summaryrefslogtreecommitdiffstats
path: root/drivers/scsi/lpfc/lpfc_mbox.c
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-27 10:05:51 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-27 10:05:51 +0000
commit5d1646d90e1f2cceb9f0828f4b28318cd0ec7744 (patch)
treea94efe259b9009378be6d90eb30d2b019d95c194 /drivers/scsi/lpfc/lpfc_mbox.c
parentInitial commit. (diff)
downloadlinux-upstream.tar.xz
linux-upstream.zip
Adding upstream version 5.10.209.upstream/5.10.209upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'drivers/scsi/lpfc/lpfc_mbox.c')
-rw-r--r--drivers/scsi/lpfc/lpfc_mbox.c2626
1 files changed, 2626 insertions, 0 deletions
diff --git a/drivers/scsi/lpfc/lpfc_mbox.c b/drivers/scsi/lpfc/lpfc_mbox.c
new file mode 100644
index 000000000..8764fdfc4
--- /dev/null
+++ b/drivers/scsi/lpfc/lpfc_mbox.c
@@ -0,0 +1,2626 @@
+/*******************************************************************
+ * This file is part of the Emulex Linux Device Driver for *
+ * Fibre Channel Host Bus Adapters. *
+ * Copyright (C) 2017-2018 Broadcom. All Rights Reserved. The term *
+ * “Broadcom” refers to Broadcom Inc. and/or its subsidiaries. *
+ * Copyright (C) 2004-2016 Emulex. All rights reserved. *
+ * EMULEX and SLI are trademarks of Emulex. *
+ * www.broadcom.com *
+ * Portions Copyright (C) 2004-2005 Christoph Hellwig *
+ * *
+ * This program is free software; you can redistribute it and/or *
+ * modify it under the terms of version 2 of the GNU General *
+ * Public License as published by the Free Software Foundation. *
+ * This program is distributed in the hope that it will be useful. *
+ * ALL EXPRESS OR IMPLIED CONDITIONS, REPRESENTATIONS AND *
+ * WARRANTIES, INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY, *
+ * FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT, ARE *
+ * DISCLAIMED, EXCEPT TO THE EXTENT THAT SUCH DISCLAIMERS ARE HELD *
+ * TO BE LEGALLY INVALID. See the GNU General Public License for *
+ * more details, a copy of which can be found in the file COPYING *
+ * included with this package. *
+ *******************************************************************/
+
+#include <linux/blkdev.h>
+#include <linux/pci.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_transport_fc.h>
+#include <scsi/scsi.h>
+#include <scsi/fc/fc_fs.h>
+
+#include "lpfc_hw4.h"
+#include "lpfc_hw.h"
+#include "lpfc_sli.h"
+#include "lpfc_sli4.h"
+#include "lpfc_nl.h"
+#include "lpfc_disc.h"
+#include "lpfc_scsi.h"
+#include "lpfc.h"
+#include "lpfc_logmsg.h"
+#include "lpfc_crtn.h"
+#include "lpfc_compat.h"
+
+/**
+ * lpfc_dump_static_vport - Dump HBA's static vport information.
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @offset: offset for dumping vport info.
+ *
+ * The dump mailbox command provides a method for the device driver to obtain
+ * various types of information from the HBA device.
+ *
+ * This routine prepares the mailbox command for dumping list of static
+ * vports to be created.
+ **/
+int
+lpfc_dump_static_vport(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb,
+ uint16_t offset)
+{
+ MAILBOX_t *mb;
+ struct lpfc_dmabuf *mp;
+
+ mb = &pmb->u.mb;
+
+ /* Setup to dump vport info region */
+ memset(pmb, 0, sizeof(LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_DUMP_MEMORY;
+ mb->un.varDmp.type = DMP_NV_PARAMS;
+ mb->un.varDmp.entry_index = offset;
+ mb->un.varDmp.region_id = DMP_REGION_VPORT;
+ mb->mbxOwner = OWN_HOST;
+
+ /* For SLI3 HBAs data is embedded in mailbox */
+ if (phba->sli_rev != LPFC_SLI_REV4) {
+ mb->un.varDmp.cv = 1;
+ mb->un.varDmp.word_cnt = DMP_RSP_SIZE/sizeof(uint32_t);
+ return 0;
+ }
+
+ /* For SLI4 HBAs driver need to allocate memory */
+ mp = kmalloc(sizeof(struct lpfc_dmabuf), GFP_KERNEL);
+ if (mp)
+ mp->virt = lpfc_mbuf_alloc(phba, 0, &mp->phys);
+
+ if (!mp || !mp->virt) {
+ kfree(mp);
+ lpfc_printf_log(phba, KERN_ERR, LOG_MBOX,
+ "2605 lpfc_dump_static_vport: memory"
+ " allocation failed\n");
+ return 1;
+ }
+ memset(mp->virt, 0, LPFC_BPL_SIZE);
+ INIT_LIST_HEAD(&mp->list);
+ /* save address for completion */
+ pmb->ctx_buf = (uint8_t *)mp;
+ mb->un.varWords[3] = putPaddrLow(mp->phys);
+ mb->un.varWords[4] = putPaddrHigh(mp->phys);
+ mb->un.varDmp.sli4_length = sizeof(struct static_vport_info);
+
+ return 0;
+}
+
+/**
+ * lpfc_down_link - Bring down HBAs link.
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * This routine prepares a mailbox command to bring down HBA link.
+ **/
+void
+lpfc_down_link(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t *mb;
+ memset(pmb, 0, sizeof(LPFC_MBOXQ_t));
+ mb = &pmb->u.mb;
+ mb->mbxCommand = MBX_DOWN_LINK;
+ mb->mbxOwner = OWN_HOST;
+}
+
+/**
+ * lpfc_dump_mem - Prepare a mailbox command for reading a region.
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @offset: offset into the region.
+ * @region_id: config region id.
+ *
+ * The dump mailbox command provides a method for the device driver to obtain
+ * various types of information from the HBA device.
+ *
+ * This routine prepares the mailbox command for dumping HBA's config region.
+ **/
+void
+lpfc_dump_mem(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb, uint16_t offset,
+ uint16_t region_id)
+{
+ MAILBOX_t *mb;
+ void *ctx;
+
+ mb = &pmb->u.mb;
+ ctx = pmb->ctx_buf;
+
+ /* Setup to dump VPD region */
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_DUMP_MEMORY;
+ mb->un.varDmp.cv = 1;
+ mb->un.varDmp.type = DMP_NV_PARAMS;
+ mb->un.varDmp.entry_index = offset;
+ mb->un.varDmp.region_id = region_id;
+ mb->un.varDmp.word_cnt = (DMP_RSP_SIZE / sizeof (uint32_t));
+ mb->un.varDmp.co = 0;
+ mb->un.varDmp.resp_offset = 0;
+ pmb->ctx_buf = ctx;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_dump_wakeup_param - Prepare mailbox command for retrieving wakeup params
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * This function create a dump memory mailbox command to dump wake up
+ * parameters.
