diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-05-06 01:02:30 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-05-06 01:02:30 +0000 |
commit | 76cb841cb886eef6b3bee341a2266c76578724ad (patch) | |
tree | f5892e5ba6cc11949952a6ce4ecbe6d516d6ce58 /drivers/scsi/bfa/bfa_core.c | |
parent | Initial commit. (diff) | |
download | linux-76cb841cb886eef6b3bee341a2266c76578724ad.tar.xz linux-76cb841cb886eef6b3bee341a2266c76578724ad.zip |
Adding upstream version 4.19.249.upstream/4.19.249
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'drivers/scsi/bfa/bfa_core.c')
-rw-r--r-- | drivers/scsi/bfa/bfa_core.c | 2016 |
1 files changed, 2016 insertions, 0 deletions
diff --git a/drivers/scsi/bfa/bfa_core.c b/drivers/scsi/bfa/bfa_core.c new file mode 100644 index 000000000..10a63be92 --- /dev/null +++ b/drivers/scsi/bfa/bfa_core.c @@ -0,0 +1,2016 @@ +/* + * Copyright (c) 2005-2014 Brocade Communications Systems, Inc. + * Copyright (c) 2014- QLogic Corporation. + * All rights reserved + * www.qlogic.com + * + * Linux driver for QLogic BR-series Fibre Channel Host Bus Adapter. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License (GPL) Version 2 as + * published by the Free Software Foundation + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#include "bfad_drv.h" +#include "bfa_modules.h" +#include "bfi_reg.h" + +BFA_TRC_FILE(HAL, CORE); + +/* + * Message handlers for various modules. + */ +static bfa_isr_func_t bfa_isrs[BFI_MC_MAX] = { + bfa_isr_unhandled, /* NONE */ + bfa_isr_unhandled, /* BFI_MC_IOC */ + bfa_fcdiag_intr, /* BFI_MC_DIAG */ + bfa_isr_unhandled, /* BFI_MC_FLASH */ + bfa_isr_unhandled, /* BFI_MC_CEE */ + bfa_fcport_isr, /* BFI_MC_FCPORT */ + bfa_isr_unhandled, /* BFI_MC_IOCFC */ + bfa_isr_unhandled, /* BFI_MC_LL */ + bfa_uf_isr, /* BFI_MC_UF */ + bfa_fcxp_isr, /* BFI_MC_FCXP */ + bfa_lps_isr, /* BFI_MC_LPS */ + bfa_rport_isr, /* BFI_MC_RPORT */ + bfa_itn_isr, /* BFI_MC_ITN */ + bfa_isr_unhandled, /* BFI_MC_IOIM_READ */ + bfa_isr_unhandled, /* BFI_MC_IOIM_WRITE */ + bfa_isr_unhandled, /* BFI_MC_IOIM_IO */ + bfa_ioim_isr, /* BFI_MC_IOIM */ + bfa_ioim_good_comp_isr, /* BFI_MC_IOIM_IOCOM */ + bfa_tskim_isr, /* BFI_MC_TSKIM */ + bfa_isr_unhandled, /* BFI_MC_SBOOT */ + bfa_isr_unhandled, /* BFI_MC_IPFC */ + bfa_isr_unhandled, /* BFI_MC_PORT */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ + bfa_isr_unhandled, /* --------- */ +}; +/* + * Message handlers for mailbox command classes + */ +static bfa_ioc_mbox_mcfunc_t bfa_mbox_isrs[BFI_MC_MAX] = { + NULL, + NULL, /* BFI_MC_IOC */ + NULL, /* BFI_MC_DIAG */ + NULL, /* BFI_MC_FLASH */ + NULL, /* BFI_MC_CEE */ + NULL, /* BFI_MC_PORT */ + bfa_iocfc_isr, /* BFI_MC_IOCFC */ + NULL, +}; + + + +void +__bfa_trc(struct bfa_trc_mod_s *trcm, int fileno, int line, u64 data) +{ + int tail = trcm->tail; + struct bfa_trc_s *trc = &trcm->trc[tail]; + + if (trcm->stopped) + return; + + trc->fileno = (u16) fileno; + trc->line = (u16) line; + trc->data.u64 = data; + trc->timestamp = BFA_TRC_TS(trcm); + + trcm->tail = (trcm->tail + 1) & (BFA_TRC_MAX - 1); + if (trcm->tail == trcm->head) + trcm->head = (trcm->head + 1) & (BFA_TRC_MAX - 1); +} + +static void +bfa_com_port_attach(struct bfa_s *bfa) +{ + struct bfa_port_s *port = &bfa->modules.port; + struct bfa_mem_dma_s *port_dma = BFA_MEM_PORT_DMA(bfa); + + bfa_port_attach(port, &bfa->ioc, bfa, bfa->trcmod); + bfa_port_mem_claim(port, port_dma->kva_curp, port_dma->dma_curp); +} + +/* + * ablk module attach + */ +static void +bfa_com_ablk_attach(struct bfa_s *bfa) +{ + struct bfa_ablk_s *ablk = &bfa->modules.ablk; + struct bfa_mem_dma_s *ablk_dma = BFA_MEM_ABLK_DMA(bfa); + + bfa_ablk_attach(ablk, &bfa->ioc); + bfa_ablk_memclaim(ablk, ablk_dma->kva_curp, ablk_dma->dma_curp); +} + +static void +bfa_com_cee_attach(struct bfa_s *bfa) +{ + struct bfa_cee_s *cee = &bfa->modules.cee; + struct bfa_mem_dma_s *cee_dma = BFA_MEM_CEE_DMA(bfa); + + cee->trcmod = bfa->trcmod; + bfa_cee_attach(cee, &bfa->ioc, bfa); + bfa_cee_mem_claim(cee, cee_dma->kva_curp, cee_dma->dma_curp); +} + +static void +bfa_com_sfp_attach(struct bfa_s *bfa) +{ + struct bfa_sfp_s *sfp = BFA_SFP_MOD(bfa); + struct bfa_mem_dma_s *sfp_dma = BFA_MEM_SFP_DMA(bfa); + + bfa_sfp_attach(sfp, &bfa->ioc, bfa, bfa->trcmod); + bfa_sfp_memclaim(sfp, sfp_dma->kva_curp, sfp_dma->dma_curp); +} + +static void +bfa_com_flash_attach(struct bfa_s *bfa, bfa_boolean_t mincfg) +{ + struct bfa_flash_s *flash = BFA_FLASH(bfa); + struct bfa_mem_dma_s *flash_dma = BFA_MEM_FLASH_DMA(bfa); + + bfa_flash_attach(flash, &bfa->ioc, bfa, bfa->trcmod, mincfg); + bfa_flash_memclaim(flash, flash_dma->kva_curp, + flash_dma->dma_curp, mincfg); +} + +static void +bfa_com_diag_attach(struct bfa_s *bfa) +{ + struct bfa_diag_s *diag = BFA_DIAG_MOD(bfa); + struct bfa_mem_dma_s *diag_dma = BFA_MEM_DIAG_DMA(bfa); + + bfa_diag_attach(diag, &bfa->ioc, bfa, bfa_fcport_beacon, bfa->trcmod); + bfa_diag_memclaim(diag, diag_dma->kva_curp, diag_dma->dma_curp); +} + +static void +bfa_com_phy_attach(struct bfa_s *bfa, bfa_boolean_t mincfg) +{ + struct bfa_phy_s *phy = BFA_PHY(bfa); + struct bfa_mem_dma_s *phy_dma = BFA_MEM_PHY_DMA(bfa); + + bfa_phy_attach(phy, &bfa->ioc, bfa, bfa->trcmod, mincfg); + bfa_phy_memclaim(phy, phy_dma->kva_curp, phy_dma->dma_curp, mincfg); +} + +static void +bfa_com_fru_attach(struct bfa_s *bfa, bfa_boolean_t mincfg) +{ + struct bfa_fru_s *fru = BFA_FRU(bfa); + struct bfa_mem_dma_s *fru_dma = BFA_MEM_FRU_DMA(bfa); + + bfa_fru_attach(fru, &bfa->ioc, bfa, bfa->trcmod, mincfg); + bfa_fru_memclaim(fru, fru_dma->kva_curp, fru_dma->dma_curp, mincfg); +} + +/* + * BFA IOC FC related definitions + */ + +/* + * IOC local definitions + */ +#define BFA_IOCFC_TOV 5000 /* msecs */ + +enum { + BFA_IOCFC_ACT_NONE = 0, + BFA_IOCFC_ACT_INIT = 1, + BFA_IOCFC_ACT_STOP = 2, + BFA_IOCFC_ACT_DISABLE = 3, + BFA_IOCFC_ACT_ENABLE = 4, +}; + +#define DEF_CFG_NUM_FABRICS 1 +#define DEF_CFG_NUM_LPORTS 256 +#define DEF_CFG_NUM_CQS 4 +#define DEF_CFG_NUM_IOIM_REQS (BFA_IOIM_MAX) +#define DEF_CFG_NUM_TSKIM_REQS 128 +#define DEF_CFG_NUM_FCXP_REQS 64 +#define DEF_CFG_NUM_UF_BUFS 64 +#define DEF_CFG_NUM_RPORTS 1024 +#define DEF_CFG_NUM_ITNIMS (DEF_CFG_NUM_RPORTS) +#define DEF_CFG_NUM_TINS 256 + +#define DEF_CFG_NUM_SGPGS 2048 +#define DEF_CFG_NUM_REQQ_ELEMS 256 +#define DEF_CFG_NUM_RSPQ_ELEMS 64 +#define DEF_CFG_NUM_SBOOT_TGTS 16 +#define DEF_CFG_NUM_SBOOT_LUNS 16 + +/* + * IOCFC state machine definitions/declarations + */ +bfa_fsm_state_decl(bfa_iocfc, stopped, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, initing, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, dconf_read, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, init_cfg_wait, + struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, init_cfg_done, + struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, operational, + struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, dconf_write, + struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, stopping, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, enabling, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, cfg_wait, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, disabling, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, disabled, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, failed, struct bfa_iocfc_s, enum iocfc_event); +bfa_fsm_state_decl(bfa_iocfc, init_failed, + struct bfa_iocfc_s, enum iocfc_event); + +/* + * forward declaration for IOC FC functions + */ +static void bfa_iocfc_start_submod(struct bfa_s *bfa); +static void bfa_iocfc_disable_submod(struct bfa_s *bfa); +static void bfa_iocfc_send_cfg(void *bfa_arg); +static void bfa_iocfc_enable_cbfn(void *bfa_arg, enum bfa_status status); +static void bfa_iocfc_disable_cbfn(void *bfa_arg); +static void bfa_iocfc_hbfail_cbfn(void *bfa_arg); +static void bfa_iocfc_reset_cbfn(void *bfa_arg); +static struct bfa_ioc_cbfn_s bfa_iocfc_cbfn; +static void bfa_iocfc_init_cb(void *bfa_arg, bfa_boolean_t complete); +static void bfa_iocfc_stop_cb(void *bfa_arg, bfa_boolean_t compl); +static void bfa_iocfc_enable_cb(void *bfa_arg, bfa_boolean_t compl); +static void bfa_iocfc_disable_cb(void *bfa_arg, bfa_boolean_t compl); + +static void +bfa_iocfc_sm_stopped_entry(struct bfa_iocfc_s *iocfc) +{ +} + +static void +bfa_iocfc_sm_stopped(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_INIT: + case IOCFC_E_ENABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_initing); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_initing_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_ioc_enable(&iocfc->bfa->ioc); +} + +static void +bfa_iocfc_sm_initing(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_IOC_ENABLED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_read); + break; + + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopping); + break; + + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_init_failed); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_dconf_read_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_dconf_modinit(iocfc->bfa); +} + +static void +bfa_iocfc_sm_dconf_read(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_DCONF_DONE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_init_cfg_wait); + break; + + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopping); + break; + + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_init_failed); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_init_cfg_wait_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_iocfc_send_cfg(iocfc->bfa); +} + +static void +bfa_iocfc_sm_init_cfg_wait(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_CFG_DONE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_init_cfg_done); + break; + + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopping); + break; + + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_init_failed); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_init_cfg_done_entry(struct bfa_iocfc_s *iocfc) +{ + iocfc->bfa->iocfc.op_status = BFA_STATUS_OK; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.init_hcb_qe, + bfa_iocfc_init_cb, iocfc->bfa); +} + +static void +bfa_iocfc_sm_init_cfg_done(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_START: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_operational); + break; + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopping); + break; + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_failed); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_operational_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_fcport_init(iocfc->bfa); + bfa_iocfc_start_submod(iocfc->bfa); +} + +static void +bfa_iocfc_sm_operational(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_write); + break; + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_failed); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_dconf_write_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_dconf_modexit(iocfc->bfa); +} + +static void +bfa_iocfc_sm_dconf_write(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_DCONF_DONE: + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopping); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_stopping_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_ioc_disable(&iocfc->bfa->ioc); +} + +static void +bfa_iocfc_sm_stopping(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_IOC_DISABLED: + bfa_isr_disable(iocfc->bfa); + bfa_iocfc_disable_submod(iocfc->bfa); + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopped); + iocfc->bfa->iocfc.op_status = BFA_STATUS_OK; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.stop_hcb_qe, + bfa_iocfc_stop_cb, iocfc->bfa); + break; + + case IOCFC_E_IOC_ENABLED: + case IOCFC_E_DCONF_DONE: + case IOCFC_E_CFG_DONE: + break; + + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_enabling_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_ioc_enable(&iocfc->bfa->ioc); +} + +static void +bfa_iocfc_sm_enabling(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_IOC_ENABLED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_cfg_wait); + break; + + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_write); + break; + + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_failed); + + if (iocfc->bfa->iocfc.cb_reqd == BFA_FALSE) + break; + + iocfc->bfa->iocfc.op_status = BFA_STATUS_FAILED; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.en_hcb_qe, + bfa_iocfc_enable_cb, iocfc->bfa); + iocfc->bfa->iocfc.cb_reqd = BFA_FALSE; + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_cfg_wait_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_iocfc_send_cfg(iocfc->bfa); +} + +static void +bfa_iocfc_sm_cfg_wait(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_CFG_DONE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_operational); + if (iocfc->bfa->iocfc.cb_reqd == BFA_FALSE) + break; + + iocfc->bfa->iocfc.op_status = BFA_STATUS_OK; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.en_hcb_qe, + bfa_iocfc_enable_cb, iocfc->bfa); + iocfc->bfa->iocfc.cb_reqd = BFA_FALSE; + break; + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_write); + break; + case IOCFC_E_IOC_FAILED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_failed); + if (iocfc->bfa->iocfc.cb_reqd == BFA_FALSE) + break; + + iocfc->bfa->iocfc.op_status = BFA_STATUS_FAILED; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.en_hcb_qe, + bfa_iocfc_enable_cb, iocfc->bfa); + iocfc->bfa->iocfc.cb_reqd = BFA_FALSE; + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_disabling_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_ioc_disable(&iocfc->bfa->ioc); +} + +static void +bfa_iocfc_sm_disabling(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_IOC_DISABLED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabled); + break; + case IOCFC_E_IOC_ENABLED: + case IOCFC_E_DCONF_DONE: + case IOCFC_E_CFG_DONE: + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_disabled_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_isr_disable(iocfc->bfa); + bfa_iocfc_disable_submod(iocfc->bfa); + iocfc->bfa->iocfc.op_status = BFA_STATUS_OK; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.dis_hcb_qe, + bfa_iocfc_disable_cb, iocfc->bfa); +} + +static void +bfa_iocfc_sm_disabled(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_write); + break; + case IOCFC_E_ENABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_enabling); + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_failed_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_isr_disable(iocfc->bfa); + bfa_iocfc_disable_submod(iocfc->bfa); +} + +static void +bfa_iocfc_sm_failed(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_write); + break; + case IOCFC_E_DISABLE: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_disabling); + break; + case IOCFC_E_IOC_ENABLED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_cfg_wait); + break; + case IOCFC_E_IOC_FAILED: + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +static void +bfa_iocfc_sm_init_failed_entry(struct bfa_iocfc_s *iocfc) +{ + bfa_isr_disable(iocfc->bfa); + iocfc->bfa->iocfc.op_status = BFA_STATUS_FAILED; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.init_hcb_qe, + bfa_iocfc_init_cb, iocfc->bfa); +} + +static void +bfa_iocfc_sm_init_failed(struct bfa_iocfc_s *iocfc, enum iocfc_event event) +{ + bfa_trc(iocfc->bfa, event); + + switch (event) { + case IOCFC_E_STOP: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopping); + break; + case IOCFC_E_DISABLE: + bfa_ioc_disable(&iocfc->bfa->ioc); + break; + case IOCFC_E_IOC_ENABLED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_dconf_read); + break; + case IOCFC_E_IOC_DISABLED: + bfa_fsm_set_state(iocfc, bfa_iocfc_sm_stopped); + iocfc->bfa->iocfc.op_status = BFA_STATUS_OK; + bfa_cb_queue(iocfc->bfa, &iocfc->bfa->iocfc.dis_hcb_qe, + bfa_iocfc_disable_cb, iocfc->bfa); + break; + case IOCFC_E_IOC_FAILED: + break; + default: + bfa_sm_fault(iocfc->bfa, event); + break; + } +} + +/* + * BFA Interrupt handling functions + */ +static void +bfa_reqq_resume(struct bfa_s *bfa, int qid) +{ + struct list_head *waitq, *qe, *qen; + struct bfa_reqq_wait_s *wqe; + + waitq = bfa_reqq(bfa, qid); + list_for_each_safe(qe, qen, waitq) { + /* + * Callback only as long as there is room in request queue + */ + if (bfa_reqq_full(bfa, qid)) + break; + + list_del(qe); + wqe = (struct bfa_reqq_wait_s *) qe; + wqe->qresume(wqe->cbarg); + } +} + +bfa_boolean_t +bfa_isr_rspq(struct bfa_s *bfa, int qid) +{ + struct bfi_msg_s *m; + u32 pi, ci; + struct list_head *waitq; + bfa_boolean_t ret; + + ci = bfa_rspq_ci(bfa, qid); + pi = bfa_rspq_pi(bfa, qid); + + ret = (ci != pi); + + while (ci != pi) { + m = bfa_rspq_elem(bfa, qid, ci); + WARN_ON(m->mhdr.msg_class >= BFI_MC_MAX); + + bfa_isrs[m->mhdr.msg_class] (bfa, m); + CQ_INCR(ci, bfa->iocfc.cfg.drvcfg.num_rspq_elems); + } + + /* + * acknowledge RME completions and update CI + */ + bfa_isr_rspq_ack(bfa, qid, ci); + + /* + * Resume any pending requests in the corresponding reqq. + */ + waitq = bfa_reqq(bfa, qid); + if (!list_empty(waitq)) + bfa_reqq_resume(bfa, qid); + + return ret; +} + +static inline void +bfa_isr_reqq(struct bfa_s *bfa, int qid) +{ + struct list_head *waitq; + + bfa_isr_reqq_ack(bfa, qid); + + /* + * Resume any pending requests in the corresponding reqq. + */ + waitq = bfa_reqq(bfa, qid); + if (!list_empty(waitq)) + bfa_reqq_resume(bfa, qid); +} + +void +bfa_msix_all(struct bfa_s *bfa, int vec) +{ + u32 intr, qintr; + int queue; + + intr = readl(bfa->iocfc.bfa_regs.intr_status); + if (!intr) + return; + + /* + * RME completion queue interrupt + */ + qintr = intr & __HFN_INT_RME_MASK; + if (qintr && bfa->queue_process) { + for (queue = 0; queue < BFI_IOC_MAX_CQS; queue++) + bfa_isr_rspq(bfa, queue); + } + + intr &= ~qintr; + if (!intr) + return; + + /* + * CPE completion queue interrupt + */ + qintr = intr & __HFN_INT_CPE_MASK; + if (qintr && bfa->queue_process) { + for (queue = 0; queue < BFI_IOC_MAX_CQS; queue++) + bfa_isr_reqq(bfa, queue); + } + intr &= ~qintr; + if (!intr) + return; + + bfa_msix_lpu_err(bfa, intr); +} + +bfa_boolean_t +bfa_intx(struct bfa_s *bfa) +{ + u32 intr, qintr; + int queue; + bfa_boolean_t rspq_comp = BFA_FALSE; + + intr = readl(bfa->iocfc.bfa_regs.intr_status); + + qintr = intr & (__HFN_INT_RME_MASK | __HFN_INT_CPE_MASK); + if (qintr) + writel(qintr, bfa->iocfc.bfa_regs.intr_status); + + /* + * Unconditional RME completion queue interrupt + */ + if (bfa->queue_process) { + for (queue = 0; queue < BFI_IOC_MAX_CQS; queue++) + if (bfa_isr_rspq(bfa, queue)) + rspq_comp = BFA_TRUE; + } + + if (!intr) + return (qintr | rspq_comp) ? BFA_TRUE : BFA_FALSE; + + /* + * CPE completion queue interrupt + */ + qintr = intr & __HFN_INT_CPE_MASK; + if (qintr && bfa->queue_process) { + for (queue = 0; queue < BFI_IOC_MAX_CQS; queue++) + bfa_isr_reqq(bfa, queue); + } + intr &= ~qintr; + if (!intr) + return BFA_TRUE; + + if (bfa->intr_enabled) + bfa_msix_lpu_err(bfa, intr); + + return BFA_TRUE; +} + +void +bfa_isr_enable(struct bfa_s *bfa) +{ + u32 umsk; + int port_id = bfa_ioc_portid(&bfa->ioc); + + bfa_trc(bfa, bfa_ioc_pcifn(&bfa->ioc)); + bfa_trc(bfa, port_id); + + bfa_msix_ctrl_install(bfa); + + if (bfa_asic_id_ct2(bfa->ioc.pcidev.device_id)) { + umsk = __HFN_INT_ERR_MASK_CT2; + umsk |= port_id == 0 ? + __HFN_INT_FN0_MASK_CT2 : __HFN_INT_FN1_MASK_CT2; + } else { + umsk = __HFN_INT_ERR_MASK; + umsk |= port_id == 0 ? __HFN_INT_FN0_MASK : __HFN_INT_FN1_MASK; + } + + writel(umsk, bfa->iocfc.bfa_regs.intr_status); + writel(~umsk, bfa->iocfc.bfa_regs.intr_mask); + bfa->iocfc.intr_mask = ~umsk; + bfa_isr_mode_set(bfa, bfa->msix.nvecs != 0); + + /* + * Set the flag indicating successful enabling of interrupts + */ + bfa->intr_enabled = BFA_TRUE; +} + +void +bfa_isr_disable(struct bfa_s *bfa) +{ + bfa->intr_enabled = BFA_FALSE; + bfa_isr_mode_set(bfa, BFA_FALSE); + writel(-1L, bfa->iocfc.bfa_regs.intr_mask); + bfa_msix_uninstall(bfa); +} + +void +bfa_msix_reqq(struct bfa_s *bfa, int vec) +{ + bfa_isr_reqq(bfa, vec - bfa->iocfc.hwif.cpe_vec_q0); +} + +void +bfa_isr_unhandled(struct bfa_s *bfa, struct bfi_msg_s *m) +{ + bfa_trc(bfa, m->mhdr.msg_class); + bfa_trc(bfa, m->mhdr.msg_id); + bfa_trc(bfa, m->mhdr.mtag.i2htok); + WARN_ON(1); + bfa_trc_stop(bfa->trcmod); +} + +void +bfa_msix_rspq(struct bfa_s *bfa, int vec) +{ + bfa_isr_rspq(bfa, vec - bfa->iocfc.hwif.rme_vec_q0); +} + +void +bfa_msix_lpu_err(struct bfa_s *bfa, int vec) +{ + u32 intr, curr_value; + bfa_boolean_t lpu_isr, halt_isr, pss_isr; + + intr = readl(bfa->iocfc.bfa_regs.intr_status); + + if (bfa_asic_id_ct2(bfa->ioc.pcidev.device_id)) { + halt_isr = intr & __HFN_INT_CPQ_HALT_CT2; + pss_isr = intr & __HFN_INT_ERR_PSS_CT2; + lpu_isr = intr & (__HFN_INT_MBOX_LPU0_CT2 | + __HFN_INT_MBOX_LPU1_CT2); + intr &= __HFN_INT_ERR_MASK_CT2; + } else { + halt_isr = bfa_asic_id_ct(bfa->ioc.pcidev.device_id) ? + (intr & __HFN_INT_LL_HALT) : 0; + pss_isr = intr & __HFN_INT_ERR_PSS; + lpu_isr = intr & (__HFN_INT_MBOX_LPU0 | __HFN_INT_MBOX_LPU1); + intr &= __HFN_INT_ERR_MASK; + } + + if (lpu_isr) + bfa_ioc_mbox_isr(&bfa->ioc); + + if (intr) { + if (halt_isr) { + /* + * If LL_HALT bit is set then FW Init Halt LL Port + * Register needs to be cleared as well so Interrupt + * Status Register will be cleared. + */ + curr_value = readl(bfa->ioc.ioc_regs.ll_halt); + curr_value &= ~__FW_INIT_HALT_P; + writel(curr_value, bfa->ioc.ioc_regs.ll_halt); + } + + if (pss_isr) { + /* + * ERR_PSS bit needs to be cleared as well in case + * interrups are shared so driver's interrupt handler is + * still called even though it is already masked out. + */ + curr_value = readl( + bfa->ioc.ioc_regs.pss_err_status_reg); + writel(curr_value, + bfa->ioc.ioc_regs.pss_err_status_reg); + } + + writel(intr, bfa->iocfc.bfa_regs.intr_status); + bfa_ioc_error_isr(&bfa->ioc); + } +} + +/* + * BFA IOC FC related functions + */ + +/* + * BFA IOC private functions + */ + +/* + * Use the Mailbox interface to send BFI_IOCFC_H2I_CFG_REQ + */ +static void +bfa_iocfc_send_cfg(void *bfa_arg) +{ + struct bfa_s *bfa = bfa_arg; + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_cfg_req_s cfg_req; + struct bfi_iocfc_cfg_s *cfg_info = iocfc->cfginfo; + struct bfa_iocfc_cfg_s *cfg = &iocfc->cfg; + int i; + + WARN_ON(cfg->fwcfg.num_cqs > BFI_IOC_MAX_CQS); + bfa_trc(bfa, cfg->fwcfg.num_cqs); + + bfa_iocfc_reset_queues(bfa); + + /* + * initialize IOC configuration info + */ + cfg_info->single_msix_vec = 0; + if (bfa->msix.nvecs == 1) + cfg_info->single_msix_vec = 1; + cfg_info->endian_sig = BFI_IOC_ENDIAN_SIG; + cfg_info->num_cqs = cfg->fwcfg.num_cqs; + cfg_info->num_ioim_reqs = cpu_to_be16(bfa_fcpim_get_throttle_cfg(bfa, + cfg->fwcfg.num_ioim_reqs)); + cfg_info->num_fwtio_reqs = cpu_to_be16(cfg->fwcfg.num_fwtio_reqs); + + bfa_dma_be_addr_set(cfg_info->cfgrsp_addr, iocfc->cfgrsp_dma.pa); + /* + * dma map REQ and RSP circular queues and shadow pointers + */ + for (i = 0; i < cfg->fwcfg.num_cqs; i++) { + bfa_dma_be_addr_set(cfg_info->req_cq_ba[i], + iocfc->req_cq_ba[i].pa); + bfa_dma_be_addr_set(cfg_info->req_shadow_ci[i], + iocfc->req_cq_shadow_ci[i].pa); + cfg_info->req_cq_elems[i] = + cpu_to_be16(cfg->drvcfg.num_reqq_elems); + + bfa_dma_be_addr_set(cfg_info->rsp_cq_ba[i], + iocfc->rsp_cq_ba[i].pa); + bfa_dma_be_addr_set(cfg_info->rsp_shadow_pi[i], + iocfc->rsp_cq_shadow_pi[i].pa); + cfg_info->rsp_cq_elems[i] = + cpu_to_be16(cfg->drvcfg.num_rspq_elems); + } + + /* + * Enable interrupt coalescing if it is driver init path + * and not ioc disable/enable path. + */ + if (bfa_fsm_cmp_state(iocfc, bfa_iocfc_sm_init_cfg_wait)) + cfg_info->intr_attr.coalesce = BFA_TRUE; + + /* + * dma map IOC configuration itself + */ + bfi_h2i_set(cfg_req.mh, BFI_MC_IOCFC, BFI_IOCFC_H2I_CFG_REQ, + bfa_fn_lpu(bfa)); + bfa_dma_be_addr_set(cfg_req.ioc_cfg_dma_addr, iocfc->cfg_info.pa); + + bfa_ioc_mbox_send(&bfa->ioc, &cfg_req, + sizeof(struct bfi_iocfc_cfg_req_s)); +} + +static void +bfa_iocfc_init_mem(struct bfa_s *bfa, void *bfad, struct bfa_iocfc_cfg_s *cfg, + struct bfa_pcidev_s *pcidev) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + + bfa->bfad = bfad; + iocfc->bfa = bfa; + iocfc->cfg = *cfg; + + /* + * Initialize chip specific handlers. + */ + if (bfa_asic_id_ctc(bfa_ioc_devid(&bfa->ioc))) { + iocfc->hwif.hw_reginit = bfa_hwct_reginit; + iocfc->hwif.hw_reqq_ack = bfa_hwct_reqq_ack; + iocfc->hwif.hw_rspq_ack = bfa_hwct_rspq_ack; + iocfc->hwif.hw_msix_init = bfa_hwct_msix_init; + iocfc->hwif.hw_msix_ctrl_install = bfa_hwct_msix_ctrl_install; + iocfc->hwif.hw_msix_queue_install = bfa_hwct_msix_queue_install; + iocfc->hwif.hw_msix_uninstall = bfa_hwct_msix_uninstall; + iocfc->hwif.hw_isr_mode_set = bfa_hwct_isr_mode_set; + iocfc->hwif.hw_msix_getvecs = bfa_hwct_msix_getvecs; + iocfc->hwif.hw_msix_get_rme_range = bfa_hwct_msix_get_rme_range; + iocfc->hwif.rme_vec_q0 = BFI_MSIX_RME_QMIN_CT; + iocfc->hwif.cpe_vec_q0 = BFI_MSIX_CPE_QMIN_CT; + } else { + iocfc->hwif.hw_reginit = bfa_hwcb_reginit; + iocfc->hwif.hw_reqq_ack = NULL; + iocfc->hwif.hw_rspq_ack = bfa_hwcb_rspq_ack; + iocfc->hwif.hw_msix_init = bfa_hwcb_msix_init; + iocfc->hwif.hw_msix_ctrl_install = bfa_hwcb_msix_ctrl_install; + iocfc->hwif.hw_msix_queue_install = bfa_hwcb_msix_queue_install; + iocfc->hwif.hw_msix_uninstall = bfa_hwcb_msix_uninstall; + iocfc->hwif.hw_isr_mode_set = bfa_hwcb_isr_mode_set; + iocfc->hwif.hw_msix_getvecs = bfa_hwcb_msix_getvecs; + iocfc->hwif.hw_msix_get_rme_range = bfa_hwcb_msix_get_rme_range; + iocfc->hwif.rme_vec_q0 = BFI_MSIX_RME_QMIN_CB + + bfa_ioc_pcifn(&bfa->ioc) * BFI_IOC_MAX_CQS; + iocfc->hwif.cpe_vec_q0 = BFI_MSIX_CPE_QMIN_CB + + bfa_ioc_pcifn(&bfa->ioc) * BFI_IOC_MAX_CQS; + } + + if (bfa_asic_id_ct2(bfa_ioc_devid(&bfa->ioc))) { + iocfc->hwif.hw_reginit = bfa_hwct2_reginit; + iocfc->hwif.hw_isr_mode_set = NULL; + iocfc->hwif.hw_rspq_ack = bfa_hwct2_rspq_ack; + } + + iocfc->hwif.hw_reginit(bfa); + bfa->msix.nvecs = 0; +} + +static void +bfa_iocfc_mem_claim(struct bfa_s *bfa, struct