* The number of bits used per block in the bbt on the device */ + #define NAND_BBT_NRBITS_MSK 0x0000000F + #define NAND_BBT_1BIT 0x00000001 + #define NAND_BBT_2BIT 0x00000002 + #define NAND_BBT_4BIT 0x00000004 + #define NAND_BBT_8BIT 0x00000008 + /* The bad block table is in the last good block of the device */ + #define NAND_BBT_LASTBLOCK 0x00000010 + /* The bbt is at the given page, else we must scan for the bbt */ + #define NAND_BBT_ABSPAGE 0x00000020 + /* bbt is stored per chip on multichip devices */ + #define NAND_BBT_PERCHIP 0x00000080 + /* bbt has a version counter at offset veroffs */ + #define NAND_BBT_VERSION 0x00000100 + /* Create a bbt if none axists */ + #define NAND_BBT_CREATE 0x00000200 + /* Write bbt if necessary */ + #define NAND_BBT_WRITE 0x00001000 + /* Read and write back block contents when writing bbt */ + #define NAND_BBT_SAVECONTENT 0x00002000 + + +Structures +========== + +This chapter contains the autogenerated documentation of the structures +which are used in the NAND driver and might be relevant for a driver +developer. Each struct member has a short description which is marked +with an [XXX] identifier. See the chapter "Documentation hints" for an +explanation. + +.. kernel-doc:: include/linux/mtd/rawnand.h + :internal: + +Public Functions Provided +========================= + +This chapter contains the autogenerated documentation of the NAND kernel +API functions which are exported. Each function has a short description +which is marked with an [XXX] identifier. See the chapter "Documentation +hints" for an explanation. + +.. kernel-doc:: drivers/mtd/nand/raw/nand_base.c + :export: + +.. kernel-doc:: drivers/mtd/nand/raw/nand_ecc.c + :export: + +Internal Functions Provided +=========================== + +This chapter contains the autogenerated documentation of the NAND driver +internal functions. Each function has a short description which is +marked with an [XXX] identifier. See the chapter "Documentation hints" +for an explanation. The functions marked with [DEFAULT] might be +relevant for a board driver developer. + +.. kernel-doc:: drivers/mtd/nand/raw/nand_base.c + :internal: + +.. kernel-doc:: drivers/mtd/nand/raw/nand_bbt.c + :internal: + +Credits +======= + +The following people have contributed to the NAND driver: + +1. Steven J. Hill\ sjhill@realitydiluted.com + +2. David Woodhouse\ dwmw2@infradead.org + +3. Thomas Gleixner\ tglx@linutronix.de + +A lot of users have provided bugfixes, improvements and helping hands +for testing. Thanks a lot. + +The following people have contributed to this document: + +1. Thomas Gleixner\ tglx@linutronix.de diff --git a/Documentation/driver-api/pci.rst b/Documentation/driver-api/pci.rst new file mode 100644 index 000000000..ca85e5e78 --- /dev/null +++ b/Documentation/driver-api/pci.rst @@ -0,0 +1,47 @@ +PCI Support Library +------------------- + +.. kernel-doc:: drivers/pci/pci.c + :export: + +.. kernel-doc:: drivers/pci/pci-driver.c + :export: + +.. kernel-doc:: drivers/pci/remove.c + :export: + +.. kernel-doc:: drivers/pci/search.c + :export: + +.. kernel-doc:: drivers/pci/msi.c + :export: + +.. kernel-doc:: drivers/pci/bus.c + :export: + +.. kernel-doc:: drivers/pci/access.c + :export: + +.. kernel-doc:: drivers/pci/irq.c + :export: + +.. kernel-doc:: drivers/pci/probe.c + :export: + +.. kernel-doc:: drivers/pci/slot.c + :export: + +.. kernel-doc:: drivers/pci/rom.c + :export: + +.. kernel-doc:: drivers/pci/iov.c + :export: + +.. kernel-doc:: drivers/pci/pci-sysfs.c + :internal: + +PCI Hotplug Support Library +--------------------------- + +.. kernel-doc:: drivers/pci/hotplug/pci_hotplug_core.c + :export: diff --git a/Documentation/driver-api/pinctl.rst b/Documentation/driver-api/pinctl.rst new file mode 100644 index 000000000..6cb68d67f --- /dev/null +++ b/Documentation/driver-api/pinctl.rst @@ -0,0 +1,1439 @@ +=============================== +PINCTRL (PIN CONTROL) subsystem +=============================== + +This document outlines the pin control subsystem in Linux + +This subsystem deals with: + +- Enumerating and naming controllable pins + +- Multiplexing of pins, pads, fingers (etc) see below for details + +- Configuration of pins, pads, fingers (etc), such as software-controlled + biasing and driving mode specific pins, such as pull-up/down, open drain, + load capacitance etc. + +Top-level interface +=================== + +Definition of PIN CONTROLLER: + +- A pin controller is a piece of hardware, usually a set of registers, that + can control PINs. It may be able to multiplex, bias, set load capacitance, + set drive strength, etc. for individual pins or groups of pins. + +Definition of PIN: + +- PINS are equal to pads, fingers, balls or whatever packaging input or + output line you want to control and these are denoted by unsigned integers + in the range 0..maxpin. This numberspace is local to each PIN CONTROLLER, so + there may be several such number spaces in a system. This pin space may + be sparse - i.e. there may be gaps in the space with numbers where no + pin exists. + +When a PIN CONTROLLER is instantiated, it will register a descriptor to the +pin control framework, and this descriptor contains an array of pin descriptors +describing the pins handled by this specific pin controller. + +Here is an example of a PGA (Pin Grid Array) chip seen from underneath:: + + A B C D E F G H + + 8 o o o o o o o o + + 7 o o o o o o o o + + 6 o o o o o o o o + + 5 o o o o o o o o + + 4 o o o o o o o o + + 3 o o o o o o o o + + 2 o o o o o o o o + + 1 o o o o o o o o + +To register a pin controller and name all the pins on this package we can do +this in our driver:: + + #include <linux/pinctrl/pinctrl.h> + + const struct pinctrl_pin_desc foo_pins[] = { + PINCTRL_PIN(0, "A8"), + PINCTRL_PIN(1, "B8"), + PINCTRL_PIN(2, "C8"), + ... + PINCTRL_PIN(61, "F1"), + PINCTRL_PIN(62, "G1"), + PINCTRL_PIN(63, "H1"), + }; + + static struct pinctrl_desc foo_desc = { + .name = "foo", + .pins = foo_pins, + .npins = ARRAY_SIZE(foo_pins), + .owner = THIS_MODULE, + }; + + int __init foo_probe(void) + { + int error; + + struct pinctrl_dev *pctl; + + error = pinctrl_register_and_init(&foo_desc, <PARENT>, + NULL, &pctl); + if (error) + return error; + + return pinctrl_enable(pctl); + } + +To enable the pinctrl subsystem and the subgroups for PINMUX and PINCONF and +selected drivers, you need to select them from your machine's Kconfig entry, +since these are so tightly integrated with the machines they are used on. +See for example arch/arm/mach-u300/Kconfig for an example. + +Pins usually have fancier names than this. You can find these in the datasheet +for your chip. Notice that the core pinctrl.h file provides a fancy macro +called PINCTRL_PIN() to create the struct entries. As you can see I enumerated +the pins from 0 in the upper left corner to 63 in the lower right corner. +This enumeration was arbitrarily chosen, in practice you need to think +through your numbering system so that it matches the layout of registers +and such things in your driver, or the code may become complicated. You must +also consider matching of offsets to the GPIO ranges that may be handled by +the pin controller. + +For a padring with 467 pads, as opposed to actual pins, I used an enumeration +like this, walking around the edge of the chip, which seems to be industry +standard too (all these pads had names, too):: + + + 0 ..... 104 + 466 105 + . . + . . + 358 224 + 357 .... 225 + + +Pin groups +========== + +Many controllers need to deal with groups of pins, so the pin controller +subsystem has a mechanism for enumerating groups of pins and retrieving the +actual enumerated pins that are part of a certain group. + +For example, say that we have a group of pins dealing with an SPI interface +on { 0, 8, 16, 24 }, and a group of pins dealing with an I2C interface on pins +on { 24, 25 }. + +These two groups are presented to the pin control subsystem by implementing +some generic pinctrl_ops like this:: + + #include <linux/pinctrl/pinctrl.h> + + struct foo_group { + const char *name; + const unsigned int *pins; + const unsigned num_pins; + }; + + static const unsigned int spi0_pins[] = { 0, 8, 16, 24 }; + static const unsigned int i2c0_pins[] = { 24, 25 }; + + static const struct foo_group foo_groups[] = { + { + .name = "spi0_grp", + .pins = spi0_pins, + .num_pins = ARRAY_SIZE(spi0_pins), + }, + { + .name = "i2c0_grp", + .pins = i2c0_pins, + .num_pins = ARRAY_SIZE(i2c0_pins), + }, + }; + + + static int foo_get_groups_count(struct pinctrl_dev *pctldev) + { + return ARRAY_SIZE(foo_groups); + } + + static const char *foo_get_group_name(struct pinctrl_dev *pctldev, + unsigned selector) + { + return foo_groups[selector].name; + } + + static int foo_get_group_pins(struct pinctrl_dev *pctldev, unsigned selector, + const unsigned **pins, + unsigned *num_pins) + { + *pins = (unsigned *) foo_groups[selector].pins; + *num_pins = foo_groups[selector].num_pins; + return 0; + } + + static struct pinctrl_ops foo_pctrl_ops = { + .get_groups_count = foo_get_groups_count, + .get_group_name = foo_get_group_name, + .get_group_pins = foo_get_group_pins, + }; + + + static struct pinctrl_desc foo_desc = { + ... + .pctlops = &foo_pctrl_ops, + }; + +The pin control subsystem will call the .get_groups_count() function to +determine the total number of legal selectors, then it will call the other functions +to retrieve the name and pins of the group. Maintaining the data structure of +the groups is up to the driver, this is just a simple example - in practice you +may need more entries in your group structure, for example specific register +ranges associated with each group and so on. + + +Pin configuration +================= + +Pins can sometimes be software-configured in various ways, mostly related +to their electronic properties when used as inputs or outputs. For example you +may be able to make an output pin high impedance, or "tristate" meaning it is +effectively disconnected. You may be able to connect an input pin to VDD or GND +using a certain resistor value - pull up and pull down - so that the pin has a +stable value when nothing is driving the rail it is connected to, or when it's +unconnected. + +Pin configuration can be programmed by adding configuration entries into the +mapping table; see section "Board/machine configuration" below. + +The format and meaning of the configuration parameter, PLATFORM_X_PULL_UP +above, is entirely defined by the pin controller driver. + +The pin configuration driver implements callbacks for changing pin +configuration in the pin controller ops like this:: + + #include <linux/pinctrl/pinctrl.h> + #include <linux/pinctrl/pinconf.h> + #include "platform_x_pindefs.h" + + static int foo_pin_config_get(struct pinctrl_dev *pctldev, + unsigned offset, + unsigned long *config) + { + struct my_conftype conf; + + ... Find setting for pin @ offset ... + + *config = (unsigned long) conf; + } + + static int foo_pin_config_set(struct pinctrl_dev *pctldev, + unsigned offset, + unsigned long config) + { + struct my_conftype *conf = (struct my_conftype *) config; + + switch (conf) { + case PLATFORM_X_PULL_UP: + ... + } + } + } + + static int foo_pin_config_group_get (struct pinctrl_dev *pctldev, + unsigned selector, + unsigned long *config) + { + ... + } + + static int foo_pin_config_group_set (struct pinctrl_dev *pctldev, + unsigned selector, + unsigned long config) + { + ... + } + + static struct pinconf_ops foo_pconf_ops = { + .pin_config_get = foo_pin_config_get, + .pin_config_set = foo_pin_config_set, + .pin_config_group_get = foo_pin_config_group_get, + .pin_config_group_set = foo_pin_config_group_set, + }; + + /* Pin config operations are handled by some pin controller */ + static struct pinctrl_desc foo_desc = { + ... + .confops = &foo_pconf_ops, + }; + +Since some controllers have special logic for handling entire groups of pins +they can exploit the special whole-group pin control function. The +pin_config_group_set() callback is allowed to return the error code -EAGAIN, +for groups it does not want to handle, or if it just wants to do some +group-level handling and then fall through to iterate over all pins, in which +case each individual pin will be treated by separate pin_config_set() calls as +well. + + +Interaction with the GPIO subsystem +=================================== + +The GPIO drivers may want to perform operations of various types on the same +physical pins that are also registered as pin controller pins. + +First and foremost, the two subsystems can be used as completely orthogonal, +see the section named "pin control requests from drivers" and +"drivers needing both pin control and GPIOs" below for details. But in some +situations a cross-subsystem mapping between pins and GPIOs is needed. + +Since the pin controller subsystem has its pinspace local to the pin controller +we need a mapping so that the pin control subsystem can figure out which pin +controller handles control of a certain GPIO pin. Since a single pin controller +may be muxing several GPIO ranges (typically SoCs that have one set of pins, +but internally several GPIO silicon blocks, each modelled as a struct +gpio_chip) any number of GPIO ranges can be added to a pin controller instance +like this:: + + struct gpio_chip chip_a; + struct gpio_chip chip_b; + + static struct pinctrl_gpio_range gpio_range_a = { + .name = "chip a", + .id = 0, + .base = 32, + .pin_base = 32, + .npins = 16, + .gc = &chip_a; + }; + + static struct pinctrl_gpio_range gpio_range_b = { + .name = "chip b", + .id = 0, + .base = 48, + .pin_base = 64, + .npins = 8, + .gc = &chip_b; + }; + + { + struct pinctrl_dev *pctl; + ... + pinctrl_add_gpio_range(pctl, &gpio_range_a); + pinctrl_add_gpio_range(pctl, &gpio_range_b); + } + +So this complex system has one pin controller handling two different +GPIO chips. "chip a" has 16 pins and "chip b" has 8 pins. The "chip a" and +"chip b" have different .pin_base, which means a start pin number of the +GPIO range. + +The GPIO range of "chip a" starts from the GPIO base of 32 and actual +pin range also starts from 32. However "chip b" has different starting +offset for the GPIO range and pin range. The GPIO range of "chip b" starts +from GPIO number 48, while the pin range of "chip b" starts from 64. + +We can convert a gpio number to actual pin number using this "pin_base". +They are mapped in the global GPIO pin space at: + +chip a: + - GPIO range : [32 .. 47] + - pin range : [32 .. 47] +chip b: + - GPIO range : [48 .. 55] + - pin range : [64 .. 71] + +The above examples assume the mapping between the GPIOs and pins is +linear. If the mapping is sparse or haphazard, an array of arbitrary pin +numbers can be encoded in the range like this:: + + static const unsigned range_pins[] = { 14, 1, 22, 17, 10, 8, 6, 2 }; + + static struct pinctrl_gpio_range gpio_range = { + .name = "chip", + .id = 0, + .base = 32, + .pins = &range_pins, + .npins = ARRAY_SIZE(range_pins), + .gc = &chip; + }; + +In this case the pin_base property will be ignored. If the name of a pin +group is known, the pins and npins elements of the above structure can be +initialised using the function pinctrl_get_group_pins(), e.g. for pin +group "foo":: + + pinctrl_get_group_pins(pctl, "foo", &gpio_range.pins, + &gpio_range.npins); + +When GPIO-specific functions in the pin control subsystem are called, these +ranges will be used to look up the appropriate pin controller by inspecting +and matching the pin to the pin ranges across all controllers. When a +pin controller handling the matching range is found, GPIO-specific functions +will be called on that specific pin controller. + +For all functionalities dealing with pin biasing, pin muxing etc, the pin +controller subsystem will look up the corresponding pin number from the passed +in gpio number, and use the range's internals to retrieve a pin number. After +that, the subsystem passes it on to the pin control driver, so the driver +will get a pin number into its handled number range. Further it is also passed +the range ID value, so that the pin controller knows which range it should +deal with. + +Calling pinctrl_add_gpio_range from pinctrl driver is DEPRECATED. Please see +section 2.1 of Documentation/devicetree/bindings/gpio/gpio.txt on how to bind +pinctrl and gpio drivers. + + +PINMUX interfaces +================= + +These calls use the pinmux_* naming prefix. No other calls should use that +prefix. + + +What is pinmuxing? +================== + +PINMUX, also known as padmux, ballmux, alternate functions or mission modes +is a way for chip vendors producing some kind of electrical packages to use +a certain physical pin (ball, pad, finger, etc) for multiple mutually exclusive +functions, depending on the application. By "application" in this context +we usually mean a way of soldering or wiring the package into an electronic +system, even though the framework makes it possible to also change the function +at runtime. + +Here is an example of a PGA (Pin Grid Array) chip seen from underneath:: + + A B C D E F G H + +---+ + 8 | o | o o o o o o o + | | + 7 | o | o o o o o o o + | | + 6 | o | o o o o o o o + +---+---+ + 5 | o | o | o o o o o o + +---+---+ +---+ + 4 o o o o o o | o | o + | | + 3 o o o o o o | o | o + | | + 2 o o o o o o | o | o + +-------+-------+-------+---+---+ + 1 | o o | o o | o o | o | o | + +-------+-------+-------+---+---+ + +This is not tetris. The game to think of is chess. Not all PGA/BGA packages +are chessboard-like, big ones have "holes" in some arrangement according to +different design patterns, but we're using this as a simple example. Of the +pins you see some will be taken by things like a few VCC and GND to feed power +to the chip, and quite a few will be taken by large ports like an external +memory interface. The remaining pins will often be subject to pin multiplexing. + +The example 8x8 PGA package above will have pin numbers 0 through 63 assigned +to its physical pins. It will name the pins { A1, A2, A3 ... H6, H7, H8 } using +pinctrl_register_pins() and a suitable data set as shown earlier. + +In this 8x8 BGA package the pins { A8, A7, A6, A5 } can be used as an SPI port +(these are four pins: CLK, RXD, TXD, FRM). In that case, pin B5 can be used as +some general-purpose GPIO pin. However, in another setting, pins { A5, B5 } can +be used as an I2C port (these are just two pins: SCL, SDA). Needless to say, +we cannot use the SPI port and I2C port at the same time. However in the inside +of the package the silicon performing the SPI logic can alternatively be routed +out on pins { G4, G3, G2, G1 }. + +On the bottom row at { A1, B1, C1, D1, E1, F1, G1, H1 } we have something +special - it's an external MMC bus that can be 2, 4 or 8 bits wide, and it will +consume 2, 4 or 8 pins respectively, so either { A1, B1 } are taken or +{ A1, B1, C1, D1 } or all of them. If we use all 8 bits, we cannot use the SPI +port on pins { G4, G3, G2, G1 } of course. + +This way the silicon blocks present inside the chip can be multiplexed "muxed" +out on different pin ranges. Often contemporary SoC (systems on chip) will +contain several I2C, SPI, SDIO/MMC, etc silicon blocks that can be routed to +different pins by pinmux settings. + +Since general-purpose I/O pins (GPIO) are typically always in shortage, it is +common to be able to use almost any pin as a GPIO pin if it is not currently +in use by some other I/O port. + + +Pinmux conventions +================== + +The purpose of the pinmux functionality in the pin controller subsystem is to +abstract and provide pinmux settings to the devices you choose to instantiate +in your machine configuration. It is inspired by the clk, GPIO and regulator +subsystems, so devices will request their mux setting, but it's also possible +to request a single pin for e.g. GPIO. + +Definitions: + +- FUNCTIONS can be switched in and out by a driver residing with the pin + control subsystem in the drivers/pinctrl/* directory of the kernel. The + pin control driver knows the possible functions. In the example above you can + identify three pinmux functions, one for spi, one for i2c and one for mmc. + +- FUNCTIONS are assumed to be enumerable from zero in a one-dimensional array. + In this case the array could be something like: { spi0, i2c0, mmc0 } + for the three available functions. + +- FUNCTIONS have PIN GROUPS as defined on the generic level - so a certain + function is *always* associated with a certain set of pin groups, could + be just a single one, but could also be many. In the example above the + function i2c is associated with the pins { A5, B5 }, enumerated as + { 24, 25 } in the controller pin space. + + The Function spi is associated with pin groups { A8, A7, A6, A5 } + and { G4, G3, G2, G1 }, which are enumerated as { 0, 8, 16, 24 } and + { 38, 46, 54, 62 } respectively. + + Group names must be unique per pin controller, no two groups on the same + controller may have the same name. + +- The combination of a FUNCTION and a PIN GROUP determine a certain function + for a certain set of pins. The knowledge of the functions and pin groups + and their machine-specific particulars are kept inside the pinmux driver, + from the outside only the enumerators are known, and the driver core can + request: + + - The name of a function with a certain selector (>= 0) + - A list of groups associated with a certain function + - That a certain group in that list to be activated for a certain function + + As already described above, pin groups are in turn self-descriptive, so + the core will retrieve the actual pin range in a certain group from the + driver. + +- FUNCTIONS and GROUPS on a certain PIN CONTROLLER are MAPPED to a certain + device by the board file, device tree or similar machine setup configuration + mechanism, similar to how regulators are connected to devices, usually by + name. Defining a pin controller, function and group thus uniquely identify + the set of pins to be used by a certain device. (If only one possible group + of pins is available for the function, no group name need to be supplied - + the core will simply select the first and only group available.) + + In the example case we can define that this particular machine shall + use device spi0 with pinmux function fspi0 group gspi0 and i2c0 on function + fi2c0 group gi2c0, on the primary pin controller, we get mappings + like these:: + + { + {"map-spi0", spi0, pinctrl0, fspi0, gspi0}, + {"map-i2c0", i2c0, pinctrl0, fi2c0, gi2c0} + } + + Every map must be assigned a state name, pin controller, device and + function. The group is not compulsory - if it is omitted the first group + presented by the driver as applicable for the function will be selected, + which is useful for simple cases. + + It is possible to map several groups to the same combination of device, + pin controller and function. This is for cases where a certain function on + a certain pin controller may use different sets of pins in different + configurations. + +- PINS for a certain FUNCTION using a certain PIN GROUP on a certain + PIN CONTROLLER are provided on a first-come first-serve basis, so if some + other device mux setting or GPIO pin request has already taken your physical + pin, you will be denied the use of it. To get (activate) a new setting, the + old one has to be put (deactivated) first. + +Sometimes the documentation and hardware registers will be oriented around +pads (or "fingers") rather than pins - these are the soldering surfaces on the +silicon inside the package, and may or may not match the actual number of +pins/balls underneath the capsule. Pick some enumeration that makes sense to +you. Define enumerators only for the pins you can control if that makes sense. + +Assumptions: + +We assume that the number of possible function maps to pin groups is limited by +the hardware. I.e. we assume that there is no system where any function can be +mapped to any pin, like in a phone exchange. So the available pin groups for +a certain function will be limited to a few choices (say up to eight or so), +not hundreds or any amount of choices. This is the characteristic we have found +by inspecting available pinmux hardware, and a necessary assumption since we +expect pinmux drivers to present *all* possible function vs pin group mappings +to the subsystem. + + +Pinmux drivers +============== + +The pinmux core takes care of preventing conflicts on pins and calling +the pin controller driver to execute different settings. + +It is the responsibility of the pinmux driver to impose further restrictions +(say for example infer electronic limitations due to load, etc.) to determine +whether or not the requested function can actually be allowed, and in case it +is possible to perform the requested mux setting, poke the hardware so that +this happens. + +Pinmux drivers are required to supply a few callback functions, some are +optional. Usually the set_mux() function is implemented, writing values into +some certain registers to activate a certain mux setting for a certain pin. + +A simple driver for the above example will work by setting bits 0, 1, 2, 3 or 4 +into some register named MUX to select a certain function with a certain +group of pins would work something like this:: + + #include <linux/pinctrl/pinctrl.h> + #include <linux/pinctrl/pinmux.h> + + struct foo_group { + const char *name; + const unsigned int *pins; + const unsigned num_pins; + }; + + static const unsigned spi0_0_pins[] = { 0, 8, 16, 24 }; + static const unsigned spi0_1_pins[] = { 38, 46, 54, 62 }; + static const unsigned i2c0_pins[] = { 24, 25 }; + static const unsigned mmc0_1_pins[] = { 56, 57 }; + static const unsigned mmc0_2_pins[] = { 58, 59 }; + static const unsigned mmc0_3_pins[] = { 60, 61, 62, 63 }; + + static const struct foo_group foo_groups[] = { + { + .name = "spi0_0_grp", + .pins = spi0_0_pins, + .num_pins = ARRAY_SIZE(spi0_0_pins), + }, + { + .name = "spi0_1_grp", + .pins = spi0_1_pins, + .num_pins = ARRAY_SIZE(spi0_1_pins), + }, + { + .name = "i2c0_grp", + .pins = i2c0_pins, + .num_pins = ARRAY_SIZE(i2c0_pins), + }, + { + .name = "mmc0_1_grp", + .pins = mmc0_1_pins, + .num_pins = ARRAY_SIZE(mmc0_1_pins), + }, + { + .name = "mmc0_2_grp", + .pins = mmc0_2_pins, + .num_pins = ARRAY_SIZE(mmc0_2_pins), + }, + { + .name = "mmc0_3_grp", + .pins = mmc0_3_pins, + .num_pins = ARRAY_SIZE(mmc0_3_pins), + }, + }; + + + static int foo_get_groups_count(struct pinctrl_dev *pctldev) + { + return ARRAY_SIZE(foo_groups); + } + + static const char *foo_get_group_name(struct pinctrl_dev *pctldev, + unsigned selector) + { + return foo_groups[selector].name; + } + + static int foo_get_group_pins(struct pinctrl_dev *pctldev, unsigned selector, + unsigned ** const pins, + unsigned * const num_pins) + { + *pins = (unsigned *) foo_groups[selector].pins; + *num_pins = foo_groups[selector].num_pins; + return 0; + } + + static struct pinctrl_ops foo_pctrl_ops = { + .get_groups_count = foo_get_groups_count, + .get_group_name = foo_get_group_name, + .get_group_pins = foo_get_group_pins, + }; + + struct foo_pmx_func { + const char *name; + const char * const *groups; + const unsigned num_groups; + }; + + static const char * const spi0_groups[] = { "spi0_0_grp", "spi0_1_grp" }; + static const char * const i2c0_groups[] = { "i2c0_grp" }; + static const char * const mmc0_groups[] = { "mmc0_1_grp", "mmc0_2_grp", + "mmc0_3_grp" }; + + static const struct foo_pmx_func foo_functions[] = { + { + .name = "spi0", + .groups = spi0_groups, + .num_groups = ARRAY_SIZE(spi0_groups), + }, + { + .name = "i2c0", + .groups = i2c0_groups, + .num_groups = ARRAY_SIZE(i2c0_groups), + }, + { + .name = "mmc0", + .groups = mmc0_groups, + .num_groups = ARRAY_SIZE(mmc0_groups), + }, + }; + + static int foo_get_functions_count(struct pinctrl_dev *pctldev) + { + return ARRAY_SIZE(foo_functions); + } + + static const char *foo_get_fname(struct pinctrl_dev *pctldev, unsigned selector) + { + return foo_functions[selector].name; + } + + static int foo_get_groups(struct pinctrl_dev *pctldev, unsigned selector, + const char * const **groups, + unsigned * const num_groups) + { + *groups = foo_functions[selector].groups; + *num_groups = foo_functions[selector].num_groups; + return 0; + } + + static int foo_set_mux(struct pinctrl_dev *pctldev, unsigned selector, + unsigned group) + { + u8 regbit = (1 << selector + group); + + writeb((readb(MUX)|regbit), MUX) + return 0; + } + + static struct pinmux_ops foo_pmxops = { + .get_functions_count = foo_get_functions_count, + .get_function_name = foo_get_fname, + .get_function_groups = foo_get_groups, + .set_mux = foo_set_mux, + .strict = true, + }; + + /* Pinmux operations are handled by some pin controller */ + static struct pinctrl_desc foo_desc = { + ... + .pctlops = &foo_pctrl_ops, + .pmxops = &foo_pmxops, + }; + +In the example activating muxing 0 and 1 at the same time setting bits +0 and 1, uses one pin in common so they would collide. + +The beauty of the pinmux subsystem is that since it keeps track of all +pins and who is using them, it will already have denied an impossible +request like that, so the driver does not need to worry about such +things - when it gets a selector passed in, the pinmux subsystem makes +sure no other device or GPIO assignment is already using the selected +pins. Thus bits 0 and 1 in the control register will never be set at the +same time. + +All the above functions are mandatory to implement for a pinmux driver. + + +Pin control interaction with the GPIO subsystem +=============================================== + +Note that the following implies that the use case is to use a certain pin +from the Linux kernel using the API in <linux/gpio.h> with gpio_request() +and similar functions. There are cases where you may be using something +that your datasheet calls "GPIO mode", but actually is just an electrical +configuration for a certain device. See the section below named +"GPIO mode pitfalls" for more details on this scenario. + +The public pinmux API contains two functions named pinctrl_gpio_request() +and pinctrl_gpio_free(). These two functions shall *ONLY* be called from +gpiolib-based drivers as part of their gpio_request() and +gpio_free() semantics. Likewise the pinctrl_gpio_direction_[input|output] +shall only be called from within respective gpio_direction_[input|output] +gpiolib implementation. + +NOTE that platforms and individual drivers shall *NOT* request GPIO pins to be +controlled e.g. muxed in. Instead, implement a proper gpiolib driver and have +that driver request proper muxing and other control for its pins. + +The function list could become long, especially if you can convert every +individual pin into a GPIO pin independent of any other pins, and then try +the approach to define every pin as a function. + +In this case, the function array would become 64 entries for each GPIO +setting and then the device functions. + +For this reason there are two functions a pin control driver can implement +to enable only GPIO on an individual pin: .gpio_request_enable() and +.gpio_disable_free(). + +This function will pass in the affected GPIO range identified by the pin +controller core, so you know which GPIO pins are being affected by the request +operation. + +If your driver needs to have an indication from the framework of whether the +GPIO pin shall be used for input or output you can implement the +.gpio_set_direction() function. As described this shall be called from the +gpiolib driver and the affected GPIO range, pin offset and desired direction +will be passed along to this function. + +Alternatively to using these special functions, it is fully allowed to use +named functions for each GPIO pin, the pinctrl_gpio_request() will attempt to +obtain the function "gpioN" where "N" is the global GPIO pin number if no +special GPIO-handler is registered. + + +GPIO mode pitfalls +================== + +Due to the naming conventions used by hardware engineers, where "GPIO" +is taken to mean different things than what the kernel does, the developer +may be confused by a datasheet talking about a pin being possible to set +into "GPIO mode". It appears that what hardware engineers mean with +"GPIO mode" is not necessarily the use case that is implied in the kernel +interface <linux/gpio.h>: a pin that you grab from kernel code and then +either listen for input or drive high/low to assert/deassert some +external line. + +Rather hardware engineers think that "GPIO mode" means that you can +software-control a few electrical properties of the pin that you would +not be able to control if the pin was in some other mode, such as muxed in +for a device. + +The GPIO portions of a pin and its relation to a certain pin controller +configuration and muxing logic can be constructed in several ways. Here +are two examples:: + + (A) + pin config + logic regs + | +- SPI + Physical pins --- pad --- pinmux -+- I2C + | +- mmc + | +- GPIO + pin + multiplex + logic regs + +Here some electrical properties of the pin can be configured no matter +whether the pin is used for GPIO or not. If you multiplex a GPIO onto a +pin, you can also drive it high/low from "GPIO" registers. +Alternatively, the pin can be controlled by a certain peripheral, while +still applying desired pin config properties. GPIO functionality is thus +orthogonal to any other device using the pin. + +In this arrangement the registers for the GPIO portions of the pin controller, +or the registers for the GPIO hardware module are likely to reside in a +separate memory range only intended for GPIO driving, and the register +range dealing with pin config and pin multiplexing get placed into a +different memory range and a separate section of the data sheet. + +A flag "strict" in struct pinmux_ops is available to check and deny +simultaneous access to the same pin from GPIO and pin multiplexing +consumers on hardware of this type. The pinctrl driver should set this flag +accordingly. + +:: + + (B) + + pin config + logic regs + | +- SPI + Physical pins --- pad --- pinmux -+- I2C + | | +- mmc + | | + GPIO pin + multiplex + logic regs + +In this arrangement, the GPIO functionality can always be enabled, such that +e.g. a GPIO input can be used to "spy" on the SPI/I2C/MMC signal while it is +pulsed out. It is likely possible to disrupt the traffic on the pin by doing +wrong things on the GPIO block, as it is never really disconnected. It is +possible that the GPIO, pin config and pin multiplex registers are placed into +the same memory range and the same section of the data sheet, although that +need not be the case. + +In some pin controllers, although the physical pins are designed in the same +way as (B), the GPIO function still can't be enabled at the same time as the +peripheral functions. So again the "strict" flag should be set, denying +simultaneous activation by GPIO and other muxed in devices. + +From a kernel point of view, however, these are different aspects of the +hardware and shall be put into different subsystems: + +- Registers (or fields within registers) that control electrical + properties of the pin such as biasing and drive strength should be + exposed through the pinctrl subsystem, as "pin configuration" settings. + +- Registers (or fields within registers) that control muxing of signals + from various other HW blocks (e.g. I2C, MMC, or GPIO) onto pins should + be exposed through the pinctrl subsystem, as mux functions. + +- Registers (or fields within registers) that control GPIO functionality + such as setting a GPIO's output value, reading a GPIO's input value, or + setting GPIO pin direction should be exposed through the GPIO subsystem, + and if they also support interrupt capabilities, through the irqchip + abstraction. + +Depending on the exact HW register design, some functions exposed by the +GPIO subsystem may call into the pinctrl subsystem in order to +co-ordinate register settings across HW modules. In particular, this may +be needed for HW with separate GPIO and pin controller HW modules, where +e.g. GPIO direction is determined by a register in the pin controller HW +module rather than the GPIO HW module. + +Electrical properties of the pin such as biasing and drive strength +may be placed at some pin-specific register in all cases or as part +of the GPIO register in case (B) especially. This doesn't mean that such +properties necessarily pertain to what the Linux kernel calls "GPIO". + +Example: a pin is usually muxed in to be used as a UART TX line. But during +system sleep, we need to put this pin into "GPIO mode" and ground it. + +If you make a 1-to-1 map to the GPIO subsystem for this pin, you may start +to think that you need to come up with something really complex, that the +pin shall be used for UART TX and GPIO at the same time, that you will grab +a pin control handle and set it to a certain state to enable UART TX to be +muxed in, then twist it over to GPIO mode and use gpio_direction_output() +to drive it low during sleep, then mux it over to UART TX again when you +wake up and maybe even gpio_request/gpio_free as part of this cycle. This +all gets very complicated. + +The solution is to not think that what the datasheet calls "GPIO mode" +has to be handled by the <linux/gpio.h> interface. Instead view this as +a certain pin config setting. Look in e.g. <linux/pinctrl/pinconf-generic.h> +and you find this in the documentation: + + PIN_CONFIG_OUTPUT: + this will configure the pin in output, use argument + 1 to indicate high level, argument 0 to indicate low level. + +So it is perfectly possible to push a pin into "GPIO mode" and drive the +line low as part of the usual pin control map. So for example your UART +driver may look like this:: + + #include <linux/pinctrl/consumer.h> + + struct pinctrl *pinctrl; + struct pinctrl_state *pins_default; + struct pinctrl_state *pins_sleep; + + pins_default = pinctrl_lookup_state(uap->pinctrl, PINCTRL_STATE_DEFAULT); + pins_sleep = pinctrl_lookup_state(uap->pinctrl, PINCTRL_STATE_SLEEP); + + /* Normal mode */ + retval = pinctrl_select_state(pinctrl, pins_default); + /* Sleep mode */ + retval = pinctrl_select_state(pinctrl, pins_sleep); + +And your machine configuration may look like this: +-------------------------------------------------- + +:: + + static unsigned long uart_default_mode[] = { + PIN_CONF_PACKED(PIN_CONFIG_DRIVE_PUSH_PULL, 0), + }; + + static unsigned long uart_sleep_mode[] = { + PIN_CONF_PACKED(PIN_CONFIG_OUTPUT, 0), + }; + + static struct pinctrl_map pinmap[] __initdata = { + PIN_MAP_MUX_GROUP("uart", PINCTRL_STATE_DEFAULT, "pinctrl-foo", + "u0_group", "u0"), + PIN_MAP_CONFIGS_PIN("uart", PINCTRL_STATE_DEFAULT, "pinctrl-foo", + "UART_TX_PIN", uart_default_mode), + PIN_MAP_MUX_GROUP("uart", PINCTRL_STATE_SLEEP, "pinctrl-foo", + "u0_group", "gpio-mode"), + PIN_MAP_CONFIGS_PIN("uart", PINCTRL_STATE_SLEEP, "pinctrl-foo", + "UART_TX_PIN", uart_sleep_mode), + }; + + foo_init(void) { + pinctrl_register_mappings(pinmap, ARRAY_SIZE(pinmap)); + } + +Here the pins we want to control are in the "u0_group" and there is some +function called "u0" that can be enabled on this group of pins, and then +everything is UART business as usual. But there is also some function +named "gpio-mode" that can be mapped onto the same pins to move them into +GPIO mode. + +This will give the desired effect without any bogus interaction with the +GPIO subsystem. It is just an electrical configuration used by that device +when going to sleep, it might imply that the pin is set into something the +datasheet calls "GPIO mode", but that is not the point: it is still used +by that UART device to control the pins that pertain to that very UART +driver, putting them into modes needed by the UART. GPIO in the Linux +kernel sense are just some 1-bit line, and is a different use case. + +How the registers are poked to attain the push or pull, and output low +configuration and the muxing of the "u0" or "gpio-mode" group onto these +pins is a question for the driver. + +Some datasheets will be more helpful and refer to the "GPIO mode" as +"low power mode" rather than anything to do with GPIO. This often means +the same thing electrically speaking, but in this latter case the +software engineers will usually quickly identify that this is some +specific muxing or configuration rather than anything related to the GPIO +API. + + +Board/machine configuration +=========================== + +Boards and machines define how a certain complete running system is put +together, including how GPIOs and devices are muxed, how regulators are +constrained and how the clock tree looks. Of course pinmux settings are also +part of this. + +A pin controller configuration for a machine looks pretty much like a simple +regulator configuration, so for the example array above we want to enable i2c +and spi on the second function mapping:: + + #include <linux/pinctrl/machine.h> + + static const struct pinctrl_map mapping[] __initconst = { + { + .dev_name = "foo-spi.0", + .name = PINCTRL_STATE_DEFAULT, + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .data.mux.function = "spi0", + }, + { + .dev_name = "foo-i2c.0", + .name = PINCTRL_STATE_DEFAULT, + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .data.mux.function = "i2c0", + }, + { + .dev_name = "foo-mmc.0", + .name = PINCTRL_STATE_DEFAULT, + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .data.mux.function = "mmc0", + }, + }; + +The dev_name here matches to the unique device name that can be used to look +up the device struct (just like with clockdev or regulators). The function name +must match a function provided by the pinmux driver handling this pin range. + +As you can see we may have several pin controllers on the system and thus +we need to specify which one of them contains the functions we wish to map. + +You register this pinmux mapping to the pinmux subsystem by simply:: + + ret = pinctrl_register_mappings(mapping, ARRAY_SIZE(mapping)); + +Since the above construct is pretty common there is a helper macro to make +it even more compact which assumes you want to use pinctrl-foo and position +0 for mapping, for example:: + + static struct pinctrl_map mapping[] __initdata = { + PIN_MAP_MUX_GROUP("foo-i2c.o", PINCTRL_STATE_DEFAULT, + "pinctrl-foo", NULL, "i2c0"), + }; + +The mapping table may also contain pin configuration entries. It's common for +each pin/group to have a number of configuration entries that affect it, so +the table entries for configuration reference an array of config parameters +and values. An example using the convenience macros is shown below:: + + static unsigned long i2c_grp_configs[] = { + FOO_PIN_DRIVEN, + FOO_PIN_PULLUP, + }; + + static unsigned long i2c_pin_configs[] = { + FOO_OPEN_COLLECTOR, + FOO_SLEW_RATE_SLOW, + }; + + static struct pinctrl_map mapping[] __initdata = { + PIN_MAP_MUX_GROUP("foo-i2c.0", PINCTRL_STATE_DEFAULT, + "pinctrl-foo", "i2c0", "i2c0"), + PIN_MAP_CONFIGS_GROUP("foo-i2c.0", PINCTRL_STATE_DEFAULT, + "pinctrl-foo", "i2c0", i2c_grp_configs), + PIN_MAP_CONFIGS_PIN("foo-i2c.0", PINCTRL_STATE_DEFAULT, + "pinctrl-foo", "i2c0scl", i2c_pin_configs), + PIN_MAP_CONFIGS_PIN("foo-i2c.0", PINCTRL_STATE_DEFAULT, + "pinctrl-foo", "i2c0sda", i2c_pin_configs), + }; + +Finally, some devices expect the mapping table to contain certain specific +named states. When running on hardware that doesn't need any pin controller +configuration, the mapping table must still contain those named states, in +order to explicitly indicate that the states were provided and intended to +be empty. Table entry macro PIN_MAP_DUMMY_STATE serves the purpose of defining +a named state without causing any pin controller to be programmed:: + + static struct pinctrl_map mapping[] __initdata = { + PIN_MAP_DUMMY_STATE("foo-i2c.0", PINCTRL_STATE_DEFAULT), + }; + + +Complex mappings +================ + +As it is possible to map a function to different groups of pins an optional +.group can be specified like this:: + + ... + { + .dev_name = "foo-spi.0", + .name = "spi0-pos-A", + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "spi0", + .group = "spi0_0_grp", + }, + { + .dev_name = "foo-spi.0", + .name = "spi0-pos-B", + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "spi0", + .group = "spi0_1_grp", + }, + ... + +This example mapping is used to switch between two positions for spi0 at +runtime, as described further below under the heading "Runtime pinmuxing". + +Further it is possible for one named state to affect the muxing of several +groups of pins, say for example in the mmc0 example above, where you can +additively expand the mmc0 bus from 2 to 4 to 8 pins. If we want to use all +three groups for a total of 2+2+4 = 8 pins (for an 8-bit MMC bus as is the +case), we define a mapping like this:: + + ... + { + .dev_name = "foo-mmc.0", + .name = "2bit" + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "mmc0", + .group = "mmc0_1_grp", + }, + { + .dev_name = "foo-mmc.0", + .name = "4bit" + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "mmc0", + .group = "mmc0_1_grp", + }, + { + .dev_name = "foo-mmc.0", + .name = "4bit" + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "mmc0", + .group = "mmc0_2_grp", + }, + { + .dev_name = "foo-mmc.0", + .name = "8bit" + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "mmc0", + .group = "mmc0_1_grp", + }, + { + .dev_name = "foo-mmc.0", + .name = "8bit" + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "mmc0", + .group = "mmc0_2_grp", + }, + { + .dev_name = "foo-mmc.0", + .name = "8bit" + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "mmc0", + .group = "mmc0_3_grp", + }, + ... + +The result of grabbing this mapping from the device with something like +this (see next paragraph):: + + p = devm_pinctrl_get(dev); + s = pinctrl_lookup_state(p, "8bit"); + ret = pinctrl_select_state(p, s); + +or more simply:: + + p = devm_pinctrl_get_select(dev, "8bit"); + +Will be that you activate all the three bottom records in the mapping at +once. Since they share the same name, pin controller device, function and +device, and since we allow multiple groups to match to a single device, they +all get selected, and they all get enabled and disable simultaneously by the +pinmux core. + + +Pin control requests from drivers +================================= + +When a device driver is about to probe the device core will automatically +attempt to issue pinctrl_get_select_default() on these devices. +This way driver writers do not need to add any of the boilerplate code +of the type found below. However when doing fine-grained state selection +and not using the "default" state, you may have to do some device driver +handling of the pinctrl handles and states. + +So if you just want to put the pins for a certain device into the default +state and be done with it, there is nothing you need to do besides +providing the proper mapping table. The device core will take care of +the rest. + +Generally it is discouraged to let individual drivers get and enable pin +control. So if possible, handle the pin control in platform code or some other +place where you have access to all the affected struct device * pointers. In +some cases where a driver needs to e.g. switch between different mux mappings +at runtime this is not possible. + +A typical case is if a driver needs to switch bias of pins from normal +operation and going to sleep, moving from the PINCTRL_STATE_DEFAULT to +PINCTRL_STATE_SLEEP at runtime, re-biasing or even re-muxing pins to save +current in sleep mode. + +A driver may request a certain control state to be activated, usually just the +default state like this:: + + #include <linux/pinctrl/consumer.h> + + struct foo_state { + struct pinctrl *p; + struct pinctrl_state *s; + ... + }; + + foo_probe() + { + /* Allocate a state holder named "foo" etc */ + struct foo_state *foo = ...; + + foo->p = devm_pinctrl_get(&device); + if (IS_ERR(foo->p)) { + /* FIXME: clean up "foo" here */ + return PTR_ERR(foo->p); + } + + foo->s = pinctrl_lookup_state(foo->p, PINCTRL_STATE_DEFAULT); + if (IS_ERR(foo->s)) { + /* FIXME: clean up "foo" here */ + return PTR_ERR(s); + } + + ret = pinctrl_select_state(foo->s); + if (ret < 0) { + /* FIXME: clean up "foo" here */ + return ret; + } + } + +This get/lookup/select/put sequence can just as well be handled by bus drivers +if you don't want each and every driver to handle it and you know the +arrangement on your bus. + +The semantics of the pinctrl APIs are: + +- pinctrl_get() is called in process context to obtain a handle to all pinctrl + information for a given client device. It will allocate a struct from the + kernel memory to hold the pinmux state. All mapping table parsing or similar + slow operations take place within this API. + +- devm_pinctrl_get() is a variant of pinctrl_get() that causes pinctrl_put() + to be called automatically on the retrieved pointer when the associated + device is removed. It is recommended to use this function over plain + pinctrl_get(). + +- pinctrl_lookup_state() is called in process context to obtain a handle to a + specific state for a client device. This operation may be slow, too. + +- pinctrl_select_state() programs pin controller hardware according to the + definition of the state as given by the mapping table. In theory, this is a + fast-path operation, since it only involved blasting some register settings + into hardware. However, note that some pin controllers may have their + registers on a slow/IRQ-based bus, so client devices should not assume they + can call pinctrl_select_state() from non-blocking contexts. + +- pinctrl_put() frees all information associated with a pinctrl handle. + +- devm_pinctrl_put() is a variant of pinctrl_put() that may be used to + explicitly destroy a pinctrl object returned by devm_pinctrl_get(). + However, use of this function will be rare, due to the automatic cleanup + that will occur even without calling it. + + pinctrl_get() must be paired with a plain pinctrl_put(). + pinctrl_get() may not be paired with devm_pinctrl_put(). + devm_pinctrl_get() can optionally be paired with devm_pinctrl_put(). + devm_pinctrl_get() may not be paired with plain pinctrl_put(). + +Usually the pin control core handled the get/put pair and call out to the +device drivers bookkeeping operations, like checking available functions and +the associated pins, whereas select_state pass on to the pin controller +driver which takes care of activating and/or deactivating the mux setting by +quickly poking some registers. + +The pins are allocated for your device when you issue the devm_pinctrl_get() +call, after this you should be able to see this in the debugfs listing of all +pins. + +NOTE: the pinctrl system will return -EPROBE_DEFER if it cannot find the +requested pinctrl handles, for example if the pinctrl driver has not yet +registered. Thus make sure that the error path in your driver gracefully +cleans up and is ready to retry the probing later in the startup process. + + +Drivers needing both pin control and GPIOs +========================================== + +Again, it is discouraged to let drivers lookup and select pin control states +themselves, but again sometimes this is unavoidable. + +So say that your driver is fetching its resources like this:: + + #include <linux/pinctrl/consumer.h> + #include <linux/gpio.h> + + struct pinctrl *pinctrl; + int gpio; + + pinctrl = devm_pinctrl_get_select_default(&dev); + gpio = devm_gpio_request(&dev, 14, "foo"); + +Here we first request a certain pin state and then request GPIO 14 to be +used. If you're using the subsystems orthogonally like this, you should +nominally always get your pinctrl handle and select the desired pinctrl +state BEFORE requesting the GPIO. This is a semantic convention to avoid +situations that can be electrically unpleasant, you will certainly want to +mux in and bias pins in a certain way before the GPIO subsystems starts to +deal with them. + +The above can be hidden: using the device core, the pinctrl core may be +setting up the config and muxing for the pins right before the device is +probing, nevertheless orthogonal to the GPIO subsystem. + +But there are also situations where it makes sense for the GPIO subsystem +to communicate directly with the pinctrl subsystem, using the latter as a +back-end. This is when the GPIO driver may call out to the functions +described in the section "Pin control interaction with the GPIO subsystem" +above. This only involves per-pin multiplexing, and will be completely +hidden behind the gpio_*() function namespace. In this case, the driver +need not interact with the pin control subsystem at all. + +If a pin control driver and a GPIO driver is dealing with the same pins +and the use cases involve multiplexing, you MUST implement the pin controller +as a back-end for the GPIO driver like this, unless your hardware design +is such that the GPIO controller can override the pin controller's +multiplexing state through hardware without the need to interact with the +pin control system. + + +System pin control hogging +========================== + +Pin control map entries can be hogged by the core when the pin controller +is registered. This means that the core will attempt to call pinctrl_get(), +lookup_state() and select_state() on it immediately after the pin control +device has been registered. + +This occurs for mapping table entries where the client device name is equal +to the pin controller device name, and the state name is PINCTRL_STATE_DEFAULT:: + + { + .dev_name = "pinctrl-foo", + .name = PINCTRL_STATE_DEFAULT, + .type = PIN_MAP_TYPE_MUX_GROUP, + .ctrl_dev_name = "pinctrl-foo", + .function = "power_func", + }, + +Since it may be common to request the core to hog a few always-applicable +mux settings on the primary pin controller, there is a convenience macro for +this:: + + PIN_MAP_MUX_GROUP_HOG_DEFAULT("pinctrl-foo", NULL /* group */, + "power_func") + +This gives the exact same result as the above construction. + + +Runtime pinmuxing +================= + +It is possible to mux a certain function in and out at runtime, say to move +an SPI port from one set of pins to another set of pins. Say for example for +spi0 in the example above, we expose two different groups of pins for the same +function, but with different named in the mapping as described under +"Advanced mapping" above. So that for an SPI device, we have two states named +"pos-A" and "pos-B". + +This snippet first initializes a state object for both groups (in foo_probe()), +then muxes the function in the pins defined by group A, and finally muxes it in +on the pins defined by group B:: + + #include <linux/pinctrl/consumer.h> + + struct pinctrl *p; + struct pinctrl_state *s1, *s2; + + foo_probe() + { + /* Setup */ + p = devm_pinctrl_get(&device); + if (IS_ERR(p)) + ... + + s1 = pinctrl_lookup_state(foo->p, "pos-A"); + if (IS_ERR(s1)) + ... + + s2 = pinctrl_lookup_state(foo->p, "pos-B"); + if (IS_ERR(s2)) + ... + } + + foo_switch() + { + /* Enable on position A */ + ret = pinctrl_select_state(s1); + if (ret < 0) + ... + + ... + + /* Enable on position B */ + ret = pinctrl_select_state(s2); + if (ret < 0) + ... + + ... + } + +The above has to be done from process context. The reservation of the pins +will be done when the state is activated, so in effect one specific pin +can be used by different functions at different times on a running system. diff --git a/Documentation/driver-api/pm/conf.py b/Documentation/driver-api/pm/conf.py new file mode 100644 index 000000000..a89fac112 --- /dev/null +++ b/Documentation/driver-api/pm/conf.py @@ -0,0 +1,10 @@ +# -*- coding: utf-8; mode: python -*- + +project = "Device Power Management" + +tags.add("subproject") + +latex_documents = [ + ('index', 'pm.tex', project, + 'The kernel development community', 'manual'), +] diff --git a/Documentation/driver-api/pm/devices.rst b/Documentation/driver-api/pm/devices.rst new file