+ */
+void
+lpfc_dump_wakeup_param(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t *mb;
+ void *ctx;
+
+ mb = &pmb->u.mb;
+ /* Save context so that we can restore after memset */
+ ctx = pmb->ctx_buf;
+
+ /* Setup to dump VPD region */
+ memset(pmb, 0, sizeof(LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_DUMP_MEMORY;
+ mb->mbxOwner = OWN_HOST;
+ mb->un.varDmp.cv = 1;
+ mb->un.varDmp.type = DMP_NV_PARAMS;
+ if (phba->sli_rev < LPFC_SLI_REV4)
+ mb->un.varDmp.entry_index = 0;
+ mb->un.varDmp.region_id = WAKE_UP_PARMS_REGION_ID;
+ mb->un.varDmp.word_cnt = WAKE_UP_PARMS_WORD_SIZE;
+ mb->un.varDmp.co = 0;
+ mb->un.varDmp.resp_offset = 0;
+ pmb->ctx_buf = ctx;
+ return;
+}
+
+/**
+ * lpfc_read_nv - Prepare a mailbox command for reading HBA's NVRAM param
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The read NVRAM mailbox command returns the HBA's non-volatile parameters
+ * that are used as defaults when the Fibre Channel link is brought on-line.
+ *
+ * This routine prepares the mailbox command for reading information stored
+ * in the HBA's NVRAM. Specifically, the HBA's WWNN and WWPN.
+ **/
+void
+lpfc_read_nv(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_READ_NV;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_config_async - Prepare a mailbox command for enabling HBA async event
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @ring: ring number for the asynchronous event to be configured.
+ *
+ * The asynchronous event enable mailbox command is used to enable the
+ * asynchronous event posting via the ASYNC_STATUS_CN IOCB response and
+ * specifies the default ring to which events are posted.
+ *
+ * This routine prepares the mailbox command for enabling HBA asynchronous
+ * event support on a IOCB ring.
+ **/
+void
+lpfc_config_async(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb,
+ uint32_t ring)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_ASYNCEVT_ENABLE;
+ mb->un.varCfgAsyncEvent.ring = ring;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_heart_beat - Prepare a mailbox command for heart beat
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The heart beat mailbox command is used to detect an unresponsive HBA, which
+ * is defined as any device where no error attention is sent and both mailbox
+ * and rings are not processed.
+ *
+ * This routine prepares the mailbox command for issuing a heart beat in the
+ * form of mailbox command to the HBA. The timely completion of the heart
+ * beat mailbox command indicates the health of the HBA.
+ **/
+void
+lpfc_heart_beat(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_HEARTBEAT;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_read_topology - Prepare a mailbox command for reading HBA topology
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @mp: DMA buffer memory for reading the link attention information into.
+ *
+ * The read topology mailbox command is issued to read the link topology
+ * information indicated by the HBA port when the Link Event bit of the Host
+ * Attention (HSTATT) register is set to 1 (For SLI-3) or when an FC Link
+ * Attention ACQE is received from the port (For SLI-4). A Link Event
+ * Attention occurs based on an exception detected at the Fibre Channel link
+ * interface.
+ *
+ * This routine prepares the mailbox command for reading HBA link topology
+ * information. A DMA memory has been set aside and address passed to the
+ * HBA through @mp for the HBA to DMA link attention information into the
+ * memory as part of the execution of the mailbox command.
+ *
+ * Return codes
+ * 0 - Success (currently always return 0)
+ **/
+int
+lpfc_read_topology(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb,
+ struct lpfc_dmabuf *mp)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ INIT_LIST_HEAD(&mp->list);
+ mb->mbxCommand = MBX_READ_TOPOLOGY;
+ mb->un.varReadTop.lilpBde64.tus.f.bdeSize = LPFC_ALPA_MAP_SIZE;
+ mb->un.varReadTop.lilpBde64.addrHigh = putPaddrHigh(mp->phys);
+ mb->un.varReadTop.lilpBde64.addrLow = putPaddrLow(mp->phys);
+
+ /* Save address for later completion and set the owner to host so that
+ * the FW knows this mailbox is available for processing.
+ */
+ pmb->ctx_buf = (uint8_t *)mp;
+ mb->mbxOwner = OWN_HOST;
+ return (0);
+}
+
+/**
+ * lpfc_clear_la - Prepare a mailbox command for clearing HBA link attention
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The clear link attention mailbox command is issued to clear the link event
+ * attention condition indicated by the Link Event bit of the Host Attention
+ * (HSTATT) register. The link event attention condition is cleared only if
+ * the event tag specified matches that of the current link event counter.
+ * The current event tag is read using the read link attention event mailbox
+ * command.
+ *
+ * This routine prepares the mailbox command for clearing HBA link attention
+ * information.
+ **/
+void
+lpfc_clear_la(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->un.varClearLA.eventTag = phba->fc_eventTag;
+ mb->mbxCommand = MBX_CLEAR_LA;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_config_link - Prepare a mailbox command for configuring link on a HBA
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The configure link mailbox command is used before the initialize link
+ * mailbox command to override default value and to configure link-oriented
+ * parameters such as DID address and various timers. Typically, this
+ * command would be used after an F_Port login to set the returned DID address
+ * and the fabric timeout values. This command is not valid before a configure
+ * port command has configured the HBA port.
+ *
+ * This routine prepares the mailbox command for configuring link on a HBA.
+ **/
+void
+lpfc_config_link(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ struct lpfc_vport *vport = phba->pport;
+ MAILBOX_t *mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ /* NEW_FEATURE
+ * SLI-2, Coalescing Response Feature.
+ */
+ if (phba->cfg_cr_delay && (phba->sli_rev < LPFC_SLI_REV4)) {
+ mb->un.varCfgLnk.cr = 1;
+ mb->un.varCfgLnk.ci = 1;
+ mb->un.varCfgLnk.cr_delay = phba->cfg_cr_delay;
+ mb->un.varCfgLnk.cr_count = phba->cfg_cr_count;
+ }
+
+ mb->un.varCfgLnk.myId = vport->fc_myDID;
+ mb->un.varCfgLnk.edtov = phba->fc_edtov;
+ mb->un.varCfgLnk.arbtov = phba->fc_arbtov;
+ mb->un.varCfgLnk.ratov = phba->fc_ratov;
+ mb->un.varCfgLnk.rttov = phba->fc_rttov;
+ mb->un.varCfgLnk.altov = phba->fc_altov;
+ mb->un.varCfgLnk.crtov = phba->fc_crtov;
+ mb->un.varCfgLnk.cscn = 0;
+ if (phba->bbcredit_support && phba->cfg_enable_bbcr) {
+ mb->un.varCfgLnk.cscn = 1;
+ mb->un.varCfgLnk.bbscn = bf_get(lpfc_bbscn_def,
+ &phba->sli4_hba.bbscn_params);
+ }
+
+ if (phba->cfg_ack0 && (phba->sli_rev < LPFC_SLI_REV4))
+ mb->un.varCfgLnk.ack0_enable = 1;
+
+ mb->mbxCommand = MBX_CONFIG_LINK;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_config_msi - Prepare a mailbox command for configuring msi-x
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The configure MSI-X mailbox command is used to configure the HBA's SLI-3
+ * MSI-X multi-message interrupt vector association to interrupt attention
+ * conditions.