bfa_iocfc_cfg_s *cfg) +{ + u8 *dm_kva = NULL; + u64 dm_pa = 0; + int i, per_reqq_sz, per_rspq_sz; + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfa_mem_dma_s *ioc_dma = BFA_MEM_IOC_DMA(bfa); + struct bfa_mem_dma_s *iocfc_dma = BFA_MEM_IOCFC_DMA(bfa); + struct bfa_mem_dma_s *reqq_dma, *rspq_dma; + + /* First allocate dma memory for IOC */ + bfa_ioc_mem_claim(&bfa->ioc, bfa_mem_dma_virt(ioc_dma), + bfa_mem_dma_phys(ioc_dma)); + + /* Claim DMA-able memory for the request/response queues */ + per_reqq_sz = BFA_ROUNDUP((cfg->drvcfg.num_reqq_elems * BFI_LMSG_SZ), + BFA_DMA_ALIGN_SZ); + per_rspq_sz = BFA_ROUNDUP((cfg->drvcfg.num_rspq_elems * BFI_LMSG_SZ), + BFA_DMA_ALIGN_SZ); + + for (i = 0; i < cfg->fwcfg.num_cqs; i++) { + reqq_dma = BFA_MEM_REQQ_DMA(bfa, i); + iocfc->req_cq_ba[i].kva = bfa_mem_dma_virt(reqq_dma); + iocfc->req_cq_ba[i].pa = bfa_mem_dma_phys(reqq_dma); + memset(iocfc->req_cq_ba[i].kva, 0, per_reqq_sz); + + rspq_dma = BFA_MEM_RSPQ_DMA(bfa, i); + iocfc->rsp_cq_ba[i].kva = bfa_mem_dma_virt(rspq_dma); + iocfc->rsp_cq_ba[i].pa = bfa_mem_dma_phys(rspq_dma); + memset(iocfc->rsp_cq_ba[i].kva, 0, per_rspq_sz); + } + + /* Claim IOCFC dma memory - for shadow CI/PI */ + dm_kva = bfa_mem_dma_virt(iocfc_dma); + dm_pa = bfa_mem_dma_phys(iocfc_dma); + + for (i = 0; i < cfg->fwcfg.num_cqs; i++) { + iocfc->req_cq_shadow_ci[i].kva = dm_kva; + iocfc->req_cq_shadow_ci[i].pa = dm_pa; + dm_kva += BFA_CACHELINE_SZ; + dm_pa += BFA_CACHELINE_SZ; + + iocfc->rsp_cq_shadow_pi[i].kva = dm_kva; + iocfc->rsp_cq_shadow_pi[i].pa = dm_pa; + dm_kva += BFA_CACHELINE_SZ; + dm_pa += BFA_CACHELINE_SZ; + } + + /* Claim IOCFC dma memory - for the config info page */ + bfa->iocfc.cfg_info.kva = dm_kva; + bfa->iocfc.cfg_info.pa = dm_pa; + bfa->iocfc.cfginfo = (struct bfi_iocfc_cfg_s *) dm_kva; + dm_kva += BFA_ROUNDUP(sizeof(struct bfi_iocfc_cfg_s), BFA_CACHELINE_SZ); + dm_pa += BFA_ROUNDUP(sizeof(struct bfi_iocfc_cfg_s), BFA_CACHELINE_SZ); + + /* Claim IOCFC dma memory - for the config response */ + bfa->iocfc.cfgrsp_dma.kva = dm_kva; + bfa->iocfc.cfgrsp_dma.pa = dm_pa; + bfa->iocfc.cfgrsp = (struct bfi_iocfc_cfgrsp_s *) dm_kva; + dm_kva += BFA_ROUNDUP(sizeof(struct bfi_iocfc_cfgrsp_s), + BFA_CACHELINE_SZ); + dm_pa += BFA_ROUNDUP(sizeof(struct bfi_iocfc_cfgrsp_s), + BFA_CACHELINE_SZ); + + /* Claim IOCFC kva memory */ + bfa_ioc_debug_memclaim(&bfa->ioc, bfa_mem_kva_curp(iocfc)); + bfa_mem_kva_curp(iocfc) += BFA_DBG_FWTRC_LEN; +} + +/* + * Start BFA submodules. + */ +static void +bfa_iocfc_start_submod(struct bfa_s *bfa) +{ + int i; + + bfa->queue_process = BFA_TRUE; + for (i = 0; i < BFI_IOC_MAX_CQS; i++) + bfa_isr_rspq_ack(bfa, i, bfa_rspq_ci(bfa, i)); + + bfa_fcport_start(bfa); + bfa_uf_start(bfa); + /* + * bfa_init() with flash read is complete. now invalidate the stale + * content of lun mask like unit attention, rp tag and lp tag. + */ + bfa_ioim_lm_init(BFA_FCP_MOD(bfa)->bfa); + + bfa->iocfc.submod_enabled = BFA_TRUE; +} + +/* + * Disable BFA submodules. + */ +static void +bfa_iocfc_disable_submod(struct bfa_s *bfa) +{ + if (bfa->iocfc.submod_enabled == BFA_FALSE) + return; + + bfa_fcdiag_iocdisable(bfa); + bfa_fcport_iocdisable(bfa); + bfa_fcxp_iocdisable(bfa); + bfa_lps_iocdisable(bfa); + bfa_rport_iocdisable(bfa); + bfa_fcp_iocdisable(bfa); + bfa_dconf_iocdisable(bfa); + + bfa->iocfc.submod_enabled = BFA_FALSE; +} + +static void +bfa_iocfc_init_cb(void *bfa_arg, bfa_boolean_t complete) +{ + struct bfa_s *bfa = bfa_arg; + + if (complete) + bfa_cb_init(bfa->bfad, bfa->iocfc.op_status); +} + +static void +bfa_iocfc_stop_cb(void *bfa_arg, bfa_boolean_t compl) +{ + struct bfa_s *bfa = bfa_arg; + struct bfad_s *bfad = bfa->bfad; + + if (compl) + complete(&bfad->comp); +} + +static void +bfa_iocfc_enable_cb(void *bfa_arg, bfa_boolean_t compl) +{ + struct bfa_s *bfa = bfa_arg; + struct bfad_s *bfad = bfa->bfad; + + if (compl) + complete(&bfad->enable_comp); +} + +static void +bfa_iocfc_disable_cb(void *bfa_arg, bfa_boolean_t compl) +{ + struct bfa_s *bfa = bfa_arg; + struct bfad_s *bfad = bfa->bfad; + + if (compl) + complete(&bfad->disable_comp); +} + +/** + * configure queue registers from firmware response + */ +static void +bfa_iocfc_qreg(struct bfa_s *bfa, struct bfi_iocfc_qreg_s *qreg) +{ + int i; + struct bfa_iocfc_regs_s *r = &bfa->iocfc.bfa_regs; + void __iomem *kva = bfa_ioc_bar0(&bfa->ioc); + + for (i = 0; i < BFI_IOC_MAX_CQS; i++) { + bfa->iocfc.hw_qid[i] = qreg->hw_qid[i]; + r->cpe_q_ci[i] = kva + be32_to_cpu(qreg->cpe_q_ci_off[i]); + r->cpe_q_pi[i] = kva + be32_to_cpu(qreg->cpe_q_pi_off[i]); + r->cpe_q_ctrl[i] = kva + be32_to_cpu(qreg->cpe_qctl_off[i]); + r->rme_q_ci[i] = kva + be32_to_cpu(qreg->rme_q_ci_off[i]); + r->rme_q_pi[i] = kva + be32_to_cpu(qreg->rme_q_pi_off[i]); + r->rme_q_ctrl[i] = kva + be32_to_cpu(qreg->rme_qctl_off[i]); + } +} + +static void +bfa_iocfc_res_recfg(struct bfa_s *bfa, struct bfa_iocfc_fwcfg_s *fwcfg) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_cfg_s *cfg_info = iocfc->cfginfo; + + bfa_fcxp_res_recfg(bfa, fwcfg->num_fcxp_reqs); + bfa_uf_res_recfg(bfa, fwcfg->num_uf_bufs); + bfa_rport_res_recfg(bfa, fwcfg->num_rports); + bfa_fcp_res_recfg(bfa, cpu_to_be16(cfg_info->num_ioim_reqs), + fwcfg->num_ioim_reqs); + bfa_tskim_res_recfg(bfa, fwcfg->num_tskim_reqs); +} + +/* + * Update BFA configuration from firmware configuration. + */ +static void +bfa_iocfc_cfgrsp(struct bfa_s *bfa) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_cfgrsp_s *cfgrsp = iocfc->cfgrsp; + struct bfa_iocfc_fwcfg_s *fwcfg = &cfgrsp->fwcfg; + + fwcfg->num_cqs = fwcfg->num_cqs; + fwcfg->num_ioim_reqs = be16_to_cpu(fwcfg->num_ioim_reqs); + fwcfg->num_fwtio_reqs = be16_to_cpu(fwcfg->num_fwtio_reqs); + fwcfg->num_tskim_reqs = be16_to_cpu(fwcfg->num_tskim_reqs); + fwcfg->num_fcxp_reqs = be16_to_cpu(fwcfg->num_fcxp_reqs); + fwcfg->num_uf_bufs = be16_to_cpu(fwcfg->num_uf_bufs); + fwcfg->num_rports = be16_to_cpu(fwcfg->num_rports); + + /* + * configure queue register offsets as learnt from firmware + */ + bfa_iocfc_qreg(bfa, &cfgrsp->qreg); + + /* + * Re-configure resources as learnt from Firmware + */ + bfa_iocfc_res_recfg(bfa, fwcfg); + + /* + * Install MSIX queue handlers + */ + bfa_msix_queue_install(bfa); + + if (bfa->iocfc.cfgrsp->pbc_cfg.pbc_pwwn != 0) { + bfa->ioc.attr->pwwn = bfa->iocfc.cfgrsp->pbc_cfg.pbc_pwwn; + bfa->ioc.attr->nwwn = bfa->iocfc.cfgrsp->pbc_cfg.pbc_nwwn; + bfa_fsm_send_event(iocfc, IOCFC_E_CFG_DONE); + } +} + +void +bfa_iocfc_reset_queues(struct bfa_s *bfa) +{ + int q; + + for (q = 0; q < BFI_IOC_MAX_CQS; q++) { + bfa_reqq_ci(bfa, q) = 0; + bfa_reqq_pi(bfa, q) = 0; + bfa_rspq_ci(bfa, q) = 0; + bfa_rspq_pi(bfa, q) = 0; + } +} + +/* + * Process FAA pwwn msg from fw. + */ +static void +bfa_iocfc_process_faa_addr(struct bfa_s *bfa, struct bfi_faa_addr_msg_s *msg) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_cfgrsp_s *cfgrsp = iocfc->cfgrsp; + + cfgrsp->pbc_cfg.pbc_pwwn = msg->pwwn; + cfgrsp->pbc_cfg.pbc_nwwn = msg->nwwn; + + bfa->ioc.attr->pwwn = msg->pwwn; + bfa->ioc.attr->nwwn = msg->nwwn; + bfa_fsm_send_event(iocfc, IOCFC_E_CFG_DONE); +} + +/* Fabric Assigned Address specific functions */ + +/* + * Check whether IOC is ready before sending command down + */ +static bfa_status_t +bfa_faa_validate_request(struct bfa_s *bfa) +{ + enum bfa_ioc_type_e ioc_type = bfa_get_type(bfa); + u32 card_type = bfa->ioc.attr->card_type; + + if (bfa_ioc_is_operational(&bfa->ioc)) { + if ((ioc_type != BFA_IOC_TYPE_FC) || bfa_mfg_is_mezz(card_type)) + return BFA_STATUS_FEATURE_NOT_SUPPORTED; + } else { + return BFA_STATUS_IOC_NON_OP; + } + + return BFA_STATUS_OK; +} + +bfa_status_t +bfa_faa_query(struct bfa_s *bfa, struct bfa_faa_attr_s *attr, + bfa_cb_iocfc_t cbfn, void *cbarg) +{ + struct bfi_faa_query_s faa_attr_req; + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + bfa_status_t status; + + status = bfa_faa_validate_request(bfa); + if (status != BFA_STATUS_OK) + return status; + + if (iocfc->faa_args.busy == BFA_TRUE) + return BFA_STATUS_DEVBUSY; + + iocfc->faa_args.faa_attr = attr; + iocfc->faa_args.faa_cb.faa_cbfn = cbfn; + iocfc->faa_args.faa_cb.faa_cbarg = cbarg; + + iocfc->faa_args.busy = BFA_TRUE; + memset(&faa_attr_req, 0, sizeof(struct bfi_faa_query_s)); + bfi_h2i_set(faa_attr_req.mh, BFI_MC_IOCFC, + BFI_IOCFC_H2I_FAA_QUERY_REQ, bfa_fn_lpu(bfa)); + + bfa_ioc_mbox_send(&bfa->ioc, &faa_attr_req, + sizeof(struct bfi_faa_query_s)); + + return BFA_STATUS_OK; +} + +/* + * FAA query response + */ +static void +bfa_faa_query_reply(struct bfa_iocfc_s *iocfc, + bfi_faa_query_rsp_t *rsp) +{ + void *cbarg = iocfc->faa_args.faa_cb.faa_cbarg; + + if (iocfc->faa_args.faa_attr) { + iocfc->faa_args.faa_attr->faa = rsp->faa; + iocfc->faa_args.faa_attr->faa_state = rsp->faa_status; + iocfc->faa_args.faa_attr->pwwn_source = rsp->addr_source; + } + + WARN_ON(!iocfc->faa_args.faa_cb.faa_cbfn); + + iocfc->faa_args.faa_cb.faa_cbfn(cbarg, BFA_STATUS_OK); + iocfc->faa_args.busy = BFA_FALSE; +} + +/* + * IOC enable request is complete + */ +static void +bfa_iocfc_enable_cbfn(void *bfa_arg, enum bfa_status status) +{ + struct bfa_s *bfa = bfa_arg; + + if (status == BFA_STATUS_OK) + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_IOC_ENABLED); + else + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_IOC_FAILED); +} + +/* + * IOC disable request is complete + */ +static void +bfa_iocfc_disable_cbfn(void *bfa_arg) +{ + struct bfa_s *bfa = bfa_arg; + + bfa->queue_process = BFA_FALSE; + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_IOC_DISABLED); +} + +/* + * Notify sub-modules of hardware failure. + */ +static void +bfa_iocfc_hbfail_cbfn(void *bfa_arg) +{ + struct bfa_s *bfa = bfa_arg; + + bfa->queue_process = BFA_FALSE; + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_IOC_FAILED); +} + +/* + * Actions on chip-reset completion. + */ +static void +bfa_iocfc_reset_cbfn(void *bfa_arg) +{ + struct bfa_s *bfa = bfa_arg; + + bfa_iocfc_reset_queues(bfa); + bfa_isr_enable(bfa); +} + +/* + * Query IOC memory requirement information. + */ +void +bfa_iocfc_meminfo(struct bfa_iocfc_cfg_s *cfg, struct bfa_meminfo_s *meminfo, + struct bfa_s *bfa) +{ + int q, per_reqq_sz, per_rspq_sz; + struct bfa_mem_dma_s *ioc_dma = BFA_MEM_IOC_DMA(bfa); + struct bfa_mem_dma_s *iocfc_dma = BFA_MEM_IOCFC_DMA(bfa); + struct bfa_mem_kva_s *iocfc_kva = BFA_MEM_IOCFC_KVA(bfa); + u32 dm_len = 0; + + /* dma memory setup for IOC */ + bfa_mem_dma_setup(meminfo, ioc_dma, + BFA_ROUNDUP(sizeof(struct bfi_ioc_attr_s), BFA_DMA_ALIGN_SZ)); + + /* dma memory setup for REQ/RSP queues */ + per_reqq_sz = BFA_ROUNDUP((cfg->drvcfg.num_reqq_elems * BFI_LMSG_SZ), + BFA_DMA_ALIGN_SZ); + per_rspq_sz = BFA_ROUNDUP((cfg->drvcfg.num_rspq_elems * BFI_LMSG_SZ), + BFA_DMA_ALIGN_SZ); + + for (q = 0; q < cfg->fwcfg.num_cqs; q++) { + bfa_mem_dma_setup(meminfo, BFA_MEM_REQQ_DMA(bfa, q), + per_reqq_sz); + bfa_mem_dma_setup(meminfo, BFA_MEM_RSPQ_DMA(bfa, q), + per_rspq_sz); + } + + /* IOCFC dma memory - calculate Shadow CI/PI size */ + for (q = 0; q < cfg->fwcfg.num_cqs; q++) + dm_len += (2 * BFA_CACHELINE_SZ); + + /* IOCFC dma memory - calculate config info / rsp size */ + dm_len += BFA_ROUNDUP(sizeof(struct bfi_iocfc_cfg_s), BFA_CACHELINE_SZ); + dm_len += BFA_ROUNDUP(sizeof(struct bfi_iocfc_cfgrsp_s), + BFA_CACHELINE_SZ); + + /* dma memory setup for IOCFC */ + bfa_mem_dma_setup(meminfo, iocfc_dma, dm_len); + + /* kva memory setup for IOCFC */ + bfa_mem_kva_setup(meminfo, iocfc_kva, BFA_DBG_FWTRC_LEN); +} + +/* + * Query IOC memory requirement information. + */ +void +bfa_iocfc_attach(struct bfa_s *bfa, void *bfad, struct bfa_iocfc_cfg_s *cfg, + struct bfa_pcidev_s *pcidev) +{ + int i; + struct bfa_ioc_s *ioc = &bfa->ioc; + + bfa_iocfc_cbfn.enable_cbfn = bfa_iocfc_enable_cbfn; + bfa_iocfc_cbfn.disable_cbfn = bfa_iocfc_disable_cbfn; + bfa_iocfc_cbfn.hbfail_cbfn = bfa_iocfc_hbfail_cbfn; + bfa_iocfc_cbfn.reset_cbfn = bfa_iocfc_reset_cbfn; + + ioc->trcmod = bfa->trcmod; + bfa_ioc_attach(&bfa->ioc, bfa, &bfa_iocfc_cbfn, &bfa->timer_mod); + + bfa_ioc_pci_init(&bfa->ioc, pcidev, BFI_PCIFN_CLASS_FC); + bfa_ioc_mbox_register(&bfa->ioc, bfa_mbox_isrs); + + bfa_iocfc_init_mem(bfa, bfad, cfg, pcidev); + bfa_iocfc_mem_claim(bfa, cfg); + INIT_LIST_HEAD(&bfa->timer_mod.timer_q); + + INIT_LIST_HEAD(&bfa->comp_q); + for (i = 0; i < BFI_IOC_MAX_CQS; i++) + INIT_LIST_HEAD(&bfa->reqq_waitq[i]); + + bfa->iocfc.cb_reqd = BFA_FALSE; + bfa->iocfc.op_status = BFA_STATUS_OK; + bfa->iocfc.submod_enabled = BFA_FALSE; + + bfa_fsm_set_state(&bfa->iocfc, bfa_iocfc_sm_stopped); +} + +/* + * Query IOC memory requirement information. + */ +void +bfa_iocfc_init(struct bfa_s *bfa) +{ + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_INIT); +} + +/* + * IOC start called from bfa_start(). Called to start IOC operations + * at driver instantiation for this instance. + */ +void +bfa_iocfc_start(struct bfa_s *bfa) +{ + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_START); +} + +/* + * IOC stop called from bfa_stop(). Called