mode 100644 index 000000000..1128705a5 --- /dev/null +++ b/Documentation/driver-api/pm/devices.rst @@ -0,0 +1,827 @@ +.. |struct dev_pm_ops| replace:: :c:type:`struct dev_pm_ops <dev_pm_ops>` +.. |struct dev_pm_domain| replace:: :c:type:`struct dev_pm_domain <dev_pm_domain>` +.. |struct bus_type| replace:: :c:type:`struct bus_type <bus_type>` +.. |struct device_type| replace:: :c:type:`struct device_type <device_type>` +.. |struct class| replace:: :c:type:`struct class <class>` +.. |struct wakeup_source| replace:: :c:type:`struct wakeup_source <wakeup_source>` +.. |struct device| replace:: :c:type:`struct device <device>` + +============================== +Device Power Management Basics +============================== + +:: + + Copyright (c) 2010-2011 Rafael J. Wysocki <rjw@sisk.pl>, Novell Inc. + Copyright (c) 2010 Alan Stern <stern@rowland.harvard.edu> + Copyright (c) 2016 Intel Corp., Rafael J. Wysocki <rafael.j.wysocki@intel.com> + +Most of the code in Linux is device drivers, so most of the Linux power +management (PM) code is also driver-specific. Most drivers will do very +little; others, especially for platforms with small batteries (like cell +phones), will do a lot. + +This writeup gives an overview of how drivers interact with system-wide +power management goals, emphasizing the models and interfaces that are +shared by everything that hooks up to the driver model core. Read it as +background for the domain-specific work you'd do with any specific driver. + + +Two Models for Device Power Management +====================================== + +Drivers will use one or both of these models to put devices into low-power +states: + + System Sleep model: + + Drivers can enter low-power states as part of entering system-wide + low-power states like "suspend" (also known as "suspend-to-RAM"), or + (mostly for systems with disks) "hibernation" (also known as + "suspend-to-disk"). + + This is something that device, bus, and class drivers collaborate on + by implementing various role-specific suspend and resume methods to + cleanly power down hardware and software subsystems, then reactivate + them without loss of data. + + Some drivers can manage hardware wakeup events, which make the system + leave the low-power state. This feature may be enabled or disabled + using the relevant :file:`/sys/devices/.../power/wakeup` file (for + Ethernet drivers the ioctl interface used by ethtool may also be used + for this purpose); enabling it may cost some power usage, but let the + whole system enter low-power states more often. + + Runtime Power Management model: + + Devices may also be put into low-power states while the system is + running, independently of other power management activity in principle. + However, devices are not generally independent of each other (for + example, a parent device cannot be suspended unless all of its child + devices have been suspended). Moreover, depending on the bus type the + device is on, it may be necessary to carry out some bus-specific + operations on the device for this purpose. Devices put into low power + states at run time may require special handling during system-wide power + transitions (suspend or hibernation). + + For these reasons not only the device driver itself, but also the + appropriate subsystem (bus type, device type or device class) driver and + the PM core are involved in runtime power management. As in the system + sleep power management case, they need to collaborate by implementing + various role-specific suspend and resume methods, so that the hardware + is cleanly powered down and reactivated without data or service loss. + +There's not a lot to be said about those low-power states except that they are +very system-specific, and often device-specific. Also, that if enough devices +have been put into low-power states (at runtime), the effect may be very similar +to entering some system-wide low-power state (system sleep) ... and that +synergies exist, so that several drivers using runtime PM might put the system +into a state where even deeper power saving options are available. + +Most suspended devices will have quiesced all I/O: no more DMA or IRQs (except +for wakeup events), no more data read or written, and requests from upstream +drivers are no longer accepted. A given bus or platform may have different +requirements though. + +Examples of hardware wakeup events include an alarm from a real time clock, +network wake-on-LAN packets, keyboard or mouse activity, and media insertion +or removal (for PCMCIA, MMC/SD, USB, and so on). + +Interfaces for Entering System Sleep States +=========================================== + +There are programming interfaces provided for subsystems (bus type, device type, +device class) and device drivers to allow them to participate in the power +management of devices they are concerned with. These interfaces cover both +system sleep and runtime power management. + + +Device Power Management Operations +---------------------------------- + +Device power management operations, at the subsystem level as well as at the +device driver level, are implemented by defining and populating objects of type +|struct dev_pm_ops| defined in :file:`include/linux/pm.h`. The roles of the +methods included in it will be explained in what follows. For now, it should be +sufficient to remember that the last three methods are specific to runtime power +management while the remaining ones are used during system-wide power +transitions. + +There also is a deprecated "old" or "legacy" interface for power management +operations available at least for some subsystems. This approach does not use +|struct dev_pm_ops| objects and it is suitable only for implementing system +sleep power management methods in a limited way. Therefore it is not described +in this document, so please refer directly to the source code for more +information about it. + + +Subsystem-Level Methods +----------------------- + +The core methods to suspend and resume devices reside in +|struct dev_pm_ops| pointed to by the :c:member:`ops` member of +|struct dev_pm_domain|, or by the :c:member:`pm` member of |struct bus_type|, +|struct device_type| and |struct class|. They are mostly of interest to the +people writing infrastructure for platforms and buses, like PCI or USB, or +device type and device class drivers. They also are relevant to the writers of +device drivers whose subsystems (PM domains, device types, device classes and +bus types) don't provide all power management methods. + +Bus drivers implement these methods as appropriate for the hardware and the +drivers using it; PCI works differently from USB, and so on. Not many people +write subsystem-level drivers; most driver code is a "device driver" that builds +on top of bus-specific framework code. + +For more information on these driver calls, see the description later; +they are called in phases for every device, respecting the parent-child +sequencing in the driver model tree. + + +:file:`/sys/devices/.../power/wakeup` files +------------------------------------------- + +All device objects in the driver model contain fields that control the handling +of system wakeup events (hardware signals that can force the system out of a +sleep state). These fields are initialized by bus or device driver code using +:c:func:`device_set_wakeup_capable()` and :c:func:`device_set_wakeup_enable()`, +defined in :file:`include/linux/pm_wakeup.h`. + +The :c:member:`power.can_wakeup` flag just records whether the device (and its +driver) can physically support wakeup events. The +:c:func:`device_set_wakeup_capable()` routine affects this flag. The +:c:member:`power.wakeup` field is a pointer to an object of type +|struct wakeup_source| used for controlling whether or not the device should use +its system wakeup mechanism and for notifying the PM core of system wakeup +events signaled by the device. This object is only present for wakeup-capable +devices (i.e. devices whose :c:member:`can_wakeup` flags are set) and is created +(or removed) by :c:func:`device_set_wakeup_capable()`. + +Whether or not a device is capable of issuing wakeup events is a hardware +matter, and the kernel is responsible for keeping track of it. By contrast, +whether or not a wakeup-capable device should issue wakeup events is a policy +decision, and it is managed by user space through a sysfs attribute: the +:file:`power/wakeup` file. User space can write the "enabled" or "disabled" +strings to it to indicate whether or not, respectively, the device is supposed +to signal system wakeup. This file is only present if the +:c:member:`power.wakeup` object exists for the given device and is created (or +removed) along with that object, by :c:func:`device_set_wakeup_capable()`. +Reads from the file will return the corresponding string. + +The initial value in the :file:`power/wakeup` file is "disabled" for the +majority of devices; the major exceptions are power buttons, keyboards, and +Ethernet adapters whose WoL (wake-on-LAN) feature has been set up with ethtool. +It should also default to "enabled" for devices that don't generate wakeup +requests on their own but merely forward wakeup requests from one bus to another +(like PCI Express ports). + +The :c:func:`device_may_wakeup()` routine returns true only if the +:c:member:`power.wakeup` object exists and the corresponding :file:`power/wakeup` +file contains the "enabled" string. This information is used by subsystems, +like the PCI bus type code, to see whether or not to enable the devices' wakeup +mechanisms. If device wakeup mechanisms are enabled or disabled directly by +drivers, they also should use :c:func:`device_may_wakeup()` to decide what to do +during a system sleep transition. Device drivers, however, are not expected to +call :c:func:`device_set_wakeup_enable()` directly in any case. + +It ought to be noted that system wakeup is conceptually different from "remote +wakeup" used by runtime power management, although it may be supported by the +same physical mechanism. Remote wakeup is a feature allowing devices in +low-power states to trigger specific interrupts to signal conditions in which +they should be put into the full-power state. Those interrupts may or may not +be used to signal system wakeup events, depending on the hardware design. On +some systems it is impossible to trigger them from system sleep states. In any +case, remote wakeup should always be enabled for runtime power management for +all devices and drivers that support it. + + +:file:`/sys/devices/.../power/control` files +-------------------------------------------- + +Each device in the driver model has a flag to control whether it is subject to +runtime power management. This flag, :c:member:`runtime_auto`, is initialized +by the bus type (or generally subsystem) code using :c:func:`pm_runtime_allow()` +or :c:func:`pm_runtime_forbid()`; the default is to allow runtime power +management. + +The setting can be adjusted by user space by writing either "on" or "auto" to +the device's :file:`power/control` sysfs file. Writing "auto" calls +:c:func:`pm_runtime_allow()`, setting the flag and allowing the device to be +runtime power-managed by its driver. Writing "on" calls +:c:func:`pm_runtime_forbid()`, clearing the flag, returning the device to full +power if it was in a low-power state, and preventing the +device from being runtime power-managed. User space can check the current value +of the :c:member:`runtime_auto` flag by reading that file. + +The device's :c:member:`runtime_auto` flag has no effect on the handling of +system-wide power transitions. In particular, the device can (and in the +majority of cases should and will) be put into a low-power state during a +system-wide transition to a sleep state even though its :c:member:`runtime_auto` +flag is clear. + +For more information about the runtime power management framework, refer to +:file:`Documentation/power/runtime_pm.txt`. + + +Calling Drivers to Enter and Leave System Sleep States +====================================================== + +When the system goes into a sleep state, each device's driver is asked to +suspend the device by putting it into a state compatible with the target +system state. That's usually some version of "off", but the details are +system-specific. Also, wakeup-enabled devices will usually stay partly +functional in order to wake the system. + +When the system leaves that low-power state, the device's driver is asked to +resume it by returning it to full power. The suspend and resume operations +always go together, and both are multi-phase operations. + +For simple drivers, suspend might quiesce the device using class code +and then turn its hardware as "off" as possible during suspend_noirq. The +matching resume calls would then completely reinitialize the hardware +before reactivating its class I/O queues. + +More power-aware drivers might prepare the devices for triggering system wakeup +events. + + +Call Sequence Guarantees +------------------------ + +To ensure that bridges and similar links needing to talk to a device are +available when the device is suspended or resumed, the device hierarchy is +walked in a bottom-up order to suspend devices. A top-down order is +used to resume those devices. + +The ordering of the device hierarchy is defined by the order in which devices +get registered: a child can never be registered, probed or resumed before +its parent; and can't be removed or suspended after that parent. + +The policy is that the device hierarchy should match hardware bus topology. +[Or at least the control bus, for devices which use multiple busses.] +In particular, this means that a device registration may fail if the parent of +the device is suspending (i.e. has been chosen by the PM core as the next +device to suspend) or has already suspended, as well as after all of the other +devices have been suspended. Device drivers must be prepared to cope with such +situations. + + +System Power Management Phases +------------------------------ + +Suspending or resuming the system is done in several phases. Different phases +are used for suspend-to-idle, shallow (standby), and deep ("suspend-to-RAM") +sleep states and the hibernation state ("suspend-to-disk"). Each phase involves +executing callbacks for every device before the next phase begins. Not all +buses or classes support all these callbacks and not all drivers use all the +callbacks. The various phases always run after tasks have been frozen and +before they are unfrozen. Furthermore, the ``*_noirq`` phases run at a time +when IRQ handlers have been disabled (except for those marked with the +IRQF_NO_SUSPEND flag). + +All phases use PM domain, bus, type, class or driver callbacks (that is, methods +defined in ``dev->pm_domain->ops``, ``dev->bus->pm``, ``dev->type->pm``, +``dev->class->pm`` or ``dev->driver->pm``). These callbacks are regarded by the +PM core as mutually exclusive. Moreover, PM domain callbacks always take +precedence over all of the other callbacks and, for example, type callbacks take +precedence over bus, class and driver callbacks. To be precise, the following +rules are used to determine which callback to execute in the given phase: + + 1. If ``dev->pm_domain`` is present, the PM core will choose the callback + provided by ``dev->pm_domain->ops`` for execution. + + 2. Otherwise, if both ``dev->type`` and ``dev->type->pm`` are present, the + callback provided by ``dev->type->pm`` will be chosen for execution. + + 3. Otherwise, if both ``dev->class`` and ``dev->class->pm`` are present, + the callback provided by ``dev->class->pm`` will be chosen for + execution. + + 4. Otherwise, if both ``dev->bus`` and ``dev->bus->pm`` are present, the + callback provided by ``dev->bus->pm`` will be chosen for execution. + +This allows PM domains and device types to override callbacks provided by bus +types or device classes if necessary. + +The PM domain, type, class and bus callbacks may in turn invoke device- or +driver-specific methods stored in ``dev->driver->pm``, but they don't have to do +that. + +If the subsystem callback chosen for execution is not present, the PM core will +execute the corresponding method from the ``dev->driver->pm`` set instead if +there is one. + + +Entering System Suspend +----------------------- + +When the system goes into the freeze, standby or memory sleep state, +the phases are: ``prepare``, ``suspend``, ``suspend_late``, ``suspend_noirq``. + + 1. The ``prepare`` phase is meant to prevent races by preventing new + devices from being registered; the PM core would never know that all the + children of a device had been suspended if new children could be + registered at will. [By contrast, from the PM core's perspective, + devices may be unregistered at any time.] Unlike the other + suspend-related phases, during the ``prepare`` phase the device + hierarchy is traversed top-down. + + After the ``->prepare`` callback method returns, no new children may be + registered below the device. The method may also prepare the device or + driver in some way for the upcoming system power transition, but it + should not put the device into a low-power state. Moreover, if the + device supports runtime power management, the ``->prepare`` callback + method must not update its state in case it is necessary to resume it + from runtime suspend later on. + + For devices supporting runtime power management, the return value of the + prepare callback can be used to indicate to the PM core that it may + safely leave the device in runtime suspend (if runtime-suspended + already), provided that all of the device's descendants are also left in + runtime suspend. Namely, if the prepare callback returns a positive + number and that happens for all of the descendants of the device too, + and all of them (including the device itself) are runtime-suspended, the + PM core will skip the ``suspend``, ``suspend_late`` and + ``suspend_noirq`` phases as well as all of the corresponding phases of + the subsequent device resume for all of these devices. In that case, + the ``->complete`` callback will be invoked directly after the + ``->prepare`` callback and is entirely responsible for putting the + device into a consistent state as appropriate. + + Note that this direct-complete procedure applies even if the device is + disabled for runtime PM; only the runtime-PM status matters. It follows + that if a device has system-sleep callbacks but does not support runtime + PM, then its prepare callback must never return a positive value. This + is because all such devices are initially set to runtime-suspended with + runtime PM disabled. + + This feature also can be controlled by device drivers by using the + ``DPM_FLAG_NEVER_SKIP`` and ``DPM_FLAG_SMART_PREPARE`` driver power + management flags. [Typically, they are set at the time the driver is + probed against the device in question by passing them to the + :c:func:`dev_pm_set_driver_flags` helper function.] If the first of + these flags is set, the PM core will not apply the direct-complete + procedure described above to the given device and, consequenty, to any + of its ancestors. The second flag, when set, informs the middle layer + code (bus types, device types, PM domains, classes) that it should take + the return value of the ``->prepare`` callback provided by the driver + into account and it may only return a positive value from its own + ``->prepare`` callback if the driver's one also has returned a positive + value. + + 2. The ``->suspend`` methods should quiesce the device to stop it from + performing I/O. They also may save the device registers and put it into + the appropriate low-power state, depending on the bus type the device is + on, and they may enable wakeup events. + + However, for devices supporting runtime power management, the + ``->suspend`` methods provided by subsystems (bus types and PM domains + in particular) must follow an additional rule regarding what can be done + to the devices before their drivers' ``->suspend`` methods are called. + Namely, they can only resume the devices from runtime suspend by + calling :c:func:`pm_runtime_resume` for them, if that is necessary, and + they must not update the state of the devices in any other way at that + time (in case the drivers need to resume the devices from runtime + suspend in their ``->suspend`` methods). + + 3. For a number of devices it is convenient to split suspend into the + "quiesce device" and "save device state" phases, in which cases + ``suspend_late`` is meant to do the latter. It is always executed after + runtime power management has been disabled for the device in question. + + 4. The ``suspend_noirq`` phase occurs after IRQ handlers have been disabled, + which means that the driver's interrupt handler will not be called while + the callback method is running. The ``->suspend_noirq`` methods should + save the values of the device's registers that weren't saved previously + and finally put the device into the appropriate low-power state. + + The majority of subsystems and device drivers need not implement this + callback. However, bus types allowing devices to share interrupt + vectors, like PCI, generally need it; otherwise a driver might encounter + an error during the suspend phase by fielding a shared interrupt + generated by some other device after its own device had been set to low + power. + +At the end of these phases, drivers should have stopped all I/O transactions +(DMA, IRQs), saved enough state that they can re-initialize or restore previous +state (as needed by the hardware), and placed the device into a low-power state. +On many platforms they will gate off one or more clock sources; sometimes they +will also switch off power supplies or reduce voltages. [Drivers supporting +runtime PM may already have performed some or all of these steps.] + +If :c:func:`device_may_wakeup(dev)` returns ``true``, the device should be +prepared for generating hardware wakeup signals to trigger a system wakeup event +when the system is in the sleep state. For example, :c:func:`enable_irq_wake()` +might identify GPIO signals hooked up to a switch or other external hardware, +and :c:func:`pci_enable_wake()` does something similar for the PCI PME signal. + +If any of these callbacks returns an error, the system won't enter the desired +low-power state. Instead, the PM core will unwind its actions by resuming all +the devices that were suspended. + + +Leaving System Suspend +---------------------- + +When resuming from freeze, standby or memory sleep, the phases are: +``resume_noirq``, ``resume_early``, ``resume``, ``complete``. + + 1. The ``->resume_noirq`` callback methods should perform any actions + needed before the driver's interrupt handlers are invoked. This + generally means undoing the actions of the ``suspend_noirq`` phase. If + the bus type permits devices to share interrupt vectors, like PCI, the + method should bring the device and its driver into a state in which the + driver can recognize if the device is the source of incoming interrupts, + if any, and handle them correctly. + + For example, the PCI bus type's ``->pm.resume_noirq()`` puts the device + into the full-power state (D0 in the PCI terminology) and restores the + standard configuration registers of the device. Then it calls the + device driver's ``->pm.resume_noirq()`` method to perform device-specific + actions. + + 2. The ``->resume_early`` methods should prepare devices for the execution + of the resume methods. This generally involves undoing the actions of + the preceding ``suspend_late`` phase. + + 3. The ``->resume`` methods should bring the device back to its operating + state, so that it can perform normal I/O. This generally involves + undoing the actions of the ``suspend`` phase. + + 4. The ``complete`` phase should undo the actions of the ``prepare`` phase. + For this reason, unlike the other resume-related phases, during the + ``complete`` phase the device hierarchy is traversed bottom-up. + + Note, however, that new children may be registered below the device as + soon as the ``->resume`` callbacks occur; it's not necessary to wait + until the ``complete`` phase with that. + + Moreover, if the preceding ``->prepare`` callback returned a positive + number, the device may have been left in runtime suspend throughout the + whole system suspend and resume (the ``suspend``, ``suspend_late``, + ``suspend_noirq`` phases of system suspend and the ``resume_noirq``, + ``resume_early``, ``resume`` phases of system resume may have been + skipped for it). In that case, the ``->complete`` callback is entirely + responsible for putting the device into a consistent state after system + suspend if necessary. [For example, it may need to queue up a runtime + resume request for the device for this purpose.] To check if that is + the case, the ``->complete`` callback can consult the device's + ``power.direct_complete`` flag. Namely, if that flag is set when the + ``->complete`` callback is being run, it has been called directly after + the preceding ``->prepare`` and special actions may be required + to make the device work correctly afterward. + +At the end of these phases, drivers should be as functional as they were before +suspending: I/O can be performed using DMA and IRQs, and the relevant clocks are +gated on. + +However, the details here may again be platform-specific. For example, +some systems support multiple "run" states, and the mode in effect at +the end of resume might not be the one which preceded suspension. +That means availability of certain clocks or power supplies changed, +which could easily affect how a driver works. + +Drivers need to be able to handle hardware which has been reset since all of the +suspend methods were called, for example by complete reinitialization. +This may be the hardest part, and the one most protected by NDA'd documents +and chip errata. It's simplest if the hardware state hasn't changed since +the suspend was carried out, but that can only be guaranteed if the target +system sleep entered was suspend-to-idle. For the other system sleep states +that may not be the case (and usually isn't for ACPI-defined system sleep +states, like S3). + +Drivers must also be prepared to notice that the device has been removed +while the system was powered down, whenever that's physically possible. +PCMCIA, MMC, USB, Firewire, SCSI, and even IDE are common examples of busses +where common Linux platforms will see such removal. Details of how drivers +will notice and handle such removals are currently bus-specific, and often +involve a separate thread. + +These callbacks may return an error value, but the PM core will ignore such +errors since there's nothing it can do about them other than printing them in +the system log. + + +Entering Hibernation +-------------------- + +Hibernating the system is more complicated than putting it into sleep states, +because it involves creating and saving a system image. Therefore there are +more phases for hibernation, with a different set of callbacks. These phases +always run after tasks have been frozen and enough memory has been freed. + +The general procedure for hibernation is to quiesce all devices ("freeze"), +create an image of the system memory while everything is stable, reactivate all +devices ("thaw"), write the image to permanent storage, and finally shut down +the system ("power off"). The phases used to accomplish this are: ``prepare``, +``freeze``, ``freeze_late``, ``freeze_noirq``, ``thaw_noirq``, ``thaw_early``, +``thaw``, ``complete``, ``prepare``, ``poweroff``, ``poweroff_late``, +``poweroff_noirq``. + + 1. The ``prepare`` phase is discussed in the "Entering System Suspend" + section above. + + 2. The ``->freeze`` methods should quiesce the device so that it doesn't + generate IRQs or DMA, and they may need to save the values of device + registers. However the device does not have to be put in a low-power + state, and to save time it's best not to do so. Also, the device should + not be prepared to generate wakeup events. + + 3. The ``freeze_late`` phase is analogous to the ``suspend_late`` phase + described earlier, except that the device should not be put into a + low-power state and should not be allowed to generate wakeup events. + + 4. The ``freeze_noirq`` phase is analogous to the ``suspend_noirq`` phase + discussed earlier, except again that the device should not be put into + a low-power state and should not be allowed to generate wakeup events. + +At this point the system image is created. All devices should be inactive and +the contents of memory should remain undisturbed while this happens, so that the +image forms an atomic snapshot of the system state. + + 5. The ``thaw_noirq`` phase is analogous to the ``resume_noirq`` phase + discussed earlier. The main difference is that its methods can assume + the device is in the same state as at the end of the ``freeze_noirq`` + phase. + + 6. The ``thaw_early`` phase is analogous to the ``resume_early`` phase + described above. Its methods should undo the actions of the preceding + ``freeze_late``, if necessary. + + 7. The ``thaw`` phase is analogous to the ``resume`` phase discussed + earlier. Its methods should bring the device back to an operating + state, so that it can be used for saving the image if necessary. + + 8. The ``complete`` phase is discussed in the "Leaving System Suspend" + section above. + +At this point the system image is saved, and the devices then need to be +prepared for the upcoming system shutdown. This is much like suspending them +before putting the system into the suspend-to-idle, shallow or deep sleep state, +and the phases are similar. + + 9. The ``prepare`` phase is discussed above. + + 10. The ``poweroff`` phase is analogous to the ``suspend`` phase. + + 11. The ``poweroff_late`` phase is analogous to the ``suspend_late`` phase. + + 12. The ``poweroff_noirq`` phase is analogous to the ``suspend_noirq`` phase. + +The ``->poweroff``, ``->poweroff_late`` and ``->poweroff_noirq`` callbacks +should do essentially the same things as the ``->suspend``, ``->suspend_late`` +and ``->suspend_noirq`` callbacks, respectively. The only notable difference is +that they need not store the device register values, because the registers +should already have been stored during the ``freeze``, ``freeze_late`` or +``freeze_noirq`` phases. + + +Leaving Hibernation +------------------- + +Resuming from hibernation is, again, more complicated than resuming from a sleep +state in which the contents of main memory are preserved, because it requires +a system image to be loaded into memory and the pre-hibernation memory contents +to be restored before control can be passed back to the image kernel. + +Although in principle the image might be loaded into memory and the +pre-hibernation memory contents restored by the boot loader, in practice this +can't be done because boot loaders aren't smart enough and there is no +established protocol for passing the necessary information. So instead, the +boot loader loads a fresh instance of the kernel, called "the restore kernel", +into memory and passes control to it in the usual way. Then the restore kernel +reads the system image, restores the pre-hibernation memory contents, and passes +control to the image kernel. Thus two different kernel instances are involved +in resuming from hibernation. In fact, the restore kernel may be completely +different from the image kernel: a different configuration and even a different +version. This has important consequences for device drivers and their +subsystems. + +To be able to load the system image into memory, the restore kernel needs to +include at least a subset of device drivers allowing it to access the storage +medium containing the image, although it doesn't need to include all of the +drivers present in the image kernel. After the image has been loaded, the +devices managed by the boot kernel need to be prepared for passing control back +to the image kernel. This is very similar to the initial steps involved in +creating a system image, and it is accomplished in the same way, using +``prepare``, ``freeze``, and ``freeze_noirq`` phases. However, the devices +affected by these phases are only those having drivers in the restore kernel; +other devices will still be in whatever state the boot loader left them. + +Should the restoration of the pre-hibernation memory contents fail, the restore +kernel would go through the "thawing" procedure described above, using the +``thaw_noirq``, ``thaw_early``, ``thaw``, and ``complete`` phases, and then +continue running normally. This happens only rarely. Most often the +pre-hibernation memory contents are restored successfully and control is passed +to the image kernel, which then becomes responsible for bringing the system back +to the working state. + +To achieve this, the image kernel must restore the devices' pre-hibernation +functionality. The operation is much like waking up from a sleep state (with +the memory contents preserved), although it involves different phases: +``restore_noirq``, ``restore_early``, ``restore``, ``complete``. + + 1. The ``restore_noirq`` phase is analogous to the ``resume_noirq`` phase. + + 2. The ``restore_early`` phase is analogous to the ``resume_early`` phase. + + 3. The ``restore`` phase is analogous to the ``resume`` phase. + + 4. The ``complete`` phase is discussed above. + +The main difference from ``resume[_early|_noirq]`` is that +``restore[_early|_noirq]`` must assume the device has been accessed and +reconfigured by the boot loader or the restore kernel. Consequently, the state +of the device may be different from the state remembered from the ``freeze``, +``freeze_late`` and ``freeze_noirq`` phases. The device may even need to be +reset and completely re-initialized. In many cases this difference doesn't +matter, so the ``->resume[_early|_noirq]`` and ``->restore[_early|_norq]`` +method pointers can be set to the same routines. Nevertheless, different +callback pointers are used in case there is a situation where it actually does +matter. + + +Power Management Notifiers +========================== + +There are some operations that cannot be carried out by the power management +callbacks discussed above, because the callbacks occur too late or too early. +To handle these cases, subsystems and device drivers may register power +management notifiers that are called before tasks are frozen and after they have +been thawed. Generally speaking, the PM notifiers are suitable for performing +actions that either require user space to be available, or at least won't +interfere with user space. + +For details refer to :doc:`notifiers`. + + +Device Low-Power (suspend) States +================================= + +Device low-power states aren't standard. One device might only handle +"on" and "off", while another might support a dozen different versions of +"on" (how many engines are active?), plus a state that gets back to "on" +faster than from a full "off". + +Some buses define rules about what different suspend states mean. PCI +gives one example: after the suspend sequence completes, a non-legacy +PCI device may not perform DMA or issue IRQs, and any wakeup events it +issues would be issued through the PME# bus signal. Plus, there are +several PCI-standard device states, some of which are optional. + +In contrast, integrated system-on-chip processors often use IRQs as the +wakeup event sources (so drivers would call :c:func:`enable_irq_wake`) and +might be able to treat DMA completion as a wakeup event (sometimes DMA can stay +active too, it'd only be the CPU and some peripherals that sleep). + +Some details here may be platform-specific. Systems may have devices that +can be fully active in certain sleep states, such as an LCD display that's +refreshed using DMA while most of the system is sleeping lightly ... and +its frame buffer might even be updated by a DSP or other non-Linux CPU while +the Linux control processor stays idle. + +Moreover, the specific actions taken may depend on the target system state. +One target system state might allow a given device to be very operational; +another might require a hard shut down with re-initialization on resume. +And two different target systems might use the same device in different +ways; the aforementioned LCD might be active in one product's "standby", +but a different product using the same SOC might work differently. + + +Device Power Management Domains +=============================== + +Sometimes devices share reference clocks or other power resources. In those +cases it generally is not possible to put devices into low-power states +individually. Instead, a set of devices sharing a power resource can be put +into a low-power state together at the same time by turning off the shared +power resource. Of course, they also need to be put into the full-power state +together, by turning the shared power resource on. A set of devices with this +property is often referred to as a power domain. A power domain may also be +nested inside another power domain. The nested domain is referred to as the +sub-domain of the parent domain. + +Support for power domains is provided through the :c:member:`pm_domain` field of +|struct device|. This field is a pointer to an object of type +|struct dev_pm_domain|, defined in :file:`include/linux/pm.h`, providing a set +of power management callbacks analogous to the subsystem-level and device driver +callbacks that are executed for the given device during all power transitions, +instead of the respective subsystem-level callbacks. Specifically, if a +device's :c:member:`pm_domain` pointer is not NULL, the ``->suspend()`` callback +from the object pointed to by it will be executed instead of its subsystem's +(e.g. bus type's) ``->suspend()`` callback and analogously for all of the +remaining callbacks. In other words, power management domain callbacks, if +defined for the given device, always take precedence over the callbacks provided +by the device's subsystem (e.g. bus type). + +The support for device power management domains is only relevant to platforms +needing to use the same device driver power management callbacks in many +different power domain configurations and wanting to avoid incorporating the +support for power domains into subsystem-level callbacks, for example by +modifying the platform bus type. Other platforms need not implement it or take +it into account in any way. + +Devices may be defined as IRQ-safe which indicates to the PM core that their +runtime PM callbacks may be invoked with disabled interrupts (see +:file:`Documentation/power/runtime_pm.txt` for more information). If an +IRQ-safe device belongs to a PM domain, the runtime PM of the domain will be +disallowed, unless the domain itself is defined as IRQ-safe. However, it +makes sense to define a PM domain as IRQ-safe only if all the devices in it +are IRQ-safe. Moreover, if an IRQ-safe domain has a parent domain, the runtime +PM of the parent is only allowed if the parent itself is IRQ-safe too with the +additional restriction that all child domains of an IRQ-safe parent must also +be IRQ-safe. + + +Runtime Power Management +======================== + +Many devices are able to dynamically power down while the system is still +running. This feature is useful for devices that are not being used, and +can offer significant power savings on a running system. These devices +often support a range of runtime power states, which might use names such +as "off", "sleep", "idle", "active", and so on. Those states will in some +cases (like PCI) be partially constrained by the bus the device uses, and will +usually include hardware states that are also used in system sleep states. + +A system-wide power transition can be started while some devices are in low +power states due to runtime power management. The system sleep PM callbacks +should recognize such situations and react to them appropriately, but the +necessary actions are subsystem-specific. + +In some cases the decision may be made at the subsystem level while in other +cases the device driver may be left to decide. In some cases it may be +desirable to leave a suspended device in that state during a system-wide power +transition, but in other cases the device must be put back into the full-power +state temporarily, for example so that its system wakeup capability can be +disabled. This all depends on the hardware and the design of the subsystem and +device driver in question. + +If it is necessary to resume a device from runtime suspend during a system-wide +transition into a sleep state, that can be done by calling +:c:func:`pm_runtime_resume` for it from the ``->suspend`` callback (or its +couterpart for transitions related to hibernation) of either the device's driver +or a subsystem responsible for it (for example, a bus type or a PM domain). +That is guaranteed to work by the requirement that subsystems must not change +the state of devices (possibly except for resuming them from runtime suspend) +from their ``->prepare`` and ``->suspend`` callbacks (or equivalent) *before* +invoking device drivers' ``->suspend`` callbacks (or equivalent). + +Some bus types and PM domains have a policy to resume all devices from runtime +suspend upfront in their ``->suspend`` callbacks, but that may not be really +necessary if the driver of the device can cope with runtime-suspended devices. +The driver can indicate that by setting ``DPM_FLAG_SMART_SUSPEND`` in +:c:member:`power.driver_flags` at the probe time, by passing it to the +:c:func:`dev_pm_set_driver_flags` helper. That also may cause middle-layer code +(bus types, PM domains etc.) to skip the ``->suspend_late`` and +``->suspend_noirq`` callbacks provided by the driver if the device remains in +runtime suspend at the beginning of the ``suspend_late`` phase of system-wide +suspend (or in the ``poweroff_late`` phase of hibernation), when runtime PM +has been disabled for it, under the assumption that its state should not change +after that point until the system-wide transition is over (the PM core itself +does that for devices whose "noirq", "late" and "early" system-wide PM callbacks +are executed directly by it). If that happens, the driver's system-wide resume +callbacks, if present, may still be invoked during the subsequent system-wide +resume transition and the device's runtime power management status may be set +to "active" before enabling runtime PM for it, so the driver must be prepared to +cope with the invocation of its system-wide resume callbacks back-to-back with +its ``->runtime_suspend`` one (without the intervening ``->runtime_resume`` and +so on) and the final state of the device must reflect the "active" runtime PM +status in that case. + +During system-wide resume from a sleep state it's easiest to put devices into +the full-power state, as explained in :file:`Documentation/power/runtime_pm.txt`. +[Refer to that document for more information regarding this particular issue as +well as for information on the device runtime power management framework in +general.] + +However, it often is desirable to leave devices in suspend after system +transitions to the working state, especially if those devices had been in +runtime suspend before the preceding system-wide suspend (or analogous) +transition. Device drivers can use the ``DPM_FLAG_LEAVE_SUSPENDED`` flag to +indicate to the PM core (and middle-layer code) that they prefer the specific +devices handled by them to be left suspended and they have no problems with +skipping their system-wide resume callbacks for this reason. Whether or not the +devices will actually be left in suspend may depend on their state before the +given system suspend-resume cycle and on the type of the system transition under +way. In particular, devices are not left suspended if that transition is a +restore from hibernation, as device states are not guaranteed to be reflected +by the information stored in the hibernation image in that case. + +The middle-layer code involved in the handling of the device is expected to +indicate to the PM core if the device may be left in suspend by setting its +:c:member:`power.may_skip_resume` status bit which is checked by the PM core +during the "noirq" phase of the preceding system-wide suspend (or analogous) +transition. The middle layer is then responsible for handling the device as +appropriate in its "noirq" resume callback, which is executed regardless of +whether or not the device is left suspended, but the other resume callbacks +(except for ``->complete``) will be skipped automatically by the PM core if the +device really can be left in suspend. + +For devices whose "noirq", "late" and "early" driver callbacks are invoked +directly by the PM core, all of the system-wide resume callbacks are skipped if +``DPM_FLAG_LEAVE_SUSPENDED`` is set and the device is in runtime suspend during +the ``suspend_noirq`` (or analogous) phase or the transition under way is a +proper system suspend (rather than anything related to hibernation) and the +device's wakeup settings are suitable for runtime PM (that is, it cannot +generate wakeup signals at all or it is allowed to wake up the system from +sleep). diff --git a/Documentation/driver-api/pm/index.rst b/Documentation/driver-api/pm/index.rst new file mode 100644 index 000000000..2f6d0e9cf --- /dev/null +++ b/Documentation/driver-api/pm/index.rst @@ -0,0 +1,16 @@ +======================= +Device Power Management +======================= + +.. toctree:: + + devices + notifiers + types + +.. only:: subproject and html + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/pm/notifiers.rst b/Documentation/driver-api/pm/notifiers.rst new file mode 100644 index 000000000..62f860026 --- /dev/null +++ b/Documentation/driver-api/pm/notifiers.rst @@ -0,0 +1,70 @@ +============================= +Suspend/Hibernation Notifiers +============================= + +:: + + Copyright (c) 2016 Intel Corp., Rafael J. Wysocki <rafael.j.wysocki@intel.com> + +There are some operations that subsystems or drivers may want to carry out +before hibernation/suspend or after restore/resume, but they require the system +to be fully functional, so the drivers' and subsystems' ``->suspend()`` and +``->resume()`` or even ``->prepare()`` and ``->complete()`` callbacks are not +suitable for this purpose. + +For example, device drivers may want to upload firmware to their devices after +resume/restore, but they cannot do it by calling :c:func:`request_firmware()` +from their ``->resume()`` or ``->complete()`` callback routines (user land +processes are frozen at these points). The solution may be to load the firmware +into memory before processes are frozen and upload it from there in the +``->resume()`` routine. A suspend/hibernation notifier may be used for that. + +Subsystems or drivers having such needs can register suspend notifiers that +will be called upon the following events by the PM core: + +``PM_HIBERNATION_PREPARE`` + The system is going to hibernate, tasks will be frozen immediately. This + is different from ``PM_SUSPEND_PREPARE`` below, because in this case + additional work is done between the notifiers and the invocation of PM + callbacks for the "freeze" transition. + +``PM_POST_HIBERNATION`` + The system memory state has been restored from a hibernation image or an + error occurred during hibernation. Device restore callbacks have been + executed and tasks have been thawed. + +``PM_RESTORE_PREPARE`` + The system is going to restore a hibernation image. If all goes well, + the restored image kernel will issue a ``PM_POST_HIBERNATION`` + notification. + +``PM_POST_RESTORE`` + An error occurred during restore from hibernation. Device restore + callbacks have been executed and tasks have been thawed. + +``PM_SUSPEND_PREPARE`` + The system is preparing for suspend. + +``PM_POST_SUSPEND`` + The system has just resumed or an error occurred during suspend. Device + resume callbacks have been executed and tasks have been thawed. + +It is generally assumed that whatever the notifiers do for +``PM_HIBERNATION_PREPARE``, should be undone for ``PM_POST_HIBERNATION``. +Analogously, operations carried out for ``PM_SUSPEND_PREPARE`` should be +reversed for ``PM_POST_SUSPEND``. + +Moreover, if one of the notifiers fails for the ``PM_HIBERNATION_PREPARE`` or +``PM_SUSPEND_PREPARE`` event, the notifiers that have already succeeded for that +event will be called for ``PM_POST_HIBERNATION`` or ``PM_POST_SUSPEND``, +respectively. + +The hibernation and suspend notifiers are called with :c:data:`pm_mutex` held. +They are defined in the usual way, but their last argument is meaningless (it is +always NULL). + +To register and/or unregister a suspend notifier use +:c:func:`register_pm_notifier()` and :c:func:`unregister_pm_notifier()`, +respectively (both defined in :file:`include/linux/suspend.h`). If you don't +need to unregister the notifier, you can also use the :c:func:`pm_notifier()` +macro defined in :file:`include/linux/suspend.h`. diff --git a/Documentation/driver-api/pm/types.rst b/Documentation/driver-api/pm/types.rst new file mode 100644 index 000000000..3ebdecc54 --- /dev/null +++ b/Documentation/driver-api/pm/types.rst @@ -0,0 +1,5 @@ +================================== +Device Power Management Data Types +================================== + +.. kernel-doc:: include/linux/pm.h diff --git a/Documentation/driver-api/rapidio.rst b/Documentation/driver-api/rapidio.rst new file mode 100644 index 000000000..71ff658ab --- /dev/null +++ b/Documentation/driver-api/rapidio.rst @@ -0,0 +1,107 @@ +======================= +RapidIO Subsystem Guide +======================= + +:Author: Matt Porter + +Introduction +============ + +RapidIO is a high speed switched fabric interconnect with features aimed +at the embedded market. RapidIO provides support for memory-mapped I/O +as well as message-based transactions over the switched fabric network. +RapidIO has a standardized discovery mechanism not unlike the PCI bus +standard that allows simple detection of devices in a network. + +This documentation is provided for developers intending to support +RapidIO on new architectures, write new drivers, or to understand the +subsystem internals. + +Known Bugs and Limitations +========================== + +Bugs +---- + +None. ;) + +Limitations +----------- + +1. Access/management of RapidIO memory regions is not supported + +2. Multiple host enumeration is not supported + +RapidIO driver interface +======================== + +Drivers are provided a set of calls in order to interface with the +subsystem to gather info on devices, request/map memory region +resources, and manage mailboxes/doorbells. + +Functions +--------- + +.. kernel-doc:: include/linux/rio_drv.h + :internal: + +.. kernel-doc:: drivers/rapidio/rio-driver.c + :export: + +.. kernel-doc:: drivers/rapidio/rio.c + :export: + +Internals +========= + +This chapter contains the autogenerated documentation of the RapidIO +subsystem. + +Structures +---------- + +.. kernel-doc:: include/linux/rio.h + :internal: + +Enumeration and Discovery +------------------------- + +.. kernel-doc:: drivers/rapidio/rio-scan.c + :internal: + +Driver functionality +-------------------- + +.. kernel-doc:: drivers/rapidio/rio.c + :internal: + +.. kernel-doc:: drivers/rapidio/rio-access.c + :internal: + +Device model support +-------------------- + +.. kernel-doc:: drivers/rapidio/rio-driver.c + :internal: + +PPC32 support +------------- + +.. kernel-doc:: arch/powerpc/sysdev/fsl_rio.c + :internal: + +Credits +======= + +The following people have contributed to the RapidIO subsystem directly +or indirectly: + +1. Matt Porter\ mporter@kernel.crashing.org + +2. Randy Vinson\ rvinson@mvista.com + +3. Dan Malek\ dan@embeddedalley.com + +The following people have contributed to this document: + +1. Matt Porter\ mporter@kernel.crashing.org diff --git a/Documentation/driver-api/regulator.rst b/Documentation/driver-api/regulator.rst new file mode 100644 index 000000000..520da0a52 --- /dev/null +++ b/Documentation/driver-api/regulator.rst @@ -0,0 +1,170 @@ +.. Copyright 2007-2008 Wolfson Microelectronics + +.. This documentation is free software; you can redistribute +.. it and/or modify it under the terms of the GNU General Public +.. License version 2 as published by the Free Software Foundation. + +================================= +Voltage and current regulator API +================================= + +:Author: Liam Girdwood +:Author: Mark Brown + +Introduction +============ + +This framework is designed to provide a standard kernel interface to +control voltage and current regulators. + +The intention is to allow systems to dynamically control regulator power +output in order to save power and prolong battery life. This applies to +both voltage regulators (where voltage output is controllable) and +current sinks (where current limit is controllable). + +Note that additional (and currently more complete) documentation is +available in the Linux kernel source under +``Documentation/power/regulator``. + +Glossary +-------- + +The regulator API uses a number of terms which may not be familiar: + +Regulator + + Electronic device that supplies power to other devices. Most regulators + can enable and disable their output and some can also control their + output voltage or current. + +Consumer + + Electronic device which consumes power provided by a regulator. These + may either be static, requiring only a fixed supply, or dynamic, + requiring active management of the regulator at runtime. + +Power Domain + + The electronic circuit supplied by a given regulator, including the + regulator and all consumer devices. The configuration of the regulator + is shared between all the components in the circuit. + +Power Management Integrated Circuit (PMIC) + + An IC which contains numerous regulators and often also other + subsystems. In an embedded system the primary PMIC is often equivalent + to a combination of the PSU and southbridge in a desktop system. + +Consumer driver interface +========================= + +This offers a similar API to the kernel clock framework. Consumer +drivers use `get <#API-regulator-get>`__ and +`put <#API-regulator-put>`__ operations to acquire and release +regulators. Functions are provided to `enable <#API-regulator-enable>`__ +and `disable <#API-regulator-disable>`__ the regulator and to get and +set the runtime parameters of the regulator. + +When requesting regulators consumers use symbolic names for their +supplies, such as "Vcc", which are mapped into actual regulator devices +by the machine interface. + +A stub version of this API is provided when the regulator framework is +not in use in order to minimise the need to use ifdefs. + +Enabling and disabling +---------------------- + +The regulator API provides reference counted enabling and disabling of +regulators. Consumer devices use the :c:func:`regulator_enable()` and +:c:func:`regulator_disable()` functions to enable and disable +regulators. Calls to the two functions must be balanced. + +Note that since multiple consumers may be using a regulator and machine +constraints may not allow the regulator to be disabled there is no +guarantee that calling :c:func:`regulator_disable()` will actually +cause the supply provided by the regulator to be disabled. Consumer +drivers should assume that the regulator may be enabled at all times. + +Configuration +------------- + +Some consumer devices may need to be able to dynamically configure their +supplies. For example, MMC drivers may need to select the correct +operating voltage for their cards. This may be done while the regulator +is enabled or disabled. + +The :c:func:`regulator_set_voltage()` and +:c:func:`regulator_set_current_limit()` functions