+ *
+ * Return codes
+ * 0 - Success
+ * -EINVAL - Failure
+ **/
+int
+lpfc_config_msi(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+ uint32_t attentionConditions[2];
+
+ /* Sanity check */
+ if (phba->cfg_use_msi != 2) {
+ lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
+ "0475 Not configured for supporting MSI-X "
+ "cfg_use_msi: 0x%x\n", phba->cfg_use_msi);
+ return -EINVAL;
+ }
+
+ if (phba->sli_rev < 3) {
+ lpfc_printf_log(phba, KERN_ERR, LOG_INIT,
+ "0476 HBA not supporting SLI-3 or later "
+ "SLI Revision: 0x%x\n", phba->sli_rev);
+ return -EINVAL;
+ }
+
+ /* Clear mailbox command fields */
+ memset(pmb, 0, sizeof(LPFC_MBOXQ_t));
+
+ /*
+ * SLI-3, Message Signaled Interrupt Fearure.
+ */
+
+ /* Multi-message attention configuration */
+ attentionConditions[0] = (HA_R0ATT | HA_R1ATT | HA_R2ATT | HA_ERATT |
+ HA_LATT | HA_MBATT);
+ attentionConditions[1] = 0;
+
+ mb->un.varCfgMSI.attentionConditions[0] = attentionConditions[0];
+ mb->un.varCfgMSI.attentionConditions[1] = attentionConditions[1];
+
+ /*
+ * Set up message number to HA bit association
+ */
+#ifdef __BIG_ENDIAN_BITFIELD
+ /* RA0 (FCP Ring) */
+ mb->un.varCfgMSI.messageNumberByHA[HA_R0_POS] = 1;
+ /* RA1 (Other Protocol Extra Ring) */
+ mb->un.varCfgMSI.messageNumberByHA[HA_R1_POS] = 1;
+#else /* __LITTLE_ENDIAN_BITFIELD */
+ /* RA0 (FCP Ring) */
+ mb->un.varCfgMSI.messageNumberByHA[HA_R0_POS^3] = 1;
+ /* RA1 (Other Protocol Extra Ring) */
+ mb->un.varCfgMSI.messageNumberByHA[HA_R1_POS^3] = 1;
+#endif
+ /* Multi-message interrupt autoclear configuration*/
+ mb->un.varCfgMSI.autoClearHA[0] = attentionConditions[0];
+ mb->un.varCfgMSI.autoClearHA[1] = attentionConditions[1];
+
+ /* For now, HBA autoclear does not work reliably, disable it */
+ mb->un.varCfgMSI.autoClearHA[0] = 0;
+ mb->un.varCfgMSI.autoClearHA[1] = 0;
+
+ /* Set command and owner bit */
+ mb->mbxCommand = MBX_CONFIG_MSI;
+ mb->mbxOwner = OWN_HOST;
+
+ return 0;
+}
+
+/**
+ * lpfc_init_link - Prepare a mailbox command for initialize link on a HBA
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @topology: the link topology for the link to be initialized to.
+ * @linkspeed: the link speed for the link to be initialized to.
+ *
+ * The initialize link mailbox command is used to initialize the Fibre
+ * Channel link. This command must follow a configure port command that
+ * establishes the mode of operation.
+ *
+ * This routine prepares the mailbox command for initializing link on a HBA
+ * with the specified link topology and speed.
+ **/
+void
+lpfc_init_link(struct lpfc_hba * phba,
+ LPFC_MBOXQ_t * pmb, uint32_t topology, uint32_t linkspeed)
+{
+ lpfc_vpd_t *vpd;
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ switch (topology) {
+ case FLAGS_TOPOLOGY_MODE_LOOP_PT:
+ mb->un.varInitLnk.link_flags = FLAGS_TOPOLOGY_MODE_LOOP;
+ mb->un.varInitLnk.link_flags |= FLAGS_TOPOLOGY_FAILOVER;
+ break;
+ case FLAGS_TOPOLOGY_MODE_PT_PT:
+ mb->un.varInitLnk.link_flags = FLAGS_TOPOLOGY_MODE_PT_PT;
+ break;
+ case FLAGS_TOPOLOGY_MODE_LOOP:
+ mb->un.varInitLnk.link_flags = FLAGS_TOPOLOGY_MODE_LOOP;
+ break;
+ case FLAGS_TOPOLOGY_MODE_PT_LOOP:
+ mb->un.varInitLnk.link_flags = FLAGS_TOPOLOGY_MODE_PT_PT;
+ mb->un.varInitLnk.link_flags |= FLAGS_TOPOLOGY_FAILOVER;
+ break;
+ case FLAGS_LOCAL_LB:
+ mb->un.varInitLnk.link_flags = FLAGS_LOCAL_LB;
+ break;
+ }
+
+ if ((phba->pcidev->device == PCI_DEVICE_ID_LANCER_G6_FC ||
+ phba->pcidev->device == PCI_DEVICE_ID_LANCER_G7_FC) &&
+ !(phba->sli4_hba.pc_sli4_params.pls) &&
+ mb->un.varInitLnk.link_flags & FLAGS_TOPOLOGY_MODE_LOOP) {
+ mb->un.varInitLnk.link_flags = FLAGS_TOPOLOGY_MODE_PT_PT;
+ phba->cfg_topology = FLAGS_TOPOLOGY_MODE_PT_PT;
+ }
+
+ /* Enable asynchronous ABTS responses from firmware */
+ mb->un.varInitLnk.link_flags |= FLAGS_IMED_ABORT;
+
+ /* NEW_FEATURE
+ * Setting up the link speed
+ */
+ vpd = &phba->vpd;
+ if (vpd->rev.feaLevelHigh >= 0x02){
+ switch(linkspeed){
+ case LPFC_USER_LINK_SPEED_1G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_1G;
+ break;
+ case LPFC_USER_LINK_SPEED_2G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_2G;
+ break;
+ case LPFC_USER_LINK_SPEED_4G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_4G;
+ break;
+ case LPFC_USER_LINK_SPEED_8G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_8G;
+ break;
+ case LPFC_USER_LINK_SPEED_10G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_10G;
+ break;
+ case LPFC_USER_LINK_SPEED_16G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_16G;
+ break;
+ case LPFC_USER_LINK_SPEED_32G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_32G;
+ break;
+ case LPFC_USER_LINK_SPEED_64G:
+ mb->un.varInitLnk.link_flags |= FLAGS_LINK_SPEED;
+ mb->un.varInitLnk.link_speed = LINK_SPEED_64G;
+ break;
+ case LPFC_USER_LINK_SPEED_AUTO:
+ default:
+ mb->un.varInitLnk.link_speed = LINK_SPEED_AUTO;
+ break;
+ }
+
+ }
+ else
+ mb->un.varInitLnk.link_speed = LINK_SPEED_AUTO;
+
+ mb->mbxCommand = (volatile uint8_t)MBX_INIT_LINK;
+ mb->mbxOwner = OWN_HOST;
+ mb->un.varInitLnk.fabric_AL_PA = phba->fc_pref_ALPA;
+ return;
+}
+
+/**
+ * lpfc_read_sparam - Prepare a mailbox command for reading HBA parameters
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @vpi: virtual N_Port identifier.