only when driver is unloaded + * for this instance. + */ +void +bfa_iocfc_stop(struct bfa_s *bfa) +{ + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_STOP); +} + +void +bfa_iocfc_isr(void *bfaarg, struct bfi_mbmsg_s *m) +{ + struct bfa_s *bfa = bfaarg; + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + union bfi_iocfc_i2h_msg_u *msg; + + msg = (union bfi_iocfc_i2h_msg_u *) m; + bfa_trc(bfa, msg->mh.msg_id); + + switch (msg->mh.msg_id) { + case BFI_IOCFC_I2H_CFG_REPLY: + bfa_iocfc_cfgrsp(bfa); + break; + case BFI_IOCFC_I2H_UPDATEQ_RSP: + iocfc->updateq_cbfn(iocfc->updateq_cbarg, BFA_STATUS_OK); + break; + case BFI_IOCFC_I2H_ADDR_MSG: + bfa_iocfc_process_faa_addr(bfa, + (struct bfi_faa_addr_msg_s *)msg); + break; + case BFI_IOCFC_I2H_FAA_QUERY_RSP: + bfa_faa_query_reply(iocfc, (bfi_faa_query_rsp_t *)msg); + break; + default: + WARN_ON(1); + } +} + +void +bfa_iocfc_get_attr(struct bfa_s *bfa, struct bfa_iocfc_attr_s *attr) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + + attr->intr_attr.coalesce = iocfc->cfginfo->intr_attr.coalesce; + + attr->intr_attr.delay = iocfc->cfginfo->intr_attr.delay ? + be16_to_cpu(iocfc->cfginfo->intr_attr.delay) : + be16_to_cpu(iocfc->cfgrsp->intr_attr.delay); + + attr->intr_attr.latency = iocfc->cfginfo->intr_attr.latency ? + be16_to_cpu(iocfc->cfginfo->intr_attr.latency) : + be16_to_cpu(iocfc->cfgrsp->intr_attr.latency); + + attr->config = iocfc->cfg; +} + +bfa_status_t +bfa_iocfc_israttr_set(struct bfa_s *bfa, struct bfa_iocfc_intr_attr_s *attr) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_set_intr_req_s *m; + + iocfc->cfginfo->intr_attr.coalesce = attr->coalesce; + iocfc->cfginfo->intr_attr.delay = cpu_to_be16(attr->delay); + iocfc->cfginfo->intr_attr.latency = cpu_to_be16(attr->latency); + + if (!bfa_iocfc_is_operational(bfa)) + return BFA_STATUS_OK; + + m = bfa_reqq_next(bfa, BFA_REQQ_IOC); + if (!m) + return BFA_STATUS_DEVBUSY; + + bfi_h2i_set(m->mh, BFI_MC_IOCFC, BFI_IOCFC_H2I_SET_INTR_REQ, + bfa_fn_lpu(bfa)); + m->coalesce = iocfc->cfginfo->intr_attr.coalesce; + m->delay = iocfc->cfginfo->intr_attr.delay; + m->latency = iocfc->cfginfo->intr_attr.latency; + + bfa_trc(bfa, attr->delay); + bfa_trc(bfa, attr->latency); + + bfa_reqq_produce(bfa, BFA_REQQ_IOC, m->mh); + return BFA_STATUS_OK; +} + +void +bfa_iocfc_set_snsbase(struct bfa_s *bfa, int seg_no, u64 snsbase_pa) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + + iocfc->cfginfo->sense_buf_len = (BFI_IOIM_SNSLEN - 1); + bfa_dma_be_addr_set(iocfc->cfginfo->ioim_snsbase[seg_no], snsbase_pa); +} +/* + * Enable IOC after it is disabled. + */ +void +bfa_iocfc_enable(struct bfa_s *bfa) +{ + bfa_plog_str(bfa->plog, BFA_PL_MID_HAL, BFA_PL_EID_MISC, 0, + "IOC Enable"); + bfa->iocfc.cb_reqd = BFA_TRUE; + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_ENABLE); +} + +void +bfa_iocfc_disable(struct bfa_s *bfa) +{ + bfa_plog_str(bfa->plog, BFA_PL_MID_HAL, BFA_PL_EID_MISC, 0, + "IOC Disable"); + + bfa_fsm_send_event(&bfa->iocfc, IOCFC_E_DISABLE); +} + +bfa_boolean_t +bfa_iocfc_is_operational(struct bfa_s *bfa) +{ + return bfa_ioc_is_operational(&bfa->ioc) && + bfa_fsm_cmp_state(&bfa->iocfc, bfa_iocfc_sm_operational); +} + +/* + * Return boot target port wwns -- read from boot information in flash. + */ +void +bfa_iocfc_get_bootwwns(struct bfa_s *bfa, u8 *nwwns, wwn_t *wwns) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_cfgrsp_s *cfgrsp = iocfc->cfgrsp; + int i; + + if (cfgrsp->pbc_cfg.boot_enabled && cfgrsp->pbc_cfg.nbluns) { + bfa_trc(bfa, cfgrsp->pbc_cfg.nbluns); + *nwwns = cfgrsp->pbc_cfg.nbluns; + for (i = 0; i < cfgrsp->pbc_cfg.nbluns; i++) + wwns[i] = cfgrsp->pbc_cfg.blun[i].tgt_pwwn; + + return; + } + + *nwwns = cfgrsp->bootwwns.nwwns; + memcpy(wwns, cfgrsp->bootwwns.wwn, sizeof(cfgrsp->bootwwns.wwn)); +} + +int +bfa_iocfc_get_pbc_vports(struct bfa_s *bfa, struct bfi_pbc_vport_s *pbc_vport) +{ + struct bfa_iocfc_s *iocfc = &bfa->iocfc; + struct bfi_iocfc_cfgrsp_s *cfgrsp = iocfc->cfgrsp; + + memcpy(pbc_vport, cfgrsp->pbc_cfg.vport, sizeof(cfgrsp->pbc_cfg.vport)); + return cfgrsp->pbc_cfg.nvports; +} + + +/* + * Use this function query the memory requirement of the BFA library. + * This function needs to be called before bfa_attach() to get the + * memory required of the BFA layer for a given driver configuration. + * + * This call will fail, if the cap is out of range compared to pre-defined + * values within the BFA library + * + * @param[in] cfg - pointer to bfa_ioc_cfg_t. Driver layer should indicate + * its configuration in this structure. + * The default values for struct bfa_iocfc_cfg_s can be + * fetched using bfa_cfg_get_default() API. + * + * If cap's boundary check fails, the library will use + * the default bfa_cap_t values (and log a warning msg). + * + * @param[out] meminfo - pointer to bfa_meminfo_t. This content + * indicates the memory type (see bfa_mem_type_t) and + * amount of memory required. + * + * Driver should allocate the memory, populate the + * starting address for each block and provide the same + * structure as input parameter to bfa_attach() call. + * + * @param[in] bfa - pointer to the bfa structure, used while fetching the + * dma, kva memory information of the bfa sub-modules. + * + * @return void + * + * Special Considerations: @note + */ +void +bfa_cfg_get_meminfo(struct bfa_iocfc_cfg_s *cfg, struct bfa_meminfo_s *meminfo, + struct bfa_s *bfa) +{ + struct bfa_mem_dma_s *port_dma = BFA_MEM_PORT_DMA(bfa); + struct bfa_mem_dma_s *ablk_dma = BFA_MEM_ABLK_DMA(bfa); + struct bfa_mem_dma_s *cee_dma = BFA_MEM_CEE_DMA(bfa); + struct bfa_mem_dma_s *sfp_dma = BFA_MEM_SFP_DMA(bfa); + struct bfa_mem_dma_s *flash_dma = BFA_MEM_FLASH_DMA(bfa); + struct bfa_mem_dma_s *diag_dma = BFA_MEM_DIAG_DMA(bfa); + struct bfa_mem_dma_s *phy_dma = BFA_MEM_PHY_DMA(bfa); + struct bfa_mem_dma_s *fru_dma = BFA_MEM_FRU_DMA(bfa); + + WARN_ON((cfg == NULL) || (meminfo == NULL)); + + memset((void *)meminfo, 0, sizeof(struct bfa_meminfo_s)); + + /* Initialize the DMA & KVA meminfo queues */ + INIT_LIST_HEAD(&meminfo->dma_info.qe); + INIT_LIST_HEAD(&meminfo->kva_info.qe); + + bfa_iocfc_meminfo(cfg, meminfo, bfa); + bfa_sgpg_meminfo(cfg, meminfo, bfa); + bfa_fcport_meminfo(cfg, meminfo, bfa); + bfa_fcxp_meminfo(cfg, meminfo, bfa); + bfa_lps_meminfo(cfg, meminfo, bfa); + bfa_uf_meminfo(cfg, meminfo, bfa); + bfa_rport_meminfo(cfg, meminfo, bfa); + bfa_fcp_meminfo(cfg, meminfo, bfa); + bfa_dconf_meminfo(cfg, meminfo, bfa); + + /* dma info setup */ + bfa_mem_dma_setup(meminfo, port_dma, bfa_port_meminfo()); + bfa_mem_dma_setup(meminfo, ablk_dma, bfa_ablk_meminfo()); + bfa_mem_dma_setup(meminfo, cee_dma, bfa_cee_meminfo()); + bfa_mem_dma_setup(meminfo, sfp_dma, bfa_sfp_meminfo()); + bfa_mem_dma_setup(meminfo, flash_dma, + bfa_flash_meminfo(cfg->drvcfg.min_cfg)); + bfa_mem_dma_setup(meminfo, diag_dma, bfa_diag_meminfo()); + bfa_mem_dma_setup(meminfo, phy_dma, + bfa_phy_meminfo(cfg->drvcfg.min_cfg)); + bfa_mem_dma_setup(meminfo, fru_dma, + bfa_fru_meminfo(cfg->drvcfg.min_cfg)); +} + +/* + * Use this function to do attach the driver instance with the BFA + * library. This function will not trigger any HW initialization + * process (which will be done in bfa_init() call) + * + * This call will fail, if the cap is out of range compared to + * pre-defined values within the BFA library + * + * @param[out] bfa Pointer to bfa_t. + * @param[in] bfad Opaque handle back to the driver's IOC structure + * @param[in] cfg Pointer to bfa_ioc_cfg_t. Should be same structure + * that was used in bfa_cfg_get_meminfo(). + * @param[in] meminfo Pointer to bfa_meminfo_t. The driver should + * use the bfa_cfg_get_meminfo() call to + * find the memory blocks required, allocate the + * required memory and provide the starting addresses. + * @param[in] pcidev pointer to struct bfa_pcidev_s + * + * @return + * void + * + * Special Considerations: + * + * @note + * + */ +void +bfa_attach(struct bfa_s *bfa, void *bfad, struct bfa_iocfc_cfg_s *cfg, + struct bfa_meminfo_s *meminfo, struct bfa_pcidev_s *pcidev) +{ + struct bfa_mem_dma_s *dma_info, *dma_elem; + struct bfa_mem_kva_s *kva_info, *kva_elem; + struct list_head *dm_qe, *km_qe; + + bfa->fcs = BFA_FALSE; + + WARN_ON((cfg == NULL) || (meminfo == NULL)); + + /* Initialize memory pointers for iterative allocation */ + dma_info = &meminfo->dma_info; + dma_info->kva_curp = dma_info->kva; + dma_info->dma_curp = dma_info->dma; + + kva_info = &meminfo->kva_info; + kva_info->kva_curp = kva_info->kva; + + list_for_each(dm_qe, &dma_info->qe) { + dma_elem = (struct bfa_mem_dma_s *) dm_qe; + dma_elem->kva_curp = dma_elem->kva; + dma_elem->dma_curp = dma_elem->dma; + } + + list_for_each(km_qe, &kva_info->qe) { + kva_elem = (struct bfa_mem_kva_s *) km_qe; + kva_elem->kva_curp = kva_elem->kva; + } + + bfa_iocfc_attach(bfa, bfad, cfg, pcidev); + bfa_fcdiag_attach(bfa, bfad, cfg, pcidev); + bfa_sgpg_attach(bfa, bfad, cfg, pcidev); + bfa_fcport_attach(bfa, bfad, cfg, pcidev); + bfa_fcxp_attach(bfa, bfad, cfg, pcidev); + bfa_lps_attach(bfa, bfad, cfg, pcidev); + bfa_uf_attach(bfa, bfad, cfg, pcidev); + bfa_rport_attach(bfa, bfad, cfg, pcidev); + bfa_fcp_attach(bfa, bfad, cfg, pcidev); + bfa_dconf_attach(bfa, bfad, cfg); + bfa_com_port_attach(bfa); + bfa_com_ablk_attach(bfa); + bfa_com_cee_attach(bfa); + bfa_com_sfp_attach(bfa); + bfa_com_flash_attach(bfa, cfg->drvcfg.min_cfg); + bfa_com_diag_attach(bfa); + bfa_com_phy_attach(bfa, cfg->drvcfg.min_cfg); + bfa_com_fru_attach(bfa, cfg->drvcfg.min_cfg); +} + +/* + * Use this function to delete a BFA IOC. IOC should be stopped (by + * calling bfa_stop()) before this function call. + * + * @param[in] bfa - pointer to bfa_t. + * + * @return + * void + * + * Special Considerations: + * + * @note + */ +void +bfa_detach(struct bfa_s *bfa) +{ + bfa_ioc_detach(&bfa->ioc); +} + +void +bfa_comp_deq(struct bfa_s *bfa, struct list_head *comp_q) +{ + INIT_LIST_HEAD(comp_q); + list_splice_tail_init(&bfa->comp_q, comp_q); +} + +void +bfa_comp_process(struct bfa_s *bfa, struct list_head *comp_q) +{ + struct list_head *qe; + struct list_head *qen; + struct bfa_cb_qe_s *hcb_qe; + bfa_cb_cbfn_status_t cbfn; + + list_for_each_safe(qe, qen, comp_q) { + hcb_qe = (struct bfa_cb_qe_s *) qe; + if (hcb_qe->pre_rmv) { + /* qe is invalid after return, dequeue before cbfn() */ + list_del(qe); + cbfn = (bfa_cb_cbfn_status_t)(hcb_qe->cbfn); + cbfn(hcb_qe->cbarg, hcb_qe->fw_status); + } else + hcb_qe->cbfn(hcb_qe->cbarg, BFA_TRUE); + } +} + +void +bfa_comp_free(struct bfa_s *bfa, struct list_head *comp_q) +{ + struct list_head *qe; + struct bfa_cb_qe_s *hcb_qe; + + while (!list_empty(comp_q)) { + bfa_q_deq(comp_q, &qe); + hcb_qe = (struct bfa_cb_qe_s *) qe; + WARN_ON(hcb_qe->pre_rmv); + hcb_qe->cbfn(hcb_qe->cbarg, BFA_FALSE); + } +} + +/* + * Return the list of PCI vendor/device id lists supported by this + * BFA instance. + */ +void +bfa_get_pciids(struct bfa_pciid_s **pciids, int *npciids) +{ + static struct bfa_pciid_s __pciids[] = { + {BFA_PCI_VENDOR_ID_BROCADE, BFA_PCI_DEVICE_ID_FC_8G2P}, + {BFA_PCI_VENDOR_ID_BROCADE, BFA_PCI_DEVICE_ID_FC_8G1P}, + {BFA_PCI_VENDOR_ID_BROCADE, BFA_PCI_DEVICE_ID_CT}, + {BFA_PCI_VENDOR_ID_BROCADE, BFA_PCI_DEVICE_ID_CT_FC}, + }; + + *npciids = ARRAY_SIZE(__pciids); + *pciids = __pciids; +} + +/* + * Use this function query the default struct bfa_iocfc_cfg_s value (compiled + * into BFA layer). The OS driver can then turn back and overwrite entries that + * have been configured by the user. + * + * @param[in] cfg - pointer to bfa_ioc_cfg_t + * + * @return + * void + * + * Special Considerations: + * note + */ +void +bfa_cfg_get_default(struct bfa_iocfc_cfg_s *cfg) +{ + cfg->fwcfg.num_fabrics = DEF_CFG_NUM_FABRICS; + cfg->fwcfg.num_lports = DEF_CFG_NUM_LPORTS; + cfg->fwcfg.num_rports = DEF_CFG_NUM_RPORTS; + cfg->fwcfg.num_ioim_reqs = DEF_CFG_NUM_IOIM_REQS; + cfg->fwcfg.num_tskim_reqs = DEF_CFG_NUM_TSKIM_REQS; + cfg->fwcfg.num_fcxp_reqs = DEF_CFG_NUM_FCXP_REQS; + cfg->fwcfg.num_uf_bufs = DEF_CFG_NUM_UF_BUFS; + cfg->fwcfg.num_cqs = DEF_CFG_NUM_CQS; + cfg->fwcfg.num_fwtio_reqs = 0; + + cfg->drvcfg.num_reqq_elems = DEF_CFG_NUM_REQQ_ELEMS; + cfg->drvcfg.num_rspq_elems = DEF_CFG_NUM_RSPQ_ELEMS; + cfg->drvcfg.num_sgpgs = DEF_CFG_NUM_SGPGS; + cfg->drvcfg.num_sboot_tgts = DEF_CFG_NUM_SBOOT_TGTS; + cfg->drvcfg.num_sboot_luns = DEF_CFG_NUM_SBOOT_LUNS; + cfg->drvcfg.path_tov = BFA_FCPIM_PATHTOV_DEF; + cfg->drvcfg.ioc_recover = BFA_FALSE; + cfg->drvcfg.delay_comp = BFA_FALSE; + +} + +void +bfa_cfg_get_min(struct bfa_iocfc_cfg_s *cfg) +{ + bfa_cfg_get_default(cfg); + cfg->fwcfg.num_ioim_reqs = BFA_IOIM_MIN; + cfg->fwcfg.num_tskim_reqs = BFA_TSKIM_MIN; + cfg->fwcfg.num_fcxp_reqs = BFA_FCXP_MIN; + cfg->fwcfg.num_uf_bufs = BFA_UF_MIN; + cfg->fwcfg.num_rports = BFA_RPORT_MIN; + cfg->fwcfg.num_fwtio_reqs = 0; + + cfg->drvcfg.num_sgpgs = BFA_SGPG_MIN; + cfg->drvcfg.num_reqq_elems = BFA_REQQ_NELEMS_MIN; + cfg->drvcfg.num_rspq_elems = BFA_RSPQ_NELEMS_MIN; + cfg->drvcfg.min_cfg = BFA_TRUE; +} |