provide the primary +interface for this. Both take ranges of voltages and currents, supporting +drivers that do not require a specific value (eg, CPU frequency scaling +normally permits the CPU to use a wider range of supply voltages at lower +frequencies but does not require that the supply voltage be lowered). Where +an exact value is required both minimum and maximum values should be +identical. + +Callbacks +--------- + +Callbacks may also be registered for events such as regulation failures. + +Regulator driver interface +========================== + +Drivers for regulator chips register the regulators with the regulator +core, providing operations structures to the core. A notifier interface +allows error conditions to be reported to the core. + +Registration should be triggered by explicit setup done by the platform, +supplying a struct :c:type:`regulator_init_data` for the regulator +containing constraint and supply information. + +Machine interface +================= + +This interface provides a way to define how regulators are connected to +consumers on a given system and what the valid operating parameters are +for the system. + +Supplies +-------- + +Regulator supplies are specified using struct +:c:type:`regulator_consumer_supply`. This is done at driver registration +time as part of the machine constraints. + +Constraints +----------- + +As well as defining the connections the machine interface also provides +constraints defining the operations that clients are allowed to perform +and the parameters that may be set. This is required since generally +regulator devices will offer more flexibility than it is safe to use on +a given system, for example supporting higher supply voltages than the +consumers are rated for. + +This is done at driver registration time` by providing a +struct :c:type:`regulation_constraints`. + +The constraints may also specify an initial configuration for the +regulator in the constraints, which is particularly useful for use with +static consumers. + +API reference +============= + +Due to limitations of the kernel documentation framework and the +existing layout of the source code the entire regulator API is +documented here. + +.. kernel-doc:: include/linux/regulator/consumer.h + :internal: + +.. kernel-doc:: include/linux/regulator/machine.h + :internal: + +.. kernel-doc:: include/linux/regulator/driver.h + :internal: + +.. kernel-doc:: drivers/regulator/core.c + :export: diff --git a/Documentation/driver-api/s390-drivers.rst b/Documentation/driver-api/s390-drivers.rst new file mode 100644 index 000000000..30e6aa7e1 --- /dev/null +++ b/Documentation/driver-api/s390-drivers.rst @@ -0,0 +1,135 @@ +=================================== +Writing s390 channel device drivers +=================================== + +:Author: Cornelia Huck + +Introduction +============ + +This document describes the interfaces available for device drivers that +drive s390 based channel attached I/O devices. This includes interfaces +for interaction with the hardware and interfaces for interacting with +the common driver core. Those interfaces are provided by the s390 common +I/O layer. + +The document assumes a familarity with the technical terms associated +with the s390 channel I/O architecture. For a description of this +architecture, please refer to the "z/Architecture: Principles of +Operation", IBM publication no. SA22-7832. + +While most I/O devices on a s390 system are typically driven through the +channel I/O mechanism described here, there are various other methods +(like the diag interface). These are out of the scope of this document. + +The s390 common I/O layer also provides access to some devices that are +not strictly considered I/O devices. They are considered here as well, +although they are not the focus of this document. + +Some additional information can also be found in the kernel source under +Documentation/s390/driver-model.txt. + +The css bus +=========== + +The css bus contains the subchannels available on the system. They fall +into several categories: + +* Standard I/O subchannels, for use by the system. They have a child + device on the ccw bus and are described below. +* I/O subchannels bound to the vfio-ccw driver. See + Documentation/s390/vfio-ccw.txt. +* Message subchannels. No Linux driver currently exists. +* CHSC subchannels (at most one). The chsc subchannel driver can be used + to send asynchronous chsc commands. +* eADM subchannels. Used for talking to storage class memory. + +The ccw bus +=========== + +The ccw bus typically contains the majority of devices available to a +s390 system. Named after the channel command word (ccw), the basic +command structure used to address its devices, the ccw bus contains +so-called channel attached devices. They are addressed via I/O +subchannels, visible on the css bus. A device driver for +channel-attached devices, however, will never interact with the +subchannel directly, but only via the I/O device on the ccw bus, the ccw +device. + +I/O functions for channel-attached devices +------------------------------------------ + +Some hardware structures have been translated into C structures for use +by the common I/O layer and device drivers. For more information on the +hardware structures represented here, please consult the Principles of +Operation. + +.. kernel-doc:: arch/s390/include/asm/cio.h + :internal: + +ccw devices +----------- + +Devices that want to initiate channel I/O need to attach to the ccw bus. +Interaction with the driver core is done via the common I/O layer, which +provides the abstractions of ccw devices and ccw device drivers. + +The functions that initiate or terminate channel I/O all act upon a ccw +device structure. Device drivers must not bypass those functions or +strange side effects may happen. + +.. kernel-doc:: arch/s390/include/asm/ccwdev.h + :internal: + +.. kernel-doc:: drivers/s390/cio/device.c + :export: + +.. kernel-doc:: drivers/s390/cio/device_ops.c + :export: + +The channel-measurement facility +-------------------------------- + +The channel-measurement facility provides a means to collect measurement +data which is made available by the channel subsystem for each channel +attached device. + +.. kernel-doc:: arch/s390/include/uapi/asm/cmb.h + :internal: + +.. kernel-doc:: drivers/s390/cio/cmf.c + :export: + +The ccwgroup bus +================ + +The ccwgroup bus only contains artificial devices, created by the user. +Many networking devices (e.g. qeth) are in fact composed of several ccw +devices (like read, write and data channel for qeth). The ccwgroup bus +provides a mechanism to create a meta-device which contains those ccw +devices as slave devices and can be associated with the netdevice. + +ccw group devices +----------------- + +.. kernel-doc:: arch/s390/include/asm/ccwgroup.h + :internal: + +.. kernel-doc:: drivers/s390/cio/ccwgroup.c + :export: + +Generic interfaces +================== + +The following section contains interfaces in use not only by drivers +dealing with ccw devices, but drivers for various other s390 hardware +as well. + +Adapter interrupts +------------------ + +The common I/O layer provides helper functions for dealing with adapter +interrupts and interrupt vectors. + +.. kernel-doc:: drivers/s390/cio/airq.c + :export: diff --git a/Documentation/driver-api/scsi.rst b/Documentation/driver-api/scsi.rst new file mode 100644 index 000000000..64b231d12 --- /dev/null +++ b/Documentation/driver-api/scsi.rst @@ -0,0 +1,338 @@ +===================== +SCSI Interfaces Guide +===================== + +:Author: James Bottomley +:Author: Rob Landley + +Introduction +============ + +Protocol vs bus +--------------- + +Once upon a time, the Small Computer Systems Interface defined both a +parallel I/O bus and a data protocol to connect a wide variety of +peripherals (disk drives, tape drives, modems, printers, scanners, +optical drives, test equipment, and medical devices) to a host computer. + +Although the old parallel (fast/wide/ultra) SCSI bus has largely fallen +out of use, the SCSI command set is more widely used than ever to +communicate with devices over a number of different busses. + +The `SCSI protocol <http://www.t10.org/scsi-3.htm>`__ is a big-endian +peer-to-peer packet based protocol. SCSI commands are 6, 10, 12, or 16 +bytes long, often followed by an associated data payload. + +SCSI commands can be transported over just about any kind of bus, and +are the default protocol for storage devices attached to USB, SATA, SAS, +Fibre Channel, FireWire, and ATAPI devices. SCSI packets are also +commonly exchanged over Infiniband, +`I2O <http://i2o.shadowconnect.com/faq.php>`__, TCP/IP +(`iSCSI <https://en.wikipedia.org/wiki/ISCSI>`__), even `Parallel +ports <http://cyberelk.net/tim/parport/parscsi.html>`__. + +Design of the Linux SCSI subsystem +---------------------------------- + +The SCSI subsystem uses a three layer design, with upper, mid, and low +layers. Every operation involving the SCSI subsystem (such as reading a +sector from a disk) uses one driver at each of the 3 levels: one upper +layer driver, one lower layer driver, and the SCSI midlayer. + +The SCSI upper layer provides the interface between userspace and the +kernel, in the form of block and char device nodes for I/O and ioctl(). +The SCSI lower layer contains drivers for specific hardware devices. + +In between is the SCSI mid-layer, analogous to a network routing layer +such as the IPv4 stack. The SCSI mid-layer routes a packet based data +protocol between the upper layer's /dev nodes and the corresponding +devices in the lower layer. It manages command queues, provides error +handling and power management functions, and responds to ioctl() +requests. + +SCSI upper layer +================ + +The upper layer supports the user-kernel interface by providing device +nodes. + +sd (SCSI Disk) +-------------- + +sd (sd_mod.o) + +sr (SCSI CD-ROM) +---------------- + +sr (sr_mod.o) + +st (SCSI Tape) +-------------- + +st (st.o) + +sg (SCSI Generic) +----------------- + +sg (sg.o) + +ch (SCSI Media Changer) +----------------------- + +ch (ch.c) + +SCSI mid layer +============== + +SCSI midlayer implementation +---------------------------- + +include/scsi/scsi_device.h +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: include/scsi/scsi_device.h + :internal: + +drivers/scsi/scsi.c +~~~~~~~~~~~~~~~~~~~ + +Main file for the SCSI midlayer. + +.. kernel-doc:: drivers/scsi/scsi.c + :export: + +drivers/scsi/scsicam.c +~~~~~~~~~~~~~~~~~~~~~~ + +`SCSI Common Access +Method <http://www.t10.org/ftp/t10/drafts/cam/cam-r12b.pdf>`__ support +functions, for use with HDIO_GETGEO, etc. + +.. kernel-doc:: drivers/scsi/scsicam.c + :export: + +drivers/scsi/scsi_error.c +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Common SCSI error/timeout handling routines. + +.. kernel-doc:: drivers/scsi/scsi_error.c + :export: + +drivers/scsi/scsi_devinfo.c +~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Manage scsi_dev_info_list, which tracks blacklisted and whitelisted +devices. + +.. kernel-doc:: drivers/scsi/scsi_devinfo.c + :internal: + +drivers/scsi/scsi_ioctl.c +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Handle ioctl() calls for SCSI devices. + +.. kernel-doc:: drivers/scsi/scsi_ioctl.c + :export: + +drivers/scsi/scsi_lib.c +~~~~~~~~~~~~~~~~~~~~~~~~ + +SCSI queuing library. + +.. kernel-doc:: drivers/scsi/scsi_lib.c + :export: + +drivers/scsi/scsi_lib_dma.c +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +SCSI library functions depending on DMA (map and unmap scatter-gather +lists). + +.. kernel-doc:: drivers/scsi/scsi_lib_dma.c + :export: + +drivers/scsi/scsi_proc.c +~~~~~~~~~~~~~~~~~~~~~~~~~ + +The functions in this file provide an interface between the PROC file +system and the SCSI device drivers It is mainly used for debugging, +statistics and to pass information directly to the lowlevel driver. I.E. +plumbing to manage /proc/scsi/\* + +.. kernel-doc:: drivers/scsi/scsi_proc.c + :internal: + +drivers/scsi/scsi_netlink.c +~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Infrastructure to provide async events from transports to userspace via +netlink, using a single NETLINK_SCSITRANSPORT protocol for all +transports. See `the original patch +submission <http://marc.info/?l=linux-scsi&m=115507374832500&w=2>`__ for +more details. + +.. kernel-doc:: drivers/scsi/scsi_netlink.c + :internal: + +drivers/scsi/scsi_scan.c +~~~~~~~~~~~~~~~~~~~~~~~~~ + +Scan a host to determine which (if any) devices are attached. The +general scanning/probing algorithm is as follows, exceptions are made to +it depending on device specific flags, compilation options, and global +variable (boot or module load time) settings. A specific LUN is scanned +via an INQUIRY command; if the LUN has a device attached, a scsi_device +is allocated and setup for it. For every id of every channel on the +given host, start by scanning LUN 0. Skip hosts that don't respond at +all to a scan of LUN 0. Otherwise, if LUN 0 has a device attached, +allocate and setup a scsi_device for it. If target is SCSI-3 or up, +issue a REPORT LUN, and scan all of the LUNs returned by the REPORT LUN; +else, sequentially scan LUNs up until some maximum is reached, or a LUN +is seen that cannot have a device attached to it. + +.. kernel-doc:: drivers/scsi/scsi_scan.c + :internal: + +drivers/scsi/scsi_sysctl.c +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Set up the sysctl entry: "/dev/scsi/logging_level" +(DEV_SCSI_LOGGING_LEVEL) which sets/returns scsi_logging_level. + +drivers/scsi/scsi_sysfs.c +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +SCSI sysfs interface routines. + +.. kernel-doc:: drivers/scsi/scsi_sysfs.c + :export: + +drivers/scsi/hosts.c +~~~~~~~~~~~~~~~~~~~~ + +mid to lowlevel SCSI driver interface + +.. kernel-doc:: drivers/scsi/hosts.c + :export: + +drivers/scsi/scsi_common.c +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +general support functions + +.. kernel-doc:: drivers/scsi/scsi_common.c + :export: + +Transport classes +----------------- + +Transport classes are service libraries for drivers in the SCSI lower +layer, which expose transport attributes in sysfs. + +Fibre Channel transport +~~~~~~~~~~~~~~~~~~~~~~~ + +The file drivers/scsi/scsi_transport_fc.c defines transport attributes +for Fibre Channel. + +.. kernel-doc:: drivers/scsi/scsi_transport_fc.c + :export: + +iSCSI transport class +~~~~~~~~~~~~~~~~~~~~~ + +The file drivers/scsi/scsi_transport_iscsi.c defines transport +attributes for the iSCSI class, which sends SCSI packets over TCP/IP +connections. + +.. kernel-doc:: drivers/scsi/scsi_transport_iscsi.c + :export: + +Serial Attached SCSI (SAS) transport class +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The file drivers/scsi/scsi_transport_sas.c defines transport +attributes for Serial Attached SCSI, a variant of SATA aimed at large +high-end systems. + +The SAS transport class contains common code to deal with SAS HBAs, an +aproximated representation of SAS topologies in the driver model, and +various sysfs attributes to expose these topologies and management +interfaces to userspace. + +In addition to the basic SCSI core objects this transport class +introduces two additional intermediate objects: The SAS PHY as +represented by struct sas_phy defines an "outgoing" PHY on a SAS HBA or +Expander, and the SAS remote PHY represented by struct sas_rphy defines +an "incoming" PHY on a SAS Expander or end device. Note that this is +purely a software concept, the underlying hardware for a PHY and a +remote PHY is the exactly the same. + +There is no concept of a SAS port in this code, users can see what PHYs +form a wide port based on the port_identifier attribute, which is the +same for all PHYs in a port. + +.. kernel-doc:: drivers/scsi/scsi_transport_sas.c + :export: + +SATA transport class +~~~~~~~~~~~~~~~~~~~~ + +The SATA transport is handled by libata, which has its own book of +documentation in this directory. + +Parallel SCSI (SPI) transport class +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The file drivers/scsi/scsi_transport_spi.c defines transport +attributes for traditional (fast/wide/ultra) SCSI busses. + +.. kernel-doc:: drivers/scsi/scsi_transport_spi.c + :export: + +SCSI RDMA (SRP) transport class +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The file drivers/scsi/scsi_transport_srp.c defines transport +attributes for SCSI over Remote Direct Memory Access. + +.. kernel-doc:: drivers/scsi/scsi_transport_srp.c + :export: + +SCSI lower layer +================ + +Host Bus Adapter transport types +-------------------------------- + +Many modern device controllers use the SCSI command set as a protocol to +communicate with their devices through many different types of physical +connections. + +In SCSI language a bus capable of carrying SCSI commands is called a +"transport", and a controller connecting to such a bus is called a "host +bus adapter" (HBA). + +Debug transport +~~~~~~~~~~~~~~~ + +The file drivers/scsi/scsi_debug.c simulates a host adapter with a +variable number of disks (or disk like devices) attached, sharing a +common amount of RAM. Does a lot of checking to make sure that we are +not getting blocks mixed up, and panics the kernel if anything out of +the ordinary is seen. + +To be more realistic, the simulated devices have the transport +attributes of SAS disks. + +For documentation see http://sg.danny.cz/sg/sdebug26.html + +todo +~~~~ + +Parallel (fast/wide/ultra) SCSI, USB, SATA, SAS, Fibre Channel, +FireWire, ATAPI devices, Infiniband, I2O, Parallel ports, +netlink... diff --git a/Documentation/driver-api/slimbus.rst b/Documentation/driver-api/slimbus.rst new file mode 100644 index 000000000..410eec79b --- /dev/null +++ b/Documentation/driver-api/slimbus.rst @@ -0,0 +1,132 @@ +============================ +Linux kernel SLIMbus support +============================ + +Overview +======== + +What is SLIMbus? +---------------- +SLIMbus (Serial Low Power Interchip Media Bus) is a specification developed by +MIPI (Mobile Industry Processor Interface) alliance. The bus uses master/slave +configuration, and is a 2-wire multi-drop implementation (clock, and data). + +Currently, SLIMbus is used to interface between application processors of SoCs +(System-on-Chip) and peripheral components (typically codec). SLIMbus uses +Time-Division-Multiplexing to accommodate multiple data channels, and +a control channel. + +The control channel is used for various control functions such as bus +management, configuration and status updates. These messages can be unicast (e.g. +reading/writing device specific values), or multicast (e.g. data channel +reconfiguration sequence is a broadcast message announced to all devices) + +A data channel is used for data-transfer between 2 SLIMbus devices. Data +channel uses dedicated ports on the device. + +Hardware description: +--------------------- +SLIMbus specification has different types of device classifications based on +their capabilities. +A manager device is responsible for enumeration, configuration, and dynamic +channel allocation. Every bus has 1 active manager. + +A generic device is a device providing application functionality (e.g. codec). + +Framer device is responsible for clocking the bus, and transmitting frame-sync +and framing information on the bus. + +Each SLIMbus component has an interface device for monitoring physical layer. + +Typically each SoC contains SLIMbus component having 1 manager, 1 framer device, +1 generic device (for data channel support), and 1 interface device. +External peripheral SLIMbus component usually has 1 generic device (for +functionality/data channel support), and an associated interface device. +The generic device's registers are mapped as 'value elements' so that they can +be written/read using SLIMbus control channel exchanging control/status type of +information. +In case there are multiple framer devices on the same bus, manager device is +responsible to select the active-framer for clocking the bus. + +Per specification, SLIMbus uses "clock gears" to do power management based on +current frequency and bandwidth requirements. There are 10 clock gears and each +gear changes the SLIMbus frequency to be twice its previous gear. + +Each device has a 6-byte enumeration-address and the manager assigns every +device with a 1-byte logical address after the devices report presence on the +bus. + +Software description: +--------------------- +There are 2 types of SLIMbus drivers: + +slim_controller represents a 'controller' for SLIMbus. This driver should +implement duties needed by the SoC (manager device, associated +interface device for monitoring the layers and reporting errors, default +framer device). + +slim_device represents the 'generic device/component' for SLIMbus, and a +slim_driver should implement driver for that slim_device. + +Device notifications to the driver: +----------------------------------- +Since SLIMbus devices have mechanisms for reporting their presence, the +framework allows drivers to bind when corresponding devices report their +presence on the bus. +However, it is possible that the driver needs to be probed +first so that it can enable corresponding SLIMbus device (e.g. power it up and/or +take it out of reset). To support that behavior, the framework allows drivers +to probe first as well (e.g. using standard DeviceTree compatibility field). +This creates the necessity for the driver to know when the device is functional +(i.e. reported present). device_up callback is used for that reason when the +device reports present and is assigned a logical address by the controller. + +Similarly, SLIMbus devices 'report absent' when they go down. A 'device_down' +callback notifies the driver when the device reports absent and its logical +address assignment is invalidated by the controller. + +Another notification "boot_device" is used to notify the slim_driver when +controller resets the bus. This notification allows the driver to take necessary +steps to boot the device so that it's functional after the bus has been reset. + +Driver and Controller APIs: +--------------------------- +.. kernel-doc:: include/linux/slimbus.h + :internal: + +.. kernel-doc:: drivers/slimbus/slimbus.h + :internal: + +.. kernel-doc:: drivers/slimbus/core.c + :export: + +Clock-pause: +------------ +SLIMbus mandates that a reconfiguration sequence (known as clock-pause) be +broadcast to all active devices on the bus before the bus can enter low-power +mode. Controller uses this sequence when it decides to enter low-power mode so +that corresponding clocks and/or power-rails can be turned off to save power. +Clock-pause is exited by waking up framer device (if controller driver initiates +exiting low power mode), or by toggling the data line (if a slave device wants +to initiate it). + +Clock-pause APIs: +~~~~~~~~~~~~~~~~~ +.. kernel-doc:: drivers/slimbus/sched.c + :export: + +Messaging: +---------- +The framework supports regmap and read/write apis to exchange control-information +with a SLIMbus device. APIs can be synchronous or asynchronous. +The header file <linux/slimbus.h> has more documentation about messaging APIs. + +Messaging APIs: +~~~~~~~~~~~~~~~ +.. kernel-doc:: drivers/slimbus/messaging.c + :export: + +Streaming APIs: +~~~~~~~~~~~~~~~ +.. kernel-doc:: drivers/slimbus/stream.c + :export: diff --git a/Documentation/driver-api/sound.rst b/Documentation/driver-api/sound.rst new file mode 100644 index 000000000..afef6eabc --- /dev/null +++ b/Documentation/driver-api/sound.rst @@ -0,0 +1,54 @@ +Sound Devices +============= + +.. kernel-doc:: include/sound/core.h + :internal: + +.. kernel-doc:: sound/sound_core.c + :export: + +.. kernel-doc:: include/sound/pcm.h + :internal: + +.. kernel-doc:: sound/core/pcm.c + :export: + +.. kernel-doc:: sound/core/device.c + :export: + +.. kernel-doc:: sound/core/info.c + :export: + +.. kernel-doc:: sound/core/rawmidi.c + :export: + +.. kernel-doc:: sound/core/sound.c + :export: + +.. kernel-doc:: sound/core/memory.c + :export: + +.. kernel-doc:: sound/core/pcm_memory.c + :export: + +.. kernel-doc:: sound/core/init.c + :export: + +.. kernel-doc:: sound/core/isadma.c + :export: + +.. kernel-doc:: sound/core/control.c + :export: + +.. kernel-doc:: sound/core/pcm_lib.c + :export: + +.. kernel-doc:: sound/core/hwdep.c + :export: + +.. kernel-doc:: sound/core/pcm_native.c + :export: + +.. kernel-doc:: sound/core/memalloc.c + :export: + diff --git a/Documentation/driver-api/soundwire/error_handling.rst b/Documentation/driver-api/soundwire/error_handling.rst new file mode 100644 index 000000000..aa3a0a23a --- /dev/null +++ b/Documentation/driver-api/soundwire/error_handling.rst @@ -0,0 +1,65 @@ +======================== +SoundWire Error Handling +======================== + +The SoundWire PHY was designed with care and errors on the bus are going to +be very unlikely, and if they happen it should be limited to single bit +errors. Examples of this design can be found in the synchronization +mechanism (sync loss after two errors) and short CRCs used for the Bulk +Register Access. + +The errors can be detected with multiple mechanisms: + +1. Bus clash or parity errors: This mechanism relies on low-level detectors + that are independent of the payload and usages, and they cover both control + and audio data. The current implementation only logs such errors. + Improvements could be invalidating an entire programming sequence and + restarting from a known position. In the case of such errors outside of a + control/command sequence, there is no concealment or recovery for audio + data enabled by the SoundWire protocol, the location of the error will also + impact its audibility (most-significant bits will be more impacted in PCM), + and after a number of such errors are detected the bus might be reset. Note + that bus clashes due to programming errors (two streams using the same bit + slots) or electrical issues during the transmit/receive transition cannot + be distinguished, although a recurring bus clash when audio is enabled is a + indication of a bus allocation issue. The interrupt mechanism can also help + identify Slaves which detected a Bus Clash or a Parity Error, but they may + not be responsible for the errors so resetting them individually is not a + viable recovery strategy. + +2. Command status: Each command is associated with a status, which only + covers transmission of the data between devices. The ACK status indicates + that the command was received and will be executed by the end of the + current frame. A NAK indicates that the command was in error and will not + be applied. In case of a bad programming (command sent to non-existent + Slave or to a non-implemented register) or electrical issue, no response + signals the command was ignored. Some Master implementations allow for a + command to be retransmitted several times. If the retransmission fails, + backtracking and restarting the entire programming sequence might be a + solution. Alternatively some implementations might directly issue a bus + reset and re-enumerate all devices. + +3. Timeouts: In a number of cases such as ChannelPrepare or + ClockStopPrepare, the bus driver is supposed to poll a register field until + it transitions to a NotFinished value of zero. The MIPI SoundWire spec 1.1 + does not define timeouts but the MIPI SoundWire DisCo document adds + recommendation on timeouts. If such configurations do not complete, the + driver will return a -ETIMEOUT. Such timeouts are symptoms of a faulty + Slave device and are likely impossible to recover from. + +Errors during global reconfiguration sequences are extremely difficult to +handle: + +1. BankSwitch: An error during the last command issuing a BankSwitch is + difficult to backtrack from. Retransmitting the Bank Switch command may be + possible in a single segment setup, but this can lead to synchronization + problems when enabling multiple bus segments (a command with side effects + such as frame reconfiguration would be handled at different times). A global + hard-reset might be the best solution. + +Note that SoundWire does not provide a mechanism to detect illegal values +written in valid registers. In a number of cases the standard even mentions +that the Slave might behave in implementation-defined ways. The bus +implementation does not provide a recovery mechanism for such errors, Slave +or Master driver implementers are responsible for writing valid values in +valid registers and implement additional range checking if needed. diff --git a/Documentation/driver-api/soundwire/index.rst b/Documentation/driver-api/soundwire/index.rst new file mode 100644 index 000000000..6db026028 --- /dev/null +++ b/Documentation/driver-api/soundwire/index.rst @@ -0,0 +1,18 @@ +======================= +SoundWire Documentation +======================= + +.. toctree:: + :maxdepth: 1 + + summary + stream + error_handling + locking + +.. only:: subproject + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/soundwire/locking.rst b/Documentation/driver-api/soundwire/locking.rst new file mode 100644 index 000000000..253f73555 --- /dev/null +++ b/Documentation/driver-api/soundwire/locking.rst @@ -0,0 +1,106 @@ +================= +SoundWire Locking +================= + +This document explains locking mechanism of the SoundWire Bus. Bus uses +following locks in order to avoid race conditions in Bus operations on +shared resources. + + - Bus lock + + - Message lock + +Bus lock +======== + +SoundWire Bus lock is a mutex and is part of Bus data structure +(sdw_bus) which is used for every Bus instance. This lock is used to +serialize each of the following operations(s) within SoundWire Bus instance. + + - Addition and removal of Slave(s), changing Slave status. + + - Prepare, Enable, Disable and De-prepare stream operations. + + - Access of Stream data structure. + +Message lock +============ + +SoundWire message transfer lock. This mutex is part of +Bus data structure (sdw_bus). This lock is used to serialize the message +transfers (read/write) within a SoundWire Bus instance. + +Below examples show how locks are acquired. + +Example 1 +--------- + +Message transfer. + + 1. For every message transfer + + a. Acquire Message lock. + + b. Transfer message (Read/Write) to Slave1 or broadcast message on + Bus in case of bank switch. + + c. Release Message lock :: + + +----------+ +---------+ + | | | | + | Bus | | Master | + | | | Driver | + | | | | + +----+-----+ +----+----+ + | | + | bus->ops->xfer_msg() | + <-------------------------------+ a. Acquire Message lock + | | b. Transfer message + | | + +-------------------------------> c. Release Message lock + | return success/error | d. Return success/error + | | + + + + +Example 2 +--------- + +Prepare operation. + + 1. Acquire lock for Bus instance associated with Master 1. + + 2. For every message transfer in Prepare operation + + a. Acquire Message lock. + + b. Transfer message (Read/Write) to Slave1 or broadcast message on + Bus in case of bank switch. + + c. Release Message lock. + + 3. Release lock for Bus instance associated with Master 1 :: + + +----------+ +---------+ + | | | | + | Bus | | Master | + | | | Driver | + | | | | + +----+-----+ +----+----+ + | | + | sdw_prepare_stream() | + <-------------------------------+ 1. Acquire bus lock + | | 2. Perform stream prepare + | | + | | + | bus->ops->xfer_msg() | + <-------------------------------+ a. Acquire Message lock + | | b. Transfer message + | | + +-------------------------------> c. Release Message lock + | return success/error | d. Return success/error + | | + | | + | return success/error | 3. Release bus lock + +-------------------------------> 4. Return success/error + | | + + + diff --git a/Documentation/driver-api/soundwire/stream.rst b/Documentation/driver-api/soundwire/stream.rst new file mode 100644 index 000000000..29121aa55 --- /dev/null +++ b/Documentation/driver-api/soundwire/stream.rst @@ -0,0 +1,372 @@ +========================= +Audio Stream in SoundWire +========================= + +An audio stream is a logical or virtual connection created between + + (1) System memory buffer(s) and Codec(s) + + (2) DSP memory buffer(s) and Codec(s) + + (3) FIFO(s) and Codec(s) + + (4) Codec(s) and Codec(s) + +which is typically driven by a DMA(s) channel through the data link. An +audio stream contains one or more channels of data. All channels within +stream must have same sample rate and same sample size. + +Assume a stream with two channels (Left & Right) is opened using SoundWire +interface. Below are some ways a stream can be represented in SoundWire. + +Stream Sample in memory (System memory, DSP memory or FIFOs) :: + + ------------------------- + | L | R | L | R | L | R | + ------------------------- + +Example 1: Stereo Stream with L and R channels is rendered from Master to +Slave. Both Master and Slave is using single port. :: + + +---------------+ Clock Signal +---------------+ + | Master +----------------------------------+ Slave | + | Interface | | Interface | + | | | 1 | + | | Data Signal | | + | L + R +----------------------------------+ L + R | + | (Data) | Data Direction | (Data) | + +---------------+ +-----------------------> +---------------+ + + +Example 2: Stereo Stream with L and R channels is captured from Slave to +Master. Both Master and Slave is using single port. :: + + + +---------------+ Clock Signal +---------------+ + | Master +----------------------------------+ Slave | + | Interface | | Interface | + | | | 1 | + | | Data Signal | | + | L + R +----------------------------------+ L + R | + | (Data) | Data Direction | (Data) | + +---------------+ <-----------------------+ +---------------+ + + +Example 3: Stereo Stream with L and R channels is rendered by Master. Each +of the L and R channel is received by two different Slaves. Master and both +Slaves are using single port. :: + + +---------------+ Clock Signal +---------------+ + | Master +---------+------------------------+ Slave | + | Interface | | | Interface | + | | | | 1 | + | | | Data Signal | | + | L + R +---+------------------------------+ L | + | (Data) | | | Data Direction | (Data) | + +---------------+ | | +-------------> +---------------+ + | | + | | + | | +---------------+ + | +----------------------> | Slave | + | | Interface | + | | 2 | + | | | + +----------------------------> | R | + | (Data) | + +---------------+ + + +Example 4: Stereo Stream with L and R channel is rendered by two different +Ports of the Master and is received by only single Port of the Slave +interface. :: + + +--------------------+ + | | + | +--------------+ +----------------+ + | | || | | + | | Data Port || L Channel | | + | | 1 |------------+ | | + | | L Channel || | +-----+----+ | + | | (Data) || | L + R Channel || Data | | + | Master +----------+ | +---+---------> || Port | | + | Interface | | || 1 | | + | +--------------+ | || | | + | | || | +----------+ | + | | Data Port |------------+ | | + | | 2 || R Channel | Slave | + | | R Channel || | Interface | + | | (Data) || | 1 | + | +--------------+ Clock Signal | L + R | + | +---------------------------> | (Data) | + +--------------------+ | | + +----------------+ + +SoundWire Stream Management flow +================================ + +Stream definitions +------------------ + + (1) Current stream: This is classified as the stream on which operation has + to be performed like prepare, enable, disable, de-prepare etc. + + (2) Active stream: This is classified as the stream which is already active + on Bus other than current stream. There can be multiple active streams + on the Bus. + +SoundWire Bus manages stream operations for each stream getting +rendered/captured on the SoundWire Bus. This section explains Bus operations +done for each of the stream allocated/released on Bus. Following are the +stream states maintained by the Bus for each of the audio stream. + + +SoundWire stream states +----------------------- + +Below shows the SoundWire stream states and state transition diagram. :: + + +-----------+ +------------+ +----------+ +----------+ + | ALLOCATED +---->| CONFIGURED +---->| PREPARED +---->| ENABLED | + | STATE | | STATE | | STATE | | STATE | + +-----------+ +------------+ +----------+ +----+-----+ + ^ + | + | + v + +----------+ +------------+ +----+-----+ + | RELEASED |<----------+ DEPREPARED |<-------+ DISABLED | + | STATE | | STATE | | STATE | + +----------+ +------------+ +----------+ + +NOTE: State transition between prepare and deprepare is supported in Spec +but not in the software (subsystem) + +NOTE2: Stream state transition checks need to be handled by caller +framework, for example ALSA/ASoC. No checks for stream transition exist in +SoundWire subsystem. + +Stream State Operations +----------------------- + +Below section explains the operations done by the Bus on Master(s) and +Slave(s) as part of stream state transitions. + +SDW_STREAM_ALLOCATED +~~~~~~~~~~~~~~~~~~~~ + +Allocation state for stream. This is the entry state +of the stream. Operations performed before entering in this state: + + (1) A stream runtime is allocated for the stream. This stream + runtime is used as a reference for all the operations performed + on the stream. + + (2) The resources required for holding stream runtime information are + allocated and initialized. This holds all stream related information + such as stream type (PCM/PDM) and parameters, Master and Slave + interface associated with the stream, stream state etc. + +After all above operations are successful, stream state is set to +``SDW_STREAM_ALLOCATED``. + +Bus implements below API for allocate a stream which needs to be called once +per stream. From ASoC DPCM framework, this stream state maybe linked to +.startup() operation. + + .. code-block:: c + int sdw_alloc_stream(char * stream_name); + + +SDW_STREAM_CONFIGURED +~~~~~~~~~~~~~~~~~~~~~ + +Configuration state of stream. Operations performed before entering in +this state: + + (1) The resources allocated for stream information in SDW_STREAM_ALLOCATED + state are updated here. This includes stream parameters, Master(s) + and Slave(s) runtime information associated with current stream. + + (2) All the Master(s) and Slave(s) associated with current stream provide + the port information to Bus which includes port numbers allocated by + Master(s) and Slave(s) for current stream and their channel mask. + +After all above operations are successful, stream state is set to +``SDW_STREAM_CONFIGURED``. + +Bus implements below APIs for CONFIG state which needs to be called by +the respective Master(s) and Slave(s) associated with stream. These APIs can +only be invoked once by respective Master(s) and Slave(s). From ASoC DPCM +framework, this stream state is linked to .hw_params() operation. + + .. code-block:: c + int sdw_stream_add_master(struct sdw_bus * bus, + struct sdw_stream_config * stream_config, + struct sdw_ports_config * ports_config, + struct sdw_stream_runtime * stream); + + int sdw_stream_add_slave(struct sdw_slave * slave, + struct sdw_stream_config * stream_config, + struct sdw_ports_config * ports_config, + struct sdw_stream_runtime * stream); + + +SDW_STREAM_PREPARED +~~~~~~~~~~~~~~~~~~~ + +Prepare state of stream. Operations performed before entering in this state: + + (1) Bus parameters such as bandwidth, frame shape, clock frequency, + are computed based on current stream as well as already active + stream(s) on Bus. Re-computation is required to accommodate current + stream on the Bus. + + (2) Transport and port parameters of all Master(s) and Slave(s) port(s) are + computed for the current as well as already active stream based on frame + shape and clock frequency computed in step 1. + + (3) Computed Bus and transport parameters are programmed in Master(s) and + Slave(s) registers. The banked registers programming is done on the + alternate bank (bank currently unused). Port(s) are enabled for the + already active stream(s) on the alternate bank (bank currently unused). + This is done in order to not disrupt already active stream(s). + + (4) Once all the values are programmed, Bus initiates switch to alternate + bank where all new values programmed gets into effect. + + (5) Ports of Master(s) and Slave(s) for current stream are prepared by + programming PrepareCtrl register. + +After all above operations are successful, stream state is set to +``SDW_STREAM_PREPARED``. + +Bus implements below API for PREPARE state which needs to be called once per +stream. From ASoC DPCM framework, this stream state is linked to +.prepare() operation. + + .. code-block:: c + int sdw_prepare_stream(struct sdw_stream_runtime * stream); + + +SDW_STREAM_ENABLED +~~~~~~~~~~~~~~~~~~ + +Enable state of stream. The data port(s) are enabled upon entering this state. +Operations performed before entering in this state: + + (1) All the values computed in SDW_STREAM_PREPARED state are programmed + in alternate bank (bank currently unused). It includes programming of + already active stream(s) as well. + + (2) All the Master(s) and Slave(s) port(s) for the current stream are + enabled on alternate bank (bank currently unused) by programming + ChannelEn register. + + (3) Once all the values are programmed, Bus initiates switch to alternate + bank where all new values programmed gets into effect and port(s) + associated with current stream are enabled. + +After all above operations are successful, stream state is set to +``SDW_STREAM_ENABLED``. + +Bus implements below API for ENABLE state which needs to be called once per +stream. From ASoC DPCM framework, this stream state is linked to +.trigger() start operation. + + .. code-block:: c + int sdw_enable_stream(struct sdw_stream_runtime * stream); + +SDW_STREAM_DISABLED +~~~~~~~~~~~~~~~~~~~ + +Disable state of stream. The data port(s) are disabled upon exiting this state. +Operations performed before entering in this state: + + (1) All the Master(s) and Slave(s) port(s) for the current stream are + disabled on alternate bank (bank currently unused) by programming + ChannelEn register. + + (2) All the current configuration of Bus and active stream(s) are programmed + into alternate bank (bank currently unused). + + (3) Once all the values are programmed, Bus initiates switch to alternate + bank where all new values programmed gets into effect and port(s) associated + with current stream are disabled. + +After all above operations are successful, stream state is set to +``SDW_STREAM_DISABLED``. + +Bus implements below API for DISABLED state which needs to be called once +per stream. From ASoC DPCM framework, this stream state is linked to +.trigger() stop operation. + + .. code-block:: c + int sdw_disable_stream(struct sdw_stream_runtime * stream); + + +SDW_STREAM_DEPREPARED +~~~~~~~~~~~~~~~~~~~~~ + +De-prepare state of stream. Operations performed before entering in this +state: + + (1) All the port(s) of Master(s) and Slave(s) for current stream are + de-prepared by programming PrepareCtrl register. + + (2) The payload bandwidth of current stream is reduced from the total + bandwidth requirement of bus and new parameters calculated and + applied by performing bank switch etc. + +After all above operations are successful, stream state is set to +``SDW_STREAM_DEPREPARED``. + +Bus implements below API for DEPREPARED state which needs to be called once +per stream. From ASoC DPCM framework, this stream state is linked to +.trigger() stop operation. + + .. code-block:: c + int sdw_deprepare_stream(struct sdw_stream_runtime * stream); + + +SDW_STREAM_RELEASED +~~~~~~~~~~~~~~~~~~~ + +Release state of stream. Operations performed before entering in this state: + + (1) Release port resources for all Master(s) and Slave(s) port(s) + associated with current stream. + + (2) Release Master(s) and Slave(s) runtime resources associated with + current stream. + + (3) Release stream runtime resources associated with current stream. + +After all above operations are successful, stream state is set to +``SDW_STREAM_RELEASED``. + +Bus implements below APIs for RELEASE state which needs to be called by +all the Master(s) and Slave(s) associated with stream. From ASoC DPCM +framework, this stream state is linked to .hw_free() operation. + + .. code-block:: c + int sdw_stream_remove_master(struct sdw_bus * bus, + struct sdw_stream_runtime * stream); + int sdw_stream_remove_slave(struct sdw_slave * slave, + struct sdw_stream_runtime * stream); + + +The .shutdown() ASoC DPCM operation calls below Bus API to release +stream assigned as part of ALLOCATED state. + +In .shutdown() the data structure maintaining stream state are freed up. + + .. code-block:: c + void sdw_release_stream(struct sdw_stream_runtime * stream); + +Not Supported +============= + +1. A single port with multiple channels supported cannot be used between two +streams or across stream. For example a port with 4 channels cannot be used +to handle 2 independent stereo streams even though it's possible in theory +in SoundWire. diff --git a/Documentation/driver-api/soundwire/summary.rst b/Documentation/driver-api/soundwire/summary.rst new file mode 100644 index 000000000..8193125a2 --- /dev/null +++ b/Documentation/driver-api/soundwire/summary.rst @@ -0,0 +1,207 @@ +=========================== +SoundWire Subsystem Summary +=========================== + +SoundWire is a new interface ratified in 2015 by the MIPI Alliance. +SoundWire is used for transporting data typically related to audio +functions. SoundWire interface is optimized to integrate audio devices in +mobile or mobile inspired systems. + +SoundWire is a 2-pin multi-drop interface with data and clock line. It +facilitates development of low cost, efficient, high performance systems. +Broad level key features of SoundWire interface include: + + (1) Transporting all of payload data channels, control information, and setup + commands over a single two-pin interface. + + (2) Lower clock frequency, and hence lower power consumption, by use of DDR + (Dual Data Rate) data transmission. + + (3) Clock scaling and optional multiple data lanes to give wide flexibility + in data rate to match system requirements. + + (4) Device status monitoring, including interrupt-style alerts to the Master. + +The SoundWire protocol supports up to eleven Slave interfaces. All the +interfaces share the common Bus containing data and clock line. Each of the +Slaves can support up to 14 Data Ports. 13 Data Ports are dedicated to audio +transport. Data Port0 is dedicated to transport of Bulk control information, +each of the audio Data Ports (1..14) can support up to 8 Channels in +transmit or receiving mode (typically fixed direction but configurable +direction is enabled by the specification). Bandwidth restrictions to +~19.2..24.576Mbits/s don't however allow for 11*13*8 channels to be +transmitted simultaneously. + +Below figure shows an example of connectivity between a SoundWire Master and +two Slave devices. :: + + +---------------+ +---------------+ + | | Clock Signal | | + | Master |-------+-------------------------------| Slave | + | Interface | | Data Signal | Interface 1 | + | |-------|-------+-----------------------| | + +---------------+ | | +---------------+ + | | + | | + | | + +--+-------+--+ + | | + | Slave | + | Interface 2 | + | | + +-------------+ + + +Terminology +=========== + +The MIPI SoundWire specification uses the term 'device' to refer to a Master +or Slave interface, which of course can be confusing. In this summary and +code we use the term interface only to refer to the hardware. We follow the +Linux device model by mapping each Slave interface connected on the bus as a +device managed by a specific driver. The Linux SoundWire subsystem provides +a framework to implement a SoundWire Slave driver with an API allowing +3rd-party vendors to enable implementation-defined functionality while +common setup/configuration tasks are handled by the bus. + +Bus: +Implements SoundWire Linux Bus which handles the SoundWire protocol. +Programs all the MIPI-defined Slave registers. Represents a SoundWire +Master. Multiple instances of Bus may be present in a system. + +Slave: +Registers as SoundWire Slave device (Linux Device). Multiple Slave devices +can register to a Bus instance. + +Slave driver: +Driver controlling the Slave device. MIPI-specified registers are controlled +directly by the Bus (and transmitted through the Master driver/interface). +Any implementation-defined Slave register is controlled by Slave driver. In +practice, it is expected that the Slave driver relies on regmap and does not +request direct register access. + +Programming interfaces (SoundWire Master interface Driver) +========================================================== + +SoundWire Bus supports programming interfaces for the SoundWire Master +implementation and SoundWire Slave devices. All the code uses the "sdw" +prefix commonly used by SoC designers and 3rd party vendors. + +Each of the SoundWire Master interfaces needs to be registered to the Bus. +Bus implements API to read standard Master MIPI properties and also provides +callback in Master ops for Master driver to implement its own functions that +provides capabilities information. DT support is not implemented at this +time but should be trivial to add since capabilities are enabled with the +``device_property_`` API. + +The Master interface along with the Master interface capabilities are +registered based on board file, DT or ACPI. + +Following is the Bus API to register the SoundWire Bus: + +.. code-block:: c + + int sdw_add_bus_master(struct sdw_bus *bus) + { + if (!bus->dev) + return -ENODEV; + + mutex_init(&bus->lock); + INIT_LIST_HEAD(&bus->slaves); + + /* Check ACPI for Slave devices */ + sdw_acpi_find_slaves(bus); + + /* Check DT for Slave devices */ + sdw_of_find_slaves(bus); + + return 0; + } + +This will initialize sdw_bus object for Master device. "sdw_master_ops" and +"sdw_master_port_ops" callback functions are provided to the Bus. + +"sdw_master_ops" is used by Bus to control the Bus in the hardware specific +way. It includes Bus control functions such as sending the SoundWire +read/write messages on Bus, setting up clock frequency & Stream +Synchronization Point (SSP). The "sdw_master_ops" structure abstracts the +hardware details of the Master from the Bus. + +"sdw_master_port_ops" is used by Bus to setup the Port parameters of the +Master interface Port. Master interface Port register map is not defined by +MIPI specification, so Bus calls the "sdw_master_port_ops" callback +function to do Port operations like "Port Prepare", "Port Transport params +set", "Port enable and disable". The implementation of the Master driver can +then perform hardware-specific configurations. + +Programming interfaces (SoundWire Slave Driver) +=============================================== + +The MIPI specification requires each Slave interface to expose a unique +48-bit identifier, stored in 6 read-only dev_id registers. This dev_id +identifier contains vendor and part information, as well as a field enabling +to differentiate between identical components. An additional class field is +currently unused. Slave driver is written for a specific vendor and part +identifier, Bus enumerates the Slave device based on these two ids. +Slave device and driver match is done based on these two ids . Probe +of the Slave driver is called by Bus on successful match between device and +driver id. A parent/child relationship is enforced between Master and Slave +devices (the logical representation is aligned with the physical +connectivity). + +The information on Master/Slave dependencies is stored in platform data, +board-file, ACPI or DT. The MIPI Software specification defines additional +link_id parameters for controllers that have multiple Master interfaces. The +dev_id registers are only unique in the scope of a link, and the link_id +unique in the scope of a controller. Both dev_id and link_id are not +necessarily unique at the system level but the parent/child information is +used to avoid ambiguity. + +.. code-block:: c + + static const struct sdw_device_id slave_id[] = { + SDW_SLAVE_ENTRY(0x025d, 0x700, 0), + {}, + }; + MODULE_DEVICE_TABLE(sdw, slave_id); + + static struct sdw_driver slave_sdw_driver = { + .driver = { + .name = "slave_xxx", + .pm = &slave_runtime_pm, + }, + .probe = slave_sdw_probe, + .remove = slave_sdw_remove, + .ops = &slave_slave_ops, + .id_table = slave_id, + }; + + +For capabilities, Bus implements API to read standard Slave MIPI properties +and also provides callback in Slave ops for Slave driver to implement own +function that provides capabilities information. Bus needs to know a set of +Slave capabilities to program Slave registers and to control the Bus +reconfigurations. + +Future enhancements to be done +============================== + + (1) Bulk Register Access (BRA) transfers. + + + (2) Multiple data lane support. + +Links +===== + +SoundWire MIPI specification 1.1 is available at: +https://members.mipi.org/wg/All-Members/document/70290 + +SoundWire MIPI DisCo (Discovery and Configuration) specification is +available at: +https://www.mipi.org/specifications/mipi-disco-soundwire + +(publicly accessible with registration or directly accessible to MIPI +members) + +MIPI Alliance Manufacturer ID Page: mid.mipi.org diff --git a/Documentation/driver-api/spi.rst b/Documentation/driver-api/spi.rst new file mode 100644 index 000000000..f64cb6664 --- /dev/null +++ b/Documentation/driver-api/spi.rst @@ -0,0 +1,53 @@ +Serial Peripheral Interface (SPI) +================================= + +SPI is the "Serial Peripheral Interface", widely used with embedded +systems because it is a simple and efficient interface: basically a +multiplexed shift register. Its three signal wires hold a clock (SCK, +often in the range of 1-20 MHz), a "Master Out, Slave In" (MOSI) data +line, and a "Master In, Slave Out" (MISO) data line. SPI is a full +duplex protocol; for each bit shifted out the MOSI line (one per clock) +another is shifted in on the MISO line. Those bits are assembled into +words of various sizes on the way to and from system memory. An +additional chipselect line is usually active-low (nCS); four signals are +normally used for each peripheral, plus sometimes an interrupt. + +The SPI bus facilities listed here provide a generalized interface to +declare SPI busses and devices, manage them according to the standard +Linux driver model, and perform input/output operations. At this time, +only "master" side interfaces are supported, where Linux talks to SPI +peripherals and does not implement such a peripheral itself. (Interfaces +to support implementing SPI slaves would necessarily look different.) + +The programming interface is structured around two kinds of driver, and +two kinds of device. A "Controller Driver" abstracts the controller +hardware, which may be as simple as a set of GPIO pins or as complex as +a pair of FIFOs connected to dual DMA engines on the other side of the +SPI shift register (maximizing throughput). Such drivers bridge between +whatever bus they sit on (often the platform bus) and SPI, and expose +the SPI side of their device as a :c:type:`struct spi_master +<spi_master>`. SPI devices are children of that master, +represented as a :c:type:`struct spi_device <spi_device>` and +manufactured from :c:type:`struct spi_board_info +<spi_board_info>` descriptors which are usually provided by +board-specific initialization code. A :c:type:`struct spi_driver +<spi_driver>` is called a "Protocol Driver", and is bound to a +spi_device using normal driver model calls. + +The I/O model is a set of queued messages. Protocol drivers submit one +or more :c:type:`struct spi_message <spi_message>` objects, +which are processed and completed