+ *
+ * The read service parameter mailbox command is used to read the HBA port
+ * service parameters. The service parameters are read into the buffer
+ * specified directly by a BDE in the mailbox command. These service
+ * parameters may then be used to build the payload of an N_Port/F_POrt
+ * login request and reply (LOGI/ACC).
+ *
+ * This routine prepares the mailbox command for reading HBA port service
+ * parameters. The DMA memory is allocated in this function and the addresses
+ * are populated into the mailbox command for the HBA to DMA the service
+ * parameters into.
+ *
+ * Return codes
+ * 0 - Success
+ * 1 - DMA memory allocation failed
+ **/
+int
+lpfc_read_sparam(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb, int vpi)
+{
+ struct lpfc_dmabuf *mp;
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->mbxOwner = OWN_HOST;
+
+ /* Get a buffer to hold the HBAs Service Parameters */
+
+ mp = kmalloc(sizeof (struct lpfc_dmabuf), GFP_KERNEL);
+ if (mp)
+ mp->virt = lpfc_mbuf_alloc(phba, 0, &mp->phys);
+ if (!mp || !mp->virt) {
+ kfree(mp);
+ mb->mbxCommand = MBX_READ_SPARM64;
+ /* READ_SPARAM: no buffers */
+ lpfc_printf_log(phba, KERN_WARNING, LOG_MBOX,
+ "0301 READ_SPARAM: no buffers\n");
+ return (1);
+ }
+ INIT_LIST_HEAD(&mp->list);
+ mb->mbxCommand = MBX_READ_SPARM64;
+ mb->un.varRdSparm.un.sp64.tus.f.bdeSize = sizeof (struct serv_parm);
+ mb->un.varRdSparm.un.sp64.addrHigh = putPaddrHigh(mp->phys);
+ mb->un.varRdSparm.un.sp64.addrLow = putPaddrLow(mp->phys);
+ if (phba->sli_rev >= LPFC_SLI_REV3)
+ mb->un.varRdSparm.vpi = phba->vpi_ids[vpi];
+
+ /* save address for completion */
+ pmb->ctx_buf = mp;
+
+ return (0);
+}
+
+/**
+ * lpfc_unreg_did - Prepare a mailbox command for unregistering DID
+ * @phba: pointer to lpfc hba data structure.
+ * @vpi: virtual N_Port identifier.
+ * @did: remote port identifier.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The unregister DID mailbox command is used to unregister an N_Port/F_Port
+ * login for an unknown RPI by specifying the DID of a remote port. This
+ * command frees an RPI context in the HBA port. This has the effect of
+ * performing an implicit N_Port/F_Port logout.
+ *
+ * This routine prepares the mailbox command for unregistering a remote
+ * N_Port/F_Port (DID) login.
+ **/
+void
+lpfc_unreg_did(struct lpfc_hba * phba, uint16_t vpi, uint32_t did,
+ LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->un.varUnregDID.did = did;
+ mb->un.varUnregDID.vpi = vpi;
+ if ((vpi != 0xffff) &&
+ (phba->sli_rev == LPFC_SLI_REV4))
+ mb->un.varUnregDID.vpi = phba->vpi_ids[vpi];
+
+ mb->mbxCommand = MBX_UNREG_D_ID;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_read_config - Prepare a mailbox command for reading HBA configuration
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The read configuration mailbox command is used to read the HBA port
+ * configuration parameters. This mailbox command provides a method for
+ * seeing any parameters that may have changed via various configuration
+ * mailbox commands.
+ *
+ * This routine prepares the mailbox command for reading out HBA configuration
+ * parameters.
+ **/
+void
+lpfc_read_config(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->mbxCommand = MBX_READ_CONFIG;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_read_lnk_stat - Prepare a mailbox command for reading HBA link stats
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The read link status mailbox command is used to read the link status from
+ * the HBA. Link status includes all link-related error counters. These
+ * counters are maintained by the HBA and originated in the link hardware
+ * unit. Note that all of these counters wrap.
+ *
+ * This routine prepares the mailbox command for reading out HBA link status.
+ **/
+void
+lpfc_read_lnk_stat(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->mbxCommand = MBX_READ_LNK_STAT;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_reg_rpi - Prepare a mailbox command for registering remote login
+ * @phba: pointer to lpfc hba data structure.
+ * @vpi: virtual N_Port identifier.
+ * @did: remote port identifier.
+ * @param: pointer to memory holding the server parameters.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ * @rpi: the rpi to use in the registration (usually only used for SLI4.
+ *
+ * The registration login mailbox command is used to register an N_Port or
+ * F_Port login. This registration allows the HBA to cache the remote N_Port
+ * service parameters internally and thereby make the appropriate FC-2
+ * decisions. The remote port service parameters are handed off by the driver
+ * to the HBA using a descriptor entry that directly identifies a buffer in
+ * host memory. In exchange, the HBA returns an RPI identifier.
+ *
+ * This routine prepares the mailbox command for registering remote port login.
+ * The function allocates DMA buffer for passing the service parameters to the
+ * HBA with the mailbox command.
+ *
+ * Return codes
+ * 0 - Success
+ * 1 - DMA memory allocation failed
+ **/
+int
+lpfc_reg_rpi(struct lpfc_hba *phba, uint16_t vpi, uint32_t did,
+ uint8_t *param, LPFC_MBOXQ_t *pmb, uint16_t rpi)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+ uint8_t *sparam;
+ struct lpfc_dmabuf *mp;
+
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->un.varRegLogin.rpi = 0;
+ if (phba->sli_rev == LPFC_SLI_REV4)
+ mb->un.varRegLogin.rpi = phba->sli4_hba.rpi_ids[rpi];
+ if (phba->sli_rev >= LPFC_SLI_REV3)
+ mb->un.varRegLogin.vpi = phba->vpi_ids[vpi];
+ mb->un.varRegLogin.did = did;
+ mb->mbxOwner = OWN_HOST;
+ /* Get a buffer to hold NPorts Service Parameters */
+ mp = kmalloc(sizeof (struct lpfc_dmabuf), GFP_KERNEL);
+ if (mp)
+ mp->virt = lpfc_mbuf_alloc(phba, 0, &mp->phys);
+ if (!mp || !mp->virt) {
+ kfree(mp);
+ mb->mbxCommand = MBX_REG_LOGIN64;
+ /* REG_LOGIN: no buffers */
+ lpfc_printf_log(phba, KERN_WARNING, LOG_MBOX,
+ "0302 REG_LOGIN: no buffers, VPI:%d DID:x%x, "
+ "rpi x%x\n", vpi, did, rpi);
+ return 1;
+ }
+ INIT_LIST_HEAD(&mp->list);
+ sparam = mp->virt;
+
+ /* Copy param's into a new buffer */
+ memcpy(sparam, param, sizeof (struct serv_parm));
+
+ /* save address for completion */
+ pmb->ctx_buf = (uint8_t *)mp;
+
+ mb->mbxCommand = MBX_REG_LOGIN64;
+ mb->un.varRegLogin.un.sp64.tus.f.bdeSize = sizeof (struct serv_parm);
+ mb->un.varRegLogin.un.sp64.addrHigh = putPaddrHigh(mp->phys);
+ mb->un.varRegLogin.un.sp64.addrLow = putPaddrLow(mp->phys);
+
+ return 0;
+}
+
+/**
+ * lpfc_unreg_login - Prepare a mailbox command for unregistering remote login
+ * @phba: pointer to lpfc hba data structure.