asynchronously. (There are synchronous +wrappers, however.) Messages are built from one or more +:c:type:`struct spi_transfer <spi_transfer>` objects, each of +which wraps a full duplex SPI transfer. A variety of protocol tweaking +options are needed, because different chips adopt very different +policies for how they use the bits transferred with SPI. + +.. kernel-doc:: include/linux/spi/spi.h + :internal: + +.. kernel-doc:: drivers/spi/spi.c + :functions: spi_register_board_info + +.. kernel-doc:: drivers/spi/spi.c + :export: diff --git a/Documentation/driver-api/target.rst b/Documentation/driver-api/target.rst new file mode 100644 index 000000000..4363611dd --- /dev/null +++ b/Documentation/driver-api/target.rst @@ -0,0 +1,64 @@ +================================= +target and iSCSI Interfaces Guide +================================= + +Introduction and Overview +========================= + +TBD + +Target core device interfaces +============================= + +.. kernel-doc:: drivers/target/target_core_device.c + :export: + +Target core transport interfaces +================================ + +.. kernel-doc:: drivers/target/target_core_transport.c + :export: + +Target-supported userspace I/O +============================== + +.. kernel-doc:: drivers/target/target_core_user.c + :doc: Userspace I/O + +.. kernel-doc:: include/uapi/linux/target_core_user.h + :doc: Ring Design + +iSCSI helper functions +====================== + +.. kernel-doc:: drivers/scsi/libiscsi.c + :export: + + +iSCSI boot information +====================== + +.. kernel-doc:: drivers/scsi/iscsi_boot_sysfs.c + :export: + + +iSCSI transport class +===================== + +The file drivers/scsi/scsi_transport_iscsi.c defines transport +attributes for the iSCSI class, which sends SCSI packets over TCP/IP +connections. + +.. kernel-doc:: drivers/scsi/scsi_transport_iscsi.c + :export: + + +iSCSI TCP interfaces +==================== + +.. kernel-doc:: drivers/scsi/iscsi_tcp.c + :internal: + +.. kernel-doc:: drivers/scsi/libiscsi_tcp.c + :export: + diff --git a/Documentation/driver-api/uio-howto.rst b/Documentation/driver-api/uio-howto.rst new file mode 100644 index 000000000..fb2eb73be --- /dev/null +++ b/Documentation/driver-api/uio-howto.rst @@ -0,0 +1,723 @@ +======================= +The Userspace I/O HOWTO +======================= + +:Author: Hans-Jürgen Koch Linux developer, Linutronix +:Date: 2006-12-11 + +About this document +=================== + +Translations +------------ + +If you know of any translations for this document, or you are interested +in translating it, please email me hjk@hansjkoch.de. + +Preface +------- + +For many types of devices, creating a Linux kernel driver is overkill. +All that is really needed is some way to handle an interrupt and provide +access to the memory space of the device. The logic of controlling the +device does not necessarily have to be within the kernel, as the device +does not need to take advantage of any of other resources that the +kernel provides. One such common class of devices that are like this are +for industrial I/O cards. + +To address this situation, the userspace I/O system (UIO) was designed. +For typical industrial I/O cards, only a very small kernel module is +needed. The main part of the driver will run in user space. This +simplifies development and reduces the risk of serious bugs within a +kernel module. + +Please note that UIO is not an universal driver interface. Devices that +are already handled well by other kernel subsystems (like networking or +serial or USB) are no candidates for an UIO driver. Hardware that is +ideally suited for an UIO driver fulfills all of the following: + +- The device has memory that can be mapped. The device can be + controlled completely by writing to this memory. + +- The device usually generates interrupts. + +- The device does not fit into one of the standard kernel subsystems. + +Acknowledgments +--------------- + +I'd like to thank Thomas Gleixner and Benedikt Spranger of Linutronix, +who have not only written most of the UIO code, but also helped greatly +writing this HOWTO by giving me all kinds of background information. + +Feedback +-------- + +Find something wrong with this document? (Or perhaps something right?) I +would love to hear from you. Please email me at hjk@hansjkoch.de. + +About UIO +========= + +If you use UIO for your card's driver, here's what you get: + +- only one small kernel module to write and maintain. + +- develop the main part of your driver in user space, with all the + tools and libraries you're used to. + +- bugs in your driver won't crash the kernel. + +- updates of your driver can take place without recompiling the kernel. + +How UIO works +------------- + +Each UIO device is accessed through a device file and several sysfs +attribute files. The device file will be called ``/dev/uio0`` for the +first device, and ``/dev/uio1``, ``/dev/uio2`` and so on for subsequent +devices. + +``/dev/uioX`` is used to access the address space of the card. Just use +:c:func:`mmap()` to access registers or RAM locations of your card. + +Interrupts are handled by reading from ``/dev/uioX``. A blocking +:c:func:`read()` from ``/dev/uioX`` will return as soon as an +interrupt occurs. You can also use :c:func:`select()` on +``/dev/uioX`` to wait for an interrupt. The integer value read from +``/dev/uioX`` represents the total interrupt count. You can use this +number to figure out if you missed some interrupts. + +For some hardware that has more than one interrupt source internally, +but not separate IRQ mask and status registers, there might be +situations where userspace cannot determine what the interrupt source +was if the kernel handler disables them by writing to the chip's IRQ +register. In such a case, the kernel has to disable the IRQ completely +to leave the chip's register untouched. Now the userspace part can +determine the cause of the interrupt, but it cannot re-enable +interrupts. Another cornercase is chips where re-enabling interrupts is +a read-modify-write operation to a combined IRQ status/acknowledge +register. This would be racy if a new interrupt occurred simultaneously. + +To address these problems, UIO also implements a write() function. It is +normally not used and can be ignored for hardware that has only a single +interrupt source or has separate IRQ mask and status registers. If you +need it, however, a write to ``/dev/uioX`` will call the +:c:func:`irqcontrol()` function implemented by the driver. You have +to write a 32-bit value that is usually either 0 or 1 to disable or +enable interrupts. If a driver does not implement +:c:func:`irqcontrol()`, :c:func:`write()` will return with +``-ENOSYS``. + +To handle interrupts properly, your custom kernel module can provide its +own interrupt handler. It will automatically be called by the built-in +handler. + +For cards that don't generate interrupts but need to be polled, there is +the possibility to set up a timer that triggers the interrupt handler at +configurable time intervals. This interrupt simulation is done by +calling :c:func:`uio_event_notify()` from the timer's event +handler. + +Each driver provides attributes that are used to read or write +variables. These attributes are accessible through sysfs files. A custom +kernel driver module can add its own attributes to the device owned by +the uio driver, but not added to the UIO device itself at this time. +This might change in the future if it would be found to be useful. + +The following standard attributes are provided by the UIO framework: + +- ``name``: The name of your device. It is recommended to use the name + of your kernel module for this. + +- ``version``: A version string defined by your driver. This allows the + user space part of your driver to deal with different versions of the + kernel module. + +- ``event``: The total number of interrupts handled by the driver since + the last time the device node was read. + +These attributes appear under the ``/sys/class/uio/uioX`` directory. +Please note that this directory might be a symlink, and not a real +directory. Any userspace code that accesses it must be able to handle +this. + +Each UIO device can make one or more memory regions available for memory +mapping. This is necessary because some industrial I/O cards require +access to more than one PCI memory region in a driver. + +Each mapping has its own directory in sysfs, the first mapping appears +as ``/sys/class/uio/uioX/maps/map0/``. Subsequent mappings create +directories ``map1/``, ``map2/``, and so on. These directories will only +appear if the size of the mapping is not 0. + +Each ``mapX/`` directory contains four read-only files that show +attributes of the memory: + +- ``name``: A string identifier for this mapping. This is optional, the + string can be empty. Drivers can set this to make it easier for + userspace to find the correct mapping. + +- ``addr``: The address of memory that can be mapped. + +- ``size``: The size, in bytes, of the memory pointed to by addr. + +- ``offset``: The offset, in bytes, that has to be added to the pointer + returned by :c:func:`mmap()` to get to the actual device memory. + This is important if the device's memory is not page aligned. + Remember that pointers returned by :c:func:`mmap()` are always + page aligned, so it is good style to always add this offset. + +From userspace, the different mappings are distinguished by adjusting +the ``offset`` parameter of the :c:func:`mmap()` call. To map the +memory of mapping N, you have to use N times the page size as your +offset:: + + offset = N * getpagesize(); + +Sometimes there is hardware with memory-like regions that can not be +mapped with the technique described here, but there are still ways to +access them from userspace. The most common example are x86 ioports. On +x86 systems, userspace can access these ioports using +:c:func:`ioperm()`, :c:func:`iopl()`, :c:func:`inb()`, +:c:func:`outb()`, and similar functions. + +Since these ioport regions can not be mapped, they will not appear under +``/sys/class/uio/uioX/maps/`` like the normal memory described above. +Without information about the port regions a hardware has to offer, it +becomes difficult for the userspace part of the driver to find out which +ports belong to which UIO device. + +To address this situation, the new directory +``/sys/class/uio/uioX/portio/`` was added. It only exists if the driver +wants to pass information about one or more port regions to userspace. +If that is the case, subdirectories named ``port0``, ``port1``, and so +on, will appear underneath ``/sys/class/uio/uioX/portio/``. + +Each ``portX/`` directory contains four read-only files that show name, +start, size, and type of the port region: + +- ``name``: A string identifier for this port region. The string is + optional and can be empty. Drivers can set it to make it easier for + userspace to find a certain port region. + +- ``start``: The first port of this region. + +- ``size``: The number of ports in this region. + +- ``porttype``: A string describing the type of port. + +Writing your own kernel module +============================== + +Please have a look at ``uio_cif.c`` as an example. The following +paragraphs explain the different sections of this file. + +struct uio_info +--------------- + +This structure tells the framework the details of your driver, Some of +the members are required, others are optional. + +- ``const char *name``: Required. The name of your driver as it will + appear in sysfs. I recommend using the name of your module for this. + +- ``const char *version``: Required. This string appears in + ``/sys/class/uio/uioX/version``. + +- ``struct uio_mem mem[ MAX_UIO_MAPS ]``: Required if you have memory + that can be mapped with :c:func:`mmap()`. For each mapping you + need to fill one of the ``uio_mem`` structures. See the description + below for details. + +- ``struct uio_port port[ MAX_UIO_PORTS_REGIONS ]``: Required if you + want to pass information about ioports to userspace. For each port + region you need to fill one of the ``uio_port`` structures. See the + description below for details. + +- ``long irq``: Required. If your hardware generates an interrupt, it's + your modules task to determine the irq number during initialization. + If you don't have a hardware generated interrupt but want to trigger + the interrupt handler in some other way, set ``irq`` to + ``UIO_IRQ_CUSTOM``. If you had no interrupt at all, you could set + ``irq`` to ``UIO_IRQ_NONE``, though this rarely makes sense. + +- ``unsigned long irq_flags``: Required if you've set ``irq`` to a + hardware interrupt number. The flags given here will be used in the + call to :c:func:`request_irq()`. + +- ``int (*mmap)(struct uio_info *info, struct vm_area_struct *vma)``: + Optional. If you need a special :c:func:`mmap()` + function, you can set it here. If this pointer is not NULL, your + :c:func:`mmap()` will be called instead of the built-in one. + +- ``int (*open)(struct uio_info *info, struct inode *inode)``: + Optional. You might want to have your own :c:func:`open()`, + e.g. to enable interrupts only when your device is actually used. + +- ``int (*release)(struct uio_info *info, struct inode *inode)``: + Optional. If you define your own :c:func:`open()`, you will + probably also want a custom :c:func:`release()` function. + +- ``int (*irqcontrol)(struct uio_info *info, s32 irq_on)``: + Optional. If you need to be able to enable or disable interrupts + from userspace by writing to ``/dev/uioX``, you can implement this + function. The parameter ``irq_on`` will be 0 to disable interrupts + and 1 to enable them. + +Usually, your device will have one or more memory regions that can be +mapped to user space. For each region, you have to set up a +``struct uio_mem`` in the ``mem[]`` array. Here's a description of the +fields of ``struct uio_mem``: + +- ``const char *name``: Optional. Set this to help identify the memory + region, it will show up in the corresponding sysfs node. + +- ``int memtype``: Required if the mapping is used. Set this to + ``UIO_MEM_PHYS`` if you you have physical memory on your card to be + mapped. Use ``UIO_MEM_LOGICAL`` for logical memory (e.g. allocated + with :c:func:`kmalloc()`). There's also ``UIO_MEM_VIRTUAL`` for + virtual memory. + +- ``phys_addr_t addr``: Required if the mapping is used. Fill in the + address of your memory block. This address is the one that appears in + sysfs. + +- ``resource_size_t size``: Fill in the size of the memory block that + ``addr`` points to. If ``size`` is zero, the mapping is considered + unused. Note that you *must* initialize ``size`` with zero for all + unused mappings. + +- ``void *internal_addr``: If you have to access this memory region + from within your kernel module, you will want to map it internally by + using something like :c:func:`ioremap()`. Addresses returned by + this function cannot be mapped to user space, so you must not store + it in ``addr``. Use ``internal_addr`` instead to remember such an + address. + +Please do not touch the ``map`` element of ``struct uio_mem``! It is +used by the UIO framework to set up sysfs files for this mapping. Simply +leave it alone. + +Sometimes, your device can have one or more port regions which can not +be mapped to userspace. But if there are other possibilities for +userspace to access these ports, it makes sense to make information +about the ports available in sysfs. For each region, you have to set up +a ``struct uio_port`` in the ``port[]`` array. Here's a description of +the fields of ``struct uio_port``: + +- ``char *porttype``: Required. Set this to one of the predefined + constants. Use ``UIO_PORT_X86`` for the ioports found in x86 + architectures. + +- ``unsigned long start``: Required if the port region is used. Fill in + the number of the first port of this region. + +- ``unsigned long size``: Fill in the number of ports in this region. + If ``size`` is zero, the region is considered unused. Note that you + *must* initialize ``size`` with zero for all unused regions. + +Please do not touch the ``portio`` element of ``struct uio_port``! It is +used internally by the UIO framework to set up sysfs files for this +region. Simply leave it alone. + +Adding an interrupt handler +--------------------------- + +What you need to do in your interrupt handler depends on your hardware +and on how you want to handle it. You should try to keep the amount of +code in your kernel interrupt handler low. If your hardware requires no +action that you *have* to perform after each interrupt, then your +handler can be empty. + +If, on the other hand, your hardware *needs* some action to be performed +after each interrupt, then you *must* do it in your kernel module. Note +that you cannot rely on the userspace part of your driver. Your +userspace program can terminate at any time, possibly leaving your +hardware in a state where proper interrupt handling is still required. + +There might also be applications where you want to read data from your +hardware at each interrupt and buffer it in a piece of kernel memory +you've allocated for that purpose. With this technique you could avoid +loss of data if your userspace program misses an interrupt. + +A note on shared interrupts: Your driver should support interrupt +sharing whenever this is possible. It is possible if and only if your +driver can detect whether your hardware has triggered the interrupt or +not. This is usually done by looking at an interrupt status register. If +your driver sees that the IRQ bit is actually set, it will perform its +actions, and the handler returns IRQ_HANDLED. If the driver detects +that it was not your hardware that caused the interrupt, it will do +nothing and return IRQ_NONE, allowing the kernel to call the next +possible interrupt handler. + +If you decide not to support shared interrupts, your card won't work in +computers with no free interrupts. As this frequently happens on the PC +platform, you can save yourself a lot of trouble by supporting interrupt +sharing. + +Using uio_pdrv for platform devices +----------------------------------- + +In many cases, UIO drivers for platform devices can be handled in a +generic way. In the same place where you define your +``struct platform_device``, you simply also implement your interrupt +handler and fill your ``struct uio_info``. A pointer to this +``struct uio_info`` is then used as ``platform_data`` for your platform +device. + +You also need to set up an array of ``struct resource`` containing +addresses and sizes of your memory mappings. This information is passed +to the driver using the ``.resource`` and ``.num_resources`` elements of +``struct platform_device``. + +You now have to set the ``.name`` element of ``struct platform_device`` +to ``"uio_pdrv"`` to use the generic UIO platform device driver. This +driver will fill the ``mem[]`` array according to the resources given, +and register the device. + +The advantage of this approach is that you only have to edit a file you +need to edit anyway. You do not have to create an extra driver. + +Using uio_pdrv_genirq for platform devices +------------------------------------------ + +Especially in embedded devices, you frequently find chips where the irq +pin is tied to its own dedicated interrupt line. In such cases, where +you can be really sure the interrupt is not shared, we can take the +concept of ``uio_pdrv`` one step further and use a generic interrupt +handler. That's what ``uio_pdrv_genirq`` does. + +The setup for this driver is the same as described above for +``uio_pdrv``, except that you do not implement an interrupt handler. The +``.handler`` element of ``struct uio_info`` must remain ``NULL``. The +``.irq_flags`` element must not contain ``IRQF_SHARED``. + +You will set the ``.name`` element of ``struct platform_device`` to +``"uio_pdrv_genirq"`` to use this driver. + +The generic interrupt handler of ``uio_pdrv_genirq`` will simply disable +the interrupt line using :c:func:`disable_irq_nosync()`. After +doing its work, userspace can reenable the interrupt by writing +0x00000001 to the UIO device file. The driver already implements an +:c:func:`irq_control()` to make this possible, you must not +implement your own. + +Using ``uio_pdrv_genirq`` not only saves a few lines of interrupt +handler code. You also do not need to know anything about the chip's +internal registers to create the kernel part of the driver. All you need +to know is the irq number of the pin the chip is connected to. + +Using uio_dmem_genirq for platform devices +------------------------------------------ + +In addition to statically allocated memory ranges, they may also be a +desire to use dynamically allocated regions in a user space driver. In +particular, being able to access memory made available through the +dma-mapping API, may be particularly useful. The ``uio_dmem_genirq`` +driver provides a way to accomplish this. + +This driver is used in a similar manner to the ``"uio_pdrv_genirq"`` +driver with respect to interrupt configuration and handling. + +Set the ``.name`` element of ``struct platform_device`` to +``"uio_dmem_genirq"`` to use this driver. + +When using this driver, fill in the ``.platform_data`` element of +``struct platform_device``, which is of type +``struct uio_dmem_genirq_pdata`` and which contains the following +elements: + +- ``struct uio_info uioinfo``: The same structure used as the + ``uio_pdrv_genirq`` platform data + +- ``unsigned int *dynamic_region_sizes``: Pointer to list of sizes of + dynamic memory regions to be mapped into user space. + +- ``unsigned int num_dynamic_regions``: Number of elements in + ``dynamic_region_sizes`` array. + +The dynamic regions defined in the platform data will be appended to the +`` mem[] `` array after the platform device resources, which implies +that the total number of static and dynamic memory regions cannot exceed +``MAX_UIO_MAPS``. + +The dynamic memory regions will be allocated when the UIO device file, +``/dev/uioX`` is opened. Similar to static memory resources, the memory +region information for dynamic regions is then visible via sysfs at +``/sys/class/uio/uioX/maps/mapY/*``. The dynamic memory regions will be +freed when the UIO device file is closed. When no processes are holding +the device file open, the address returned to userspace is ~0. + +Writing a driver in userspace +============================= + +Once you have a working kernel module for your hardware, you can write +the userspace part of your driver. You don't need any special libraries, +your driver can be written in any reasonable language, you can use +floating point numbers and so on. In short, you can use all the tools +and libraries you'd normally use for writing a userspace application. + +Getting information about your UIO device +----------------------------------------- + +Information about all UIO devices is available in sysfs. The first thing +you should do in your driver is check ``name`` and ``version`` to make +sure your talking to the right device and that its kernel driver has the +version you expect. + +You should also make sure that the memory mapping you need exists and +has the size you expect. + +There is a tool called ``lsuio`` that lists UIO devices and their +attributes. It is available here: + +http://www.osadl.org/projects/downloads/UIO/user/ + +With ``lsuio`` you can quickly check if your kernel module is loaded and +which attributes it exports. Have a look at the manpage for details. + +The source code of ``lsuio`` can serve as an example for getting +information about an UIO device. The file ``uio_helper.c`` contains a +lot of functions you could use in your userspace driver code. + +mmap() device memory +-------------------- + +After you made sure you've got the right device with the memory mappings +you need, all you have to do is to call :c:func:`mmap()` to map the +device's memory to userspace. + +The parameter ``offset`` of the :c:func:`mmap()` call has a special +meaning for UIO devices: It is used to select which mapping of your +device you want to map. To map the memory of mapping N, you have to use +N times the page size as your offset:: + + offset = N * getpagesize(); + +N starts from zero, so if you've got only one memory range to map, set +``offset = 0``. A drawback of this technique is that memory is always +mapped beginning with its start address. + +Waiting for interrupts +---------------------- + +After you successfully mapped your devices memory, you can access it +like an ordinary array. Usually, you will perform some initialization. +After that, your hardware starts working and will generate an interrupt +as soon as it's finished, has some data available, or needs your +attention because an error occurred. + +``/dev/uioX`` is a read-only file. A :c:func:`read()` will always +block until an interrupt occurs. There is only one legal value for the +``count`` parameter of :c:func:`read()`, and that is the size of a +signed 32 bit integer (4). Any other value for ``count`` causes +:c:func:`read()` to fail. The signed 32 bit integer read is the +interrupt count of your device. If the value is one more than the value +you read the last time, everything is OK. If the difference is greater +than one, you missed interrupts. + +You can also use :c:func:`select()` on ``/dev/uioX``. + +Generic PCI UIO driver +====================== + +The generic driver is a kernel module named uio_pci_generic. It can +work with any device compliant to PCI 2.3 (circa 2002) and any compliant +PCI Express device. Using this, you only need to write the userspace +driver, removing the need to write a hardware-specific kernel module. + +Making the driver recognize the device +-------------------------------------- + +Since the driver does not declare any device ids, it will not get loaded +automatically and will not automatically bind to any devices, you must +load it and allocate id to the driver yourself. For example:: + + modprobe uio_pci_generic + echo "8086 10f5" > /sys/bus/pci/drivers/uio_pci_generic/new_id + +If there already is a hardware specific kernel driver for your device, +the generic driver still won't bind to it, in this case if you want to +use the generic driver (why would you?) you'll have to manually unbind +the hardware specific driver and bind the generic driver, like this:: + + echo -n 0000:00:19.0 > /sys/bus/pci/drivers/e1000e/unbind + echo -n 0000:00:19.0 > /sys/bus/pci/drivers/uio_pci_generic/bind + +You can verify that the device has been bound to the driver by looking +for it in sysfs, for example like the following:: + + ls -l /sys/bus/pci/devices/0000:00:19.0/driver + +Which if successful should print:: + + .../0000:00:19.0/driver -> ../../../bus/pci/drivers/uio_pci_generic + +Note that the generic driver will not bind to old PCI 2.2 devices. If +binding the device failed, run the following command:: + + dmesg + +and look in the output for failure reasons. + +Things to know about uio_pci_generic +------------------------------------ + +Interrupts are handled using the Interrupt Disable bit in the PCI +command register and Interrupt Status bit in the PCI status register. +All devices compliant to PCI 2.3 (circa 2002) and all compliant PCI +Express devices should support these bits. uio_pci_generic detects +this support, and won't bind to devices which do not support the +Interrupt Disable Bit in the command register. + +On each interrupt, uio_pci_generic sets the Interrupt Disable bit. +This prevents the device from generating further interrupts until the +bit is cleared. The userspace driver should clear this bit before +blocking and waiting for more interrupts. + +Writing userspace driver using uio_pci_generic +------------------------------------------------ + +Userspace driver can use pci sysfs interface, or the libpci library that +wraps it, to talk to the device and to re-enable interrupts by writing +to the command register. + +Example code using uio_pci_generic +---------------------------------- + +Here is some sample userspace driver code using uio_pci_generic:: + + #include <stdlib.h> + #include <stdio.h> + #include <unistd.h> + #include <sys/types.h> + #include <sys/stat.h> + #include <fcntl.h> + #include <errno.h> + + int main() + { + int uiofd; + int configfd; + int err; + int i; + unsigned icount; + unsigned char command_high; + + uiofd = open("/dev/uio0", O_RDONLY); + if (uiofd < 0) { + perror("uio open:"); + return errno; + } + configfd = open("/sys/class/uio/uio0/device/config", O_RDWR); + if (configfd < 0) { + perror("config open:"); + return errno; + } + + /* Read and cache command value */ + err = pread(configfd, &command_high, 1, 5); + if (err != 1) { + perror("command config read:"); + return errno; + } + command_high &= ~0x4; + + for(i = 0;; ++i) { + /* Print out a message, for debugging. */ + if (i == 0) + fprintf(stderr, "Started uio test driver.\n"); + else + fprintf(stderr, "Interrupts: %d\n", icount); + + /****************************************/ + /* Here we got an interrupt from the + device. Do something to it. */ + /****************************************/ + + /* Re-enable interrupts. */ + err = pwrite(configfd, &command_high, 1, 5); + if (err != 1) { + perror("config write:"); + break; + } + + /* Wait for next interrupt. */ + err = read(uiofd, &icount, 4); + if (err != 4) { + perror("uio read:"); + break; + } + + } + return errno; + } + +Generic Hyper-V UIO driver +========================== + +The generic driver is a kernel module named uio_hv_generic. It +supports devices on the Hyper-V VMBus similar to uio_pci_generic on +PCI bus. + +Making the driver recognize the device +-------------------------------------- + +Since the driver does not declare any device GUID's, it will not get +loaded automatically and will not automatically bind to any devices, you +must load it and allocate id to the driver yourself. For example, to use +the network device class GUID:: + + modprobe uio_hv_generic + echo "f8615163-df3e-46c5-913f-f2d2f965ed0e" > /sys/bus/vmbus/drivers/uio_hv_generic/new_id + +If there already is a hardware specific kernel driver for the device, +the generic driver still won't bind to it, in this case if you want to +use the generic driver for a userspace library you'll have to manually unbind +the hardware specific driver and bind the generic driver, using the device specific GUID +like this:: + + echo -n ed963694-e847-4b2a-85af-bc9cfc11d6f3 > /sys/bus/vmbus/drivers/hv_netvsc/unbind + echo -n ed963694-e847-4b2a-85af-bc9cfc11d6f3 > /sys/bus/vmbus/drivers/uio_hv_generic/bind + +You can verify that the device has been bound to the driver by looking +for it in sysfs, for example like the following:: + + ls -l /sys/bus/vmbus/devices/ed963694-e847-4b2a-85af-bc9cfc11d6f3/driver + +Which if successful should print:: + + .../ed963694-e847-4b2a-85af-bc9cfc11d6f3/driver -> ../../../bus/vmbus/drivers/uio_hv_generic + +Things to know about uio_hv_generic +----------------------------------- + +On each interrupt, uio_hv_generic sets the Interrupt Disable bit. This +prevents the device from generating further interrupts until the bit is +cleared. The userspace driver should clear this bit before blocking and +waiting for more interrupts. + +When host rescinds a device, the interrupt file descriptor is marked down +and any reads of the interrupt file descriptor will return -EIO. Similar +to a closed socket or disconnected serial device. + +The vmbus device regions are mapped into uio device resources: + 0) Channel ring buffers: guest to host and host to guest + 1) Guest to host interrupt signalling pages + 2) Guest to host monitor page + 3) Network receive buffer region + 4) Network send buffer region + +If a subchannel is created by a request to host, then the uio_hv_generic +device driver will create a sysfs binary file for the per-channel ring buffer. +For example:: + + /sys/bus/vmbus/devices/3811fe4d-0fa0-4b62-981a-74fc1084c757/channels/21/ring + +Further information +=================== + +- `OSADL homepage. <http://www.osadl.org>`_ + +- `Linutronix homepage. <http://www.linutronix.de>`_ diff --git a/Documentation/driver-api/usb/URB.rst b/Documentation/driver-api/usb/URB.rst new file mode 100644 index 000000000..61a54da9f --- /dev/null +++ b/Documentation/driver-api/usb/URB.rst @@ -0,0 +1,290 @@ +.. _usb-urb: + +USB Request Block (URB) +~~~~~~~~~~~~~~~~~~~~~~~ + +:Revised: 2000-Dec-05 +:Again: 2002-Jul-06 +:Again: 2005-Sep-19 +:Again: 2017-Mar-29 + + +.. note:: + + The USB subsystem now has a substantial section at :ref:`usb-hostside-api` + section, generated from the current source code. + This particular documentation file isn't complete and may not be + updated to the last version; don't rely on it except for a quick + overview. + +Basic concept or 'What is an URB?' +================================== + +The basic idea of the new driver is message passing, the message itself is +called USB Request Block, or URB for short. + +- An URB consists of all relevant information to execute any USB transaction + and deliver the data and status back. + +- Execution of an URB is inherently an asynchronous operation, i.e. the + :c:func:`usb_submit_urb` call returns immediately after it has successfully + queued the requested action. + +- Transfers for one URB can be canceled with :c:func:`usb_unlink_urb` + at any time. + +- Each URB has a completion handler, which is called after the action + has been successfully completed or canceled. The URB also contains a + context-pointer for passing information to the completion handler. + +- Each endpoint for a device logically supports a queue of requests. + You can fill that queue, so that the USB hardware can still transfer + data to an endpoint while your driver handles completion of another. + This maximizes use of USB bandwidth, and supports seamless streaming + of data to (or from) devices when using periodic transfer modes. + + +The URB structure +================= + +Some of the fields in struct :c:type:`urb` are:: + + struct urb + { + // (IN) device and pipe specify the endpoint queue + struct usb_device *dev; // pointer to associated USB device + unsigned int pipe; // endpoint information + + unsigned int transfer_flags; // URB_ISO_ASAP, URB_SHORT_NOT_OK, etc. + + // (IN) all urbs need completion routines + void *context; // context for completion routine + usb_complete_t complete; // pointer to completion routine + + // (OUT) status after each completion + int status; // returned status + + // (IN) buffer used for data transfers + void *transfer_buffer; // associated data buffer + u32 transfer_buffer_length; // data buffer length + int number_of_packets; // size of iso_frame_desc + + // (OUT) sometimes only part of CTRL/BULK/INTR transfer_buffer is used + u32 actual_length; // actual data buffer length + + // (IN) setup stage for CTRL (pass a struct usb_ctrlrequest) + unsigned char *setup_packet; // setup packet (control only) + + // Only for PERIODIC transfers (ISO, INTERRUPT) + // (IN/OUT) start_frame is set unless URB_ISO_ASAP isn't set + int start_frame; // start frame + int interval; // polling interval + + // ISO only: packets are only "best effort"; each can have errors + int error_count; // number of errors + struct usb_iso_packet_descriptor iso_frame_desc[0]; + }; + +Your driver must create the "pipe" value using values from the appropriate +endpoint descriptor in an interface that it's claimed. + + +How to get an URB? +================== + +URBs are allocated by calling :c:func:`usb_alloc_urb`:: + + struct urb *usb_alloc_urb(int isoframes, int mem_flags) + +Return value is a pointer to the allocated URB, 0 if allocation failed. +The parameter isoframes specifies the number of isochronous transfer frames +you want to schedule. For CTRL/BULK/INT, use 0. The mem_flags parameter +holds standard memory allocation flags, letting you control (among other +things) whether the underlying code may block or not. + +To free an URB, use :c:func:`usb_free_urb`:: + + void usb_free_urb(struct urb *urb) + +You may free an urb that you've submitted, but which hasn't yet been +returned to you in a completion callback. It will automatically be +deallocated when it is no longer in use. + + +What has to be filled in? +========================= + +Depending on the type of transaction, there are some inline functions +defined in ``linux/usb.h`` to simplify the initialization, such as +:c:func:`usb_fill_control_urb`, :c:func:`usb_fill_bulk_urb` and +:c:func:`usb_fill_int_urb`. In general, they need the usb device pointer, +the pipe (usual format from usb.h), the transfer buffer, the desired transfer +length, the completion handler, and its context. Take a look at the some +existing drivers to see how they're used. + +Flags: + +- For ISO there are two startup behaviors: Specified start_frame or ASAP. +- For ASAP set ``URB_ISO_ASAP`` in transfer_flags. + +If short packets should NOT be tolerated, set ``URB_SHORT_NOT_OK`` in +transfer_flags. + + +How to submit an URB? +===================== + +Just call :c:func:`usb_submit_urb`:: + + int usb_submit_urb(struct urb *urb, int mem_flags) + +The ``mem_flags`` parameter, such as ``GFP_ATOMIC``, controls memory +allocation, such as whether the lower levels may block when memory is tight. + +It immediately returns, either with status 0 (request queued) or some +error code, usually caused by the following: + +- Out of memory (``-ENOMEM``) +- Unplugged device (``-ENODEV``) +- Stalled endpoint (``-EPIPE``) +- Too many queued ISO transfers (``-EAGAIN``) +- Too many requested ISO frames (``-EFBIG``) +- Invalid INT interval (``-EINVAL``) +- More than one packet for INT (``-EINVAL``) + +After submission, ``urb->status`` is ``-EINPROGRESS``; however, you should +never look at that value except in your completion callback. + +For isochronous endpoints, your completion handlers should (re)submit +URBs to the same endpoint with the ``URB_ISO_ASAP`` flag, using +multi-buffering, to get seamless ISO streaming. + + +How to cancel an already running URB? +===================================== + +There are two ways to cancel an URB you've submitted but which hasn't +been returned to your driver yet. For an asynchronous cancel, call +:c:func:`usb_unlink_urb`:: + + int usb_unlink_urb(struct urb *urb) + +It removes the urb from the internal list and frees all allocated +HW descriptors. The status is changed to reflect unlinking. Note +that the URB will not normally have finished when :c:func:`usb_unlink_urb` +returns; you must still wait for the completion handler to be called. + +To cancel an URB synchronously, call :c:func:`usb_kill_urb`:: + + void usb_kill_urb(struct urb *urb) + +It does everything :c:func:`usb_unlink_urb` does, and in addition it waits +until after the URB has been returned and the completion handler +has finished. It also marks the URB as temporarily unusable, so +that if the completion handler or anyone else tries to resubmit it +they will get a ``-EPERM`` error. Thus you can be sure that when +:c:func:`usb_kill_urb` returns, the URB is totally idle. + +There is a lifetime issue to consider. An URB may complete at any +time, and the completion handler may free the URB. If this happens +while :c:func:`usb_unlink_urb` or :c:func:`usb_kill_urb` is running, it will +cause a memory-access violation. The driver is responsible for avoiding this, +which often means some sort of lock will be needed to prevent the URB +from being deallocated while it is still in use. + +On the other hand, since usb_unlink_urb may end up calling the +completion handler, the handler must not take any lock that is held +when usb_unlink_urb is invoked. The general solution to this problem +is to increment the URB's reference count while holding the lock, then +drop the lock and call usb_unlink_urb or usb_kill_urb, and then +decrement the URB's reference count. You increment the reference +count by calling :c:func`usb_get_urb`:: + + struct urb *usb_get_urb(struct urb *urb) + +(ignore the return value; it is the same as the argument) and +decrement the reference count by calling :c:func:`usb_free_urb`. Of course, +none of this is necessary if there's no danger of the URB being freed +by the completion handler. + + +What about the completion handler? +================================== + +The handler is of the following type:: + + typedef void (*usb_complete_t)(struct urb *) + +I.e., it gets the URB that caused the completion call. In the completion +handler, you should have a look at ``urb->status`` to detect any USB errors. +Since the context parameter is included in the URB, you can pass +information to the completion handler. + +Note that even when an error (or unlink) is reported, data may have been +transferred. That's because USB transfers are packetized; it might take +sixteen packets to transfer your 1KByte buffer, and ten of them might +have transferred successfully before the completion was called. + + +.. warning:: + + NEVER SLEEP IN A COMPLETION HANDLER. + + These are often called in atomic context. + +In the current kernel, completion handlers run with local interrupts +disabled, but in the future this will be changed, so don't assume that +local IRQs are always disabled inside completion handlers. + +How to do isochronous (ISO) transfers? +====================================== + +Besides the fields present on a bulk transfer, for ISO, you also +also have to set ``urb->interval`` to say how often to make transfers; it's +often one per frame (which is once every microframe for highspeed devices). +The actual interval used will be a power of two that's no bigger than what +you specify. You can use the :c:func:`usb_fill_int_urb` macro to fill +most ISO transfer fields. + +For ISO transfers you also have to fill a :c:type:`usb_iso_packet_descriptor` +structure, allocated at the end of the URB by :c:func:`usb_alloc_urb`, for +each packet you want to schedule. + +The :c:func:`usb_submit_urb` call modifies ``urb->interval`` to the implemented +interval value that is less than or equal to the requested interval value. If +``URB_ISO_ASAP`` scheduling is used, ``urb->start_frame`` is also updated. + +For each entry you have to specify the data offset for this frame (base is +transfer_buffer), and the length you want to write/expect to read. +After completion, actual_length contains the actual transferred length and +status contains the resulting status for the ISO transfer for this frame. +It is allowed to specify a varying length from frame to frame (e.g. for +audio synchronisation/adaptive transfer rates). You can also use the length +0 to omit one or more frames (striping). + +For scheduling you can choose your own start frame or ``URB_ISO_ASAP``. As +explained earlier, if you always keep at least one URB queued and your +completion keeps (re)submitting a later URB, you'll get smooth ISO streaming +(if usb bandwidth utilization allows). + +If you specify your own start frame, make sure it's several frames in advance +of the current frame. You might want this model if you're synchronizing +ISO data with some other event stream. + + +How to start interrupt (INT) transfers? +======================================= + +Interrupt transfers, like isochronous transfers, are periodic, and happen +in intervals that are powers of two (1, 2, 4 etc) units. Units are frames +for full and low speed devices, and microframes for high speed ones. +You can use the :c:func:`usb_fill_int_urb` macro to fill INT transfer fields. + +The :c:func:`usb_submit_urb` call modifies ``urb->interval`` to the implemented +interval value that is less than or equal to the requested interval value. + +In Linux 2.6, unlike earlier versions, interrupt URBs are not automagically +restarted when they complete. They end when the completion handler is +called, just like other URBs. If you want an interrupt URB to be restarted, +your completion handler must resubmit it. +s diff --git a/Documentation/driver-api/usb/anchors.rst b/Documentation/driver-api/usb/anchors.rst new file mode 100644 index 000000000..4b248e691 --- /dev/null +++ b/Documentation/driver-api/usb/anchors.rst @@ -0,0 +1,83 @@ +USB Anchors +~~~~~~~~~~~ + +What is anchor? +=============== + +A USB driver needs to support some callbacks requiring +a driver to cease all IO to an interface. To do so, a +driver has to keep track of the URBs it has submitted +to know they've all completed or to call usb_kill_urb +for them. The anchor is a data structure takes care of +keeping track of URBs and provides methods to deal with +multiple URBs. + +Allocation and Initialisation +============================= + +There's no API to allocate an anchor. It is simply declared +as struct usb_anchor. :c:func:`init_usb_anchor` must be called to +initialise the data structure. + +Deallocation +============ + +Once it has no more URBs associated with it, the anchor can be +freed with normal memory management operations. + +Association and disassociation of URBs with anchors +=================================================== + +An association of URBs to an anchor is made by an explicit +call to :c:func:`usb_anchor_urb`. The association is maintained until +an URB is finished by (successful) completion. Thus disassociation +is automatic. A function is provided to forcibly finish (kill) +all URBs associated with an anchor. +Furthermore, disassociation can be made with :c:func:`usb_unanchor_urb` + +Operations on multitudes of URBs +================================ + +:c:func:`usb_kill_anchored_urbs` +-------------------------------- + +This function kills all URBs associated with an anchor. The URBs +are called in the reverse temporal order they were submitted. +This way no data can be reordered. + +:c:func:`usb_unlink_anchored_urbs` +---------------------------------- + + +This function unlinks all URBs associated with an anchor. The URBs +are processed in the reverse temporal order they were submitted. +This is similar to :c:func:`usb_kill_anchored_urbs`, but it will not sleep. +Therefore no guarantee is made that the URBs have been unlinked when +the call returns. They may be unlinked later but will be unlinked in +finite time. + +:c:func:`usb_scuttle_anchored_urbs` +----------------------------------- + +All URBs of an anchor are unanchored en masse. + +:c:func:`usb_wait_anchor_empty_timeout` +--------------------------------------- + +This function waits for all URBs associated with an anchor to finish +or a timeout, whichever comes first. Its return value will tell you +whether the timeout was reached. + +:c:func:`usb_anchor_empty` +-------------------------- + +Returns true if no URBs are associated with an anchor. Locking +is the caller's responsibility. + +:c:func:`usb_get_from_anchor` +----------------------------- + +Returns the oldest anchored URB of an anchor. The URB is unanchored +and returned with a reference. As you may mix URBs to several +destinations in one anchor you have no guarantee the chronologically +first submitted URB is returned. diff --git a/Documentation/driver-api/usb/bulk-streams.rst b/Documentation/driver-api/usb/bulk-streams.rst new file mode 100644 index 000000000..99b515bab --- /dev/null +++ b/Documentation/driver-api/usb/bulk-streams.rst @@ -0,0 +1,83 @@ +USB bulk streams +~~~~~~~~~~~~~~~~ + +Background +========== + +Bulk endpoint streams were added in the USB 3.0 specification. Streams allow a +device driver to overload a bulk endpoint so that multiple transfers can be +queued at once. + +Streams are defined in sections 4.4.6.4 and 8.12.1.4 of the Universal Serial Bus +3.0 specification at http://www.usb.org/developers/docs/ The USB Attached SCSI +Protocol, which uses streams to queue multiple SCSI commands, can be found on +the T10 website (http://t10.org/). + + +Device-side implications +======================== + +Once a buffer has been queued to a stream ring, the device is notified (through +an out-of-band mechanism on another endpoint) that data is ready for that stream +ID. The device then tells the host which "stream" it wants to start. The host +can also initiate a transfer on a stream without the device asking, but the +device can refuse that transfer. Devices can switch between streams at any +time. + + +Driver implications +=================== + +:: + + int usb_alloc_streams(struct usb_interface *interface, + struct usb_host_endpoint **eps, unsigned int num_eps, + unsigned int num_streams, gfp_t mem_flags); + +Device drivers will call this API to request that the host controller driver +allocate memory so the driver can use up to num_streams stream IDs. They must +pass an array of usb_host_endpoints that need to be setup with similar stream +IDs. This is to ensure that a UASP driver will be able to use the same stream +ID for the bulk IN and OUT endpoints used in a Bi-directional command sequence. + +The return value is an error condition (if one of the endpoints doesn't support +streams, or the xHCI driver ran out of memory), or the number of streams the +host controller allocated for this endpoint. The xHCI host controller hardware +declares how many stream IDs it can support, and each bulk endpoint on a +SuperSpeed device will say how many stream IDs it can handle. Therefore, +drivers should be able to deal with being allocated less stream IDs than they +requested. + +Do NOT call this function if you have URBs enqueued for any of the endpoints +passed in as arguments. Do not call this function to request less than two +streams. + +Drivers will only be allowed to call this API once for the same endpoint +without calling usb_free_streams(). This is a simplification for the xHCI host +controller driver, and may change in the future. + + +Picking new Stream IDs to use +============================= + +Stream ID 0 is reserved, and should not be used to communicate with devices. If +usb_alloc_streams() returns with a value of N, you may use streams 1 though N. +To queue an URB for a specific stream, set the urb->stream_id value. If the +endpoint does not support streams, an error will be returned. + +Note that new API to choose the next stream ID will have to be added if the xHCI +driver supports secondary stream IDs. + + +Clean up +======== + +If a driver wishes to stop using streams to communicate with the device, it +should call:: + + void usb_free_streams(struct usb_interface *interface, + struct usb_host_endpoint **eps, unsigned int num_eps, + gfp_t mem_flags); + +All stream IDs will be deallocated when the driver releases the interface, to +ensure that drivers that don't support streams will be able to use the endpoint. diff --git a/Documentation/driver-api/usb/callbacks.rst b/Documentation/driver-api/usb/callbacks.rst new file mode 100644 index 000000000..2b80cf54b --- /dev/null +++ b/Documentation/driver-api/usb/callbacks.rst @@ -0,0 +1,157 @@ +USB core callbacks +~~~~~~~~~~~~~~~~~~ + +What callbacks will usbcore do? +=============================== + +Usbcore will call into a driver through callbacks defined in the driver +structure and through the completion handler of URBs a driver submits. +Only the former are in the scope of this document. These two kinds of +callbacks are completely independent of each other. Information on the +completion callback can be found in :ref:`usb-urb`. + +The callbacks defined in the driver structure are: + +1. Hotplugging callbacks: + + - @probe: + Called to see if the driver is willing to manage a particular + interface on a device. + + - @disconnect: + Called when the interface is no longer accessible, usually + because its device has been (or is being) disconnected or the + driver module is being unloaded. + +2. Odd backdoor through usbfs: + + - @ioctl: + Used for drivers that want to talk to userspace through + the "usbfs" filesystem. This lets devices provide ways to + expose information to user space regardless of where they + do (or don't) show up otherwise in the filesystem. + +3. Power management (PM) callbacks: + + - @suspend: + Called when the device is going to be suspended. + + - @resume: + Called when the device is being resumed. + + - @reset_resume: + Called when the suspended device has been reset instead + of being resumed. + +4. Device level operations: + + - @pre_reset: + Called when the device is about to be reset. + + - @post_reset: + Called after the device has been reset + +The ioctl interface (2) should be used only if you have a very good +reason. Sysfs is preferred these days. The PM callbacks are covered +separately in :ref:`usb-power-management`. + +Calling conventions +=================== + +All callbacks are mutually exclusive. There's no need for locking +against other USB callbacks. All callbacks are called from a task +context. You may sleep. However, it is important that all sleeps have a +small fixed upper limit in time. In particular you must not call out to +user space and await results. + +Hotplugging callbacks +===================== + +These callbacks are intended to associate and disassociate a driver with +an interface. A driver's bond to an interface is exclusive. + +The probe() callback +-------------------- + +:: + + int (*probe) (struct usb_interface *intf, + const struct usb_device_id *id); + +Accept or decline an interface. If you accept the device return 0, +otherwise -ENODEV or -ENXIO. Other error codes should be used only if a +genuine error occurred during initialisation which prevented a driver +from accepting a device that would else have been accepted. +You are strongly encouraged to use usbcore's facility, +usb_set_intfdata(), to associate a data structure with an interface, so +that you know which internal state and identity you associate with a +particular interface. The device will not be suspended and you may do IO +to the interface you are called for and endpoint 0 of the device. Device +initialisation that doesn't take too long is a good idea here. + +The disconnect() callback +------------------------- + +:: + + void (*disconnect) (struct usb_interface *intf); + +This callback is a signal to break any connection with an interface. +You are not allowed any IO to a device after returning from this +callback. You also may not do any other operation that may interfere +with another driver bound the interface, eg. a power management +operation. +If you are called due to a physical disconnection, all your URBs will be +killed by usbcore. Note that in this case disconnect will be called some +time after the physical disconnection. Thus your driver must be prepared +to deal with failing IO even prior to the callback. + +Device level callbacks +====================== + +pre_reset +--------- + +:: + + int (*pre_reset)(struct usb_interface *intf); + +A driver or user space is triggering a reset on the device which +contains the interface passed as an argument. Cease IO, wait for all +outstanding URBs to complete, and save any device state you need to +restore. No more URBs may be submitted until the post_reset method +is called. + +If you need to allocate memory here, use GFP_NOIO or GFP_ATOMIC, if you +are in atomic context. + +post_reset +---------- + +:: + + int (*post_reset)(struct usb_interface *intf); + +The reset has completed. Restore any saved device state and begin +using the device again. + +If you need to allocate memory here, use GFP_NOIO or GFP_ATOMIC, if you +are in atomic context. + +Call sequences +============== + +No callbacks other than probe will be invoked for an interface +that isn't bound to your driver. + +Probe will never be called for an interface bound to a driver. +Hence following a successful probe, disconnect will be called +before there is another probe for the same interface. + +Once your driver is bound to an interface, disconnect can be +called at any time except in between pre_reset and post_reset. +pre_reset is always followed by post_reset, even if the reset +failed or the device has been unplugged. + +suspend