+ * @vpi: virtual N_Port identifier.
+ * @rpi: remote port identifier
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The unregistration login mailbox command is used to unregister an N_Port
+ * or F_Port login. This command frees an RPI context in the HBA. It has the
+ * effect of performing an implicit N_Port/F_Port logout.
+ *
+ * This routine prepares the mailbox command for unregistering remote port
+ * login.
+ *
+ * For SLI4 ports, the rpi passed to this function must be the physical
+ * rpi value, not the logical index.
+ **/
+void
+lpfc_unreg_login(struct lpfc_hba *phba, uint16_t vpi, uint32_t rpi,
+ LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb;
+
+ mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->un.varUnregLogin.rpi = rpi;
+ mb->un.varUnregLogin.rsvd1 = 0;
+ if (phba->sli_rev >= LPFC_SLI_REV3)
+ mb->un.varUnregLogin.vpi = phba->vpi_ids[vpi];
+
+ mb->mbxCommand = MBX_UNREG_LOGIN;
+ mb->mbxOwner = OWN_HOST;
+
+ return;
+}
+
+/**
+ * lpfc_sli4_unreg_all_rpis - unregister all RPIs for a vport on SLI4 HBA.
+ * @vport: pointer to a vport object.
+ *
+ * This routine sends mailbox command to unregister all active RPIs for
+ * a vport.
+ **/
+void
+lpfc_sli4_unreg_all_rpis(struct lpfc_vport *vport)
+{
+ struct lpfc_hba *phba = vport->phba;
+ LPFC_MBOXQ_t *mbox;
+ int rc;
+
+ mbox = mempool_alloc(phba->mbox_mem_pool, GFP_KERNEL);
+ if (mbox) {
+ /*
+ * For SLI4 functions, the rpi field is overloaded for
+ * the vport context unreg all. This routine passes
+ * 0 for the rpi field in lpfc_unreg_login for compatibility
+ * with SLI3 and then overrides the rpi field with the
+ * expected value for SLI4.
+ */
+ lpfc_unreg_login(phba, vport->vpi, phba->vpi_ids[vport->vpi],
+ mbox);
+ mbox->u.mb.un.varUnregLogin.rsvd1 = 0x4000;
+ mbox->vport = vport;
+ mbox->mbox_cmpl = lpfc_sli_def_mbox_cmpl;
+ mbox->ctx_ndlp = NULL;
+ rc = lpfc_sli_issue_mbox(phba, mbox, MBX_NOWAIT);
+ if (rc == MBX_NOT_FINISHED)
+ mempool_free(mbox, phba->mbox_mem_pool);
+ }
+}
+
+/**
+ * lpfc_reg_vpi - Prepare a mailbox command for registering vport identifier
+ * @vport: pointer to a vport object.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The registration vport identifier mailbox command is used to activate a
+ * virtual N_Port after it has acquired an N_Port_ID. The HBA validates the
+ * N_Port_ID against the information in the selected virtual N_Port context
+ * block and marks it active to allow normal processing of IOCB commands and
+ * received unsolicited exchanges.
+ *
+ * This routine prepares the mailbox command for registering a virtual N_Port.
+ **/
+void
+lpfc_reg_vpi(struct lpfc_vport *vport, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+ struct lpfc_hba *phba = vport->phba;
+
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ /*
+ * Set the re-reg VPI bit for f/w to update the MAC address.
+ */
+ if ((phba->sli_rev == LPFC_SLI_REV4) &&
+ !(vport->fc_flag & FC_VPORT_NEEDS_REG_VPI))
+ mb->un.varRegVpi.upd = 1;
+
+ mb->un.varRegVpi.vpi = phba->vpi_ids[vport->vpi];
+ mb->un.varRegVpi.sid = vport->fc_myDID;
+ if (phba->sli_rev == LPFC_SLI_REV4)
+ mb->un.varRegVpi.vfi = phba->sli4_hba.vfi_ids[vport->vfi];
+ else
+ mb->un.varRegVpi.vfi = vport->vfi + vport->phba->vfi_base;
+ memcpy(mb->un.varRegVpi.wwn, &vport->fc_portname,
+ sizeof(struct lpfc_name));
+ mb->un.varRegVpi.wwn[0] = cpu_to_le32(mb->un.varRegVpi.wwn[0]);
+ mb->un.varRegVpi.wwn[1] = cpu_to_le32(mb->un.varRegVpi.wwn[1]);
+
+ mb->mbxCommand = MBX_REG_VPI;
+ mb->mbxOwner = OWN_HOST;
+ return;
+
+}
+
+/**
+ * lpfc_unreg_vpi - Prepare a mailbox command for unregistering vport id
+ * @phba: pointer to lpfc hba data structure.
+ * @vpi: virtual N_Port identifier.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The unregistration vport identifier mailbox command is used to inactivate
+ * a virtual N_Port. The driver must have logged out and unregistered all
+ * remote N_Ports to abort any activity on the virtual N_Port. The HBA will
+ * unregisters any default RPIs associated with the specified vpi, aborting
+ * any active exchanges. The HBA will post the mailbox response after making
+ * the virtual N_Port inactive.
+ *
+ * This routine prepares the mailbox command for unregistering a virtual
+ * N_Port.
+ **/
+void
+lpfc_unreg_vpi(struct lpfc_hba *phba, uint16_t vpi, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ if (phba->sli_rev == LPFC_SLI_REV3)
+ mb->un.varUnregVpi.vpi = phba->vpi_ids[vpi];
+ else if (phba->sli_rev >= LPFC_SLI_REV4)
+ mb->un.varUnregVpi.sli4_vpi = phba->vpi_ids[vpi];
+
+ mb->mbxCommand = MBX_UNREG_VPI;
+ mb->mbxOwner = OWN_HOST;
+ return;
+
+}
+
+/**
+ * lpfc_config_pcb_setup - Set up IOCB rings in the Port Control Block (PCB)
+ * @phba: pointer to lpfc hba data structure.
+ *
+ * This routine sets up and initializes the IOCB rings in the Port Control
+ * Block (PCB).