is always followed by one of: resume, reset_resume, or +disconnect. diff --git a/Documentation/driver-api/usb/dma.rst b/Documentation/driver-api/usb/dma.rst new file mode 100644 index 000000000..59d5aee89 --- /dev/null +++ b/Documentation/driver-api/usb/dma.rst @@ -0,0 +1,136 @@ +USB DMA +~~~~~~~ + +In Linux 2.5 kernels (and later), USB device drivers have additional control +over how DMA may be used to perform I/O operations. The APIs are detailed +in the kernel usb programming guide (kerneldoc, from the source code). + +API overview +============ + +The big picture is that USB drivers can continue to ignore most DMA issues, +though they still must provide DMA-ready buffers (see +``Documentation/DMA-API-HOWTO.txt``). That's how they've worked through +the 2.4 (and earlier) kernels, or they can now be DMA-aware. + +DMA-aware usb drivers: + +- New calls enable DMA-aware drivers, letting them allocate dma buffers and + manage dma mappings for existing dma-ready buffers (see below). + +- URBs have an additional "transfer_dma" field, as well as a transfer_flags + bit saying if it's valid. (Control requests also have "setup_dma", but + drivers must not use it.) + +- "usbcore" will map this DMA address, if a DMA-aware driver didn't do + it first and set ``URB_NO_TRANSFER_DMA_MAP``. HCDs + don't manage dma mappings for URBs. + +- There's a new "generic DMA API", parts of which are usable by USB device + drivers. Never use dma_set_mask() on any USB interface or device; that + would potentially break all devices sharing that bus. + +Eliminating copies +================== + +It's good to avoid making CPUs copy data needlessly. The costs can add up, +and effects like cache-trashing can impose subtle penalties. + +- If you're doing lots of small data transfers from the same buffer all + the time, that can really burn up resources on systems which use an + IOMMU to manage the DMA mappings. It can cost MUCH more to set up and + tear down the IOMMU mappings with each request than perform the I/O! + + For those specific cases, USB has primitives to allocate less expensive + memory. They work like kmalloc and kfree versions that give you the right + kind of addresses to store in urb->transfer_buffer and urb->transfer_dma. + You'd also set ``URB_NO_TRANSFER_DMA_MAP`` in urb->transfer_flags:: + + void *usb_alloc_coherent (struct usb_device *dev, size_t size, + int mem_flags, dma_addr_t *dma); + + void usb_free_coherent (struct usb_device *dev, size_t size, + void *addr, dma_addr_t dma); + + Most drivers should **NOT** be using these primitives; they don't need + to use this type of memory ("dma-coherent"), and memory returned from + :c:func:`kmalloc` will work just fine. + + The memory buffer returned is "dma-coherent"; sometimes you might need to + force a consistent memory access ordering by using memory barriers. It's + not using a streaming DMA mapping, so it's good for small transfers on + systems where the I/O would otherwise thrash an IOMMU mapping. (See + ``Documentation/DMA-API-HOWTO.txt`` for definitions of "coherent" and + "streaming" DMA mappings.) + + Asking for 1/Nth of a page (as well as asking for N pages) is reasonably + space-efficient. + + On most systems the memory returned will be uncached, because the + semantics of dma-coherent memory require either bypassing CPU caches + or using cache hardware with bus-snooping support. While x86 hardware + has such bus-snooping, many other systems use software to flush cache + lines to prevent DMA conflicts. + +- Devices on some EHCI controllers could handle DMA to/from high memory. + + Unfortunately, the current Linux DMA infrastructure doesn't have a sane + way to expose these capabilities ... and in any case, HIGHMEM is mostly a + design wart specific to x86_32. So your best bet is to ensure you never + pass a highmem buffer into a USB driver. That's easy; it's the default + behavior. Just don't override it; e.g. with ``NETIF_F_HIGHDMA``. + + This may force your callers to do some bounce buffering, copying from + high memory to "normal" DMA memory. If you can come up with a good way + to fix this issue (for x86_32 machines with over 1 GByte of memory), + feel free to submit patches. + +Working with existing buffers +============================= + +Existing buffers aren't usable for DMA without first being mapped into the +DMA address space of the device. However, most buffers passed to your +driver can safely be used with such DMA mapping. (See the first section +of Documentation/DMA-API-HOWTO.txt, titled "What memory is DMA-able?") + +- When you're using scatterlists, you can map everything at once. On some + systems, this kicks in an IOMMU and turns the scatterlists into single + DMA transactions:: + + int usb_buffer_map_sg (struct usb_device *dev, unsigned pipe, + struct scatterlist *sg, int nents); + + void usb_buffer_dmasync_sg (struct usb_device *dev, unsigned pipe, + struct scatterlist *sg, int n_hw_ents); + + void usb_buffer_unmap_sg (struct usb_device *dev, unsigned pipe, + struct scatterlist *sg, int n_hw_ents); + + It's probably easier to use the new ``usb_sg_*()`` calls, which do the DMA + mapping and apply other tweaks to make scatterlist i/o be fast. + +- Some drivers may prefer to work with the model that they're mapping large + buffers, synchronizing their safe re-use. (If there's no re-use, then let + usbcore do the map/unmap.) Large periodic transfers make good examples + here, since it's cheaper to just synchronize the buffer than to unmap it + each time an urb completes and then re-map it on during resubmission. + + These calls all work with initialized urbs: ``urb->dev``, ``urb->pipe``, + ``urb->transfer_buffer``, and ``urb->transfer_buffer_length`` must all be + valid when these calls are used (``urb->setup_packet`` must be valid too + if urb is a control request):: + + struct urb *usb_buffer_map (struct urb *urb); + + void usb_buffer_dmasync (struct urb *urb); + + void usb_buffer_unmap (struct urb *urb); + + The calls manage ``urb->transfer_dma`` for you, and set + ``URB_NO_TRANSFER_DMA_MAP`` so that usbcore won't map or unmap the buffer. + They cannot be used for setup_packet buffers in control requests. + +Note that several of those interfaces are currently commented out, since +they don't have current users. See the source code. Other than the dmasync +calls (where the underlying DMA primitives have changed), most of them can +easily be commented back in if you want to use them. diff --git a/Documentation/driver-api/usb/dwc3.rst b/Documentation/driver-api/usb/dwc3.rst new file mode 100644 index 000000000..8b36ff11c --- /dev/null +++ b/Documentation/driver-api/usb/dwc3.rst @@ -0,0 +1,711 @@ +=============================================================== +Synopsys DesignWare Core SuperSpeed USB 3.0 Controller +=============================================================== + +:Author: Felipe Balbi <felipe.balbi@linux.intel.com> +:Date: April 2017 + +Introduction +============ + +The *Synopsys DesignWare Core SuperSpeed USB 3.0 Controller* +(hereinafter referred to as *DWC3*) is a USB SuperSpeed compliant +controller which can be configured in one of 4 ways: + + 1. Peripheral-only configuration + 2. Host-only configuration + 3. Dual-Role configuration + 4. Hub configuration + +Linux currently supports several versions of this controller. In all +likelyhood, the version in your SoC is already supported. At the time +of this writing, known tested versions range from 2.02a to 3.10a. As a +rule of thumb, anything above 2.02a should work reliably well. + +Currently, we have many known users for this driver. In alphabetical +order: + + 1. Cavium + 2. Intel Corporation + 3. Qualcomm + 4. Rockchip + 5. ST + 6. Samsung + 7. Texas Instruments + 8. Xilinx + +Summary of Features +====================== + +For details about features supported by your version of DWC3, consult +your IP team and/or *Synopsys DesignWare Core SuperSpeed USB 3.0 +Controller Databook*. Following is a list of features supported by the +driver at the time of this writing: + + 1. Up to 16 bidirectional endpoints (including the control + pipe - ep0) + 2. Flexible endpoint configuration + 3. Simultaneous IN and OUT transfer support + 4. Scatter-list support + 5. Up to 256 TRBs [#trb]_ per endpoint + 6. Support for all transfer types (*Control*, *Bulk*, + *Interrupt*, and *Isochronous*) + 7. SuperSpeed Bulk Streams + 8. Link Power Management + 9. Trace Events for debugging + 10. DebugFS [#debugfs]_ interface + +These features have all been exercised with many of the **in-tree** +gadget drivers. We have verified both *ConfigFS* [#configfs]_ and +legacy gadget drivers. + +Driver Design +============== + +The DWC3 driver sits on the *drivers/usb/dwc3/* directory. All files +related to this driver are in this one directory. This makes it easy +for new-comers to read the code and understand how it behaves. + +Because of DWC3's configuration flexibility, the driver is a little +complex in some places but it should be rather straightforward to +understand. + +The biggest part of the driver refers to the Gadget API. + +Known Limitations +=================== + +Like any other HW, DWC3 has its own set of limitations. To avoid +constant questions about such problems, we decided to document them +here and have a single location to where we could point users. + +OUT Transfer Size Requirements +--------------------------------- + +According to Synopsys Databook, all OUT transfer TRBs [#trb]_ must +have their *size* field set to a value which is integer divisible by +the endpoint's *wMaxPacketSize*. This means that *e.g.* in order to +receive a Mass Storage *CBW* [#cbw]_, req->length must either be set +to a value that's divisible by *wMaxPacketSize* (1024 on SuperSpeed, +512 on HighSpeed, etc), or DWC3 driver must add a Chained TRB pointing +to a throw-away buffer for the remaining length. Without this, OUT +transfers will **NOT** start. + +Note that as of this writing, this won't be a problem because DWC3 is +fully capable of appending a chained TRB for the remaining length and +completely hide this detail from the gadget driver. It's still worth +mentioning because this seems to be the largest source of queries +about DWC3 and *non-working transfers*. + +TRB Ring Size Limitation +------------------------- + +We, currently, have a hard limit of 256 TRBs [#trb]_ per endpoint, +with the last TRB being a Link TRB [#link_trb]_ pointing back to the +first. This limit is arbitrary but it has the benefit of adding up to +exactly 4096 bytes, or 1 Page. + +DWC3 driver will try its best to cope with more than 255 requests and, +for the most part, it should work normally. However this is not +something that has been exercised very frequently. If you experience +any problems, see section **Reporting Bugs** below. + +Reporting Bugs +================ + +Whenever you encounter a problem with DWC3, first and foremost you +should make sure that: + + 1. You're running latest tag from `Linus' tree`_ + 2. You can reproduce the error without any out-of-tree changes + to DWC3 + 3. You have checked that it's not a fault on the host machine + +After all these are verified, then here's how to capture enough +information so we can be of any help to you. + +Required Information +--------------------- + +DWC3 relies exclusively on Trace Events for debugging. Everything is +exposed there, with some extra bits being exposed to DebugFS +[#debugfs]_. + +In order to capture DWC3's Trace Events you should run the following +commands **before** plugging the USB cable to a host machine: + +.. code-block:: sh + + # mkdir -p /d + # mkdir -p /t + # mount -t debugfs none /d + # mount -t tracefs none /t + # echo 81920 > /t/buffer_size_kb + # echo 1 > /t/events/dwc3/enable + +After this is done, you can connect your USB cable and reproduce the +problem. As soon as the fault is reproduced, make a copy of files +``trace`` and ``regdump``, like so: + +.. code-block:: sh + + # cp /t/trace /root/trace.txt + # cat /d/*dwc3*/regdump > /root/regdump.txt + +Make sure to compress ``trace.txt`` and ``regdump.txt`` in a tarball +and email it to `me`_ with `linux-usb`_ in Cc. If you want to be extra +sure that I'll help you, write your subject line in the following +format: + + **[BUG REPORT] usb: dwc3: Bug while doing XYZ** + +On the email body, make sure to detail what you doing, which gadget +driver you were using, how to reproduce the problem, what SoC you're +using, which OS (and its version) was running on the Host machine. + +With all this information, we should be able to understand what's +going on and be helpful to you. + +Debugging +=========== + +First and foremost a disclaimer:: + + DISCLAIMER: The information available on DebugFS and/or TraceFS can + change at any time at any Major Linux Kernel Release. If writing + scripts, do **NOT** assume information to be available in the + current format. + +With that out of the way, let's carry on. + +If you're willing to debug your own problem, you deserve a round of +applause :-) + +Anyway, there isn't much to say here other than Trace Events will be +really helpful in figuring out issues with DWC3. Also, access to +Synopsys Databook will be **really** valuable in this case. + +A USB Sniffer can be helpful at times but it's not entirely required, +there's a lot that can be understood without looking at the wire. + +Feel free to email `me`_ and Cc `linux-usb`_ if you need any help. + +``DebugFS`` +------------- + +``DebugFS`` is very good for gathering snapshots of what's going on +with DWC3 and/or any endpoint. + +On DWC3's ``DebugFS`` directory, you will find the following files and +directories: + +``ep[0..15]{in,out}/`` +``link_state`` +``regdump`` +``testmode`` + +``link_state`` +`````````````` + +When read, ``link_state`` will print out one of ``U0``, ``U1``, +``U2``, ``U3``, ``SS.Disabled``, ``RX.Detect``, ``SS.Inactive``, +``Polling``, ``Recovery``, ``Hot Reset``, ``Compliance``, +``Loopback``, ``Reset``, ``Resume`` or ``UNKNOWN link state``. + +This file can also be written to in order to force link to one of the +states above. + +``regdump`` +````````````` + +File name is self-explanatory. When read, ``regdump`` will print out a +register dump of DWC3. Note that this file can be grepped to find the +information you want. + +``testmode`` +`````````````` + +When read, ``testmode`` will print out a name of one of the specified +USB 2.0 Testmodes (``test_j``, ``test_k``, ``test_se0_nak``, +``test_packet``, ``test_force_enable``) or the string ``no test`` in +case no tests are currently being executed. + +In order to start any of these test modes, the same strings can be +written to the file and DWC3 will enter the requested test mode. + + +``ep[0..15]{in,out}`` +`````````````````````` + +For each endpoint we expose one directory following the naming +convention ``ep$num$dir`` *(ep0in, ep0out, ep1in, ...)*. Inside each +of these directories you will find the following files: + +``descriptor_fetch_queue`` +``event_queue`` +``rx_fifo_queue`` +``rx_info_queue`` +``rx_request_queue`` +``transfer_type`` +``trb_ring`` +``tx_fifo_queue`` +``tx_request_queue`` + +With access to Synopsys Databook, you can decode the information on +them. + +``transfer_type`` +~~~~~~~~~~~~~~~~~~ + +When read, ``transfer_type`` will print out one of ``control``, +``bulk``, ``interrupt`` or ``isochronous`` depending on what the +endpoint descriptor says. If the endpoint hasn't been enabled yet, it +will print ``--``. + +``trb_ring`` +~~~~~~~~~~~~~ + +When read, ``trb_ring`` will print out details about all TRBs on the +ring. It will also tell you where our enqueue and dequeue pointers are +located in the ring: + +.. code-block:: sh + + buffer_addr,size,type,ioc,isp_imi,csp,chn,lst,hwo + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c75c000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c75c000,481,normal,1,0,1,0,0,0 + 000000002c784000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c784000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c784000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c784000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c75c000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c75c000,481,normal,1,0,1,0,0,0 + 000000002c780000,481,normal,1,0,1,0,0,0 + 000000002c784000,481,normal,1,0,1,0,0,0 + 000000002c788000,481,normal,1,0,1,0,0,0 + 000000002c78c000,481,normal,1,0,1,0,0,0 + 000000002c790000,481,normal,1,0,1,0,0,0 + 000000002c754000,481,normal,1,0,1,0,0,0 + 000000002c758000,481,normal,1,0,1,0,0,0 + 000000002c75c000,512,normal,1,0,1,0,0,1 D + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 E + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 0000000000000000,0,UNKNOWN,0,0,0,0,0,0 + 00000000381ab000,0,link,0,0,0,0,0,1 + + +Trace Events +------------- + +DWC3 also provides several trace events which help us gathering +information about the behavior of the driver during runtime. + +In order to use these events, you must enable ``CONFIG_FTRACE`` in +your kernel config. + +For details about how enable DWC3 events, see section **Reporting +Bugs**. + +The following subsections will give details about each Event Class and +each Event defined by DWC3. + +MMIO +``````` + +It is sometimes useful to look at every MMIO access when looking for +bugs. Because of that, DWC3 offers two Trace Events (one for +dwc3_readl() and one for dwc3_writel()). ``TP_printk`` follows:: + + TP_printk("addr %p value %08x", __entry->base + __entry->offset, + __entry->value) + +Interrupt Events +```````````````` + +Every IRQ event can be logged and decoded into a human readable +string. Because every event will be different, we don't give an +example other than the ``TP_printk`` format used:: + + TP_printk("event (%08x): %s", __entry->event, + dwc3_decode_event(__entry->event, __entry->ep0state)) + +Control Request +````````````````` + +Every USB Control Request can be logged to the trace buffer. The +output format is:: + + TP_printk("%s", dwc3_decode_ctrl(__entry->bRequestType, + __entry->bRequest, __entry->wValue, + __entry->wIndex, __entry->wLength) + ) + +Note that Standard Control Requests will be decoded into +human-readable strings with their respective arguments. Class and +Vendor requests will be printed out a sequence of 8 bytes in hex +format. + +Lifetime of a ``struct usb_request`` +``````````````````````````````````````` + +The entire lifetime of a ``struct usb_request`` can be tracked on the +trace buffer. We have one event for each of allocation, free, +queueing, dequeueing, and giveback. Output format is:: + + TP_printk("%s: req %p length %u/%u %s%s%s ==> %d", + __get_str(name), __entry->req, __entry->actual, __entry->length, + __entry->zero ? "Z" : "z", + __entry->short_not_ok ? "S" : "s", + __entry->no_interrupt ? "i" : "I", + __entry->status + ) + +Generic Commands +```````````````````` + +We can log and decode every Generic Command with its completion +code. Format is:: + + TP_printk("cmd '%s' [%x] param %08x --> status: %s", + dwc3_gadget_generic_cmd_string(__entry->cmd), + __entry->cmd, __entry->param, + dwc3_gadget_generic_cmd_status_string(__entry->status) + ) + +Endpoint Commands +```````````````````` + +Endpoints commands can also be logged together with completion +code. Format is:: + + TP_printk("%s: cmd '%s' [%d] params %08x %08x %08x --> status: %s", + __get_str(name), dwc3_gadget_ep_cmd_string(__entry->cmd), + __entry->cmd, __entry->param0, + __entry->param1, __entry->param2, + dwc3_ep_cmd_status_string(__entry->cmd_status) + ) + +Lifetime of a ``TRB`` +`````````````````````` + +A ``TRB`` Lifetime is simple. We are either preparing a ``TRB`` or +completing it. With these two events, we can see how a ``TRB`` changes +over time. Format is:: + + TP_printk("%s: %d/%d trb %p buf %08x%08x size %s%d ctrl %08x (%c%c%c%c:%c%c:%s)", + __get_str(name), __entry->queued, __entry->allocated, + __entry->trb, __entry->bph, __entry->bpl, + ({char *s; + int pcm = ((__entry->size >> 24) & 3) + 1; + switch (__entry->type) { + case USB_ENDPOINT_XFER_INT: + case USB_ENDPOINT_XFER_ISOC: + switch (pcm) { + case 1: + s = "1x "; + break; + case 2: + s = "2x "; + break; + case 3: + s = "3x "; + break; + } + default: + s = ""; + } s; }), + DWC3_TRB_SIZE_LENGTH(__entry->size), __entry->ctrl, + __entry->ctrl & DWC3_TRB_CTRL_HWO ? 'H' : 'h', + __entry->ctrl & DWC3_TRB_CTRL_LST ? 'L' : 'l', + __entry->ctrl & DWC3_TRB_CTRL_CHN ? 'C' : 'c', + __entry->ctrl & DWC3_TRB_CTRL_CSP ? 'S' : 's', + __entry->ctrl & DWC3_TRB_CTRL_ISP_IMI ? 'S' : 's', + __entry->ctrl & DWC3_TRB_CTRL_IOC ? 'C' : 'c', + dwc3_trb_type_string(DWC3_TRBCTL_TYPE(__entry->ctrl)) + ) + +Lifetime of an Endpoint +``````````````````````` + +And endpoint's lifetime is summarized with enable and disable +operations, both of which can be traced. Format is:: + + TP_printk("%s: mps %d/%d streams %d burst %d ring %d/%d flags %c:%c%c%c%c%c:%c:%c", + __get_str(name), __entry->maxpacket, + __entry->maxpacket_limit, __entry->max_streams, + __entry->maxburst, __entry->trb_enqueue, + __entry->trb_dequeue, + __entry->flags & DWC3_EP_ENABLED ? 'E' : 'e', + __entry->flags & DWC3_EP_STALL ? 'S' : 's', + __entry->flags & DWC3_EP_WEDGE ? 'W' : 'w', + __entry->flags & DWC3_EP_TRANSFER_STARTED ? 'B' : 'b', + __entry->flags & DWC3_EP_PENDING_REQUEST ? 'P' : 'p', + __entry->flags & DWC3_EP_END_TRANSFER_PENDING ? 'E' : 'e', + __entry->direction ? '<' : '>' + ) + + +Structures, Methods and Definitions +==================================== + +.. kernel-doc:: drivers/usb/dwc3/core.h + :doc: main data structures + :internal: + +.. kernel-doc:: drivers/usb/dwc3/gadget.h + :doc: gadget-only helpers + :internal: + +.. kernel-doc:: drivers/usb/dwc3/gadget.c + :doc: gadget-side implementation + :internal: + +.. kernel-doc:: drivers/usb/dwc3/core.c + :doc: core driver (probe, PM, etc) + :internal: + +.. [#trb] Transfer Request Block +.. [#link_trb] Transfer Request Block pointing to another Transfer + Request Block. +.. [#debugfs] The Debug File System +.. [#configfs] The Config File System +.. [#cbw] Command Block Wrapper +.. _Linus' tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/ +.. _me: felipe.balbi@linux.intel.com +.. _linux-usb: linux-usb@vger.kernel.org diff --git a/Documentation/driver-api/usb/error-codes.rst b/Documentation/driver-api/usb/error-codes.rst new file mode 100644 index 000000000..a3e84bfac --- /dev/null +++ b/Documentation/driver-api/usb/error-codes.rst @@ -0,0 +1,207 @@ +.. _usb-error-codes: + +USB Error codes +~~~~~~~~~~~~~~~ + +:Revised: 2004-Oct-21 + +This is the documentation of (hopefully) all possible error codes (and +their interpretation) that can be returned from usbcore. + +Some of them are returned by the Host Controller Drivers (HCDs), which +device drivers only see through usbcore. As a rule, all the HCDs should +behave the same except for transfer speed dependent behaviors and the +way certain faults are reported. + + +Error codes returned by :c:func:`usb_submit_urb` +================================================ + +Non-USB-specific: + + +=============== =============================================== +0 URB submission went fine + +``-ENOMEM`` no memory for allocation of internal structures +=============== =============================================== + +USB-specific: + +======================= ======================================================= +``-EBUSY`` The URB is already active. + +``-ENODEV`` specified USB-device or bus doesn't exist + +``-ENOENT`` specified interface or endpoint does not exist or + is not enabled + +``-ENXIO`` host controller driver does not support queuing of + this type of urb. (treat as a host controller bug.) + +``-EINVAL`` a) Invalid transfer type specified (or not supported) + b) Invalid or unsupported periodic transfer interval + c) ISO: attempted to change transfer interval + d) ISO: ``number_of_packets`` is < 0 + e) various other cases + +``-EXDEV`` ISO: ``URB_ISO_ASAP`` wasn't specified and all the + frames the URB would be scheduled in have already + expired. + +``-EFBIG`` Host controller driver can't schedule that many ISO + frames. + +``-EPIPE`` The pipe type specified in the URB doesn't match the + endpoint's actual type. + +``-EMSGSIZE`` (a) endpoint maxpacket size is zero; it is not usable + in the current interface altsetting. + (b) ISO packet is larger than the endpoint maxpacket. + (c) requested data transfer length is invalid: negative + or too large for the host controller. + +``-ENOSPC`` This request would overcommit the usb bandwidth reserved + for periodic transfers (interrupt, isochronous). + +``-ESHUTDOWN`` The device or host controller has been disabled due to + some problem that could not be worked around. + +``-EPERM`` Submission failed because ``urb->reject`` was set. + +``-EHOSTUNREACH`` URB was rejected because the device is suspended. + +``-ENOEXEC`` A control URB doesn't contain a Setup packet. +======================= ======================================================= + +Error codes returned by ``in urb->status`` or in ``iso_frame_desc[n].status`` (for ISO) +======================================================================================= + +USB device drivers may only test urb status values in completion handlers. +This is because otherwise there would be a race between HCDs updating +these values on one CPU, and device drivers testing them on another CPU. + +A transfer's actual_length may be positive even when an error has been +reported. That's because transfers often involve several packets, so that +one or more packets could finish before an error stops further endpoint I/O. + +For isochronous URBs, the urb status value is non-zero only if the URB is +unlinked, the device is removed, the host controller is disabled, or the total +transferred length is less than the requested length and the +``URB_SHORT_NOT_OK`` flag is set. Completion handlers for isochronous URBs +should only see ``urb->status`` set to zero, ``-ENOENT``, ``-ECONNRESET``, +``-ESHUTDOWN``, or ``-EREMOTEIO``. Individual frame descriptor status fields +may report more status codes. + + +=============================== =============================================== +0 Transfer completed successfully + +``-ENOENT`` URB was synchronously unlinked by + :c:func:`usb_unlink_urb` + +``-EINPROGRESS`` URB still pending, no results yet + (That is, if drivers see this it's a bug.) + +``-EPROTO`` [#f1]_, [#f2]_ a) bitstuff error + b) no response packet received within the + prescribed bus turn-around time + c) unknown USB error + +``-EILSEQ`` [#f1]_, [#f2]_ a) CRC mismatch + b) no response packet received within the + prescribed bus turn-around time + c) unknown USB error + + Note that often the controller hardware does + not distinguish among cases a), b), and c), so + a driver cannot tell whether there was a + protocol error, a failure to respond (often + caused by device disconnect), or some other + fault. + +``-ETIME`` [#f2]_ No response packet received within the + prescribed bus turn-around time. This error + may instead be reported as + ``-EPROTO`` or ``-EILSEQ``. + +``-ETIMEDOUT`` Synchronous USB message functions use this code + to indicate timeout expired before the transfer + completed, and no other error was reported + by HC. + +``-EPIPE`` [#f2]_ Endpoint stalled. For non-control endpoints, + reset this status with + :c:func:`usb_clear_halt`. + +``-ECOMM`` During an IN transfer, the host controller + received data from an endpoint faster than it + could be written to system memory + +``-ENOSR`` During an OUT transfer, the host controller + could not retrieve data from system memory fast + enough to keep up with the USB data rate + +``-EOVERFLOW`` [#f1]_ The amount of data returned by the endpoint was + greater than either the max packet size of the + endpoint or the remaining buffer size. + "Babble". + +``-EREMOTEIO`` The data read from the endpoint did not fill + the specified buffer, and ``URB_SHORT_NOT_OK`` + was set in ``urb->transfer_flags``. + +``-ENODEV`` Device was removed. Often preceded by a burst + of other errors, since the hub driver doesn't + detect device removal events immediately. + +``-EXDEV`` ISO transfer only partially completed + (only set in ``iso_frame_desc[n].status``, + not ``urb->status``) + +``-EINVAL`` ISO madness, if this happens: Log off and + go home + +``-ECONNRESET`` URB was asynchronously unlinked by + :c:func:`usb_unlink_urb` + +``-ESHUTDOWN`` The device or host controller has been + disabled due to some problem that could not + be worked around, such as a physical + disconnect. +=============================== =============================================== + + +.. [#f1] + + Error codes like ``-EPROTO``, ``-EILSEQ`` and ``-EOVERFLOW`` normally + indicate hardware problems such as bad devices (including firmware) + or cables. + +.. [#f2] + + This is also one of several codes that different kinds of host + controller use to indicate a transfer has failed because of device + disconnect. In the interval before the hub driver starts disconnect + processing, devices may receive such fault reports for every request. + + + +Error codes returned by usbcore-functions +========================================= + +.. note:: expect also other submit and transfer status codes + +:c:func:`usb_register`: + +======================= =================================== +``-EINVAL`` error during registering new driver +======================= =================================== + +``usb_get_*/usb_set_*()``, +:c:func:`usb_control_msg`, +:c:func:`usb_bulk_msg()`: + +======================= ============================================== +``-ETIMEDOUT`` Timeout expired before the transfer completed. +======================= ============================================== diff --git a/Documentation/driver-api/usb/gadget.rst b/Documentation/driver-api/usb/gadget.rst new file mode 100644 index 000000000..3e8a3809c --- /dev/null +++ b/Documentation/driver-api/usb/gadget.rst @@ -0,0 +1,510 @@ +======================== +USB Gadget API for Linux +======================== + +:Author: David Brownell +:Date: 20 August 2004 + +Introduction +============ + +This document presents a Linux-USB "Gadget" kernel mode API, for use +within peripherals and other USB devices that embed Linux. It provides +an overview of the API structure, and shows how that fits into a system +development project. This is the first such API released on Linux to +address a number of important problems, including: + +- Supports USB 2.0, for high speed devices which can stream data at + several dozen megabytes per second. + +- Handles devices with dozens of endpoints just as well as ones with + just two fixed-function ones. Gadget drivers can be written so + they're easy to port to new hardware. + +- Flexible enough to expose more complex USB device capabilities such + as multiple configurations, multiple interfaces, composite devices, + and alternate interface settings. + +- USB "On-The-Go" (OTG) support, in conjunction with updates to the + Linux-USB host side. + +- Sharing data structures and API models with the Linux-USB host side + API. This helps the OTG support, and looks forward to more-symmetric + frameworks (where the same I/O model is used by both host and device + side drivers). + +- Minimalist, so it's easier to support new device controller hardware. + I/O processing doesn't imply large demands for memory or CPU + resources. + +Most Linux developers will not be able to use this API, since they have +USB ``host`` hardware in a PC, workstation, or server. Linux users with +embedded systems are more likely to have USB peripheral hardware. To +distinguish drivers running inside such hardware from the more familiar +Linux "USB device drivers", which are host side proxies for the real USB +devices, a different term is used: the drivers inside the peripherals +are "USB gadget drivers". In USB protocol interactions, the device +driver is the master (or "client driver") and the gadget driver is the +slave (or "function driver"). + +The gadget API resembles the host side Linux-USB API in that both use +queues of request objects to package I/O buffers, and those requests may +be submitted or canceled. They share common definitions for the standard +USB *Chapter 9* messages, structures, and constants. Also, both APIs +bind and unbind drivers to devices. The APIs differ in detail, since the +host side's current URB framework exposes a number of implementation +details and assumptions that are inappropriate for a gadget API. While +the model for control transfers and configuration management is +necessarily different (one side is a hardware-neutral master, the other +is a hardware-aware slave), the endpoint I/0 API used here should also +be usable for an overhead-reduced host side API. + +Structure of Gadget Drivers +=========================== + +A system running inside a USB peripheral normally has at least three +layers inside the kernel to handle USB protocol processing, and may have +additional layers in user space code. The ``gadget`` API is used by the +middle layer to interact with the lowest level (which directly handles +hardware). + +In Linux, from the bottom up, these layers are: + +*USB Controller Driver* + This is the lowest software level. It is the only layer that talks + to hardware, through registers, fifos, dma, irqs, and the like. The + ``<linux/usb/gadget.h>`` API abstracts the peripheral controller + endpoint hardware. That hardware is exposed through endpoint + objects, which accept streams of IN/OUT buffers, and through + callbacks that interact with gadget drivers. Since normal USB + devices only have one upstream port, they only have one of these + drivers. The controller driver can support any number of different + gadget drivers, but only one of them can be used at a time. + + Examples of such controller hardware include the PCI-based NetChip + 2280 USB 2.0 high speed controller, the SA-11x0 or PXA-25x UDC + (found within many PDAs), and a variety of other products. + +*Gadget Driver* + The lower boundary of this driver implements hardware-neutral USB + functions, using calls to the controller driver. Because such + hardware varies widely in capabilities and restrictions, and is used + in embedded environments where space is at a premium, the gadget + driver is often configured at compile time to work with endpoints + supported by one particular controller. Gadget drivers may be + portable to several different controllers, using conditional + compilation. (Recent kernels substantially simplify the work + involved in supporting new hardware, by *autoconfiguring* endpoints + automatically for many bulk-oriented drivers.) Gadget driver + responsibilities include: + + - handling setup requests (ep0 protocol responses) possibly + including class-specific functionality + + - returning configuration and string descriptors + + - (re)setting configurations and interface altsettings, including + enabling and configuring endpoints + + - handling life cycle events, such as managing bindings to + hardware, USB suspend/resume, remote wakeup, and disconnection + from the USB host. + + - managing IN and OUT transfers on all currently enabled endpoints + + Such drivers may be modules of proprietary code, although that + approach is discouraged in the Linux community. + +*Upper Level* + Most gadget drivers have an upper boundary that connects to some + Linux driver or framework in Linux. Through that boundary flows the + data which the gadget driver produces and/or consumes through + protocol transfers over USB. Examples include: + + - user mode code, using generic (gadgetfs) or application specific + files in ``/dev`` + + - networking subsystem (for network gadgets, like the CDC Ethernet + Model gadget driver) + + - data capture drivers, perhaps video4Linux or a scanner driver; or + test and measurement hardware. + + - input subsystem (for HID gadgets) + + - sound subsystem (for audio gadgets) + + - file system (for PTP gadgets) + + - block i/o subsystem (for usb-storage gadgets) + + - ... and more + +*Additional Layers* + Other layers may exist. These could include kernel layers, such as + network protocol stacks, as well as user mode applications building + on standard POSIX system call APIs such as ``open()``, ``close()``, + ``read()`` and ``write()``. On newer systems, POSIX Async I/O calls may + be an option. Such user mode code will not necessarily be subject to + the GNU General Public License (GPL). + +OTG-capable systems will also need to include a standard Linux-USB host +side stack, with ``usbcore``, one or more *Host Controller Drivers* +(HCDs), *USB Device Drivers* to support the OTG "Targeted Peripheral +List", and so forth. There will also be an *OTG Controller Driver*, +which is visible to gadget and device driver developers only indirectly. +That helps the host and device side USB controllers implement the two +new OTG protocols (HNP and SRP). Roles switch (host to peripheral, or +vice versa) using HNP during USB suspend processing, and SRP can be +viewed as a more battery-friendly kind of device wakeup protocol. + +Over time, reusable utilities are evolving to help make some gadget +driver tasks simpler. For example, building configuration descriptors +from vectors of descriptors for the configurations interfaces and +endpoints is now automated, and many drivers now use autoconfiguration +to choose hardware endpoints and initialize their descriptors. A +potential example of particular interest is code implementing standard +USB-IF protocols for HID, networking, storage, or audio classes. Some +developers are interested in KDB or KGDB hooks, to let target hardware +be remotely debugged. Most such USB protocol code doesn't need to be +hardware-specific, any more than network protocols like X11, HTTP, or +NFS are. Such gadget-side interface drivers should eventually be +combined, to implement composite devices. + +Kernel Mode Gadget API +====================== + +Gadget drivers declare themselves through a struct +:c:type:`usb_gadget_driver`, which is responsible for most parts of enumeration +for a struct :c:type:`usb_gadget`. The response to a set_configuration usually +involves enabling one or more of the struct :c:type:`usb_ep` objects exposed by +the gadget, and submitting one or more struct :c:type:`usb_request` buffers to +transfer data. Understand those four data types, and their operations, +and you will understand how this API works. + +.. Note:: + + Other than the "Chapter 9" data types, most of the significant data + types and functions are described here. + + However, some relevant information is likely omitted from what you + are reading. One example of such information is endpoint + autoconfiguration. You'll have to read the header file, and use + example source code (such as that for "Gadget Zero"), to fully + understand the API. + + The part of the API implementing some basic driver capabilities is + specific to the version of the Linux kernel that's in use. The 2.6 + and upper kernel versions include a *driver model* framework that has + no analogue on earlier kernels; so those parts of the gadget API are + not fully portable. (They are implemented on 2.4 kernels, but in a + different way.) The driver model state is another part of this API that is + ignored by the kerneldoc tools. + +The core API does not expose every possible hardware feature, only the +most widely available ones. There are significant hardware features, +such as device-to-device DMA (without temporary storage in a memory +buffer) that would be added using hardware-specific APIs. + +This API allows drivers to use conditional compilation to handle +endpoint capabilities of different hardware, but doesn't require that. +Hardware tends to have arbitrary restrictions, relating to transfer +types, addressing, packet sizes, buffering, and availability. As a rule, +such differences only matter for "endpoint zero" logic that handles +device configuration and management. The API supports limited run-time +detection of capabilities, through naming conventions for endpoints. +Many drivers will be able to at least partially autoconfigure +themselves. In particular, driver init sections will often have endpoint +autoconfiguration logic that scans the hardware's list of endpoints to +find ones matching the driver requirements (relying on those +conventions), to eliminate some of the most common reasons for +conditional compilation. + +Like the Linux-USB host side API, this API exposes the "chunky" nature +of USB messages: I/O requests are in terms of one or more "packets", and +packet boundaries are visible to drivers. Compared to RS-232 serial +protocols, USB resembles synchronous protocols like HDLC (N bytes per +frame, multipoint addressing, host as the primary station and devices as +secondary stations) more than asynchronous ones (tty style: 8 data bits +per frame, no parity, one stop bit). So for example the controller +drivers won't buffer two single byte writes into a single two-byte USB +IN packet, although gadget drivers may do so when they implement +protocols where packet boundaries (and "short packets") are not +significant. + +Driver Life Cycle +----------------- + +Gadget drivers make endpoint I/O requests to hardware without needing to +know many details of the hardware, but driver setup/configuration code +needs to handle some differences. Use the API like this: + +1. Register a driver for the particular device side usb controller + hardware, such as the net2280 on PCI (USB 2.0), sa11x0 or pxa25x as + found in Linux PDAs, and so on. At this point the device is logically + in the USB ch9 initial state (``attached``), drawing no power and not + usable (since it does not yet support enumeration). Any host should + not see the device, since it's not activated the data line pullup + used by the host to detect a device, even if VBUS power is available. + +2. Register a gadget driver that implements some higher level device + function. That will then bind() to a :c:type:`usb_gadget`, which activates + the data line pullup sometime after detecting VBUS. + +3. The hardware driver can now start enumerating. The steps it handles + are to accept USB ``power`` and ``set_address`` requests. Other steps are + handled by the gadget driver. If the gadget driver module is unloaded + before the host starts to enumerate, steps before step 7 are skipped. + +4. The gadget driver's ``setup()`` call returns usb descriptors, based both + on what the bus interface hardware provides and on the functionality + being implemented. That can involve alternate settings or + configurations, unless the hardware prevents such operation. For OTG + devices, each configuration descriptor includes an OTG descriptor. + +5. The gadget driver handles the last step of enumeration, when the USB + host issues a ``set_configuration`` call. It enables all endpoints used + in that configuration, with all interfaces in their default settings. + That involves using a list of the hardware's endpoints, enabling each + endpoint according to its descriptor. It may also involve using + ``usb_gadget_vbus_draw`` to let more power be drawn from VBUS, as + allowed by that configuration. For OTG devices, setting a + configuration may also involve reporting HNP capabilities through a + user interface. + +6. Do real work and perform data transfers, possibly involving changes + to interface settings or switching to new configurations, until the + device is disconnect()ed from the host. Queue any number of transfer + requests to each endpoint. It may be suspended and resumed several + times before being disconnected. On disconnect, the drivers go back + to step 3 (above). + +7. When the gadget driver module is being unloaded, the driver unbind() + callback is issued. That lets the controller driver be unloaded. + +Drivers will normally be arranged so that just loading the gadget driver +module (or statically linking it into a Linux kernel) allows the +peripheral device to be enumerated, but some drivers will defer +enumeration until some higher level component (like a user mode daemon) +enables it. Note that at this lowest level there are no policies about +how ep0 configuration logic is implemented, except that it should obey +USB specifications. Such issues are in the domain of gadget drivers, +including knowing about implementation constraints imposed by some USB +controllers or understanding that composite devices might happen to be +built by integrating reusable components. + +Note that the lifecycle above can be slightly different for OTG devices. +Other than providing an additional OTG descriptor in each configuration, +only the HNP-related differences are particularly visible to driver +code. They involve reporting requirements during the ``SET_CONFIGURATION`` +request, and the option to invoke HNP during some suspend callbacks. +Also, SRP changes the semantics of ``usb_gadget_wakeup`` slightly. + +USB 2.0 Chapter 9 Types and Constants +------------------------------------- + +Gadget drivers rely on common USB structures and constants defined in +the :ref:`linux/usb/ch9.h <usb_chapter9>` header file, which is standard in +Linux 2.6+ kernels. These are the same types and constants used by host side +drivers (and usbcore). + +Core Objects and Methods +------------------------ + +These are declared in ``<linux/usb/gadget.h>``, and are used by gadget +drivers to interact with USB peripheral controller drivers. + +.. kernel-doc:: include/linux/usb/gadget.h + :internal: + +Optional Utilities +------------------ + +The core API is sufficient for writing a USB Gadget Driver, but some +optional utilities are provided to simplify common tasks. These +utilities include endpoint autoconfiguration. + +.. kernel-doc:: drivers/usb/gadget/usbstring.c + :export: + +.. kernel-doc:: drivers/usb/gadget/config.c + :export: + +Composite Device Framework +-------------------------- + +The core API is sufficient for writing drivers for composite USB devices +(with more than one function in a given configuration), and also +multi-configuration devices (also more than one function, but not +necessarily sharing a given configuration). There is however an optional +framework which makes it easier to reuse and combine functions. + +Devices using this framework provide a struct :c:type:`usb_composite_driver`, +which in turn provides one or more struct :c:type:`usb_configuration` +instances. Each such configuration includes at least one struct +:c:type:`usb_function`, which packages a user visible role such as "network +link" or "mass storage device". Management functions may also exist, +such as "Device Firmware Upgrade". + +.. kernel-doc:: include/linux/usb/composite.h + :internal: + +.. kernel-doc:: drivers/usb/gadget/composite.c + :export: + +Composite Device Functions +-------------------------- + +At this writing, a few of the current gadget drivers have been converted +to this framework. Near-term plans include converting all of them, +except for ``gadgetfs``. + +Peripheral Controller Drivers +============================= + +The first hardware supporting this API was the NetChip 2280 controller, +which supports USB 2.0 high speed and is based on PCI. This is the +``net2280`` driver module. The driver supports Linux kernel versions 2.4 +and 2.6; contact NetChip Technologies for development boards and product +information. + +Other hardware working in the ``gadget`` framework includes: Intel's PXA +25x and IXP42x series processors (``pxa2xx_udc``), Toshiba TC86c001 +"Goku-S" (``goku_udc``), Renesas SH7705/7727 (``sh_udc``), MediaQ 11xx +(``mq11xx_udc``), Hynix HMS30C7202 (``h7202_udc``), National 9303/4 +(``n9604_udc``), Texas Instruments OMAP (``omap_udc``), Sharp LH7A40x +(``lh7a40x_udc``), and more. Most of those are full speed controllers. + +At this writing, there are people at work on drivers in this framework +for several other USB device controllers, with plans to make many of +them be widely available. + +A partial USB simulator, the ``dummy_hcd`` driver, is available. It can +act like a net2280, a pxa25x, or an sa11x0 in terms of available +endpoints and device speeds; and it simulates control, bulk, and to some +extent interrupt transfers. That lets you develop some parts of a gadget +driver on a normal PC, without any special hardware, and perhaps with +the assistance of tools such as GDB running with User Mode Linux. At +least one person has expressed interest in adapting that approach, +hooking it up to a simulator for a microcontroller. Such simulators can +help debug subsystems where the runtime hardware is unfriendly to +software development, or is not yet available. + +Support for other controllers is expected to be developed and +contributed over time, as this driver framework evolves. + +Gadget Drivers +============== + +In addition to *Gadget Zero* (used primarily for testing and development +with drivers for usb controller hardware), other gadget drivers exist. + +There's an ``ethernet`` gadget driver, which implements one of the most +useful *Communications Device Class* (CDC) models. One of the standards +for cable modem interoperability even specifies the use of this ethernet +model as one of two mandatory options. Gadgets using this code look to a +USB host as if they're an Ethernet adapter. It provides access to a +network where the gadget's CPU is one host, which could easily be +bridging, routing, or firewalling access to other networks. Since some +hardware can't fully implement the CDC Ethernet requirements, this +driver also implements a "good parts only" subset of CDC Ethernet. (That +subset doesn't advertise itself as CDC Ethernet, to avoid creating +problems.) + +Support for Microsoft's ``RNDIS`` protocol has been contributed by +Pengutronix and Auerswald GmbH. This is like CDC Ethernet, but it runs +on more slightly USB hardware (but less than the CDC subset). However, +its main claim to fame is being able to connect directly to recent +versions of Windows, using drivers that Microsoft bundles and supports, +making it much simpler to network with Windows. + +There is also support for user mode gadget drivers, using ``gadgetfs``. +This provides a *User Mode API* that presents each endpoint as a single +file descriptor. I/O is done using normal ``read()`` and ``read()`` calls. +Familiar tools like GDB and pthreads can be used to develop and debug +user mode drivers, so that once a robust controller driver is available +many applications for it won't require new kernel mode software. Linux +2.6 *Async I/O (AIO)* support is available, so that user mode software +can stream data with only slightly more overhead than a kernel driver. + +There's a USB Mass Storage class driver, which provides a different +solution for interoperability with systems such as MS-Windows and MacOS. +That *Mass Storage* driver uses a file or block device as backing store +for a drive, like the ``loop`` driver. The USB host uses the BBB, CB, or +CBI versions of the mass storage class specification, using transparent +SCSI commands to access the data from the backing store. + +There's a "serial line" driver, useful for TTY style operation over USB. +The latest version of that driver supports CDC ACM style operation, like +a USB modem, and so on most hardware it can interoperate easily with +MS-Windows. One interesting use of that driver is in boot firmware (like +a BIOS), which can sometimes use that model with very small systems +without real serial lines. + +Support for other kinds of gadget is expected to be developed and +contributed over time, as this driver framework evolves. + +USB On-The-GO (OTG) +=================== + +USB OTG support on Linux 2.6 was initially developed by Texas +Instruments for `OMAP <http://www.omap.com>`__ 16xx and 17xx series +processors. Other OTG systems should work in similar ways, but the +hardware level details could be very different. + +Systems need specialized hardware support to implement OTG, notably +including a special *Mini-AB* jack and associated transceiver to support +*Dual-Role* operation: they can act either as a host, using the standard +Linux-USB host side driver stack, or as a peripheral, using this +``gadget`` framework. To do that, the system software relies on small +additions to those programming interfaces, and on a new internal +component (here called an "OTG Controller") affecting which driver stack +connects to the OTG port. In each role, the system can re-use the +existing pool of hardware-neutral drivers, layered on top of the +controller driver interfaces (:c:type:`usb_bus` or :c:type:`usb_gadget`). +Such drivers need at most minor changes, and most of the calls added to +support OTG can also benefit non-OTG products. + +- Gadget drivers test the ``is_otg`` flag, and use it to determine + whether or not to include an OTG descriptor in each of their + configurations. + +- Gadget drivers may need changes to support the two new OTG protocols, + exposed in new gadget attributes such as ``b_hnp_enable`` flag. HNP + support should be reported through a user interface (two LEDs could + suffice), and is triggered in some cases when the host suspends the + peripheral. SRP support can be user-initiated just like remote + wakeup, probably by pressing the same button. + +- On the host side, USB device drivers need to be taught to trigger HNP + at appropriate moments, using ``usb_suspend_device()``. That also + conserves battery power, which is useful even for non-OTG + configurations. + +- Also on the host side, a driver must support the OTG "Targeted + Peripheral List". That's just a whitelist, used to reject peripherals + not supported with a given Linux OTG host. *This whitelist is + product-specific; each product must modify* ``otg_whitelist.h`` *to + match its interoperability specification.* + + Non-OTG Linux hosts, like PCs and workstations, normally have some + solution for adding drivers, so that peripherals that aren't + recognized can eventually be supported. That approach is unreasonable + for consumer products that may never have their firmware upgraded, + and where it's usually unrealistic to expect traditional + PC/workstation/server kinds of support model to work. For example, + it's often impractical to change device firmware once the product has + been distributed, so driver bugs can't normally be fixed if they're + found after shipment. + +Additional changes are needed below those hardware-neutral :c:type:`usb_bus` +and :c:type:`usb_gadget` driver interfaces; those aren't discussed here in any +detail. Those affect the hardware-specific code for each USB Host or +Peripheral controller, and how the HCD initializes (since OTG can be +active only on a single port). They also involve what may be called an +*OTG Controller Driver*, managing the OTG transceiver and the OTG state +machine logic as well as much of the root hub behavior for the OTG port. +The OTG controller driver needs to activate and deactivate USB +controllers depending on the relevant device role. Some related changes +were needed inside usbcore, so that it can identify OTG-capable devices +and respond appropriately to HNP or SRP protocols. diff --git a/Documentation/driver-api/usb/hotplug.rst b/Documentation/driver-api/usb/hotplug.rst new file mode 100644 index 000000000..79663e653 --- /dev/null +++ b/Documentation/driver-api/usb/hotplug.rst @@ -0,0 +1,154 @@ +USB hotplugging +~~~~~~~~~~~~~~~ + +Linux Hotplugging +================= + + +In hotpluggable busses like USB (and Cardbus PCI), end-users plug devices +into the bus with power on. In most cases, users expect the devices to become +immediately usable. That means the system must do many things, including: + + - Find a driver that can handle the device. That may involve + loading a kernel module; newer drivers can use module-init-tools + to publish their device (and class) support to user utilities. + + - Bind a driver to that device. Bus frameworks do that using a + device driver's probe() routine. + + - Tell other subsystems to configure the new device. Print + queues may need to be enabled, networks brought up, disk + partitions mounted, and so on. In some cases these will + be driver-specific actions. + +This involves a mix of kernel mode and user mode actions. Making devices +be immediately usable means that any user mode actions can't wait for an +administrator to do them: the kernel must trigger them, either passively +(triggering some monitoring daemon to invoke a helper program) or +actively (calling such a user mode helper program directly). + +Those triggered actions must support a system's administrative policies; +such programs are called "policy agents" here. Typically they involve +shell scripts that dispatch to more familiar administration tools. + +Because some of those actions rely on information about drivers (metadata) +that is currently available only when the drivers are dynamically linked, +you get the best hotplugging when you configure a highly modular system. + +Kernel Hotplug Helper (``/sbin/hotplug``) +========================================= + +There is a kernel parameter: ``/proc/sys/kernel/hotplug``, which normally +holds the pathname ``/sbin/hotplug``. That parameter names a program +which the kernel may invoke at various times. + +The /sbin/hotplug program can be invoked by any subsystem as part of its +reaction to a configuration change, from a thread in that subsystem. +Only one parameter is required: the name of a subsystem being notified of +some kernel event. That name is used as the first key for further event +dispatch; any other argument and environment parameters are specified by +the subsystem making that invocation. + +Hotplug software and other resources is available at: + + http://linux-hotplug.sourceforge.net + +Mailing list information is also available at that site. + + +USB Policy Agent +================ + +The USB subsystem currently invokes ``/sbin/hotplug`` when USB devices +are added or removed from system. The invocation is done by the kernel +hub workqueue [hub_wq], or else as part of root hub initialization +(done by init, modprobe, kapmd, etc). Its single command line parameter +is the string "usb", and it passes these environment variables: + +========== ============================================ +ACTION ``add``, ``remove`` +PRODUCT USB vendor, product, and version codes (hex) +TYPE device class codes (decimal) +INTERFACE interface 0 class codes (decimal) +========== ============================================ + +If "usbdevfs" is configured, DEVICE and DEVFS are also passed. DEVICE is +the pathname of the device, and is useful for devices with multiple and/or +alternate interfaces that complicate driver selection. By design, USB +hotplugging is independent of ``usbdevfs``: you can do most essential parts +of USB device setup without using that filesystem, and without running a +user mode daemon to detect changes in system configuration. + +Currently available policy agent implementations can load drivers for +modules, and can invoke driver-specific setup scripts. The newest ones +leverage USB module-init-tools support. Later agents might unload drivers. + + +USB Modutils Support +==================== + +Current versions of module-init-tools will create a ``modules.usbmap`` file +which contains the entries from each driver's ``MODULE_DEVICE_TABLE``. Such +files can be used by various user mode policy agents to make sure all the +right driver modules get loaded, either at boot time or later. + +See ``linux/usb.h`` for full information about such table entries; or look +at existing drivers. Each table entry describes one or more criteria to +be used when matching a driver to a device or class of devices. The +specific criteria are identified by bits set in "match_flags", paired +with field values. You can construct the criteria directly, or with +macros such as these, and use driver_info to store more information:: + + USB_DEVICE (vendorId, productId) + ... matching devices with specified vendor and product ids + USB_DEVICE_VER (vendorId, productId, lo, hi) + ... like USB_DEVICE with lo <= productversion <= hi + USB_INTERFACE_INFO (class, subclass, protocol) + ... matching specified interface class info + USB_DEVICE_INFO (class, subclass, protocol) + ... matching specified device class info + +A short example, for a driver that supports several specific USB devices +and their quirks, might have a MODULE_DEVICE_TABLE like this:: + + static const struct usb_device_id mydriver_id_table[] = { + { USB_DEVICE (0x9999, 0xaaaa), driver_info: QUIRK_X }, + { USB_DEVICE (0xbbbb, 0x8888), driver_info: QUIRK_Y|QUIRK_Z }, + ... + { } /* end with an all-zeroes entry */ + }; + MODULE_DEVICE_TABLE(usb, mydriver_id_table); + +Most USB device drivers should pass these tables to the USB subsystem as +well as to the module management subsystem. Not all, though: some driver +frameworks connect using interfaces layered over USB, and so they won't +need such a struct :c:type:`usb_driver`. + +Drivers that connect directly to the USB subsystem should be declared +something like this:: + + static struct usb_driver mydriver = { + .name = "mydriver", + .id_table = mydriver_id_table, + .probe = my_probe, + .disconnect = my_disconnect, + + /* + if using the usb chardev framework: + .minor = MY_USB_MINOR_START, + .fops = my_file_ops, + if exposing any operations through usbdevfs: + .ioctl = my_ioctl, + */ + }; + +When the USB subsystem knows about a driver's device ID table, it's used when +choosing drivers to probe(). The thread doing new device processing checks +drivers' device ID entries from the ``MODULE_DEVICE_TABLE`` against interface +and device descriptors for the device. It will only call ``probe()`` if there +is a match, and the third argument to ``probe()`` will be the entry that +matched. + +If you don't provide an ``id_table`` for your driver, then your driver may get +probed for each new device; the third parameter to ``probe()`` will be +``NULL``. diff --git a/Documentation/driver-api/usb/index.rst b/Documentation/driver-api/usb/index.rst new file mode 100644 index 000000000..8fe995a1e --- /dev/null +++ b/Documentation/driver-api/usb/index.rst @@ -0,0 +1,29 @@ +============= +Linux USB API +============= + +.. toctree:: + + usb + gadget + anchors + bulk-streams + callbacks + dma + URB + power-management + hotplug + persist + error-codes + writing_usb_driver + dwc3 + writing_musb_glue_layer + typec + usb3-debug-port + +.. only:: subproject and html + + Indices + ======= + + * :ref:`genindex` diff --git a/Documentation/driver-api/usb/persist.rst b/Documentation/driver-api/usb/persist.rst new file mode 100644 index 000000000..08cafc629 --- /dev/null +++ b/Documentation/driver-api/usb/persist.rst @@ -0,0 +1,171 @@ +.. _usb-persist: + +USB device persistence during system suspend +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:Author: Alan Stern <stern@rowland.harvard.edu> +:Date: September 2, 2006 (Updated February 25, 2008) + + +What is the problem? +==================== + +According to the USB specification, when a USB bus is suspended the +bus must continue to supply suspend current (around 1-5 mA). This +is so that devices can maintain their internal state and hubs can +detect connect-change events (devices being plugged in or unplugged). +The technical term is "power session". + +If a USB device's power session is interrupted then the system is +required to behave as though the device has been unplugged. It's a +conservative approach; in the absence of suspend current the computer +has no way to know what has actually happened. Perhaps the same +device is still attached or perhaps it was removed and a different +device plugged into the port. The system must assume the worst. + +By default, Linux behaves according to the spec. If a USB host +controller loses power during a system suspend, then when the system +wakes up all the devices attached to that controller are treated as +though they had disconnected. This is always safe and it is the +"officially correct" thing to do. + +For many sorts of devices this behavior doesn't matter in the least. +If the kernel wants to believe that your USB keyboard was unplugged +while the system was asleep and a new keyboard was plugged in when the +system woke up, who cares? It'll still work the same when you type on +it. + +Unfortunately problems _can_ arise, particularly with mass-storage +devices. The effect is exactly the same as if the device really had +been unplugged while the system was suspended. If you had a mounted +filesystem on the device, you're out of luck -- everything in that +filesystem is now inaccessible. This is especially annoying if your +root filesystem was located on the device, since your system will +instantly crash. + +Loss of power isn't the only mechanism to worry about. Anything that +interrupts a power session will have the same effect. For example, +even though suspend current may have been maintained while the system +was asleep, on many systems during the initial stages of wakeup the +firmware (i.e., the BIOS) resets the motherboard's USB host +controllers. Result: all the power sessions are destroyed and again +it's as though you had unplugged all the USB devices. Yes, it's +entirely the BIOS's fault, but that doesn't do _you_ any good unless +you can convince the BIOS supplier to fix the problem (lots of luck!). + +On many systems the USB host controllers will get reset after a +suspend-to-RAM. On almost all systems, no suspend current is +available during hibernation (also known as swsusp or suspend-to-disk). +You can check the kernel log after resuming to see if either of these +has happened; look for lines saying "root hub lost power or was reset". + +In practice, people are forced to unmount any filesystems on a USB +device before suspending. If the root filesystem is on a USB device, +the system can't be suspended at all. (All right, it _can_ be +suspended -- but it will crash as soon as it wakes up, which isn't +much better.) + + +What is the solution? +===================== + +The kernel includes a feature called USB-persist. It tries to work +around these issues by allowing the core USB device data structures to +persist across a power-session disruption. + +It works like this. If the kernel sees that a USB host controller is +not in the expected state during resume (i.e., if the controller was +reset or otherwise had lost power) then it applies a persistence check +to each of the USB devices below that controller for which the +"persist" attribute is set. It doesn't try to resume the device; that +can't work once the power session is gone. Instead it issues a USB +port reset and then re-enumerates the device. (This is exactly the +same thing that happens whenever a USB device is reset.) If the +re-enumeration shows that the device now attached to that port has the +same descriptors as before, including the Vendor and Product IDs, then +the kernel continues to use the same device structure. In effect, the +kernel treats the device as though it had merely been reset instead of +unplugged. + +The same thing happens if the host controller is in the expected state +but a USB device was unplugged and then replugged, or if a USB device +fails to carry out a normal resume. + +If no device is now attached to the port, or if the descriptors are +different from what the kernel remembers, then the treatment is what +you would expect. The kernel destroys the old device structure and +behaves as though the old device had been unplugged and a new device +plugged in. + +The end result is that the USB device remains available and usable. +Filesystem mounts and memory mappings are unaffected, and the world is +now a good and happy place. + +Note that the "USB-persist" feature will be applied only to those +devices for which it is enabled. You can enable the feature by doing +(as root):: + + echo 1 >/sys/bus/usb/devices/.../power/persist + +where the "..." should be filled in the with the device's ID. Disable +the feature by writing 0 instead of 1. For hubs the feature is +automatically and permanently enabled and the power/persist file +doesn't even exist, so you only have to worry about setting it for +devices where it really matters. + + +Is this the best solution? +========================== + +Perhaps not. Arguably, keeping track of mounted filesystems and +memory mappings across device disconnects should be handled by a +centralized Logical Volume Manager. Such a solution would allow you +to plug in a USB flash device, create a persistent volume associated +with it, unplug the flash device, plug it back in later, and still +have the same persistent volume associated with the device. As such +it would be more far-reaching than USB-persist. + +On the other hand, writing a persistent volume manager would be a big +job and using it would require significant input from the user. This +solution is much quicker and easier -- and it exists now, a giant +point in its favor! + +Furthermore, the USB-persist feature applies to _all_ USB devices, not +just mass-storage devices. It might turn out to be equally useful for +other device types, such as network interfaces. + + +WARNING: USB-persist can be dangerous!! +======================================= + +When recovering an interrupted power session the kernel does its best +to make sure the USB device hasn't been changed; that is, the same +device is still plugged into the port as before. But the checks +aren't guaranteed to be 100% accurate. + +If you replace one USB device with another of the same type (same +manufacturer, same IDs, and so on) there's an excellent chance the +kernel won't detect the change. The serial number string and other +descriptors are compared with the kernel's stored values, but this +might not help since manufacturers frequently omit serial numbers +entirely in their devices. + +Furthermore it's quite possible to leave a USB device exactly the same +while changing its media. If you replace the flash memory card in a +USB card reader while the system is asleep, the kernel will have no +way to know you did it. The kernel will assume that nothing has +happened and will continue to use the partition tables, inodes, and +memory mappings for the old card. + +If the kernel gets fooled in this way, it's almost certain to cause +data corruption and to crash your system. You'll have no one to blame +but yourself. + +For those devices with avoid_reset_quirk attribute being set, persist +maybe fail because they may morph after reset. + +YOU HAVE BEEN WARNED! USE AT YOUR OWN RISK! + +That having been said, most of the time there shouldn't be any trouble +at all. The USB-persist feature can be extremely useful. Make the +most of it. diff --git a/Documentation/driver-api/usb/power-management.rst b/Documentation/driver-api/usb/power-management.rst new file mode 100644 index 000000000..4a74cf6f2 --- /dev/null +++ b/Documentation/driver-api/usb/power-management.rst @@ -0,0 +1,798 @@ +.. _usb-power-management: + +Power Management for USB +~~~~~~~~~~~~~~~~~~~~~~~~ + +:Author: Alan Stern <stern@rowland.harvard.edu> +:Date: Last-updated: February 2014 + +.. + Contents: + --------- + * What is Power Management? + * What is Remote Wakeup? + * When is a USB device idle? + * Forms of dynamic PM + * The user interface for dynamic PM + * Changing the default idle-delay time + * Warnings + * The driver interface for Power Management + * The driver interface for autosuspend and autoresume + * Other parts of the driver interface + * Mutual exclusion + * Interaction between dynamic PM and system PM + * xHCI hardware link PM + * USB Port Power Control + * User Interface for Port Power Control + * Suggested Userspace Port Power Policy + + +What is Power Management? +------------------------- + +Power Management (PM) is the practice of saving energy by suspending +parts of a computer system when they aren't being used. While a +component is ``suspended`` it is in a nonfunctional low-power state; it +might even be turned off completely. A suspended component can be +``resumed`` (returned to a functional full-power state) when the kernel +needs to use it. (There also are forms of PM in which components are +placed in a less functional but still usable state instead of being +suspended; an example would be reducing the CPU's clock rate. This +document will not discuss those other forms.) + +When the parts being suspended include the CPU and most of the rest of +the system, we speak of it as a "system suspend". When a particular +device is turned off while the system as a whole remains running, we +call it a "dynamic suspend" (also known as a "runtime suspend" or +"selective suspend"). This document concentrates mostly on how +dynamic PM is implemented in the USB subsystem, although system PM is +covered to some extent (see ``Documentation/power/*.txt`` for more +information about system PM). + +System PM support is present only if the kernel was built with +``CONFIG_SUSPEND`` or ``CONFIG_HIBERNATION`` enabled. Dynamic PM support + +for USB is present whenever +the kernel was built with ``CONFIG_PM`` enabled. + +[Historically, dynamic PM support for USB was present only if the +kernel had been built with ``CONFIG_USB_SUSPEND`` enabled (which depended on +``CONFIG_PM_RUNTIME``). Starting with the 3.10 kernel release, dynamic PM +support for USB was present whenever the kernel was built with +``CONFIG_PM_RUNTIME`` enabled. The ``CONFIG_USB_SUSPEND`` option had been +eliminated.] + + +What is Remote Wakeup? +---------------------- + +When a device has been suspended, it generally doesn't resume until +the computer tells it to. Likewise, if the entire computer has been +suspended, it generally doesn't resume until the user tells it to, say +by pressing a power button or opening the cover. + +However some devices have the capability of resuming by themselves, or +asking the kernel to resume them, or even telling the entire computer +to resume. This capability goes by several names such as "Wake On +LAN"; we will refer to it generically as "remote wakeup". When a +device is enabled for remote wakeup and it is suspended, it may resume +itself (or send a request to be resumed) in response to some external +event. Examples include a suspended keyboard resuming when a key is +pressed, or a suspended USB hub resuming when a device is plugged in. + + +When is a USB device idle? +-------------------------- + +A device is idle whenever the kernel thinks it's not busy doing +anything important and thus is a candidate for being suspended. The +exact definition depends on the device's driver; drivers are allowed +to declare that a device isn't idle even when there's no actual +communication taking place. (For example, a hub isn't considered idle +unless all the devices plugged into that hub are already suspended.) +In addition, a device isn't considered idle so long as a program keeps +its usbfs file open, whether or not any I/O is going on. + +If a USB device has no driver, its usbfs file isn't open, and it isn't +being accessed through sysfs, then it definitely is idle. + + +Forms of dynamic PM +------------------- + +Dynamic suspends occur when the kernel decides to suspend an idle +device. This is called ``autosuspend`` for short. In general, a device +won't be autosuspended unless it has been idle for some minimum period +of time, the so-called idle-delay time. + +Of course, nothing the kernel does on its own initiative should +prevent the computer or its devices from working properly. If a +device has been autosuspended and a program tries to use it, the +kernel will automatically resume the device (autoresume). For the +same reason, an autosuspended device will usually have remote wakeup +enabled, if the device supports remote wakeup. + +It is worth mentioning that many USB drivers don't support +autosuspend. In fact, at the time of this writing (Linux 2.6.23) the +only drivers which do support it are the hub driver, kaweth, asix, +usblp, usblcd, and usb-skeleton (which doesn't count). If a +non-supporting driver is bound to a device, the device won't be +autosuspended. In effect, the kernel pretends the device is never +idle. + +We can categorize power management events in two broad classes: +external and internal. External events are those triggered by some +agent outside the USB stack: system suspend/resume (triggered by +userspace), manual dynamic resume (also triggered by userspace), and +remote wakeup (triggered by the device). Internal events are those +triggered within the USB stack: autosuspend and autoresume. Note that +all dynamic suspend events are internal; external agents are not +allowed to issue dynamic suspends. + + +The user interface for dynamic PM +--------------------------------- + +The user interface for controlling dynamic PM is located in the ``power/`` +subdirectory of each USB device's sysfs directory, that is, in +``/sys/bus/usb/devices/.../power/`` where "..." is the device's ID. The +relevant attribute files are: wakeup, control, and +``autosuspend_delay_ms``. (There may also be a file named ``level``; this +file was deprecated as of the 2.6.35 kernel and replaced by the +``control`` file. In 2.6.38 the ``autosuspend`` file will be deprecated +and replaced by the ``autosuspend_delay_ms`` file. The only difference +is that the newer file expresses the delay in milliseconds whereas the +older file uses seconds. Confusingly, both files are present in 2.6.37 +but only ``autosuspend`` works.) + + ``power/wakeup`` + + This file is empty if the device does not support + remote wakeup. Otherwise the file contains either the + word ``enabled`` or the word ``disabled``, and you can + write those words to the file. The setting determines + whether or not remote wakeup will be enabled when the + device is next suspended. (If the setting is changed + while the device is suspended, the change won't take + effect until the following suspend.) + + ``power/control`` + + This file contains one of two words: ``on`` or ``auto``. + You can write those words to the file to change the + device's setting. + + - ``on`` means that the device should be resumed and + autosuspend is not allowed. (Of course, system + suspends are still allowed.) + + - ``auto`` is the normal state in which the kernel is + allowed to autosuspend and autoresume the device. + + (In kernels up to 2.6.32, you could also specify + ``suspend``, meaning that the device should remain + suspended and autoresume was not allowed. This + setting is no longer supported.) + + ``power/autosuspend_delay_ms`` + + This file contains an integer value, which is the + number of milliseconds the device should remain idle + before the kernel will autosuspend it (the idle-delay + time). The default is 2000. 0 means to autosuspend + as soon as the device becomes idle, and negative + values mean never to autosuspend. You can write a + number to the file to change the autosuspend + idle-delay time. + +Writing ``-1`` to ``power/autosuspend_delay_ms`` and writing ``on`` to +``power/control`` do essentially the same thing -- they both prevent the +device from being autosuspended. Yes, this is a redundancy in the +API. + +(In 2.6.21 writing ``0`` to ``power/autosuspend`` would prevent the device +from being autosuspended; the behavior was changed in 2.6.22. The +``power/autosuspend`` attribute did not exist prior to 2.6.21, and the +``power/level`` attribute did not exist prior to 2.6.22. ``power/control`` +was added in 2.6.34, and ``power/autosuspend_delay_ms`` was added in +2.6.37 but did not become functional until 2.6.38.) + + +Changing the default idle-delay time +------------------------------------ + +The default autosuspend idle-delay time (in seconds) is controlled by +a module parameter in usbcore. You can specify the value when usbcore +is loaded. For example, to set it to 5 seconds instead of 2 you would +do:: + + modprobe usbcore autosuspend=5 + +Equivalently, you could add to a configuration file in /etc/modprobe.d +a line saying:: + + options usbcore autosuspend=5 + +Some distributions load the usbcore module very early during the boot +process, by means of a program or script running from an initramfs +image. To alter the parameter value you would have to rebuild that +image. + +If usbcore is compiled into the kernel rather than built as a loadable +module, you can add:: + + usbcore.autosuspend=5 + +to the kernel's boot command line. + +Finally, the parameter value can be changed while the system is +running. If you do:: + + echo 5 >/sys/module/usbcore/parameters/autosuspend + +then each new USB device will have its autosuspend idle-delay +initialized to 5. (The idle-delay values for already existing devices +will not be affected.) + +Setting the initial default idle-delay to -1 will prevent any +autosuspend of any USB device. This has the benefit of allowing you +then to enable autosuspend for selected devices. + + +Warnings +-------- + +The USB specification states that all USB devices must support power +management. Nevertheless, the sad fact is that many devices do not +support it very well. You can suspend them all right, but when you +try to resume them they disconnect themselves from the USB bus or +they stop working entirely. This seems to be especially prevalent +among printers and scanners, but plenty of other types of device have +the same deficiency. + +For this reason, by default the kernel disables autosuspend (the +``power/control`` attribute is initialized to ``on``) for all devices other +than hubs. Hubs, at least, appear to be reasonably well-behaved in +this regard. + +(In 2.6.21 and 2.6.22 this wasn't the case. Autosuspend was enabled +by default for almost all USB devices. A number of people experienced +problems as a result.) + +This means that non-hub devices won't be autosuspended unless the user +or a program explicitly enables it. As of this writing there aren't +any widespread programs which will do this; we hope that in the near +future device managers such as HAL will take on this added +responsibility. In the meantime you can always carry out the +necessary operations by hand or add them to a udev script. You can +also change the idle-delay time; 2 seconds is not the best choice for +every device. + +If a driver knows that its device has proper suspend/resume support, +it can enable autosuspend all by itself. For example, the video +driver for a laptop's webcam might do this (in recent kernels they +do), since these devices are rarely used and so should normally be +autosuspended. + +Sometimes it turns out that even when a device does work okay with +autosuspend there are still problems. For example, the usbhid driver, +which manages keyboards and mice, has autosuspend support. Tests with +a number of keyboards show that typing on a suspended keyboard, while +causing the keyboard to do a remote wakeup all right, will nonetheless +frequently result in lost keystrokes. Tests with mice show that some +of them will issue a remote-wakeup request in response to button +presses but not to motion, and some in response to neither. + +The kernel will not prevent you from enabling autosuspend on devices +that can't handle it. It is even possible in theory to damage a +device by suspending it at the wrong time. (Highly unlikely, but +possible.) Take care. + + +The driver interface for Power Management +----------------------------------------- + +The requirements for a USB driver to support external power management +are pretty modest; the driver need only define:: + + .suspend + .resume + .reset_resume + +methods in its :c:type:`usb_driver` structure, and the ``reset_resume`` method +is optional. The methods' jobs are quite simple: + + - The ``suspend`` method is called to warn the driver that the + device is going to be suspended. If the driver returns a + negative error code, the suspend will be aborted. Normally + the driver will return 0, in which case it must cancel all + outstanding URBs (:c:func:`usb_kill_urb`) and not submit any more. + + - The ``resume`` method is called to tell the driver that the + device has been resumed and the driver can return to normal + operation. URBs may once more be submitted. + + - The ``reset_resume`` method is called to tell the driver that + the device has been resumed and it also has been reset. + The driver should redo any necessary device initialization, + since the device has probably lost most or all of its state + (although the interfaces will be in the same altsettings as + before the suspend). + +If the device is disconnected or powered down while it is suspended, +the ``disconnect`` method will be called instead of the ``resume`` or +``reset_resume`` method. This is also quite likely to happen when +waking up from hibernation, as many systems do not maintain suspend +current to the USB host controllers during hibernation. (It's +possible to work around the hibernation-forces-disconnect problem by +using the USB Persist facility.) + +The ``reset_resume`` method is used by the USB Persist facility (see +:ref:`usb-persist`) and it can also be used under certain +circumstances when ``CONFIG_USB_PERSIST`` is not enabled. Currently, if a +device is reset during a resume and the driver does not have a +``reset_resume`` method, the driver won't receive any notification about +the resume. Later kernels will call the driver's ``disconnect`` method; +2.6.23 doesn't do this. + +USB drivers are bound to interfaces, so their ``suspend`` and ``resume`` +methods get called when the interfaces are suspended or resumed. In +principle one might want to suspend some interfaces on a device (i.e., +force the drivers for those interface to stop all activity) without +suspending the other interfaces. The USB core doesn't allow this; all +interfaces are suspended when the device itself is suspended and all +interfaces are resumed when the device is resumed. It isn't possible +to suspend or resume some but not all of a device's interfaces. The +closest you can come is to unbind the interfaces' drivers. + + +The driver interface for autosuspend and autoresume +--------------------------------------------------- + +To support autosuspend and autoresume, a driver should implement all +three of the methods listed above. In addition, a driver indicates +that it supports autosuspend by setting the ``.supports_autosuspend`` flag +in its usb_driver structure. It is then responsible for informing the +USB core whenever one of its interfaces becomes busy or idle. The +driver does so by calling these six functions:: + + int usb_autopm_get_interface(struct usb_interface *intf); + void usb_autopm_put_interface(struct usb_interface *intf); + int usb_autopm_get_interface_async(struct usb_interface *intf); + void usb_autopm_put_interface_async(struct usb_interface *intf); + void usb_autopm_get_interface_no_resume(struct usb_interface *intf); + void usb_autopm_put_interface_no_suspend(struct usb_interface *intf); + +The functions work by maintaining a usage counter in the +usb_interface's embedded device structure. When the counter is > 0 +then the interface is deemed to be busy, and the kernel will not +autosuspend the interface's device. When the usage counter is = 0 +then the interface is considered to be idle, and the kernel may +autosuspend the device. + +Drivers must be careful to balance their overall changes to the usage +counter. Unbalanced "get"s will remain in effect when a driver is +unbound from its interface, preventing the device from going into +runtime suspend should the interface be bound to a driver again. On +the other hand, drivers are allowed to achieve this balance by calling +the ``usb_autopm_*`` functions even after their ``disconnect`` routine +has returned -- say from within a work-queue routine -- provided they +retain an active reference to the interface (via ``usb_get_intf`` and +``usb_put_intf``). + +Drivers using the async routines are responsible for their own +synchronization and mutual exclusion. + + :c:func:`usb_autopm_get_interface` increments the usage counter and + does an autoresume if the device is suspended. If the + autoresume fails, the counter is decremented back. + + :c:func:`usb_autopm_put_interface` decrements the usage counter and + attempts an autosuspend if the new value is = 0. + + :c:func:`usb_autopm_get_interface_async` and + :c:func:`usb_autopm_put_interface_async` do almost the same things as + their non-async counterparts. The big difference is that they + use a workqueue to do the resume or suspend part of their + jobs. As a result they can be called in an atomic context, + such as an URB's completion handler, but when they return the + device will generally not yet be in the desired state. + + :c:func:`usb_autopm_get_interface_no_resume` and + :c:func:`usb_autopm_put_interface_no_suspend` merely increment or + decrement the usage counter; they do not attempt to carry out + an autoresume or an autosuspend. Hence they can be called in + an atomic context. + +The simplest usage pattern is that a driver calls +:c:func:`usb_autopm_get_interface` in its open routine and +:c:func:`usb_autopm_put_interface` in its close or release routine. But other +patterns are possible. + +The autosuspend attempts mentioned above will often fail for one +reason or another. For example, the ``power/control`` attribute might be +set to ``on``, or another interface in the same device might not be +idle. This is perfectly normal. If the reason for failure was that +the device hasn't been idle for long enough, a timer is scheduled to +carry out the operation automatically when the autosuspend idle-delay +has expired. + +Autoresume attempts also can fail, although failure would mean that +the device is no longer present or operating properly. Unlike +autosuspend, there's no idle-delay for an autoresume. + + +Other parts of the driver interface +----------------------------------- + +Drivers can enable autosuspend for their devices by calling:: + + usb_enable_autosuspend(struct usb_device *udev); + +in their :c:func:`probe` routine, if they know that the device is capable of +suspending and resuming correctly. This is exactly equivalent to +writing ``auto`` to the device's ``power/control`` attribute. Likewise, +drivers can disable autosuspend by calling:: + + usb_disable_autosuspend(struct usb_device *udev); + +This is exactly the same as writing ``on`` to the ``power/control`` attribute. + +Sometimes a driver needs to make sure that remote wakeup is enabled +during autosuspend. For example, there's not much point +autosuspending a keyboard if the user can't cause the keyboard to do a +remote wakeup by typing on it. If the driver sets +``intf->needs_remote_wakeup`` to 1, the kernel won't autosuspend the +device if remote wakeup isn't available. (If the device is already +autosuspended, though, setting this flag won't cause the kernel to +autoresume it. Normally a driver would set this flag in its ``probe`` +method, at which time the device is guaranteed not to be +autosuspended.) + +If a driver does its I/O asynchronously in interrupt context, it +should call :c:func:`usb_autopm_get_interface_async` before starting output and +:c:func:`usb_autopm_put_interface_async` when the output queue drains. When +it receives an input event, it should call:: + + usb_mark_last_busy(struct usb_device *udev); + +in the event handler. This tells the PM core that the device was just +busy and therefore the next autosuspend idle-delay expiration should +be pushed back. Many of the usb_autopm_* routines also make this call, +so drivers need to worry only when interrupt-driven input arrives. + +Asynchronous operation is always subject to races. For example, a +driver may call the :c:func:`usb_autopm_get_interface_async` routine at a time +when the core has just finished deciding the device has been idle for +long enough but not yet gotten around to calling the driver's ``suspend`` +method. The ``suspend`` method must be responsible for synchronizing with +the I/O request routine and the URB completion handler; it should +cause autosuspends to fail with -EBUSY if the driver needs to use the +device. + +External suspend calls should never be allowed to fail in this way, +only autosuspend calls. The driver can tell them apart by applying +the :c:func:`PMSG_IS_AUTO` macro to the message argument to the ``suspend`` +method; it will return True for internal PM events (autosuspend) and +False for external PM events. + + +Mutual exclusion +---------------- + +For external events -- but not necessarily for autosuspend or +autoresume -- the device semaphore (udev->dev.sem) will be held when a +``suspend`` or ``resume`` method is called. This implies that external +suspend/resume events are mutually exclusive with calls to ``probe``, +``disconnect``, ``pre_reset``, and ``post_reset``; the USB core guarantees that +this is true of autosuspend/autoresume events as well. + +If a driver wants to block all suspend/resume calls during some +critical section, the best way is to lock the device and call +:c:func:`usb_autopm_get_interface` (and do the reverse at the end of the +critical section). Holding the device semaphore will block all +external PM calls, and the :c:func:`usb_autopm_get_interface` will prevent any +internal PM calls, even if it fails. (Exercise: Why?) + + +Interaction between dynamic PM and system PM +-------------------------------------------- + +Dynamic power management and system power management can interact in +a couple of ways. + +Firstly, a device may already be autosuspended when a system suspend +occurs. Since system suspends are supposed to be as transparent as +possible, the device should remain suspended following the system +resume. But this theory may not work out well in practice; over time +the kernel's behavior in this regard has changed. As of 2.6.37 the +policy is to resume all devices during a system resume and let them +handle their own runtime suspends afterward. + +Secondly, a dynamic power-management event may occur as a system +suspend is underway. The window for this is short, since system +suspends don't take long (a few seconds usually), but it can happen. +For example, a suspended device may send a remote-wakeup signal while +the system is suspending. The remote wakeup may succeed, which would +cause the system suspend to abort. If the remote wakeup doesn't +succeed, it may still remain active and thus cause the system to +resume as soon as the system suspend is complete. Or the remote +wakeup may fail and get lost. Which outcome occurs depends on timing +and on the hardware and firmware design. + + +xHCI hardware link PM +--------------------- + +xHCI host controller provides hardware link power management to usb2.0 +(xHCI 1.0 feature) and usb3.0 devices which support link PM. By +enabling hardware LPM, the host can automatically put the device into +lower power state(L1 for usb2.0 devices, or U1/U2 for usb3.0 devices), +which state device can enter and resume very quickly. + +The user interface for controlling hardware LPM is located in the +``power/`` subdirectory of each USB device's sysfs directory, that is, in +``/sys/bus/usb/devices/.../power/`` where "..." is the device's ID. The +relevant attribute files are ``usb2_hardware_lpm`` and ``usb3_hardware_lpm``. + + ``power/usb2_hardware_lpm`` + + When a USB2 device which support LPM is plugged to a + xHCI host root hub which support software LPM, the + host will run a software LPM test for it; if the device + enters L1 state and resume successfully and the host + supports USB2 hardware LPM, this file will show up and + driver will enable hardware LPM for the device. You + can write y/Y/1 or n/N/0 to the file to enable/disable + USB2 hardware LPM manually. This is for test purpose mainly. + + ``power/usb3_hardware_lpm_u1`` + ``power/usb3_hardware_lpm_u2`` + + When a USB 3.0 lpm-capable device is plugged in to a + xHCI host which supports link PM, it will check if U1 + and U2 exit latencies have been set in the BOS + descriptor; if the check is passed and the host + supports USB3 hardware LPM, USB3 hardware LPM will be + enabled for the device and these files will be created. + The files hold a string value (enable or disable) + indicating whether or not USB3 hardware LPM U1 or U2 + is enabled for the device. + +USB Port Power Control +---------------------- + +In addition to suspending endpoint devices and enabling hardware +controlled link power management, the USB subsystem also has the +capability to disable power to ports under some conditions. Power is +controlled through ``Set/ClearPortFeature(PORT_POWER)`` requests to a hub. +In the case of a root or platform-internal hub the host controller +driver translates ``PORT_POWER`` requests into platform firmware (ACPI) +method calls to set the port power state. For more background see the +Linux Plumbers Conference 2012 slides [#f1]_ and video [#f2]_: + +Upon receiving a ``ClearPortFeature(PORT_POWER)`` request a USB port is +logically off, and may trigger the actual loss of VBUS to the port [#f3]_. +VBUS may be maintained in the case where a hub gangs multiple ports into +a shared power well causing power to remain until all ports in the gang +are turned off. VBUS may also be maintained by hub ports configured for +a charging application. In any event a logically off port will lose +connection with its device, not respond to hotplug events, and not +respond to remote wakeup events. + +.. warning:: + + turning off a port may result in the inability to hot add a device. + Please see "User Interface for Port Power Control" for details. + +As far as the effect on the device itself it is similar to what a device +goes through during system suspend, i.e. the power session is lost. Any +USB device or driver that misbehaves with system suspend will be +similarly affected by a port power cycle event. For this reason the +implementation shares the same device recovery path (and honors the same +quirks) as the system resume path for the hub. + +.. [#f1] + + http://dl.dropbox.com/u/96820575/sarah-sharp-lpt-port-power-off2-mini.pdf + +.. [#f2] + + http://linuxplumbers.ubicast.tv/videos/usb-port-power-off-kerneluserspace-api/ + +.. [#f3] + + USB 3.1 Section 10.12 + + wakeup note: if a device is configured to send wakeup events the port + power control implementation will block poweroff attempts on that + port. + + +User Interface for Port Power Control +------------------------------------- + +The port power control mechanism uses the PM runtime system. Poweroff is +requested by clearing the ``power/pm_qos_no_power_off`` flag of the port device +(defaults to 1). If the port is disconnected it will immediately receive a +``ClearPortFeature(PORT_POWER)`` request. Otherwise, it will honor the pm +runtime rules and require the attached child device and all descendants to be +suspended. This mechanism is dependent on the hub advertising port power +switching in its hub descriptor (wHubCharacteristics logical power switching +mode field). + +Note, some interface devices/drivers do not support autosuspend. Userspace may +need to unbind the interface drivers before the :c:type:`usb_device` will +suspend. An unbound interface device is suspended by default. When unbinding, +be careful to unbind interface drivers, not the driver of the parent usb +device. Also, leave hub interface drivers bound. If the driver for the usb +device (not interface) is unbound the kernel is no longer able to resume the +device. If a hub interface driver is unbound, control of its child ports is +lost and all attached child-devices will disconnect. A good rule of thumb is +that if the 'driver/module' link for a device points to +``/sys/module/usbcore`` then unbinding it will interfere with port power +control. + +Example of the relevant files for port power control. Note, in this example +these files are relative to a usb hub device (prefix):: + + prefix=/sys/devices/pci0000:00/0000:00:14.0/usb3/3-1 + + attached child device + + hub port device + | + hub interface device + | | + v v v + $prefix/3-1:1.0/3-1-port1/device + + $prefix/3-1:1.0/3-1-port1/power/pm_qos_no_power_off + $prefix/3-1:1.0/3-1-port1/device/power/control + $prefix/3-1:1.0/3-1-port1/device/3-1.1:<intf0>/driver/unbind + $prefix/3-1:1.0/3-1-port1/device/3-1.1:<intf1>/driver/unbind + ... + $prefix/3-1:1.0/3-1-port1/device/3-1.1:<intfN>/driver/unbind + +In addition to these files some ports may have a 'peer' link to a port on +another hub. The expectation is that all superspeed ports have a +hi-speed peer:: + + $prefix/3-1:1.0/3-1-port1/peer -> ../../../../usb2/2-1/2-1:1.0/2-1-port1 + ../../../../usb2/2-1/2-1:1.0/2-1-port1/peer -> ../../../../usb3/3-1/3-1:1.0/3-1-port1 + +Distinct from 'companion ports', or 'ehci/xhci shared switchover ports' +peer ports are simply the hi-speed and superspeed interface pins that +are combined into a single usb3 connector. Peer ports share the same +ancestor XHCI device. + +While a superspeed port is powered off a device may downgrade its +connection and attempt to connect to the hi-speed pins. The +implementation takes steps to prevent this: + +1. Port suspend is sequenced to guarantee that hi-speed ports are powered-off + before their superspeed peer is permitted to power-off. The implication is + that the setting ``pm_qos_no_power_off`` to zero on a superspeed port may + not cause the port to power-off until its highspeed peer has gone to its + runtime suspend state. Userspace must take care to order the suspensions + if it wants to guarantee that a superspeed port will power-off. + +2. Port resume is sequenced to force a superspeed port to power-on prior to its + highspeed peer. + +3. Port resume always triggers an attached child device to resume. After a + power session is lost the device may have been removed, or need reset. + Resuming the child device when the parent port regains power resolves those + states and clamps the maximum port power cycle frequency at the rate the + child device can suspend (autosuspend-delay) and resume (reset-resume + latency). + +Sysfs files relevant for port power control: + + ``<hubdev-portX>/power/pm_qos_no_power_off``: + This writable flag controls the state of an idle port. + Once all children and descendants have suspended the + port may suspend/poweroff provided that + pm_qos_no_power_off is '0'. If pm_qos_no_power_off is + '1' the port will remain active/powered regardless of + the stats of descendants. Defaults to 1. + + ``<hubdev-portX>/power/runtime_status``: + This file reflects whether the port is 'active' (power is on) + or 'suspended' (logically off). There is no indication to + userspace whether VBUS is still supplied. + + ``<hubdev-portX>/connect_type``: + An advisory read-only flag to userspace indicating the + location and connection type of the port. It returns + one of four values 'hotplug', 'hardwired', 'not used', + and 'unknown'. All values, besides unknown, are set by + platform firmware. + + ``hotplug`` indicates an externally connectable/visible + port on the platform. Typically userspace would choose + to keep such a port powered to handle new device + connection events. + + ``hardwired`` refers to a port that is not visible but + connectable. Examples are internal ports for USB + bluetooth that can be disconnected via an external + switch or a port with a hardwired USB camera. It is + expected to be safe to allow these ports to suspend + provided pm_qos_no_power_off is coordinated with any + switch that gates connections. Userspace must arrange + for the device to be connected prior to the port + powering off, or to activate the port prior to enabling + connection via a switch. + + ``not used`` refers to an internal port that is expected + to never have a device connected to it. These may be + empty internal ports, or ports that are not physically + exposed on a platform. Considered safe to be + powered-off at all times. + + ``unknown`` means platform firmware does not provide + information for this port. Most commonly refers to + external hub ports which should be considered 'hotplug' + for policy decisions. + + .. note:: + + - since we are relying on the BIOS to get this ACPI + information correct, the USB port descriptions may + be missing or wrong. + + - Take care in clearing ``pm_qos_no_power_off``. Once + power is off this port will + not respond to new connect events. + + Once a child device is attached additional constraints are + applied before the port is allowed to poweroff. + + ``<child>/power/control``: + Must be ``auto``, and the port will not + power down until ``<child>/power/runtime_status`` + reflects the 'suspended' state. Default + value is controlled by child device driver. + + ``<child>/power/persist``: + This defaults to ``1`` for most devices and indicates if + kernel can persist the device's configuration across a + power session loss (suspend / port-power event). When + this value is ``0`` (quirky devices), port poweroff is + disabled. + + ``<child>/driver/unbind``: + Wakeup capable devices will block port poweroff. At + this time the only mechanism to clear the usb-internal + wakeup-capability for an interface device is to unbind + its driver. + +Summary of poweroff pre-requisite settings relative to a port device:: + + echo 0 > power/pm_qos_no_power_off + echo 0 > peer/power/pm_qos_no_power_off # if it exists + echo auto > power/control # this is the default value + echo auto > <child>/power/control + echo 1 > <child>/power/persist # this is the default value + +Suggested Userspace Port Power Policy +------------------------------------- + +As noted above userspace needs to be careful and deliberate about what +ports are enabled for poweroff. + +The default configuration is that all ports start with +``power/pm_qos_no_power_off`` set to ``1`` causing ports to always remain +active. + +Given confidence in the platform firmware's description of the ports +(ACPI _PLD record for a port populates 'connect_type') userspace can +clear pm_qos_no_power_off for all 'not used' ports. The same can be +done for 'hardwired' ports provided poweroff is coordinated with any +connection switch for the port. + +A more aggressive userspace policy is to enable USB port power off for +all ports (set ``<hubdev-portX>/power/pm_qos_no_power_off`` to ``0``) when +some external factor indicates the user has stopped interacting with the +system. For example, a distro may want to enable power off all USB +ports when the screen blanks, and re-power them when the screen becomes +active. Smart phones and tablets may want to power off USB ports when +the user pushes the power button. diff --git a/Documentation/driver-api/usb/typec.rst b/Documentation/driver-api/usb/typec.rst new file mode 100644 index 000000000..48ff58095 --- /dev/null +++ b/Documentation/driver-api/usb/typec.rst @@ -0,0 +1,233 @@ + +USB Type-C connector class +========================== + +Introduction +------------ + +The typec class is meant for describing the USB Type-C ports in a system to the +user space in unified fashion. The class is designed to provide nothing else +except the user space interface implementation in hope that it can be utilized +on as many platforms as possible. + +The platforms are expected to register every USB Type-C port they have with the +class. In a normal case the registration will be done by a USB Type-C or PD PHY +driver, but it may be a driver for firmware interface such as UCSI, driver for +USB PD controller or even driver for Thunderbolt3 controller. This document +considers the component registering the USB Type-C ports with the class as "port +driver". + +On top of showing the capabilities, the class also offer user space control over +the roles and alternate modes of ports, partners and cable plugs when the port +driver is capable of supporting those features. + +The class provides an API for the port drivers described in this document. The +attributes are described in Documentation/ABI/testing/sysfs-class-typec. + +User space interface +-------------------- +Every port will be presented as its own device under /sys/class/typec/. The +first port will be named "port0", the second "port1" and so on. + +When connected, the partner will be presented also as its own device under +/sys/class/typec/. The parent of the partner device will always be the port it +is attached to. The partner attached to port "port0" will be named +"port0-partner". Full path to the device would be +/sys/class/typec/port0/port0-partner/. + +The cable and the two plugs on it may also be optionally presented as their own +devices under /sys/class/typec/. The cable attached to the port "port0" port +will be named port0-cable and the plug on the SOP Prime end (see USB Power +Delivery Specification ch. 2.4) will be named "port0-plug0" and on the SOP +Double Prime end "port0-plug1". The parent of a cable will always be the port, +and the parent of the cable plugs will always be the cable. + +If the port, partner or cable plug supports Alternate Modes, every supported +Alternate Mode SVID will have their own device describing them. Note that the +Alternate Mode devices will not be attached to the typec class. The parent of an +alternate mode will be the device that supports it, so for example an alternate +mode of port0-partner will be presented under /sys/class/typec/port0-partner/. +Every mode that is supported will have its own group under the Alternate Mode +device named "mode<index>", for example /sys/class/typec/port0/<alternate +mode>/mode1/. The requests for entering/exiting a mode can be done with "active" +attribute file in that group. + +Driver API +---------- + +Registering the ports +~~~~~~~~~~~~~~~~~~~~~ + +The port drivers will describe every Type-C port they control with struct +typec_capability data structure, and register them with the following API: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_register_port typec_unregister_port + +When registering the ports, the prefer_role member in struct typec_capability +deserves special notice. If the port that is being registered does not have +initial role preference, which means the port does not execute Try.SNK or +Try.SRC by default, the member must have value TYPEC_NO_PREFERRED_ROLE. +Otherwise if the port executes Try.SNK by default, the member must have value +TYPEC_DEVICE, and with Try.SRC the value must be TYPEC_HOST. + +Registering Partners +~~~~~~~~~~~~~~~~~~~~ + +After successful connection of a partner, the port driver needs to register the +partner with the class. Details about the partner need to be described in struct +typec_partner_desc. The class copies the details of the partner during +registration. The class offers the following API for registering/unregistering +partners. + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_register_partner typec_unregister_partner + +The class will provide a handle to struct typec_partner if the registration was +successful, or NULL. + +If the partner is USB Power Delivery capable, and the port driver is able to +show the result of Discover Identity command, the partner descriptor structure +should include handle to struct usb_pd_identity instance. The class will then +create a sysfs directory for the identity under the partner device. The result +of Discover Identity command can then be reported with the following API: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_partner_set_identity + +Registering Cables +~~~~~~~~~~~~~~~~~~ + +After successful connection of a cable that supports USB Power Delivery +Structured VDM "Discover Identity", the port driver needs to register the cable +and one or two plugs, depending if there is CC Double Prime controller present +in the cable or not. So a cable capable of SOP Prime communication, but not SOP +Double Prime communication, should