+ **/
+static void
+lpfc_config_pcb_setup(struct lpfc_hba * phba)
+{
+ struct lpfc_sli *psli = &phba->sli;
+ struct lpfc_sli_ring *pring;
+ PCB_t *pcbp = phba->pcb;
+ dma_addr_t pdma_addr;
+ uint32_t offset;
+ uint32_t iocbCnt = 0;
+ int i;
+
+ pcbp->maxRing = (psli->num_rings - 1);
+
+ for (i = 0; i < psli->num_rings; i++) {
+ pring = &psli->sli3_ring[i];
+
+ pring->sli.sli3.sizeCiocb =
+ phba->sli_rev == 3 ? SLI3_IOCB_CMD_SIZE :
+ SLI2_IOCB_CMD_SIZE;
+ pring->sli.sli3.sizeRiocb =
+ phba->sli_rev == 3 ? SLI3_IOCB_RSP_SIZE :
+ SLI2_IOCB_RSP_SIZE;
+ /* A ring MUST have both cmd and rsp entries defined to be
+ valid */
+ if ((pring->sli.sli3.numCiocb == 0) ||
+ (pring->sli.sli3.numRiocb == 0)) {
+ pcbp->rdsc[i].cmdEntries = 0;
+ pcbp->rdsc[i].rspEntries = 0;
+ pcbp->rdsc[i].cmdAddrHigh = 0;
+ pcbp->rdsc[i].rspAddrHigh = 0;
+ pcbp->rdsc[i].cmdAddrLow = 0;
+ pcbp->rdsc[i].rspAddrLow = 0;
+ pring->sli.sli3.cmdringaddr = NULL;
+ pring->sli.sli3.rspringaddr = NULL;
+ continue;
+ }
+ /* Command ring setup for ring */
+ pring->sli.sli3.cmdringaddr = (void *)&phba->IOCBs[iocbCnt];
+ pcbp->rdsc[i].cmdEntries = pring->sli.sli3.numCiocb;
+
+ offset = (uint8_t *) &phba->IOCBs[iocbCnt] -
+ (uint8_t *) phba->slim2p.virt;
+ pdma_addr = phba->slim2p.phys + offset;
+ pcbp->rdsc[i].cmdAddrHigh = putPaddrHigh(pdma_addr);
+ pcbp->rdsc[i].cmdAddrLow = putPaddrLow(pdma_addr);
+ iocbCnt += pring->sli.sli3.numCiocb;
+
+ /* Response ring setup for ring */
+ pring->sli.sli3.rspringaddr = (void *) &phba->IOCBs[iocbCnt];
+
+ pcbp->rdsc[i].rspEntries = pring->sli.sli3.numRiocb;
+ offset = (uint8_t *)&phba->IOCBs[iocbCnt] -
+ (uint8_t *)phba->slim2p.virt;
+ pdma_addr = phba->slim2p.phys + offset;
+ pcbp->rdsc[i].rspAddrHigh = putPaddrHigh(pdma_addr);
+ pcbp->rdsc[i].rspAddrLow = putPaddrLow(pdma_addr);
+ iocbCnt += pring->sli.sli3.numRiocb;
+ }
+}
+
+/**
+ * lpfc_read_rev - Prepare a mailbox command for reading HBA revision
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The read revision mailbox command is used to read the revision levels of
+ * the HBA components. These components include hardware units, resident
+ * firmware, and available firmware. HBAs that supports SLI-3 mode of
+ * operation provide different response information depending on the version
+ * requested by the driver.
+ *
+ * This routine prepares the mailbox command for reading HBA revision
+ * information.
+ **/
+void
+lpfc_read_rev(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ mb->un.varRdRev.cv = 1;
+ mb->un.varRdRev.v3req = 1; /* Request SLI3 info */
+ mb->mbxCommand = MBX_READ_REV;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+void
+lpfc_sli4_swap_str(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+ struct lpfc_mqe *mqe;
+
+ switch (mb->mbxCommand) {
+ case MBX_READ_REV:
+ mqe = &pmb->u.mqe;
+ lpfc_sli_pcimem_bcopy(mqe->un.read_rev.fw_name,
+ mqe->un.read_rev.fw_name, 16);
+ lpfc_sli_pcimem_bcopy(mqe->un.read_rev.ulp_fw_name,
+ mqe->un.read_rev.ulp_fw_name, 16);
+ break;
+ default:
+ break;
+ }
+ return;
+}
+
+/**
+ * lpfc_build_hbq_profile2 - Set up the HBQ Selection Profile 2
+ * @hbqmb: pointer to the HBQ configuration data structure in mailbox command.
+ * @hbq_desc: pointer to the HBQ selection profile descriptor.
+ *
+ * The Host Buffer Queue (HBQ) Selection Profile 2 specifies that the HBA
+ * tests the incoming frames' R_CTL/TYPE fields with works 10:15 and performs
+ * the Sequence Length Test using the fields in the Selection Profile 2
+ * extension in words 20:31.
+ **/
+static void
+lpfc_build_hbq_profile2(struct config_hbq_var *hbqmb,
+ struct lpfc_hbq_init *hbq_desc)
+{
+ hbqmb->profiles.profile2.seqlenbcnt = hbq_desc->seqlenbcnt;
+ hbqmb->profiles.profile2.maxlen = hbq_desc->maxlen;
+ hbqmb->profiles.profile2.seqlenoff = hbq_desc->seqlenoff;
+}
+
+/**
+ * lpfc_build_hbq_profile3 - Set up the HBQ Selection Profile 3
+ * @hbqmb: pointer to the HBQ configuration data structure in mailbox command.
+ * @hbq_desc: pointer to the HBQ selection profile descriptor.
+ *
+ * The Host Buffer Queue (HBQ) Selection Profile 3 specifies that the HBA
+ * tests the incoming frame's R_CTL/TYPE fields with words 10:15 and performs
+ * the Sequence Length Test and Byte Field Test using the fields in the
+ * Selection Profile 3 extension in words 20:31.
+ **/
+static void
+lpfc_build_hbq_profile3(struct config_hbq_var *hbqmb,
+ struct lpfc_hbq_init *hbq_desc)
+{
+ hbqmb->profiles.profile3.seqlenbcnt = hbq_desc->seqlenbcnt;
+ hbqmb->profiles.profile3.maxlen = hbq_desc->maxlen;
+ hbqmb->profiles.profile3.cmdcodeoff = hbq_desc->cmdcodeoff;
+ hbqmb->profiles.profile3.seqlenoff = hbq_desc->seqlenoff;
+ memcpy(&hbqmb->profiles.profile3.cmdmatch, hbq_desc->cmdmatch,
+ sizeof(hbqmb->profiles.profile3.cmdmatch));
+}
+
+/**
+ * lpfc_build_hbq_profile5 - Set up the HBQ Selection Profile 5
+ * @hbqmb: pointer to the HBQ configuration data structure in mailbox command.
+ * @hbq_desc: pointer to the HBQ selection profile descriptor.
+ *
+ * The Host Buffer Queue (HBQ) Selection Profile 5 specifies a header HBQ. The
+ * HBA tests the initial frame of an incoming sequence using the frame's
+ * R_CTL/TYPE fields with words 10:15 and performs the Sequence Length Test
+ * and Byte Field Test using the fields in the Selection Profile 5 extension
+ * words 20:31.
+ **/
+static void
+lpfc_build_hbq_profile5(struct config_hbq_var *hbqmb,
+ struct lpfc_hbq_init *hbq_desc)
+{
+ hbqmb->profiles.profile5.seqlenbcnt = hbq_desc->seqlenbcnt;
+ hbqmb->profiles.profile5.maxlen = hbq_desc->maxlen;
+ hbqmb->profiles.profile5.cmdcodeoff = hbq_desc->cmdcodeoff;
+ hbqmb->profiles.profile5.seqlenoff = hbq_desc->seqlenoff;
+ memcpy(&hbqmb->profiles.profile5.cmdmatch, hbq_desc->cmdmatch,
+ sizeof(hbqmb->profiles.profile5.cmdmatch));
+}
+
+/**
+ * lpfc_config_hbq - Prepare a mailbox command for configuring an HBQ
+ * @phba: pointer to lpfc hba data structure.
+ * @id: HBQ identifier.
+ * @hbq_desc: pointer to the HBA descriptor data structure.
+ * @hbq_entry_index: index of the HBQ entry data structures.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The configure HBQ (Host Buffer Queue) mailbox command is used to configure
+ * an HBQ. The configuration binds events that require buffers to a particular
+ * ring and HBQ based on a selection profile.
+ *
+ * This routine prepares the mailbox command for configuring an HBQ.
+ **/
+void
+lpfc_config_hbq(struct lpfc_hba *phba, uint32_t id,
+ struct lpfc_hbq_init *hbq_desc,
+ uint32_t hbq_entry_index, LPFC_MBOXQ_t *pmb)
+{
+ int i;
+ MAILBOX_t *mb = &pmb->u.mb;
+ struct config_hbq_var *hbqmb = &mb->un.varCfgHbq;
+
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+ hbqmb->hbqId = id;
+ hbqmb->entry_count = hbq_desc->entry_count; /* # entries in HBQ */
+ hbqmb->recvNotify = hbq_desc->rn; /* Receive
+ * Notification */
+ hbqmb->numMask = hbq_desc->mask_count; /* # R_CTL/TYPE masks
+ * # in words 0-19 */
+ hbqmb->profile = hbq_desc->profile; /* Selection profile:
+ * 0 = all,
+ * 7 = logentry */
+ hbqmb->ringMask = hbq_desc->ring_mask; /* Binds HBQ to a ring
+ * e.g. Ring0=b0001,
+ * ring2=b0100 */
+ hbqmb->headerLen = hbq_desc->headerLen; /* 0 if not profile 4
+ * or 5 */
+ hbqmb->logEntry = hbq_desc->logEntry; /* Set to 1 if this
+ * HBQ will be used
+ * for LogEntry
+ * buffers */
+ hbqmb->hbqaddrLow = putPaddrLow(phba->hbqslimp.phys) +
+ hbq_entry_index * sizeof(struct lpfc_hbq_entry);
+ hbqmb->hbqaddrHigh = putPaddrHigh(phba->hbqslimp.phys);
+
+ mb->mbxCommand = MBX_CONFIG_HBQ;
+ mb->mbxOwner = OWN_HOST;
+
+ /* Copy info for profiles 2,3,5. Other
+ * profiles this area is reserved
+ */
+ if (hbq_desc->profile == 2)
+ lpfc_build_hbq_profile2(hbqmb, hbq_desc);
+ else if (hbq_desc->profile == 3)
+ lpfc_build_hbq_profile3(hbqmb, hbq_desc);
+ else if (hbq_desc->profile == 5)
+ lpfc_build_hbq_profile5(hbqmb, hbq_desc);
+
+ /* Return if no rctl / type masks for this HBQ */
+ if (!hbq_desc->mask_count)
+ return;
+
+ /* Otherwise we setup specific rctl / type masks for this HBQ */
+ for (i = 0; i < hbq_desc->mask_count; i++) {
+ hbqmb->hbqMasks[i].tmatch = hbq_desc->hbqMasks[i].tmatch;
+ hbqmb->hbqMasks[i].tmask = hbq_desc->hbqMasks[i].tmask;
+ hbqmb->hbqMasks[i].rctlmatch = hbq_desc->hbqMasks[i].rctlmatch;
+ hbqmb->hbqMasks[i].rctlmask = hbq_desc->hbqMasks[i].rctlmask;
+ }
+
+ return;
+}
+
+/**
+ * lpfc_config_ring - Prepare a mailbox command for configuring an IOCB ring
+ * @phba: pointer to lpfc hba data structure.
+ * @ring: ring number/index
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The configure ring mailbox command is used to configure an IOCB ring. This
+ * configuration binds from one to six of HBA RC_CTL/TYPE mask entries to the
+ * ring. This is used to map incoming sequences to a particular ring whose
+ * RC_CTL/TYPE mask entry matches that of the sequence. The driver should not
+ * attempt to configure a ring whose number is greater than the number
+ * specified in the Port Control Block (PCB). It is an error to issue the
+ * configure ring command more than once with the same ring number. The HBA
+ * returns an error if the driver attempts this.
+ *
+ * This routine prepares the mailbox command for configuring IOCB ring.
+ **/
+void
+lpfc_config_ring(struct lpfc_hba * phba, int ring, LPFC_MBOXQ_t * pmb)
+{
+ int i;
+ MAILBOX_t *mb = &pmb->u.mb;
+ struct lpfc_sli *psli;
+ struct lpfc_sli_ring *pring;
+
+ memset(pmb, 0, sizeof (LPFC_MBOXQ_t));
+
+ mb->un.varCfgRing.ring = ring;
+ mb->un.varCfgRing.maxOrigXchg = 0;
+ mb->un.varCfgRing.maxRespXchg = 0;
+ mb->un.varCfgRing.recvNotify = 1;
+
+ psli = &phba->sli;
+ pring = &psli->sli3_ring[ring];
+ mb->un.varCfgRing.numMask = pring->num_mask;
+ mb->mbxCommand = MBX_CONFIG_RING;
+ mb->mbxOwner = OWN_HOST;
+
+ /* Is this ring configured for a specific profile */
+ if (pring->prt[0].profile) {
+ mb->un.varCfgRing.profile = pring->prt[0].profile;
+ return;
+ }
+
+ /* Otherwise we setup specific rctl / type masks for this ring */
+ for (i = 0; i < pring->num_mask; i++) {
+ mb->un.varCfgRing.rrRegs[i].rval = pring->prt[i].rctl;
+ if (mb->un.varCfgRing.rrRegs[i].rval != FC_RCTL_ELS_REQ)
+ mb->un.varCfgRing.rrRegs[i].rmask = 0xff;
+ else
+ mb->un.varCfgRing.rrRegs[i].rmask = 0xfe;
+ mb->un.varCfgRing.rrRegs[i].tval = pring->prt[i].type;
+ mb->un.varCfgRing.rrRegs[i].tmask = 0xff;
+ }
+
+ return;
+}
+
+/**
+ * lpfc_config_port - Prepare a mailbox command for configuring port
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The configure port mailbox command is used to identify the Port Control
+ * Block (PCB) in the driver memory. After this command is issued, the
+ * driver must not access the mailbox in the HBA without first resetting
+ * the HBA. The HBA may copy the PCB information to internal storage for
+ * subsequent use; the driver can not change the PCB information unless it
+ * resets the HBA.