only have one plug registered. For more +information about SOP communication, please read chapter about it from the +latest USB Power Delivery specification. + +The plugs are represented as their own devices. The cable is registered first, +followed by registration of the cable plugs. The cable will be the parent device +for the plugs. Details about the cable need to be described in struct +typec_cable_desc and about a plug in struct typec_plug_desc. The class copies +the details during registration. The class offers the following API for +registering/unregistering cables and their plugs: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_register_cable typec_unregister_cable typec_register_plug typec_unregister_plug + +The class will provide a handle to struct typec_cable and struct typec_plug if +the registration is successful, or NULL if it isn't. + +If the cable is USB Power Delivery capable, and the port driver is able to show +the result of Discover Identity command, the cable descriptor structure should +include handle to struct usb_pd_identity instance. The class will then create a +sysfs directory for the identity under the cable device. The result of Discover +Identity command can then be reported with the following API: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_cable_set_identity + +Notifications +~~~~~~~~~~~~~ + +When the partner has executed a role change, or when the default roles change +during connection of a partner or cable, the port driver must use the following +APIs to report it to the class: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_set_data_role typec_set_pwr_role typec_set_vconn_role typec_set_pwr_opmode + +Alternate Modes +~~~~~~~~~~~~~~~ + +USB Type-C ports, partners and cable plugs may support Alternate Modes. Each +Alternate Mode will have identifier called SVID, which is either a Standard ID +given by USB-IF or vendor ID, and each supported SVID can have 1 - 6 modes. The +class provides struct typec_mode_desc for describing individual mode of a SVID, +and struct typec_altmode_desc which is a container for all the supported modes. + +Ports that support Alternate Modes need to register each SVID they support with +the following API: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_port_register_altmode + +If a partner or cable plug provides a list of SVIDs as response to USB Power +Delivery Structured VDM Discover SVIDs message, each SVID needs to be +registered. + +API for the partners: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_partner_register_altmode + +API for the Cable Plugs: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_plug_register_altmode + +So ports, partners and cable plugs will register the alternate modes with their +own functions, but the registration will always return a handle to struct +typec_altmode on success, or NULL. The unregistration will happen with the same +function: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_unregister_altmode + +If a partner or cable plug enters or exits a mode, the port driver needs to +notify the class with the following API: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_altmode_update_active + +Multiplexer/DeMultiplexer Switches +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +USB Type-C connectors may have one or more mux/demux switches behind them. Since +the plugs can be inserted right-side-up or upside-down, a switch is needed to +route the correct data pairs from the connector to the USB controllers. If +Alternate or Accessory Modes are supported, another switch is needed that can +route the pins on the connector to some other component besides USB. USB Type-C +Connector Class supplies an API for registering those switches. + +.. kernel-doc:: drivers/usb/typec/mux.c + :functions: typec_switch_register typec_switch_unregister typec_mux_register typec_mux_unregister + +In most cases the same physical mux will handle both the orientation and mode. +However, as the port drivers will be responsible for the orientation, and the +alternate mode drivers for the mode, the two are always separated into their +own logical components: "mux" for the mode and "switch" for the orientation. + +When a port is registered, USB Type-C Connector Class requests both the mux and +the switch for the port. The drivers can then use the following API for +controlling them: + +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_set_orientation typec_set_mode + +If the connector is dual-role capable, there may also be a switch for the data +role. USB Type-C Connector Class does not supply separate API for them. The +port drivers can use USB Role Class API with those. + +Illustration of the muxes behind a connector that supports an alternate mode:: + + ------------------------ + | Connector | + ------------------------ + | | + ------------------------ + \ Orientation / + -------------------- + | + -------------------- + / Mode \ + ------------------------ + / \ + ------------------------ -------------------- + | Alt Mode | / USB Role \ + ------------------------ ------------------------ + / \ + ------------------------ ------------------------ + | USB Host | | USB Device | + ------------------------ ------------------------ diff --git a/Documentation/driver-api/usb/typec_bus.rst b/Documentation/driver-api/usb/typec_bus.rst new file mode 100644 index 000000000..d5eec1715 --- /dev/null +++ b/Documentation/driver-api/usb/typec_bus.rst @@ -0,0 +1,136 @@ + +API for USB Type-C Alternate Mode drivers +========================================= + +Introduction +------------ + +Alternate modes require communication with the partner using Vendor Defined +Messages (VDM) as defined in USB Type-C and USB Power Delivery Specifications. +The communication is SVID (Standard or Vendor ID) specific, i.e. specific for +every alternate mode, so every alternate mode will need a custom driver. + +USB Type-C bus allows binding a driver to the discovered partner alternate +modes by using the SVID and the mode number. + +USB Type-C Connector Class provides a device for every alternate mode a port +supports, and separate device for every alternate mode the partner supports. +The drivers for the alternate modes are bound to the partner alternate mode +devices, and the port alternate mode devices must be handled by the port +drivers. + +When a new partner alternate mode device is registered, it is linked to the +alternate mode device of the port that the partner is attached to, that has +matching SVID and mode. Communication between the port driver and alternate mode +driver will happen using the same API. + +The port alternate mode devices are used as a proxy between the partner and the +alternate mode drivers, so the port drivers are only expected to pass the SVID +specific commands from the alternate mode drivers to the partner, and from the +partners to the alternate mode drivers. No direct SVID specific communication is +needed from the port drivers, but the port drivers need to provide the operation +callbacks for the port alternate mode devices, just like the alternate mode +drivers need to provide them for the partner alternate mode devices. + +Usage: +------ + +General +~~~~~~~ + +By default, the alternate mode drivers are responsible for entering the mode. +It is also possible to leave the decision about entering the mode to the user +space (See Documentation/ABI/testing/sysfs-class-typec). Port drivers should not +enter any modes on their own. + +``->vdm`` is the most important callback in the operation callbacks vector. It +will be used to deliver all the SVID specific commands from the partner to the +alternate mode driver, and vice versa in case of port drivers. The drivers send +the SVID specific commands to each other using :c:func:`typec_altmode_vmd()`. + +If the communication with the partner using the SVID specific commands results +in need to reconfigure the pins on the connector, the alternate mode driver +needs to notify the bus using :c:func:`typec_altmode_notify()`. The driver +passes the negotiated SVID specific pin configuration value to the function as +parameter. The bus driver will then configure the mux behind the connector using +that value as the state value for the mux, and also call blocking notification +chain to notify the external drivers about the state of the connector that need +to know it. + +NOTE: The SVID specific pin configuration values must always start from +``TYPEC_STATE_MODAL``. USB Type-C specification defines two default states for +the connector: ``TYPEC_STATE_USB`` and ``TYPEC_STATE_SAFE``. These values are +reserved by the bus as the first possible values for the state. When the +alternate mode is entered, the bus will put the connector into +``TYPEC_STATE_SAFE`` before sending Enter or Exit Mode command as defined in USB +Type-C Specification, and also put the connector back to ``TYPEC_STATE_USB`` +after the mode has been exited. + +An example of working definitions for SVID specific pin configurations would +look like this: + +enum { + ALTMODEX_CONF_A = TYPEC_STATE_MODAL, + ALTMODEX_CONF_B, + ... +}; + +Helper macro ``TYPEC_MODAL_STATE()`` can also be used: + +#define ALTMODEX_CONF_A = TYPEC_MODAL_STATE(0); +#define ALTMODEX_CONF_B = TYPEC_MODAL_STATE(1); + +Notification chain +~~~~~~~~~~~~~~~~~~ + +The drivers for the components that the alternate modes are designed for need to +get details regarding the results of the negotiation with the partner, and the +pin configuration of the connector. In case of DisplayPort alternate mode for +example, the GPU drivers will need to know those details. In case of +Thunderbolt alternate mode, the thunderbolt drivers will need to know them, and +so on. + +The notification chain is designed for this purpose. The drivers can register +notifiers with :c:func:`typec_altmode_register_notifier()`. + +Cable plug alternate modes +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The alternate mode drivers are not bound to cable plug alternate mode devices, +only to the partner alternate mode devices. If the alternate mode supports, or +requires, a cable that responds to SOP Prime, and optionally SOP Double Prime +messages, the driver for that alternate mode must request handle to the cable +plug alternate modes using :c:func:`typec_altmode_get_plug()`, and take over +their control. + +Driver API +---------- + +Alternate mode driver registering/unregistering +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/usb/typec/bus.c + :functions: typec_altmode_register_driver typec_altmode_unregister_driver + +Alternate mode driver operations +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/usb/typec/bus.c + :functions: typec_altmode_enter typec_altmode_exit typec_altmode_attention typec_altmode_vdm typec_altmode_notify + +API for the port drivers +~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/usb/typec/bus.c + :functions: typec_match_altmode + +Cable Plug operations +~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/usb/typec/bus.c + :functions: typec_altmode_get_plug typec_altmode_put_plug + +Notifications +~~~~~~~~~~~~~ +.. kernel-doc:: drivers/usb/typec/class.c + :functions: typec_altmode_register_notifier typec_altmode_unregister_notifier diff --git a/Documentation/driver-api/usb/usb.rst b/Documentation/driver-api/usb/usb.rst new file mode 100644 index 000000000..078e981e2 --- /dev/null +++ b/Documentation/driver-api/usb/usb.rst @@ -0,0 +1,1045 @@ +.. _usb-hostside-api: + +=========================== +The Linux-USB Host Side API +=========================== + +Introduction to USB on Linux +============================ + +A Universal Serial Bus (USB) is used to connect a host, such as a PC or +workstation, to a number of peripheral devices. USB uses a tree +structure, with the host as the root (the system's master), hubs as +interior nodes, and peripherals as leaves (and slaves). Modern PCs +support several such trees of USB devices, usually +a few USB 3.0 (5 GBit/s) or USB 3.1 (10 GBit/s) and some legacy +USB 2.0 (480 MBit/s) busses just in case. + +That master/slave asymmetry was designed-in for a number of reasons, one +being ease of use. It is not physically possible to mistake upstream and +downstream or it does not matter with a type C plug (or they are built into the +peripheral). Also, the host software doesn't need to deal with +distributed auto-configuration since the pre-designated master node +manages all that. + +Kernel developers added USB support to Linux early in the 2.2 kernel +series and have been developing it further since then. Besides support +for each new generation of USB, various host controllers gained support, +new drivers for peripherals have been added and advanced features for latency +measurement and improved power management introduced. + +Linux can run inside USB devices as well as on the hosts that control +the devices. But USB device drivers running inside those peripherals +don't do the same things as the ones running inside hosts, so they've +been given a different name: *gadget drivers*. This document does not +cover gadget drivers. + +USB Host-Side API Model +======================= + +Host-side drivers for USB devices talk to the "usbcore" APIs. There are +two. One is intended for *general-purpose* drivers (exposed through +driver frameworks), and the other is for drivers that are *part of the +core*. Such core drivers include the *hub* driver (which manages trees +of USB devices) and several different kinds of *host controller +drivers*, which control individual busses. + +The device model seen by USB drivers is relatively complex. + +- USB supports four kinds of data transfers (control, bulk, interrupt, + and isochronous). Two of them (control and bulk) use bandwidth as + it's available, while the other two (interrupt and isochronous) are + scheduled to provide guaranteed bandwidth. + +- The device description model includes one or more "configurations" + per device, only one of which is active at a time. Devices are supposed + to be capable of operating at lower than their top + speeds and may provide a BOS descriptor showing the lowest speed they + remain fully operational at. + +- From USB 3.0 on configurations have one or more "functions", which + provide a common functionality and are grouped together for purposes + of power management. + +- Configurations or functions have one or more "interfaces", each of which may have + "alternate settings". Interfaces may be standardized by USB "Class" + specifications, or may be specific to a vendor or device. + + USB device drivers actually bind to interfaces, not devices. Think of + them as "interface drivers", though you may not see many devices + where the distinction is important. *Most USB devices are simple, + with only one function, one configuration, one interface, and one alternate + setting.* + +- Interfaces have one or more "endpoints", each of which supports one + type and direction of data transfer such as "bulk out" or "interrupt + in". The entire configuration may have up to sixteen endpoints in + each direction, allocated as needed among all the interfaces. + +- Data transfer on USB is packetized; each endpoint has a maximum + packet size. Drivers must often be aware of conventions such as + flagging the end of bulk transfers using "short" (including zero + length) packets. + +- The Linux USB API supports synchronous calls for control and bulk + messages. It also supports asynchronous calls for all kinds of data + transfer, using request structures called "URBs" (USB Request + Blocks). + +Accordingly, the USB Core API exposed to device drivers covers quite a +lot of territory. You'll probably need to consult the USB 3.0 +specification, available online from www.usb.org at no cost, as well as +class or device specifications. + +The only host-side drivers that actually touch hardware (reading/writing +registers, handling IRQs, and so on) are the HCDs. In theory, all HCDs +provide the same functionality through the same API. In practice, that's +becoming more true, but there are still differences +that crop up especially with fault handling on the less common controllers. +Different controllers don't +necessarily report the same aspects of failures, and recovery from +faults (including software-induced ones like unlinking an URB) isn't yet +fully consistent. Device driver authors should make a point of doing +disconnect testing (while the device is active) with each different host +controller driver, to make sure drivers don't have bugs of their own as +well as to make sure they aren't relying on some HCD-specific behavior. + +.. _usb_chapter9: + +USB-Standard Types +================== + +In ``<linux/usb/ch9.h>`` you will find the USB data types defined in +chapter 9 of the USB specification. These data types are used throughout +USB, and in APIs including this host side API, gadget APIs, usb character +devices and debugfs interfaces. + +.. kernel-doc:: include/linux/usb/ch9.h + :internal: + +.. _usb_header: + +Host-Side Data Types and Macros +=============================== + +The host side API exposes several layers to drivers, some of which are +more necessary than others. These support lifecycle models for host side +drivers and devices, and support passing buffers through usbcore to some +HCD that performs the I/O for the device driver. + +.. kernel-doc:: include/linux/usb.h + :internal: + +USB Core APIs +============= + +There are two basic I/O models in the USB API. The most elemental one is +asynchronous: drivers submit requests in the form of an URB, and the +URB's completion callback handles the next step. All USB transfer types +support that model, although there are special cases for control URBs +(which always have setup and status stages, but may not have a data +stage) and isochronous URBs (which allow large packets and include +per-packet fault reports). Built on top of that is synchronous API +support, where a driver calls a routine that allocates one or more URBs, +submits them, and waits until they complete. There are synchronous +wrappers for single-buffer control and bulk transfers (which are awkward +to use in some driver disconnect scenarios), and for scatterlist based +streaming i/o (bulk or interrupt). + +USB drivers need to provide buffers that can be used for DMA, although +they don't necessarily need to provide the DMA mapping themselves. There +are APIs to use used when allocating DMA buffers, which can prevent use +of bounce buffers on some systems. In some cases, drivers may be able to +rely on 64bit DMA to eliminate another kind of bounce buffer. + +.. kernel-doc:: drivers/usb/core/urb.c + :export: + +.. kernel-doc:: drivers/usb/core/message.c + :export: + +.. kernel-doc:: drivers/usb/core/file.c + :export: + +.. kernel-doc:: drivers/usb/core/driver.c + :export: + +.. kernel-doc:: drivers/usb/core/usb.c + :export: + +.. kernel-doc:: drivers/usb/core/hub.c + :export: + +Host Controller APIs +==================== + +These APIs are only for use by host controller drivers, most of which +implement standard register interfaces such as XHCI, EHCI, OHCI, or UHCI. UHCI +was one of the first interfaces, designed by Intel and also used by VIA; +it doesn't do much in hardware. OHCI was designed later, to have the +hardware do more work (bigger transfers, tracking protocol state, and so +on). EHCI was designed with USB 2.0; its design has features that +resemble OHCI (hardware does much more work) as well as UHCI (some parts +of ISO support, TD list processing). XHCI was designed with USB 3.0. It +continues to shift support for functionality into hardware. + +There are host controllers other than the "big three", although most PCI +based controllers (and a few non-PCI based ones) use one of those +interfaces. Not all host controllers use DMA; some use PIO, and there is +also a simulator and a virtual host controller to pipe USB over the network. + +The same basic APIs are available to drivers for all those controllers. +For historical reasons they are in two layers: :c:type:`struct +usb_bus <usb_bus>` is a rather thin layer that became available +in the 2.2 kernels, while :c:type:`struct usb_hcd <usb_hcd>` +is a more featureful layer +that lets HCDs share common code, to shrink driver size and +significantly reduce hcd-specific behaviors. + +.. kernel-doc:: drivers/usb/core/hcd.c + :export: + +.. kernel-doc:: drivers/usb/core/hcd-pci.c + :export: + +.. kernel-doc:: drivers/usb/core/buffer.c + :internal: + +The USB character device nodes +============================== + +This chapter presents the Linux character device nodes. You may prefer +to avoid writing new kernel code for your USB driver. User mode device +drivers are usually packaged as applications or libraries, and may use +character devices through some programming library that wraps it. +Such libraries include: + + - `libusb <http://libusb.sourceforge.net>`__ for C/C++, and + - `jUSB <http://jUSB.sourceforge.net>`__ for Java. + +Some old information about it can be seen at the "USB Device Filesystem" +section of the USB Guide. The latest copy of the USB Guide can be found +at http://www.linux-usb.org/ + +.. note:: + + - They were used to be implemented via *usbfs*, but this is not part of + the sysfs debug interface. + + - This particular documentation is incomplete, especially with respect + to the asynchronous mode. As of kernel 2.5.66 the code and this + (new) documentation need to be cross-reviewed. + +What files are in "devtmpfs"? +----------------------------- + +Conventionally mounted at ``/dev/bus/usb/``, usbfs features include: + +- ``/dev/bus/usb/BBB/DDD`` ... magic files exposing the each device's + configuration descriptors, and supporting a series of ioctls for + making device requests, including I/O to devices. (Purely for access + by programs.) + +Each bus is given a number (``BBB``) based on when it was enumerated; within +each bus, each device is given a similar number (``DDD``). Those ``BBB/DDD`` +paths are not "stable" identifiers; expect them to change even if you +always leave the devices plugged in to the same hub port. *Don't even +think of saving these in application configuration files.* Stable +identifiers are available, for user mode applications that want to use +them. HID and networking devices expose these stable IDs, so that for +example you can be sure that you told the right UPS to power down its +second server. Pleast note that it doesn't (yet) expose those IDs. + +/dev/bus/usb/BBB/DDD +-------------------- + +Use these files in one of these basic ways: + +- *They can be read,* producing first the device descriptor (18 bytes) and + then the descriptors for the current configuration. See the USB 2.0 spec + for details about those binary data formats. You'll need to convert most + multibyte values from little endian format to your native host byte + order, although a few of the fields in the device descriptor (both of + the BCD-encoded fields, and the vendor and product IDs) will be + byteswapped for you. Note that configuration descriptors include + descriptors for interfaces, altsettings, endpoints, and maybe additional + class descriptors. + +- *Perform USB operations* using *ioctl()* requests to make endpoint I/O + requests (synchronously or asynchronously) or manage the device. These + requests need the ``CAP_SYS_RAWIO`` capability, as well as filesystem + access permissions. Only one ioctl request can be made on one of these + device files at a time. This means that if you are synchronously reading + an endpoint from one thread, you won't be able to write to a different + endpoint from another thread until the read completes. This works for + *half duplex* protocols, but otherwise you'd use asynchronous i/o + requests. + +Each connected USB device has one file. The ``BBB`` indicates the bus +number. The ``DDD`` indicates the device address on that bus. Both +of these numbers are assigned sequentially, and can be reused, so +you can't rely on them for stable access to devices. For example, +it's relatively common for devices to re-enumerate while they are +still connected (perhaps someone jostled their power supply, hub, +or USB cable), so a device might be ``002/027`` when you first connect +it and ``002/048`` sometime later. + +These files can be read as binary data. The binary data consists +of first the device descriptor, then the descriptors for each +configuration of the device. Multi-byte fields in the device descriptor +are converted to host endianness by the kernel. The configuration +descriptors are in bus endian format! The configuration descriptor +are wTotalLength bytes apart. If a device returns less configuration +descriptor data than indicated by wTotalLength there will be a hole in +the file for the missing bytes. This information is also shown +in text form by the ``/sys/kernel/debug/usb/devices`` file, described later. + +These files may also be used to write user-level drivers for the USB +devices. You would open the ``/dev/bus/usb/BBB/DDD`` file read/write, +read its descriptors to make sure it's the device you expect, and then +bind to an interface (or perhaps several) using an ioctl call. You +would issue more ioctls to the device to communicate to it using +control, bulk, or other kinds of USB transfers. The IOCTLs are +listed in the ``<linux/usbdevice_fs.h>`` file, and at this writing the +source code (``linux/drivers/usb/core/devio.c``) is the primary reference +for how to access devices through those files. + +Note that since by default these ``BBB/DDD`` files are writable only by +root, only root can write such user mode drivers. You can selectively +grant read/write permissions to other users by using ``chmod``. Also, +usbfs mount options such as ``devmode=0666`` may be helpful. + + +Life Cycle of User Mode Drivers +------------------------------- + +Such a driver first needs to find a device file for a device it knows +how to handle. Maybe it was told about it because a ``/sbin/hotplug`` +event handling agent chose that driver to handle the new device. Or +maybe it's an application that scans all the ``/dev/bus/usb`` device files, +and ignores most devices. In either case, it should :c:func:`read()` +all the descriptors from the device file, and check them against what it +knows how to handle. It might just reject everything except a particular +vendor and product ID, or need a more complex policy. + +Never assume there will only be one such device on the system at a time! +If your code can't handle more than one device at a time, at least +detect when there's more than one, and have your users choose which +device to use. + +Once your user mode driver knows what device to use, it interacts with +it in either of two styles. The simple style is to make only control +requests; some devices don't need more complex interactions than those. +(An example might be software using vendor-specific control requests for +some initialization or configuration tasks, with a kernel driver for the +rest.) + +More likely, you need a more complex style driver: one using non-control +endpoints, reading or writing data and claiming exclusive use of an +interface. *Bulk* transfers are easiest to use, but only their sibling +*interrupt* transfers work with low speed devices. Both interrupt and +*isochronous* transfers offer service guarantees because their bandwidth +is reserved. Such "periodic" transfers are awkward to use through usbfs, +unless you're using the asynchronous calls. However, interrupt transfers +can also be used in a synchronous "one shot" style. + +Your user-mode driver should never need to worry about cleaning up +request state when the device is disconnected, although it should close +its open file descriptors as soon as it starts seeing the ENODEV errors. + +The ioctl() Requests +-------------------- + +To use these ioctls, you need to include the following headers in your +userspace program:: + + #include <linux/usb.h> + #include <linux/usbdevice_fs.h> + #include <asm/byteorder.h> + +The standard USB device model requests, from "Chapter 9" of the USB 2.0 +specification, are automatically included from the ``<linux/usb/ch9.h>`` +header. + +Unless noted otherwise, the ioctl requests described here will update +the modification time on the usbfs file to which they are applied +(unless they fail). A return of zero indicates success; otherwise, a +standard USB error code is returned (These are documented in +:ref:`usb-error-codes`). + +Each of these files multiplexes access to several I/O streams, one per +endpoint. Each device has one control endpoint (endpoint zero) which +supports a limited RPC style RPC access. Devices are configured by +hub_wq (in the kernel) setting a device-wide *configuration* that +affects things like power consumption and basic functionality. The +endpoints are part of USB *interfaces*, which may have *altsettings* +affecting things like which endpoints are available. Many devices only +have a single configuration and interface, so drivers for them will +ignore configurations and altsettings. + +Management/Status Requests +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +A number of usbfs requests don't deal very directly with device I/O. +They mostly relate to device management and status. These are all +synchronous requests. + +USBDEVFS_CLAIMINTERFACE + This is used to force usbfs to claim a specific interface, which has + not previously been claimed by usbfs or any other kernel driver. The + ioctl parameter is an integer holding the number of the interface + (bInterfaceNumber from descriptor). + + Note that if your driver doesn't claim an interface before trying to + use one of its endpoints, and no other driver has bound to it, then + the interface is automatically claimed by usbfs. + + This claim will be released by a RELEASEINTERFACE ioctl, or by + closing the file descriptor. File modification time is not updated + by this request. + +USBDEVFS_CONNECTINFO + Says whether the device is lowspeed. The ioctl parameter points to a + structure like this:: + + struct usbdevfs_connectinfo { + unsigned int devnum; + unsigned char slow; + }; + + File modification time is not updated by this request. + + *You can't tell whether a "not slow" device is connected at high + speed (480 MBit/sec) or just full speed (12 MBit/sec).* You should + know the devnum value already, it's the DDD value of the device file + name. + +USBDEVFS_GETDRIVER + Returns the name of the kernel driver bound to a given interface (a + string). Parameter is a pointer to this structure, which is + modified:: + + struct usbdevfs_getdriver { + unsigned int interface; + char driver[USBDEVFS_MAXDRIVERNAME + 1]; + }; + + File modification time is not updated by this request. + +USBDEVFS_IOCTL + Passes a request from userspace through to a kernel driver that has + an ioctl entry in the *struct usb_driver* it registered:: + + struct usbdevfs_ioctl { + int ifno; + int ioctl_code; + void *data; + }; + + /* user mode call looks like this. + * 'request' becomes the driver->ioctl() 'code' parameter. + * the size of 'param' is encoded in 'request', and that data + * is copied to or from the driver->ioctl() 'buf' parameter. + */ + static int + usbdev_ioctl (int fd, int ifno, unsigned request, void *param) + { + struct usbdevfs_ioctl wrapper; + + wrapper.ifno = ifno; + wrapper.ioctl_code = request; + wrapper.data = param; + + return ioctl (fd, USBDEVFS_IOCTL, &wrapper); + } + + File modification time is not updated by this request. + + This request lets kernel drivers talk to user mode code through + filesystem operations even when they don't create a character or + block special device. It's also been used to do things like ask + devices what device special file should be used. Two pre-defined + ioctls are used to disconnect and reconnect kernel drivers, so that + user mode code can completely manage binding and configuration of + devices. + +USBDEVFS_RELEASEINTERFACE + This is used to release the claim usbfs made on interface, either + implicitly or because of a USBDEVFS_CLAIMINTERFACE call, before the + file descriptor is closed. The ioctl parameter is an integer holding + the number of the interface (bInterfaceNumber from descriptor); File + modification time is not updated by this request. + + .. warning:: + + *No security check is made to ensure that the task which made + the claim is the one which is releasing it. This means that user + mode driver may interfere other ones.* + +USBDEVFS_RESETEP + Resets the data toggle value for an endpoint (bulk or interrupt) to + DATA0. The ioctl parameter is an integer endpoint number (1 to 15, + as identified in the endpoint descriptor), with USB_DIR_IN added + if the device's endpoint sends data to the host. + + .. Warning:: + + *Avoid using this request. It should probably be removed.* Using + it typically means the device and driver will lose toggle + synchronization. If you really lost synchronization, you likely + need to completely handshake with the device, using a request + like CLEAR_HALT or SET_INTERFACE. + +USBDEVFS_DROP_PRIVILEGES + This is used to relinquish the ability to do certain operations + which are considered to be privileged on a usbfs file descriptor. + This includes claiming arbitrary interfaces, resetting a device on + which there are currently claimed interfaces from other users, and + issuing USBDEVFS_IOCTL calls. The ioctl parameter is a 32 bit mask + of interfaces the user is allowed to claim on this file descriptor. + You may issue this ioctl more than one time to narrow said mask. + +Synchronous I/O Support +~~~~~~~~~~~~~~~~~~~~~~~ + +Synchronous requests involve the kernel blocking until the user mode +request completes, either by finishing successfully or by reporting an +error. In most cases this is the simplest way to use usbfs, although as +noted above it does prevent performing I/O to more than one endpoint at +a time. + +USBDEVFS_BULK + Issues a bulk read or write request to the device. The ioctl + parameter is a pointer to this structure:: + + struct usbdevfs_bulktransfer { + unsigned int ep; + unsigned int len; + unsigned int timeout; /* in milliseconds */ + void *data; + }; + + The ``ep`` value identifies a bulk endpoint number (1 to 15, as + identified in an endpoint descriptor), masked with USB_DIR_IN when + referring to an endpoint which sends data to the host from the + device. The length of the data buffer is identified by ``len``; Recent + kernels support requests up to about 128KBytes. *FIXME say how read + length is returned, and how short reads are handled.*. + +USBDEVFS_CLEAR_HALT + Clears endpoint halt (stall) and resets the endpoint toggle. This is + only meaningful for bulk or interrupt endpoints. The ioctl parameter + is an integer endpoint number (1 to 15, as identified in an endpoint + descriptor), masked with USB_DIR_IN when referring to an endpoint + which sends data to the host from the device. + + Use this on bulk or interrupt endpoints which have stalled, + returning ``-EPIPE`` status to a data transfer request. Do not issue + the control request directly, since that could invalidate the host's + record of the data toggle. + +USBDEVFS_CONTROL + Issues a control request to the device. The ioctl parameter points + to a structure like this:: + + struct usbdevfs_ctrltransfer { + __u8 bRequestType; + __u8 bRequest; + __u16 wValue; + __u16 wIndex; + __u16 wLength; + __u32 timeout; /* in milliseconds */ + void *data; + }; + + The first eight bytes of this structure are the contents of the + SETUP packet to be sent to the device; see the USB 2.0 specification + for details. The bRequestType value is composed by combining a + ``USB_TYPE_*`` value, a ``USB_DIR_*`` value, and a ``USB_RECIP_*`` + value (from ``linux/usb.h``). If wLength is nonzero, it describes + the length of the data buffer, which is either written to the device + (USB_DIR_OUT) or read from the device (USB_DIR_IN). + + At this writing, you can't transfer more than 4 KBytes of data to or + from a device; usbfs has a limit, and some host controller drivers + have a limit. (That's not usually a problem.) *Also* there's no way + to say it's not OK to get a short read back from the device. + +USBDEVFS_RESET + Does a USB level device reset. The ioctl parameter is ignored. After + the reset, this rebinds all device interfaces. File modification + time is not updated by this request. + +.. warning:: + + *Avoid using this call* until some usbcore bugs get fixed, since + it does not fully synchronize device, interface, and driver (not + just usbfs) state. + +USBDEVFS_SETINTERFACE + Sets the alternate setting for an interface. The ioctl parameter is + a pointer to a structure like this:: + + struct usbdevfs_setinterface { + unsigned int interface; + unsigned int altsetting; + }; + + File modification time is not updated by this request. + + Those struct members are from some interface descriptor applying to + the current configuration. The interface number is the + bInterfaceNumber value, and the altsetting number is the + bAlternateSetting value. (This resets each endpoint in the + interface.) + +USBDEVFS_SETCONFIGURATION + Issues the :c:func:`usb_set_configuration()` call for the + device. The parameter is an integer holding the number of a + configuration (bConfigurationValue from descriptor). File + modification time is not updated by this request. + +.. warning:: + + *Avoid using this call* until some usbcore bugs get fixed, since + it does not fully synchronize device, interface, and driver (not + just usbfs) state. + +Asynchronous I/O Support +~~~~~~~~~~~~~~~~~~~~~~~~ + +As mentioned above, there are situations where it may be important to +initiate concurrent operations from user mode code. This is particularly +important for periodic transfers (interrupt and isochronous), but it can +be used for other kinds of USB requests too. In such cases, the +asynchronous requests described here are essential. Rather than +submitting one request and having the kernel block until it completes, +the blocking is separate. + +These requests are packaged into a structure that resembles the URB used +by kernel device drivers. (No POSIX Async I/O support here, sorry.) It +identifies the endpoint type (``USBDEVFS_URB_TYPE_*``), endpoint +(number, masked with USB_DIR_IN as appropriate), buffer and length, +and a user "context" value serving to uniquely identify each request. +(It's usually a pointer to per-request data.) Flags can modify requests +(not as many as supported for kernel drivers). + +Each request can specify a realtime signal number (between SIGRTMIN and +SIGRTMAX, inclusive) to request a signal be sent when the request +completes. + +When usbfs returns these urbs, the status value is updated, and the +buffer may have been modified. Except for isochronous transfers, the +actual_length is updated to say how many bytes were transferred; if the +USBDEVFS_URB_DISABLE_SPD flag is set ("short packets are not OK"), if +fewer bytes were read than were requested then you get an error report:: + + struct usbdevfs_iso_packet_desc { + unsigned int length; + unsigned int actual_length; + unsigned int status; + }; + + struct usbdevfs_urb { + unsigned char type; + unsigned char endpoint; + int status; + unsigned int flags; + void *buffer; + int buffer_length; + int actual_length; + int start_frame; + int number_of_packets; + int error_count; + unsigned int signr; + void *usercontext; + struct usbdevfs_iso_packet_desc iso_frame_desc[]; + }; + +For these asynchronous requests, the file modification time reflects +when the request was initiated. This contrasts with their use with the +synchronous requests, where it reflects when requests complete. + +USBDEVFS_DISCARDURB + *TBS* File modification time is not updated by this request. + +USBDEVFS_DISCSIGNAL + *TBS* File modification time is not updated by this request. + +USBDEVFS_REAPURB + *TBS* File modification time is not updated by this request. + +USBDEVFS_REAPURBNDELAY + *TBS* File modification time is not updated by this request. + +USBDEVFS_SUBMITURB + *TBS* + +The USB devices +=============== + +The USB devices are now exported via debugfs: + +- ``/sys/kernel/debug/usb/devices`` ... a text file showing each of the USB + devices on known to the kernel, and their configuration descriptors. + You can also poll() this to learn about new devices. + +/sys/kernel/debug/usb/devices +----------------------------- + +This file is handy for status viewing tools in user mode, which can scan +the text format and ignore most of it. More detailed device status +(including class and vendor status) is available from device-specific +files. For information about the current format of this file, see below. + +This file, in combination with the poll() system call, can also be used +to detect when devices are added or removed:: + + int fd; + struct pollfd pfd; + + fd = open("/sys/kernel/debug/usb/devices", O_RDONLY); + pfd = { fd, POLLIN, 0 }; + for (;;) { + /* The first time through, this call will return immediately. */ + poll(&pfd, 1, -1); + + /* To see what's changed, compare the file's previous and current + contents or scan the filesystem. (Scanning is more precise.) */ + } + +Note that this behavior is intended to be used for informational and +debug purposes. It would be more appropriate to use programs such as +udev or HAL to initialize a device or start a user-mode helper program, +for instance. + +In this file, each device's output has multiple lines of ASCII output. + +I made it ASCII instead of binary on purpose, so that someone +can obtain some useful data from it without the use of an +auxiliary program. However, with an auxiliary program, the numbers +in the first 4 columns of each ``T:`` line (topology info: +Lev, Prnt, Port, Cnt) can be used to build a USB topology diagram. + +Each line is tagged with a one-character ID for that line:: + + T = Topology (etc.) + B = Bandwidth (applies only to USB host controllers, which are + virtualized as root hubs) + D = Device descriptor info. + P = Product ID info. (from Device descriptor, but they won't fit + together on one line) + S = String descriptors. + C = Configuration descriptor info. (* = active configuration) + I = Interface descriptor info. + E = Endpoint descriptor info. + +/sys/kernel/debug/usb/devices output format +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Legend:: + d = decimal number (may have leading spaces or 0's) + x = hexadecimal number (may have leading spaces or 0's) + s = string + + + +Topology info +^^^^^^^^^^^^^ + +:: + + T: Bus=dd Lev=dd Prnt=dd Port=dd Cnt=dd Dev#=ddd Spd=dddd MxCh=dd + | | | | | | | | |__MaxChildren + | | | | | | | |__Device Speed in Mbps + | | | | | | |__DeviceNumber + | | | | | |__Count of devices at this level + | | | | |__Connector/Port on Parent for this device + | | | |__Parent DeviceNumber + | | |__Level in topology for this bus + | |__Bus number + |__Topology info tag + +Speed may be: + + ======= ====================================================== + 1.5 Mbit/s for low speed USB + 12 Mbit/s for full speed USB + 480 Mbit/s for high speed USB (added for USB 2.0); + also used for Wireless USB, which has no fixed speed + 5000 Mbit/s for SuperSpeed USB (added for USB 3.0) + ======= ====================================================== + +For reasons lost in the mists of time, the Port number is always +too low by 1. For example, a device plugged into port 4 will +show up with ``Port=03``. + +Bandwidth info +^^^^^^^^^^^^^^ + +:: + + B: Alloc=ddd/ddd us (xx%), #Int=ddd, #Iso=ddd + | | | |__Number of isochronous requests + | | |__Number of interrupt requests + | |__Total Bandwidth allocated to this bus + |__Bandwidth info tag + +Bandwidth allocation is an approximation of how much of one frame +(millisecond) is in use. It reflects only periodic transfers, which +are the only transfers that reserve bandwidth. Control and bulk +transfers use all other bandwidth, including reserved bandwidth that +is not used for transfers (such as for short packets). + +The percentage is how much of the "reserved" bandwidth is scheduled by +those transfers. For a low or full speed bus (loosely, "USB 1.1"), +90% of the bus bandwidth is reserved. For a high speed bus (loosely, +"USB 2.0") 80% is reserved. + + +Device descriptor info & Product ID info +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + +:: + + D: Ver=x.xx Cls=xx(s) Sub=xx Prot=xx MxPS=dd #Cfgs=dd + P: Vendor=xxxx ProdID=xxxx Rev=xx.xx + +where:: + + D: Ver=x.xx Cls=xx(sssss) Sub=xx Prot=xx MxPS=dd #Cfgs=dd + | | | | | | |__NumberConfigurations + | | | | | |__MaxPacketSize of Default Endpoint + | | | | |__DeviceProtocol + | | | |__DeviceSubClass + | | |__DeviceClass + | |__Device USB version + |__Device info tag #1 + +where:: + + P: Vendor=xxxx ProdID=xxxx Rev=xx.xx + | | | |__Product revision number + | | |__Product ID code + | |__Vendor ID code + |__Device info tag #2 + + +String descriptor info +^^^^^^^^^^^^^^^^^^^^^^ +:: + + S: Manufacturer=ssss + | |__Manufacturer of this device as read from the device. + | For USB host controller drivers (virtual root hubs) this may + | be omitted, or (for newer drivers) will identify the kernel + | version and the driver which provides this hub emulation. + |__String info tag + + S: Product=ssss + | |__Product description of this device as read from the device. + | For older USB host controller drivers (virtual root hubs) this + | indicates the driver; for newer ones, it's a product (and vendor) + | description that often comes from the kernel's PCI ID database. + |__String info tag + + S: SerialNumber=ssss + | |__Serial Number of this device as read from the device. + | For USB host controller drivers (virtual root hubs) this is + | some unique ID, normally a bus ID (address or slot name) that + | can't be shared with any other device. + |__String info tag + + + +Configuration descriptor info +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +:: + + C:* #Ifs=dd Cfg#=dd Atr=xx MPwr=dddmA + | | | | | |__MaxPower in mA + | | | | |__Attributes + | | | |__ConfiguratioNumber + | | |__NumberOfInterfaces + | |__ "*" indicates the active configuration (others are " ") + |__Config info tag + +USB devices may have multiple configurations, each of which act +rather differently. For example, a bus-powered configuration +might be much less capable than one that is self-powered. Only +one device configuration can be active at a time; most devices +have only one configuration. + +Each configuration consists of one or more interfaces. Each +interface serves a distinct "function", which is typically bound +to a different USB device driver. One common example is a USB +speaker with an audio interface for playback, and a HID interface +for use with software volume control. + +Interface descriptor info (can be multiple per Config) +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +:: + + I:* If#=dd Alt=dd #EPs=dd Cls=xx(sssss) Sub=xx Prot=xx Driver=ssss + | | | | | | | | |__Driver name + | | | | | | | | or "(none)" + | | | | | | | |__InterfaceProtocol + | | | | | | |__InterfaceSubClass + | | | | | |__InterfaceClass + | | | | |__NumberOfEndpoints + | | | |__AlternateSettingNumber + | | |__InterfaceNumber + | |__ "*" indicates the active altsetting (others are " ") + |__Interface info tag + +A given interface may have one or more "alternate" settings. +For example, default settings may not use more than a small +amount of periodic bandwidth. To use significant fractions +of bus bandwidth, drivers must select a non-default altsetting. + +Only one setting for an interface may be active at a time, and +only one driver may bind to an interface at a time. Most devices +have only one alternate setting per interface. + + +Endpoint descriptor info (can be multiple per Interface) +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + +:: + + E: Ad=xx(s) Atr=xx(ssss) MxPS=dddd Ivl=dddss + | | | | |__Interval (max) between transfers + | | | |__EndpointMaxPacketSize + | | |__Attributes(EndpointType) + | |__EndpointAddress(I=In,O=Out) + |__Endpoint info tag + +The interval is nonzero for all periodic (interrupt or isochronous) +endpoints. For high speed endpoints the transfer interval may be +measured in microseconds rather than milliseconds. + +For high speed periodic endpoints, the ``EndpointMaxPacketSize`` reflects +the per-microframe data transfer size. For "high bandwidth" +endpoints, that can reflect two or three packets (for up to +3KBytes every 125 usec) per endpoint. + +With the Linux-USB stack, periodic bandwidth reservations use the +transfer intervals and sizes provided by URBs, which can be less +than those found in endpoint descriptor. + +Usage examples +~~~~~~~~~~~~~~ + +If a user or script is interested only in Topology info, for +example, use something like ``grep ^T: /sys/kernel/debug/usb/devices`` +for only the Topology lines. A command like +``grep -i ^[tdp]: /sys/kernel/debug/usb/devices`` can be used to list +only the lines that begin with the characters in square brackets, +where the valid characters are TDPCIE. With a slightly more able +script, it can display any selected lines (for example, only T, D, +and P lines) and change their output format. (The ``procusb`` +Perl script is the beginning of this idea. It will list only +selected lines [selected from TBDPSCIE] or "All" lines from +``/sys/kernel/debug/usb/devices``.) + +The Topology lines can be used to generate a graphic/pictorial +of the USB devices on a system's root hub. (See more below +on how to do this.) + +The Interface lines can be used to determine what driver is +being used for each device, and which altsetting it activated. + +The Configuration lines could be used to list maximum power +(in milliamps) that a system's USB devices are using. +For example, ``grep ^C: /sys/kernel/debug/usb/devices``. + + +Here's an example, from a system which has a UHCI root hub, +an external hub connected to the root hub, and a mouse and +a serial converter connected to the external hub. + +:: + + T: Bus=00 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 2 + B: Alloc= 28/900 us ( 3%), #Int= 2, #Iso= 0 + D: Ver= 1.00 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 + P: Vendor=0000 ProdID=0000 Rev= 0.00 + S: Product=USB UHCI Root Hub + S: SerialNumber=dce0 + C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA + I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub + E: Ad=81(I) Atr=03(Int.) MxPS= 8 Ivl=255ms + + T: Bus=00 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 Spd=12 MxCh= 4 + D: Ver= 1.00 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 + P: Vendor=0451 ProdID=1446 Rev= 1.00 + C:* #Ifs= 1 Cfg#= 1 Atr=e0 MxPwr=100mA + I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub + E: Ad=81(I) Atr=03(Int.) MxPS= 1 Ivl=255ms + + T: Bus=00 Lev=02 Prnt=02 Port=00 Cnt=01 Dev#= 3 Spd=1.5 MxCh= 0 + D: Ver= 1.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 + P: Vendor=04b4 ProdID=0001 Rev= 0.00 + C:* #Ifs= 1 Cfg#= 1 Atr=80 MxPwr=100mA + I: If#= 0 Alt= 0 #EPs= 1 Cls=03(HID ) Sub=01 Prot=02 Driver=mouse + E: Ad=81(I) Atr=03(Int.) MxPS= 3 Ivl= 10ms + + T: Bus=00 Lev=02 Prnt=02 Port=02 Cnt=02 Dev#= 4 Spd=12 MxCh= 0 + D: Ver= 1.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 + P: Vendor=0565 ProdID=0001 Rev= 1.08 + S: Manufacturer=Peracom Networks, Inc. + S: Product=Peracom USB to Serial Converter + C:* #Ifs= 1 Cfg#= 1 Atr=a0 MxPwr=100mA + I: If#= 0 Alt= 0 #EPs= 3 Cls=00(>ifc ) Sub=00 Prot=00 Driver=serial + E: Ad=81(I) Atr=02(Bulk) MxPS= 64 Ivl= 16ms + E: Ad=01(O) Atr=02(Bulk) MxPS= 16 Ivl= 16ms + E: Ad=82(I) Atr=03(Int.) MxPS= 8 Ivl= 8ms + + +Selecting only the ``T:`` and ``I:`` lines from this (for example, by using +``procusb ti``), we have + +:: + + T: Bus=00 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 2 + T: Bus=00 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 Spd=12 MxCh= 4 + I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub + T: Bus=00 Lev=02 Prnt=02 Port=00 Cnt=01 Dev#= 3 Spd=1.5 MxCh= 0 + I: If#= 0 Alt= 0 #EPs= 1 Cls=03(HID ) Sub=01 Prot=02 Driver=mouse + T: Bus=00 Lev=02 Prnt=02 Port=02 Cnt=02 Dev#= 4 Spd=12 MxCh= 0 + I: If#= 0 Alt= 0 #EPs= 3 Cls=00(>ifc ) Sub=00 Prot=00 Driver=serial + + +Physically this looks like (or could be converted to):: + + +------------------+ + | PC/root_hub (12)| Dev# = 1 + +------------------+ (nn) is Mbps. + Level 0 | CN.0 | CN.1 | [CN = connector/port #] + +------------------+ + / + / + +-----------------------+ + Level 1 | Dev#2: 4-port hub (12)| + +-----------------------+ + |CN.0 |CN.1 |CN.2 |CN.3 | + +-----------------------+ + \ \____________________ + \_____ \ + \ \ + +--------------------+ +--------------------+ + Level 2 | Dev# 3: mouse (1.5)| | Dev# 4: serial (12)| + +--------------------+ +--------------------+ + + + +Or, in a more tree-like structure (ports [Connectors] without +connections could be omitted):: + + PC: Dev# 1, root hub, 2 ports, 12 Mbps + |_ CN.0: Dev# 2, hub, 4 ports, 12 Mbps + |_ CN.0: Dev #3, mouse, 1.5 Mbps + |_ CN.1: + |_ CN.2: Dev #4, serial, 12 Mbps + |_ CN.3: + |_ CN.1: diff --git a/Documentation/driver-api/usb/usb3-debug-port.rst b/Documentation/driver-api/usb/usb3-debug-port.rst new file mode 100644 index 000000000..b9fd131f4 --- /dev/null +++ b/Documentation/driver-api/usb/usb3-debug-port.rst @@ -0,0 +1,152 @@ +=============== +USB3 debug port +=============== + +:Author: Lu Baolu <baolu.lu@linux.intel.com> +:Date: March 2017 + +GENERAL +======= + +This is a HOWTO for using the USB3 debug port on x86 systems. + +Before using any kernel debugging functionality based on USB3 +debug port, you need to:: + + 1) check whether any USB3 debug port is available in + your system; + 2) check which port is used for debugging purposes; + 3) have a USB 3.0 super-speed A-to-A debugging cable. + +INTRODUCTION +============ + +The xHCI debug capability (DbC) is an optional but standalone +functionality provided by the xHCI host controller. The xHCI +specification describes DbC in the section 7.6. + +When DbC is initialized and enabled, it will present a debug +device through the debug port (normally the first USB3 +super-speed port). The debug device is fully compliant with +the USB framework and provides the equivalent of a very high +performance full-duplex serial link between the debug target +(the system under debugging) and a debug host. + +EARLY PRINTK +============ + +DbC has been designed to log early printk messages. One use for +this feature is kernel debugging. For example, when your machine +crashes very early before the regular console code is initialized. +Other uses include simpler, lockless logging instead of a full- +blown printk console driver and klogd. + +On the debug target system, you need to customize a debugging +kernel with CONFIG_EARLY_PRINTK_USB_XDBC enabled. And, add below +kernel boot parameter:: + + "earlyprintk=xdbc" + +If there are multiple xHCI controllers in your system, you can +append a host contoller index to this kernel parameter. This +index starts from 0. + +Current design doesn't support DbC runtime suspend/resume. As +the result, you'd better disable runtime power management for +USB subsystem by adding below kernel boot parameter:: + + "usbcore.autosuspend=-1" + +Before starting the debug target, you should connect the debug +port to a USB port (root port or port of any external hub) on +the debug host. The cable used to connect these two ports +should be a USB 3.0 super-speed A-to-A debugging cable. + +During early boot of the debug target, DbC will be detected and +initialized. After initialization, the debug host should be able +to enumerate the debug device in debug target. The debug host +will then bind the debug device with the usb_debug driver module +and create the /dev/ttyUSB device. + +If the debug device enumeration goes smoothly, you should be able +to see below kernel messages on the debug host:: + + # tail -f /var/log/kern.log + [ 1815.983374] usb 4-3: new SuperSpeed USB device number 4 using xhci_hcd + [ 1815.999595] usb 4-3: LPM exit latency is zeroed, disabling LPM. + [ 1815.999899] usb 4-3: New USB device found, idVendor=1d6b, idProduct=0004 + [ 1815.999902] usb 4-3: New USB device strings: Mfr=1, Product=2, SerialNumber=3 + [ 1815.999903] usb 4-3: Product: Remote GDB + [ 1815.999904] usb 4-3: Manufacturer: Linux + [ 1815.999905] usb 4-3: SerialNumber: 0001 + [ 1816.000240] usb_debug 4-3:1.0: xhci_dbc converter detected + [ 1816.000360] usb 4-3: xhci_dbc converter now attached to ttyUSB0 + +You can use any communication program, for example minicom, to +read and view the messages. Below simple bash scripts can help +you to check the sanity of the setup. + +.. code-block:: sh + + ===== start of bash scripts ============= + #!/bin/bash + + while true ; do + while [ ! -d /sys/class/tty/ttyUSB0 ] ; do + : + done + cat /dev/ttyUSB0 + done + ===== end of bash scripts =============== + +Serial TTY +========== + +The DbC support has been added to the xHCI driver. You can get a +debug device provided by the DbC at runtime. + +In order to use this, you need to make sure your kernel has been +configured to support USB_XHCI_DBGCAP. A sysfs attribute under +the xHCI device node is used to enable or disable DbC. By default, +DbC is disabled:: + + root@target:/sys/bus/pci/devices/0000:00:14.0# cat dbc + disabled + +Enable DbC with the following command:: + + root@target:/sys/bus/pci/devices/0000:00:14.0# echo enable > dbc + +You can check the DbC state at anytime:: + + root@target:/sys/bus/pci/devices/0000:00:14.0# cat dbc + enabled + +Connect the debug target to the debug host with a USB 3.0 super- +speed A-to-A debugging cable. You can see /dev/ttyDBC0 created +on the debug target. You will see below kernel message lines:: + + root@target: tail -f /var/log/kern.log + [ 182.730103] xhci_hcd 0000:00:14.0: DbC connected + [ 191.169420] xhci_hcd 0000:00:14.0: DbC configured + [ 191.169597] xhci_hcd 0000:00:14.0: DbC now attached to /dev/ttyDBC0 + +Accordingly, the DbC state has been brought up to:: + + root@target:/sys/bus/pci/devices/0000:00:14.0# cat dbc + configured + +On the debug host, you will see the debug device has been enumerated. +You will see below kernel message lines:: + + root@host: tail -f /var/log/kern.log + [ 79.454780] usb 2-2.1: new SuperSpeed USB device number 3 using xhci_hcd + [ 79.475003] usb 2-2.1: LPM exit latency is zeroed, disabling LPM. + [ 79.475389] usb 2-2.1: New USB device found, idVendor=1d6b, idProduct=0010 + [ 79.475390] usb 2-2.1: New USB device strings: Mfr=1, Product=2, SerialNumber=3 + [ 79.475391] usb 2-2.1: Product: Linux USB Debug Target + [ 79.475392] usb 2-2.1: Manufacturer: Linux Foundation + [ 79.475393] usb 2-2.1: SerialNumber: 0001 + +The debug device works now. You can use any communication or debugging +program to talk between the host and the target. diff --git a/Documentation/driver-api/usb/writing_musb_glue_layer.rst b/Documentation/driver-api/usb/writing_musb_glue_layer.rst new file mode 100644 index 000000000..5bf7152fd --- /dev/null +++ b/Documentation/driver-api/usb/writing_musb_glue_layer.rst @@ -0,0 +1,720 @@ +========================= +Writing a MUSB Glue Layer +========================= + +:Author: Apelete Seketeli + +Introduction +============ + +The Linux MUSB subsystem is part of the larger Linux USB subsystem. It +provides support for embedded USB Device Controllers (UDC) that do not +use Universal Host Controller Interface (UHCI) or Open Host Controller +Interface (OHCI). + +Instead, these embedded UDC rely on the USB On-the-Go (OTG) +specification which they implement at least partially. The silicon +reference design used in most cases is the Multipoint USB Highspeed +Dual-Role Controller (MUSB HDRC) found in the Mentor Graphics Inventra™ +design. + +As a self-taught exercise I have written an MUSB glue layer for the +Ingenic JZ4740 SoC, modelled after the many MUSB glue layers in the +kernel source tree. This layer can be found at +``drivers/usb/musb/jz4740.c``. In this documentation I will walk through the +basics of the ``jz4740.c`` glue layer, explaining the different pieces and +what needs to be done in order to write your own device glue layer. + +.. _musb-basics: + +Linux MUSB Basics +================= + +To get started on the topic, please read USB On-the-Go Basics (see +Resources) which provides an introduction of USB OTG operation at the +hardware level. A couple of wiki pages by Texas Instruments and Analog +Devices also provide an overview of the Linux kernel MUSB configuration, +albeit focused on some specific devices provided by these companies. +Finally, getting acquainted with the USB specification at USB home page +may come in handy, with practical instance provided through the Writing +USB Device Drivers documentation (again, see Resources). + +Linux USB stack is a layered architecture in which the MUSB controller +hardware sits at the lowest. The MUSB controller driver abstract the +MUSB controller hardware to the Linux USB stack:: + + ------------------------ + | | <------- drivers/usb/gadget + | Linux USB Core Stack | <------- drivers/usb/host + | | <------- drivers/usb/core + ------------------------ + ⬍ + -------------------------- + | | <------ drivers/usb/musb/musb_gadget.c + | MUSB Controller driver | <------ drivers/usb/musb/musb_host.c + | | <------ drivers/usb/musb/musb_core.c + -------------------------- + ⬍ + --------------------------------- + | MUSB Platform Specific Driver | + | | <-- drivers/usb/musb/jz4740.c + | aka "Glue Layer" | + --------------------------------- + ⬍ + --------------------------------- + | MUSB Controller Hardware | + --------------------------------- + +As outlined above, the glue layer is actually the platform specific code +sitting in between the controller driver and the controller hardware. + +Just like a Linux USB driver needs to register itself with the Linux USB +subsystem, the MUSB glue layer needs first to register itself with the +MUSB controller driver. This will allow the controller driver to know +about which device the glue layer supports and which functions to call +when a supported device is detected or released; remember we are talking +about an embedded controller chip here, so no insertion or removal at +run-time. + +All of this information is passed to the MUSB controller driver through +a :c:type:`platform_driver` structure defined in the glue layer as:: + + static struct platform_driver jz4740_driver = { + .probe = jz4740_probe, + .remove = jz4740_remove, + .driver = { + .name = "musb-jz4740", + }, + }; + +The probe and remove function pointers are called when a matching device +is detected and, respectively, released. The name string describes the +device supported by this glue layer. In the current case it matches a +platform_device structure declared in ``arch/mips/jz4740/platform.c``. Note +that we are not using device tree bindings here. + +In order to register itself to the controller driver, the glue layer +goes through a few steps, basically allocating the controller hardware +resources and initialising a couple of circuits. To do so, it needs to +keep track of the information used throughout these steps. This is done +by defining a private ``jz4740_glue`` structure:: + + struct jz4740_glue { + struct device *dev; + struct platform_device *musb; + struct clk *clk; + }; + + +The dev and musb members are both device structure variables. The first +one holds generic information about the device, since it's the basic +device structure, and the latter holds information more closely related +to the subsystem the device is registered to. The clk variable keeps +information related to the device clock operation. + +Let's go through the steps of the probe function that leads the glue +layer to register itself to the controller driver. + +.. note:: + + For the sake of readability each function will be split in logical + parts, each part being shown as if it was independent from the others. + +.. code-block:: c + :emphasize-lines: 8,12,18 + + static int jz4740_probe(struct platform_device *pdev) + { + struct platform_device *musb; + struct jz4740_glue *glue; + struct clk *clk; + int ret; + + glue = devm_kzalloc(&pdev->dev, sizeof(*glue), GFP_KERNEL); + if (!glue) + return -ENOMEM; + + musb = platform_device_alloc("musb-hdrc", PLATFORM_DEVID_AUTO); + if (!musb) { + dev_err(&pdev->dev, "failed to allocate musb device\n"); + return -ENOMEM; + } + + clk = devm_clk_get(&pdev->dev, "udc"); + if (IS_ERR(clk)) { + dev_err(&pdev->dev, "failed to get clock\n"); + ret = PTR_ERR(clk); + goto err_platform_device_put; + } + + ret = clk_prepare_enable(clk); + if (ret) { + dev_err(&pdev->dev, "failed to enable clock\n"); + goto err_platform_device_put; + } + + musb->dev.parent = &pdev->dev; + + glue->dev = &pdev->dev; + glue->musb = musb; + glue->clk = clk; + + return 0; + + err_platform_device_put: + platform_device_put(musb); + return ret; + } + +The first few lines of the probe function allocate and assign the glue, +musb and clk variables. The ``GFP_KERNEL`` flag (line 8) allows the +allocation process to sleep and wait for memory, thus being usable in a +locking situation. The ``PLATFORM_DEVID_AUTO`` flag (line 12) allows +automatic allocation and management of device IDs in order to avoid +device namespace collisions with explicit IDs. With :c:func:`devm_clk_get` +(line 18) the glue layer allocates the clock -- the ``devm_`` prefix +indicates that :c:func:`clk_get` is managed: it automatically frees the +allocated clock resource data when the device is released -- and enable +it. + + + +Then comes the registration steps: + +.. code-block:: c + :emphasize-lines: 3,5,7,9,16 + + static int jz4740_probe(struct platform_device *pdev) + { + struct musb_hdrc_platform_data *pdata = &jz4740_musb_platform_data; + + pdata->platform_ops = &jz4740_musb_ops; + + platform_set_drvdata(pdev, glue); + + ret = platform_device_add_resources(musb, pdev->resource, + pdev->num_resources); + if (ret) { + dev_err(&pdev->dev, "failed to add resources\n"); + goto err_clk_disable; + } + + ret = platform_device_add_data(musb, pdata, sizeof(*pdata)); + if (ret) { + dev_err(&pdev->dev, "failed to add platform_data\n"); + goto err_clk_disable; + } + + return 0; + + err_clk_disable: + clk_disable_unprepare(clk); + err_platform_device_put: + platform_device_put(musb); + return ret; + } + +The first step is to pass the device data privately held by the glue +layer on to the controller driver through :c:func:`platform_set_drvdata` +(line 7). Next is passing on the device resources information, also privately +held at that point, through :c:func:`platform_device_add_resources` (line 9). + +Finally comes passing on the platform specific data to the controller +driver (line 16). Platform data will be discussed in +:ref:`musb-dev-platform-data`, but here we are looking at the +``platform_ops`` function pointer (line 5) in ``musb_hdrc_platform_data`` +structure (line 3). This function pointer allows the MUSB controller +driver to know which function to call for device operation:: + + static const struct musb_platform_ops jz4740_musb_ops = { + .init = jz4740_musb_init, + .exit = jz4740_musb_exit, + }; + +Here we have the minimal case where only init and exit functions are +called by the controller driver when needed. Fact is the JZ4740 MUSB +controller is a basic controller, lacking some features found in other +controllers, otherwise we may also have pointers to a few other +functions like a power management function or a function to switch +between OTG and non-OTG modes, for instance. + +At that point of the registration process, the controller driver +actually calls the init function: + + .. code-block:: c + :emphasize-lines: 12,14 + + static int jz4740_musb_init(struct musb *musb) + { + musb->xceiv = usb_get_phy(USB_PHY_TYPE_USB2); + if (!musb->xceiv) { + pr_err("HS UDC: no transceiver configured\n"); + return -ENODEV; + } + + /* Silicon does not implement ConfigData register. + * Set dyn_fifo to avoid reading EP config from hardware. + */ + musb->dyn_fifo = true; + + musb->isr = jz4740_musb_interrupt; + + return 0; + } + +The goal of ``jz4740_musb_init()`` is to get hold of the transceiver +driver data of the MUSB controller hardware and pass it on to the MUSB +controller driver, as usual. The transceiver is the circuitry inside the +controller hardware responsible for sending/receiving the USB data. +Since it is an implementation of the physical layer of the OSI model, +the transceiver is also referred to as PHY. + +Getting hold of the ``MUSB PHY`` driver data is done with ``usb_get_phy()`` +which returns a pointer to the structure containing the driver instance +data. The next couple of instructions (line 12 and 14) are used as a +quirk and to setup IRQ handling respectively. Quirks and IRQ handling +will be discussed later in :ref:`musb-dev-quirks` and +:ref:`musb-handling-irqs`\ :: + + static int jz4740_musb_exit(struct musb *musb) + { + usb_put_phy(musb->xceiv); + + return 0; + } + +Acting as the counterpart of init, the exit function releases the MUSB +PHY driver when the controller hardware itself is about to be released. + +Again, note that init and exit are fairly simple in this case due to the +basic set of features of the JZ4740 controller hardware. When writing an +musb glue layer for a more complex controller hardware, you might need +to take care of more processing in those two functions. + +Returning from the init function, the MUSB controller driver jumps back +into the probe function:: + + static int jz4740_probe(struct platform_device *pdev) + { + ret = platform_device_add(musb); + if (ret) { + dev_err(&pdev->dev, "failed to register musb device\n"); + goto err_clk_disable; + } + + return 0; + + err_clk_disable: + clk_disable_unprepare(clk); + err_platform_device_put: + platform_device_put(musb); + return ret; + } + +This is the last part of the device registration process where the glue +layer adds the controller hardware device to Linux kernel device +hierarchy: at this stage, all known information about the device is +passed on to the Linux USB core stack: + + .. code-block:: c + :emphasize-lines: 5,6 + + static int jz4740_remove(struct platform_device *pdev) + { + struct jz4740_glue *glue = platform_get_drvdata(pdev); + + platform_device_unregister(glue->musb); + clk_disable_unprepare(glue->clk); + + return 0; + } + +Acting as the counterpart of probe, the remove function unregister the +MUSB controller hardware (line 5) and disable the clock (line 6), +allowing it to be gated. + +.. _musb-handling-irqs: + +Handling IRQs +============= + +Additionally to the MUSB controller hardware basic setup and +registration, the glue layer is also responsible for handling the IRQs: + + .. code-block:: c + :emphasize-lines: 7,9-11,14,24 + + static irqreturn_t jz4740_musb_interrupt(int irq, void *__hci) + { + unsigned long flags; + irqreturn_t retval = IRQ_NONE; + struct musb *musb = __hci; + + spin_lock_irqsave(&musb->lock, flags); + + musb->int_usb = musb_readb(musb->mregs, MUSB_INTRUSB); + musb->int_tx = musb_readw(musb->mregs, MUSB_INTRTX); + musb->int_rx = musb_readw(musb->mregs, MUSB_INTRRX); + + /* + * The controller is gadget only, the state of the host mode IRQ bits is + * undefined. Mask them to make sure that the musb driver core will + * never see them set + */ + musb->int_usb &= MUSB_INTR_SUSPEND | MUSB_INTR_RESUME | + MUSB_INTR_RESET | MUSB_INTR_SOF; + + if (musb->int_usb || musb->int_tx || musb->int_rx) + retval = musb_interrupt(musb); + + spin_unlock_irqrestore(&musb->lock, flags); + + return retval; + } + +Here the glue layer mostly has to read the relevant hardware registers +and pass their values on to the controller driver which will handle the +actual event that triggered the IRQ. + +The interrupt handler critical section is protected by the +:c:func:`spin_lock_irqsave` and counterpart :c:func:`spin_unlock_irqrestore` +functions (line 7 and 24 respectively), which prevent the interrupt +handler code to be run by two different threads at the same time. + +Then the relevant interrupt registers are read (line 9 to 11): + +- ``MUSB_INTRUSB``: indicates which USB interrupts are currently active, + +- ``MUSB_INTRTX``: indicates which of the interrupts for TX endpoints are + currently active, + +- ``MUSB_INTRRX``: indicates which of the interrupts for TX endpoints are + currently active. + +Note that :c:func:`musb_readb` is used to read 8-bit registers at most, while +:c:func:`musb_readw` allows us to read at most 16-bit registers. There are +other functions that can be used depending on the size of your device +registers. See ``musb_io.h`` for more information. + +Instruction on line 18 is another quirk specific to the JZ4740 USB +device controller, which will be discussed later in :ref:`musb-dev-quirks`. + +The glue layer still needs to register the IRQ handler though. Remember +the instruction on line 14 of the init function:: + + static int jz4740_musb_init(struct musb *musb) + { + musb->isr = jz4740_musb_interrupt; + + return 0; + } + +This instruction sets a pointer to the glue layer IRQ handler function, +in order for the controller hardware to call the handler back when an +IRQ comes from the controller hardware. The interrupt handler is now +implemented and registered. + +.. _musb-dev-platform-data: + +Device Platform Data +==================== + +In order to write an MUSB glue layer, you need to have some data +describing the hardware capabilities of your controller hardware, which +is called the platform data. + +Platform data is specific to your hardware, though it may cover a broad +range of devices, and is generally found somewhere in the ``arch/`` +directory, depending on your device architecture. + +For instance, platform data for the JZ4740 SoC is found in +``arch/mips/jz4740/platform.c``. In the ``platform.c`` file each device of the +JZ4740 SoC is described through a set of structures. + +Here is the part of ``arch/mips/jz4740/platform.c`` that covers the USB +Device Controller (UDC): + + .. code-block:: c + :emphasize-lines: 2,7,14-17,21,22,25,26,28,29 + + /* USB Device Controller */ + struct platform_device jz4740_udc_xceiv_device = { + .name = "usb_phy_gen_xceiv", + .id = 0, + }; + + static struct resource jz4740_udc_resources[] = { + [0] = { + .start = JZ4740_UDC_BASE_ADDR, + .end = JZ4740_UDC_BASE_ADDR + 0x10000 - 1, + .flags = IORESOURCE_MEM, + }, + [1] = { + .start = JZ4740_IRQ_UDC, + .end = JZ4740_IRQ_UDC, + .flags = IORESOURCE_IRQ, + .name = "mc", + }, + }; + + struct platform_device jz4740_udc_device = { + .name = "musb-jz4740", + .id = -1, + .dev = { + .dma_mask = &jz4740_udc_device.dev.coherent_dma_mask, + .coherent_dma_mask = DMA_BIT_MASK(32), + }, + .num_resources = ARRAY_SIZE(jz4740_udc_resources), + .resource = jz4740_udc_resources, + }; + +The ``jz4740_udc_xceiv_device`` platform device structure (line 2) +describes the UDC transceiver with a name and id number. + +At the time of this writing, note that ``usb_phy_gen_xceiv`` is the +specific name to be used for all transceivers that are either built-in +with reference USB IP or autonomous and doesn't require any PHY +programming. You will need to set ``CONFIG_NOP_USB_XCEIV=y`` in the +kernel configuration to make use of the corresponding transceiver +driver. The id field could be set to -1 (equivalent to +``PLATFORM_DEVID_NONE``), -2 (equivalent to ``PLATFORM_DEVID_AUTO``) or +start with 0 for the first device of this kind if we want a specific id +number. + +The ``jz4740_udc_resources`` resource structure (line 7) defines the UDC +registers base addresses. + +The first array (line 9 to 11) defines the UDC registers base memory +addresses: start points to the first register memory address, end points +to the last register memory address and the flags member defines the +type of resource we are dealing with. So ``IORESOURCE_MEM`` is used to +define the registers memory addresses. The second array (line 14 to 17) +defines the UDC IRQ registers addresses. Since there is only one IRQ +register available for the JZ4740 UDC, start and end point at the same +address. The ``IORESOURCE_IRQ`` flag tells that we are dealing with IRQ +resources, and the name ``mc`` is in fact hard-coded in the MUSB core in +order for the controller driver to retrieve this IRQ resource by +querying it by its name. + +Finally, the ``jz4740_udc_device`` platform device structure (line 21) +describes the UDC itself. + +The ``musb-jz4740`` name (line 22) defines the MUSB driver that is used +for this device; remember this is in fact the name that we used in the +``jz4740_driver`` platform driver structure in :ref:`musb-basics`. +The id field (line 23) is set to -1 (equivalent to ``PLATFORM_DEVID_NONE``) +since we do not need an id for the device: the MUSB controller driver was +already set to allocate an automatic id in :ref:`musb-basics`. In the dev field +we care for DMA related information here. The ``dma_mask`` field (line 25) +defines the width of the DMA mask that is going to be used, and +``coherent_dma_mask`` (line 26) has the same purpose but for the +``alloc_coherent`` DMA mappings: in both cases we are using a 32 bits mask. +Then the resource field (line 29) is simply a pointer to the resource +structure defined before, while the ``num_resources`` field (line 28) keeps +track of the number of arrays defined in the resource structure (in this +case there were two resource arrays defined before). + +With this quick overview of the UDC platform data at the ``arch/`` level now +done, let's get back to the MUSB glue layer specific platform data in +``drivers/usb/musb/jz4740.c``: + + .. code-block:: c + :emphasize-lines: 3,5,7-9,11 + + static struct musb_hdrc_config jz4740_musb_config = { + /* Silicon does not implement USB OTG. */ + .multipoint = 0, + /* Max EPs scanned, driver will decide which EP can be used. */ + .num_eps = 4, + /* RAMbits needed to configure EPs from table */ + .ram_bits = 9, + .fifo_cfg = jz4740_musb_fifo_cfg, + .fifo_cfg_size = ARRAY_SIZE(jz4740_musb_fifo_cfg), + }; + + static struct musb_hdrc_platform_data jz4740_musb_platform_data = { + .mode = MUSB_PERIPHERAL, + .config = &jz4740_musb_config, + }; + +First the glue layer configures some aspects of the controller driver +operation related to the controller hardware specifics. This is done +through the ``jz4740_musb_config`` :c:type:`musb_hdrc_config` structure. + +Defining the OTG capability of the controller hardware, the multipoint +member (line 3) is set to 0 (equivalent to false) since the JZ4740 UDC +is not OTG compatible. Then ``num_eps`` (line 5) defines the number of USB +endpoints of the controller hardware, including endpoint 0: here we have +3 endpoints + endpoint 0. Next is ``ram_bits`` (line 7) which is the width +of the RAM address bus for the MUSB controller hardware. This +information is needed when the controller driver cannot automatically +configure endpoints by reading the relevant controller hardware +registers. This issue will be discussed when we get to device quirks in +:ref:`musb-dev-quirks`. Last two fields (line 8 and 9) are also +about device quirks: ``fifo_cfg`` points to the USB endpoints configuration +table and ``fifo_cfg_size`` keeps track of the size of the number of +entries in that configuration table. More on that later in +:ref:`musb-dev-quirks`. + +Then this configuration is embedded inside ``jz4740_musb_platform_data`` +:c:type:`musb_hdrc_platform_data` structure (line 11): config is a pointer to +the configuration structure itself, and mode tells the controller driver +if the controller hardware may be used as ``MUSB_HOST`` only, +``MUSB_PERIPHERAL`` only or ``MUSB_OTG`` which is a dual mode. + +Remember that ``jz4740_musb_platform_data`` is then used to convey +platform data information as we have seen in the probe function in +:ref:`musb-basics`. + +.. _musb-dev-quirks: + +Device Quirks +============= + +Completing the platform data specific to your device, you may also need +to write some code in the glue layer to work around some device specific +limitations. These quirks may be due to some hardware bugs, or simply be +the result of an incomplete implementation of the USB On-the-Go +specification. + +The JZ4740 UDC exhibits such quirks, some of which we will discuss here +for the sake of insight even though these might not be found in the +controller hardware you are working on. + +Let's get back to the init function first: + + .. code-block:: c + :emphasize-lines: 12 + + static int jz4740_musb_init(struct musb *musb) + { + musb->xceiv = usb_get_phy(USB_PHY_TYPE_USB2); + if (!musb->xceiv) { + pr_err("HS UDC: no transceiver configured\n"); + return -ENODEV; + } + + /* Silicon does not implement ConfigData register. + * Set dyn_fifo to avoid reading EP config from hardware. + */ + musb->dyn_fifo = true; + + musb->isr = jz4740_musb_interrupt; + + return 0; + } + +Instruction on line 12 helps the MUSB controller driver to work around +the fact that the controller hardware is missing registers that are used +for USB endpoints configuration. + +Without these registers, the controller driver is unable to read the +endpoints configuration from the hardware, so we use line 12 instruction +to bypass reading the configuration from silicon, and rely on a +hard-coded table that describes the endpoints configuration instead:: + + static struct musb_fifo_cfg jz4740_musb_fifo_cfg[] = { + { .hw_ep_num = 1, .style = FIFO_TX, .maxpacket = 512, }, + { .hw_ep_num = 1, .style = FIFO_RX, .maxpacket = 512, }, + { .hw_ep_num = 2, .style = FIFO_TX, .maxpacket = 64, }, + }; + +Looking at the configuration table above, we see that each endpoints is +described by three fields: ``hw_ep_num`` is the endpoint number, style is +its direction (either ``FIFO_TX`` for the controller driver to send packets +in the controller hardware, or ``FIFO_RX`` to receive packets from +hardware), and maxpacket defines the maximum size of each data packet +that can be transmitted over that endpoint. Reading from the table, the +controller driver knows that endpoint 1 can be used to send and receive +USB data packets of 512 bytes at once (this is in fact a bulk in/out +endpoint), and endpoint 2 can be used to send data packets of 64 bytes +at once (this is in fact an interrupt endpoint). + +Note that there is no information about endpoint 0 here: that one is +implemented by default in every silicon design, with a predefined +configuration according to the USB specification. For more examples of +endpoint configuration tables, see ``musb_core.c``. + +Let's now get back to the interrupt handler function: + + .. code-block:: c + :emphasize-lines: 18-19 + + static irqreturn_t jz4740_musb_interrupt(int irq, void *__hci) + { + unsigned long flags; + irqreturn_t retval = IRQ_NONE; + struct musb *musb = __hci; + + spin_lock_irqsave(&musb->lock, flags); + + musb->int_usb = musb_readb(musb->mregs, MUSB_INTRUSB); + musb->int_tx = musb_readw(musb->mregs, MUSB_INTRTX); + musb->int_rx = musb_readw(musb->mregs, MUSB_INTRRX); + + /* + * The controller is gadget only, the state of the host mode IRQ bits is + * undefined. Mask them to make sure that the musb driver core will + * never see them set + */ + musb->int_usb &= MUSB_INTR_SUSPEND | MUSB_INTR_RESUME | + MUSB_INTR_RESET | MUSB_INTR_SOF; + + if (musb->int_usb || musb->int_tx || musb->int_rx) + retval = musb_interrupt(musb); + + spin_unlock_irqrestore(&musb->lock, flags); + + return retval; + } + +Instruction on line 18 above is a way for the controller driver to work +around the fact that some interrupt bits used for USB host mode +operation are missing in the ``MUSB_INTRUSB`` register, thus left in an +undefined hardware state, since this MUSB controller hardware is used in +peripheral mode only. As a consequence, the glue layer masks these +missing bits out to avoid parasite interrupts by doing a logical AND +operation between the value read from ``MUSB_INTRUSB`` and the bits that +are actually implemented in the register. + +These are only a couple of the quirks found in the JZ4740 USB device +controller. Some others were directly addressed in the MUSB core since +the fixes were generic enough to provide a better handling of the issues +for others controller hardware eventually. + +Conclusion +========== + +Writing a Linux MUSB glue layer should be a more accessible task, as +this documentation tries to show the ins and outs of this exercise. + +The JZ4740 USB device controller being fairly simple, I hope its glue +layer serves as a good example for the curious mind. Used with the +current MUSB glue layers, this documentation should provide enough +guidance to get started; should anything gets out of hand, the linux-usb +mailing list archive is another helpful resource to browse through. + +Acknowledgements +================ + +Many thanks to Lars-Peter Clausen and Maarten ter Huurne for answering +my questions while I was writing the JZ4740 glue layer and for helping +me out getting the code in good shape. + +I would also like to thank the Qi-Hardware community at large for its +cheerful guidance and support. + +Resources +========= + +USB Home Page: http://www.usb.org + +linux-usb Mailing List Archives: http://marc.info/?l=linux-usb + +USB On-the-Go Basics: +http://www.maximintegrated.com/app-notes/index.mvp/id/1822 + +:ref:`Writing USB Device Drivers <writing-usb-driver>` + +Texas Instruments USB Configuration Wiki Page: +http://processors.wiki.ti.com/index.php/Usbgeneralpage diff --git a/Documentation/driver-api/usb/writing_usb_driver.rst b/Documentation/driver-api/usb/writing_usb_driver.rst new file mode 100644 index 000000000..4fe1c06b6 --- /dev/null +++ b/Documentation/driver-api/usb/writing_usb_driver.rst @@ -0,0 +1,326 @@ +.. _writing-usb-driver: + +========================== +Writing USB Device Drivers +========================== + +:Author: Greg Kroah-Hartman + +Introduction +============ + +The Linux USB subsystem has grown from supporting only two different +types of devices in the 2.2.7 kernel (mice and keyboards), to over 20 +different types of devices in the 2.4 kernel. Linux currently supports +almost all USB class devices (standard types of devices like keyboards, +mice, modems, printers and speakers) and an ever-growing number of +vendor-specific devices (such as USB to serial converters, digital +cameras, Ethernet devices and MP3 players). For a full list of the +different USB devices currently supported, see Resources. + +The remaining kinds of USB devices that do not have support on Linux are +almost all vendor-specific devices. Each vendor decides to implement a +custom protocol to talk to their device, so a custom driver usually +needs to be created. Some vendors are open with their USB protocols and +help with the creation of Linux drivers, while others do not publish +them, and developers are forced to reverse-engineer. See Resources for +some links to handy reverse-engineering tools. + +Because each different protocol causes a new driver to be created, I +have written a generic USB driver skeleton, modelled after the +pci-skeleton.c file in the kernel source tree upon which many PCI +network drivers have been based. This USB skeleton can be found at +drivers/usb/usb-skeleton.c in the kernel source tree. In this article I +will walk through the basics of the skeleton driver, explaining the +different pieces and what needs to be done to customize it to your +specific device. + +Linux USB Basics +================ + +If you are going to write a Linux USB driver, please become familiar +with the USB protocol specification. It can be found, along with many +other useful documents, at the USB home page (see Resources). An +excellent introduction to the Linux USB subsystem can be found at the +USB Working Devices List (see Resources). It explains how the Linux USB +subsystem is structured and introduces the reader to the concept of USB +urbs (USB Request Blocks), which are essential to USB drivers. + +The first thing a Linux USB driver needs to do is register itself with +the Linux USB subsystem, giving it some information about which devices +the driver supports and which functions to call when a device supported +by the driver is inserted or removed from the system. All of this +information is passed to the USB subsystem in the :c:type:`usb_driver` +structure. The skeleton driver declares a :c:type:`usb_driver` as:: + + static struct usb_driver skel_driver = { + .name = "skeleton", + .probe = skel_probe, + .disconnect = skel_disconnect, + .fops = &skel_fops, + .minor = USB_SKEL_MINOR_BASE, + .id_table = skel_table, + }; + + +The variable name is a string that describes the driver. It is used in +informational messages printed to the system log. The probe and +disconnect function pointers are called when a device that matches the +information provided in the ``id_table`` variable is either seen or +removed. + +The fops and minor variables are optional. Most USB drivers hook into +another kernel subsystem, such as the SCSI, network or TTY subsystem. +These types of drivers register themselves with the other kernel +subsystem, and any user-space interactions are provided through that +interface. But for drivers that do not have a matching kernel subsystem, +such as MP3 players or scanners, a method of interacting with user space +is needed. The USB subsystem provides a way to register a minor device +number and a set of :c:type:`file_operations` function pointers that enable +this user-space interaction. The skeleton driver needs this kind of +interface, so it provides a minor starting number and a pointer to its +:c:type:`file_operations` functions. + +The USB driver is then registered with a call to :c:func:`usb_register`, +usually in the driver's init function, as shown here:: + + static int __init usb_skel_init(void) + { + int result; + + /* register this driver with the USB subsystem */ + result = usb_register(&skel_driver); + if (result < 0) { + err("usb_register failed for the "__FILE__ "driver." + "Error number %d", result); + return -1; + } + + return 0; + } + module_init(usb_skel_init); + + +When the driver is unloaded from the system, it needs to deregister +itself with the USB subsystem. This is done with the :c:func:`usb_deregister` +function:: + + static void __exit usb_skel_exit(void) + { + /* deregister this driver with the USB subsystem */ + usb_deregister(&skel_driver); + } + module_exit(usb_skel_exit); + + +To enable the linux-hotplug system to load the driver automatically when +the device is plugged in, you need to create a ``MODULE_DEVICE_TABLE``. +The following code tells the hotplug scripts that this module supports a +single device with a specific vendor and product ID:: + + /* table of devices that work with this driver */ + static struct usb_device_id skel_table [] = { + { USB_DEVICE(USB_SKEL_VENDOR_ID, USB_SKEL_PRODUCT_ID) }, + { } /* Terminating entry */ + }; + MODULE_DEVICE_TABLE (usb, skel_table); + + +There are other macros that can be used in describing a struct +:c:type:`usb_device_id` for drivers that support a whole class of USB +drivers. See :ref:`usb.h <usb_header>` for more information on this. + +Device operation +================ + +When a device is plugged into the USB bus that matches the device ID +pattern that your driver registered with the USB core, the probe +function is called. The :c:type:`usb_device` structure, interface number and +the interface ID are passed to the function:: + + static int skel_probe(struct usb_interface *interface, + const struct usb_device_id *id) + + +The driver now needs to verify that this device is actually one that it +can accept. If so, it returns 0. If not, or if any error occurs during +initialization, an errorcode (such as ``-ENOMEM`` or ``-ENODEV``) is +returned from the probe function. + +In the skeleton driver, we determine what end points are marked as +bulk-in and bulk-out. We create buffers to hold the data that will be +sent and received from the device, and a USB urb to write data to the +device is initialized. + +Conversely, when the device is removed from the USB bus, the disconnect +function is called with the device pointer. The driver needs to clean +any private data that has been allocated at this time and to shut down +any pending urbs that are in the USB system. + +Now that the device is plugged into the system and the driver is bound +to the device, any of the functions in the :c:type:`file_operations` structure +that were passed to the USB subsystem will be called from a user program +trying to talk to the device. The first function called will be open, as +the program tries to open the device for I/O. We increment our private +usage count and save a pointer to our internal structure in the file +structure. This is done so that future calls to file operations will +enable the driver to determine which device the user is addressing. All +of this is done with the following code:: + + /* increment our usage count for the module */ + ++skel->open_count; + + /* save our object in the file's private structure */ + file->private_data = dev; + + +After the open function is called, the read and write functions are +called to receive and send data to the device. In the ``skel_write`` +function, we receive a pointer to some data that the user wants to send +to the device and the size of the data. The function determines how much +data it can send to the device based on the size of the write urb it has +created (this size depends on the size of the bulk out end point that +the device has). Then it copies the data from user space to kernel +space, points the urb to the data and submits the urb to the USB +subsystem. This can be seen in the following code:: + + /* we can only write as much as 1 urb will hold */ + bytes_written = (count > skel->bulk_out_size) ? skel->bulk_out_size : count; + + /* copy the data from user space into our urb */ + copy_from_user(skel->write_urb->transfer_buffer, buffer, bytes_written); + + /* set up our urb */ + usb_fill_bulk_urb(skel->write_urb, + skel->dev, + usb_sndbulkpipe(skel->dev, skel->bulk_out_endpointAddr), + skel->write_urb->transfer_buffer, + bytes_written, + skel_write_bulk_callback, + skel); + + /* send the data out the bulk port */ + result = usb_submit_urb(skel->write_urb); + if (result) { + err("Failed submitting write urb, error %d", result); + } + + +When the write urb is filled up with the proper information using the +:c:func:`usb_fill_bulk_urb` function, we point the urb's completion callback +to call our own ``skel_write_bulk_callback`` function. This function is +called when the urb is finished by the USB subsystem. The callback +function is called in interrupt context, so caution must be taken not to +do very much processing at that time. Our implementation of +``skel_write_bulk_callback`` merely reports if the urb was completed +successfully or not and then returns. + +The read function works a bit differently from the write function in +that we do not use an urb to transfer data from the device to the +driver. Instead we call the :c:func:`usb_bulk_msg` function, which can be used +to send or receive data from a device without having to create urbs and +handle urb completion callback functions. We call the :c:func:`usb_bulk_msg` +function, giving it a buffer into which to place any data received from +the device and a timeout value. If the timeout period expires without +receiving any data from the device, the function will fail and return an +error message. This can be shown with the following code:: + + /* do an immediate bulk read to get data from the device */ + retval = usb_bulk_msg (skel->dev, + usb_rcvbulkpipe (skel->dev, + skel->bulk_in_endpointAddr), + skel->bulk_in_buffer, + skel->bulk_in_size, + &count, HZ*10); + /* if the read was successful, copy the data to user space */ + if (!retval) { + if (copy_to_user (buffer, skel->bulk_in_buffer, count)) + retval = -EFAULT; + else + retval = count; + } + + +The :c:func:`usb_bulk_msg` function can be very useful for doing single reads +or writes to a device; however, if you need to read or write constantly to +a device, it is recommended to set up your own urbs and submit them to +the USB subsystem. + +When the user program releases the file handle that it has been using to +talk to the device, the release function in the driver is called. In +this function we decrement our private usage count and wait for possible +pending writes:: + + /* decrement our usage count for the device */ + --skel->open_count; + + +One of the more difficult problems that USB drivers must be able to +handle smoothly is the fact that the USB device may be removed from the +system at any point in time, even if a program is currently talking to +it. It needs to be able to shut down any current reads and writes and +notify the user-space programs that the device is no longer there. The +following code (function ``skel_delete``) is an example of how to do +this:: + + static inline void skel_delete (struct usb_skel *dev) + { + kfree (dev->bulk_in_buffer); + if (dev->bulk_out_buffer != NULL) + usb_free_coherent (dev->udev, dev->bulk_out_size, + dev->bulk_out_buffer, + dev->write_urb->transfer_dma); + usb_free_urb (dev->write_urb); + kfree (dev); + } + + +If a program currently has an open handle to the device, we reset the +flag ``device_present``. For every read, write, release and other +functions that expect a device to be present, the driver first checks +this flag to see if the device is still present. If not, it releases +that the device has disappeared, and a ``-ENODEV`` error is returned to the +user-space program. When the release function is eventually called, it +determines if there is no device and if not, it does the cleanup that +the ``skel_disconnect`` function normally does if there are no open files +on the device (see Listing 5). + +Isochronous Data +================ + +This usb-skeleton driver does not have any examples of interrupt or +isochronous data being sent to or from the device. Interrupt data is +sent almost exactly as bulk data is, with a few minor exceptions. +Isochronous data works differently with continuous streams of data being +sent to or from the device. The audio and video camera drivers are very +good examples of drivers that handle isochronous data and will be useful +if you also need to do this. + +Conclusion +========== + +Writing Linux USB device drivers is not a difficult task as the +usb-skeleton driver shows. This driver, combined with the other current +USB drivers, should provide enough examples to help a beginning author +create a working driver in a minimal amount of time. The linux-usb-devel +mailing list archives also contain a lot of helpful information. + +Resources +========= + +The Linux USB Project: +http://www.linux-usb.org/ + +Linux Hotplug Project: +http://linux-hotplug.sourceforge.net/ + +Linux USB Working Devices List: +http://www.qbik.ch/usb/devices/ + +linux-usb-devel Mailing List Archives: +http://marc.theaimsgroup.com/?l=linux-usb-devel + +Programming Guide for Linux USB Device Drivers: +http://lmu.web.psi.ch/docu/manuals/software_manuals/linux_sl/usb_linux_programming_guide.pdf + +USB Home Page: http://www.usb.org diff --git a/Documentation/driver-api/vme.rst b/Documentation/driver-api/vme.rst new file mode 100644 index 000000000..def139c13 --- /dev/null +++ b/Documentation/driver-api/vme.rst @@ -0,0 +1,297 @@ +VME Device Drivers +================== + +Driver registration +------------------- + +As with other subsystems within the Linux kernel, VME device drivers register +with the VME subsystem, typically called from the devices init routine. This is +achieved via a call to :c:func:`vme_register_driver`. + +A pointer to a structure of type :c:type:`struct vme_driver <vme_driver>` must +be provided to the registration function. Along with the maximum number of +devices your driver is able to support. + +At the minimum, the '.name', '.match' and '.probe' elements of +:c:type:`struct vme_driver <vme_driver>` should be correctly set. The '.name' +element is a pointer to a string holding the device driver's name. + +The '.match' function allows control over which VME devices should be registered +with the driver. The match function should return 1 if a device should be +probed and 0 otherwise. This example match function (from vme_user.c) limits +the number of devices probed to one: + +.. code-block:: c + + #define USER_BUS_MAX 1 + ... + static int vme_user_match(struct vme_dev *vdev) + { + if (vdev->id.num >= USER_BUS_MAX) + return 0; + return 1; + } + +The '.probe' element should contain a pointer to the probe routine. The +probe routine is passed a :c:type:`struct vme_dev <vme_dev>` pointer as an +argument. + +Here, the 'num' field refers to the sequential device ID for this specific +driver. The bridge number (or bus number) can be accessed using +dev->bridge->num. + +A function is also provided to unregister the driver from the VME core called +:c:func:`vme_unregister_driver` and should usually be called from the device +driver's exit routine. + + +Resource management +------------------- + +Once a driver has registered with the VME core the provided match routine will +be called the number of times specified during the registration. If a match +succeeds, a non-zero value should be returned. A zero return value indicates +failure. For all successful matches, the probe routine of the corresponding +driver is called. The probe routine is passed a pointer to the devices +device structure. This pointer should be saved, it will be required for +requesting VME resources. + +The driver can request ownership of one or more master windows +(:c:func:`vme_master_request`), slave windows (:c:func:`vme_slave_request`) +and/or dma channels (:c:func:`vme_dma_request`). Rather than allowing the device +driver to request a specific window or DMA channel (which may be used by a +different driver) the API allows a resource to be assigned based on the required +attributes of the driver in question. For slave windows these attributes are +split into the VME address spaces that need to be accessed in 'aspace' and VME +bus cycle types required in 'cycle'. Master windows add a further set of +attributes in 'width' specifying the required data transfer widths. These +attributes are defined as bitmasks and as such any combination of the +attributes can be requested for a single window, the core will assign a window +that meets the requirements, returning a pointer of type vme_resource that +should be used to identify the allocated resource when it is used. For DMA +controllers, the request function requires the potential direction of any +transfers to be provided in the route attributes. This is typically VME-to-MEM +and/or MEM-to-VME, though some hardware can support VME-to-VME and MEM-to-MEM +transfers as well as test pattern generation. If an unallocated window fitting +the requirements can not be found a NULL pointer will be returned. + +Functions are also provided to free window allocations once they are no longer +required. These functions (:c:func:`vme_master_free`, :c:func:`vme_slave_free` +and :c:func:`vme_dma_free`) should be passed the pointer to the resource +provided during resource allocation. + + +Master windows +-------------- + +Master windows provide access from the local processor[s] out onto the VME bus. +The number of windows available and the available access modes is dependent on +the underlying chipset. A window must be configured before it can be used. + + +Master window configuration +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Once a master window has been assigned :c:func:`vme_master_set` can be used to +configure it and :c:func:`vme_master_get` to retrieve the current settings. The +address spaces, transfer widths and cycle types are the same as described +under resource management, however some of the options are mutually exclusive. +For example, only one address space may be specified. + + +Master window access +~~~~~~~~~~~~~~~~~~~~ + +The function :c:func:`vme_master_read` can be used to read from and +:c:func:`vme_master_write` used to write to configured master windows. + +In addition to simple reads and writes, :c:func:`vme_master_rmw` is provided to +do a read-modify-write transaction. Parts of a VME window can also be mapped +into user space memory using :c:func:`vme_master_mmap`. + + +Slave windows +------------- + +Slave windows provide devices on the VME bus access into mapped portions of the +local memory. The number of windows available and the access modes that can be +used is dependent on the underlying chipset. A window must be configured before +it can be used. + + +Slave window configuration +~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Once a slave window has been assigned :c:func:`vme_slave_set` can be used to +configure it and :c:func:`vme_slave_get` to retrieve the current settings. + +The address spaces, transfer widths and cycle types are the same as described +under resource management, however some of the options are mutually exclusive. +For example, only one address space may be specified. + + +Slave window buffer allocation +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Functions are provided to allow the user to allocate +(:c:func:`vme_alloc_consistent`) and free (:c:func:`vme_free_consistent`) +contiguous buffers which will be accessible by the VME bridge. These functions +do not have to be used, other methods can be used to allocate a buffer, though +care must be taken to ensure that they are contiguous and accessible by the VME +bridge. + + +Slave window access +~~~~~~~~~~~~~~~~~~~ + +Slave windows map local memory onto the VME bus, the standard methods for +accessing memory should be used. + + +DMA channels +------------ + +The VME DMA transfer provides the ability to run link-list DMA transfers. The +API introduces the concept of DMA lists. Each DMA list is a link-list which can +be passed to a DMA controller. Multiple lists can be created, extended, +executed, reused and destroyed. + + +List Management +~~~~~~~~~~~~~~~ + +The function :c:func:`vme_new_dma_list` is provided to create and +:c:func:`vme_dma_list_free` to destroy DMA lists. Execution of a list will not +automatically destroy the list, thus enabling a list to be reused for repetitive +tasks. + + +List Population +~~~~~~~~~~~~~~~ + +An item can be added to a list using :c:func:`vme_dma_list_add` (the source and +destination attributes need to be created before calling this function, this is +covered under "Transfer Attributes"). + +.. note:: + + The detailed attributes of the transfers source and destination + are not checked until an entry is added to a DMA list, the request + for a DMA channel purely checks the directions in which the + controller is expected to transfer data. As a result it is + possible for this call to return an error, for example if the + source or destination is in an unsupported VME address space. + +Transfer Attributes +~~~~~~~~~~~~~~~~~~~ + +The attributes for the source and destination are handled separately from adding +an item to a list. This is due to the diverse attributes required for each type +of source and destination. There are functions to create attributes for PCI, VME +and pattern sources and destinations (where appropriate): + + - PCI source or destination: :c:func:`vme_dma_pci_attribute` + - VME source or destination: :c:func:`vme_dma_vme_attribute` + - Pattern source: :c:func:`vme_dma_pattern_attribute` + +The function :c:func:`vme_dma_free_attribute` should be used to free an +attribute. + + +List Execution +~~~~~~~~~~~~~~ + +The function :c:func:`vme_dma_list_exec` queues a list for execution and will +return once the list has been executed. + + +Interrupts +---------- + +The VME API provides functions to attach and detach callbacks to specific VME +level and status ID combinations and for the generation of VME interrupts with +specific VME level and status IDs. + + +Attaching Interrupt Handlers +~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The function :c:func:`vme_irq_request` can be used to attach and +:c:func:`vme_irq_free` to free a specific VME level and status ID combination. +Any given combination can only be assigned a single callback function. A void +pointer parameter is provided, the value of which is passed to the callback +function, the use of this pointer is user undefined. The callback parameters are +as follows. Care must be taken in writing a callback function, callback +functions run in interrupt context: + +.. code-block:: c + + void callback(int level, int statid, void *priv); + + +Interrupt Generation +~~~~~~~~~~~~~~~~~~~~ + +The function :c:func:`vme_irq_generate` can be used to generate a VME interrupt +at a given VME level and VME status ID. + + +Location monitors +----------------- + +The VME API provides the following functionality to configure the location +monitor. + + +Location Monitor Management +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +The function :c:func:`vme_lm_request` is provided to request the use of a block +of location monitors and :c:func:`vme_lm_free` to free them after they are no +longer required. Each block may provide a number of location monitors, +monitoring adjacent locations. The function :c:func:`vme_lm_count` can be used +to determine how many locations are provided. + + +Location Monitor Configuration +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Once a bank of location monitors has been allocated, the function +:c:func:`vme_lm_set` is provided to configure the location and mode of the +location monitor. The function :c:func:`vme_lm_get` can be used to retrieve +existing settings. + + +Location Monitor Use +~~~~~~~~~~~~~~~~~~~~ + +The function :c:func:`vme_lm_attach` enables a callback to be attached and +:c:func:`vme_lm_detach` allows on to be detached from each location monitor +location. Each location monitor can monitor a number of adjacent locations. The +callback function is declared as follows. + +.. code-block:: c + + void callback(void *data); + + +Slot Detection +-------------- + +The function :c:func:`vme_slot_num` returns the slot ID of the provided bridge. + + +Bus Detection +------------- + +The function :c:func:`vme_bus_num` returns the bus ID of the provided bridge. + + +VME API +------- + +.. kernel-doc:: include/linux/vme.h + :internal: + +.. kernel-doc:: drivers/vme/vme.c + :export: diff --git a/Documentation/driver-api/w1.rst b/Documentation/driver-api/w1.rst new file mode 100644 index 000000000..9963cca78 --- /dev/null +++ b/Documentation/driver-api/w1.rst @@ -0,0 +1,70 @@ +====================== +W1: Dallas' 1-wire bus +====================== + +:Author: David Fries + +W1 API internal to the kernel +============================= + +W1 API internal to the kernel +----------------------------- + +include/linux/w1.h +~~~~~~~~~~~~~~~~~~ + +W1 kernel API functions. + +.. kernel-doc:: include/linux/w1.h + :internal: + +drivers/w1/w1.c +~~~~~~~~~~~~~~~ + +W1 core functions. + +.. kernel-doc:: drivers/w1/w1.c + :internal: + +drivers/w1/w1_family.c +~~~~~~~~~~~~~~~~~~~~~~~ + +Allows registering device family operations. + +.. kernel-doc:: drivers/w1/w1_family.c + :export: + +drivers/w1/w1_internal.h +~~~~~~~~~~~~~~~~~~~~~~~~ + +W1 internal initialization for master devices. + +.. kernel-doc:: drivers/w1/w1_internal.h + :internal: + +drivers/w1/w1_int.c +~~~~~~~~~~~~~~~~~~~~ + +W1 internal initialization for master devices. + +.. kernel-doc:: drivers/w1/w1_int.c + :export: + +drivers/w1/w1_netlink.h +~~~~~~~~~~~~~~~~~~~~~~~~ + +W1 external netlink API structures and commands. + +.. kernel-doc:: drivers/w1/w1_netlink.h + :internal: + +drivers/w1/w1_io.c +~~~~~~~~~~~~~~~~~~~ + +W1 input/output. + +.. kernel-doc:: drivers/w1/w1_io.c + :export: + +.. kernel-doc:: drivers/w1/w1_io.c + :internal: |