+ *
+ * This routine prepares the mailbox command for configuring port.
+ **/
+void
+lpfc_config_port(struct lpfc_hba *phba, LPFC_MBOXQ_t *pmb)
+{
+ MAILBOX_t __iomem *mb_slim = (MAILBOX_t __iomem *) phba->MBslimaddr;
+ MAILBOX_t *mb = &pmb->u.mb;
+ dma_addr_t pdma_addr;
+ uint32_t bar_low, bar_high;
+ size_t offset;
+ struct lpfc_hgp hgp;
+ int i;
+ uint32_t pgp_offset;
+
+ memset(pmb, 0, sizeof(LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_CONFIG_PORT;
+ mb->mbxOwner = OWN_HOST;
+
+ mb->un.varCfgPort.pcbLen = sizeof(PCB_t);
+
+ offset = (uint8_t *)phba->pcb - (uint8_t *)phba->slim2p.virt;
+ pdma_addr = phba->slim2p.phys + offset;
+ mb->un.varCfgPort.pcbLow = putPaddrLow(pdma_addr);
+ mb->un.varCfgPort.pcbHigh = putPaddrHigh(pdma_addr);
+
+ /* Always Host Group Pointer is in SLIM */
+ mb->un.varCfgPort.hps = 1;
+
+ /* If HBA supports SLI=3 ask for it */
+
+ if (phba->sli_rev == LPFC_SLI_REV3 && phba->vpd.sli3Feat.cerbm) {
+ if (phba->cfg_enable_bg)
+ mb->un.varCfgPort.cbg = 1; /* configure BlockGuard */
+ mb->un.varCfgPort.cerbm = 1; /* Request HBQs */
+ mb->un.varCfgPort.ccrp = 1; /* Command Ring Polling */
+ mb->un.varCfgPort.max_hbq = lpfc_sli_hbq_count();
+ if (phba->max_vpi && phba->cfg_enable_npiv &&
+ phba->vpd.sli3Feat.cmv) {
+ mb->un.varCfgPort.max_vpi = LPFC_MAX_VPI;
+ mb->un.varCfgPort.cmv = 1;
+ } else
+ mb->un.varCfgPort.max_vpi = phba->max_vpi = 0;
+ } else
+ phba->sli_rev = LPFC_SLI_REV2;
+ mb->un.varCfgPort.sli_mode = phba->sli_rev;
+
+ /* If this is an SLI3 port, configure async status notification. */
+ if (phba->sli_rev == LPFC_SLI_REV3)
+ mb->un.varCfgPort.casabt = 1;
+
+ /* Now setup pcb */
+ phba->pcb->type = TYPE_NATIVE_SLI2;
+ phba->pcb->feature = FEATURE_INITIAL_SLI2;
+
+ /* Setup Mailbox pointers */
+ phba->pcb->mailBoxSize = sizeof(MAILBOX_t) + MAILBOX_EXT_SIZE;
+ offset = (uint8_t *)phba->mbox - (uint8_t *)phba->slim2p.virt;
+ pdma_addr = phba->slim2p.phys + offset;
+ phba->pcb->mbAddrHigh = putPaddrHigh(pdma_addr);
+ phba->pcb->mbAddrLow = putPaddrLow(pdma_addr);
+
+ /*
+ * Setup Host Group ring pointer.
+ *
+ * For efficiency reasons, the ring get/put pointers can be
+ * placed in adapter memory (SLIM) rather than in host memory.
+ * This allows firmware to avoid PCI reads/writes when updating
+ * and checking pointers.
+ *
+ * The firmware recognizes the use of SLIM memory by comparing
+ * the address of the get/put pointers structure with that of
+ * the SLIM BAR (BAR0).
+ *
+ * Caution: be sure to use the PCI config space value of BAR0/BAR1
+ * (the hardware's view of the base address), not the OS's
+ * value of pci_resource_start() as the OS value may be a cookie
+ * for ioremap/iomap.
+ */
+
+
+ pci_read_config_dword(phba->pcidev, PCI_BASE_ADDRESS_0, &bar_low);
+ pci_read_config_dword(phba->pcidev, PCI_BASE_ADDRESS_1, &bar_high);
+
+ /*
+ * Set up HGP - Port Memory
+ *
+ * The port expects the host get/put pointers to reside in memory
+ * following the "non-diagnostic" mode mailbox (32 words, 0x80 bytes)
+ * area of SLIM. In SLI-2 mode, there's an additional 16 reserved
+ * words (0x40 bytes). This area is not reserved if HBQs are
+ * configured in SLI-3.
+ *
+ * CR0Put - SLI2(no HBQs) = 0xc0, With HBQs = 0x80
+ * RR0Get 0xc4 0x84
+ * CR1Put 0xc8 0x88
+ * RR1Get 0xcc 0x8c
+ * CR2Put 0xd0 0x90
+ * RR2Get 0xd4 0x94
+ * CR3Put 0xd8 0x98
+ * RR3Get 0xdc 0x9c
+ *
+ * Reserved 0xa0-0xbf
+ * If HBQs configured:
+ * HBQ 0 Put ptr 0xc0
+ * HBQ 1 Put ptr 0xc4
+ * HBQ 2 Put ptr 0xc8
+ * ......
+ * HBQ(M-1)Put Pointer 0xc0+(M-1)*4
+ *
+ */
+
+ if (phba->cfg_hostmem_hgp && phba->sli_rev != 3) {
+ phba->host_gp = (struct lpfc_hgp __iomem *)
+ &phba->mbox->us.s2.host[0];
+ phba->hbq_put = NULL;
+ offset = (uint8_t *)&phba->mbox->us.s2.host -
+ (uint8_t *)phba->slim2p.virt;
+ pdma_addr = phba->slim2p.phys + offset;
+ phba->pcb->hgpAddrHigh = putPaddrHigh(pdma_addr);
+ phba->pcb->hgpAddrLow = putPaddrLow(pdma_addr);
+ } else {
+ /* Always Host Group Pointer is in SLIM */
+ mb->un.varCfgPort.hps = 1;
+
+ if (phba->sli_rev == 3) {
+ phba->host_gp = &mb_slim->us.s3.host[0];
+ phba->hbq_put = &mb_slim->us.s3.hbq_put[0];
+ } else {
+ phba->host_gp = &mb_slim->us.s2.host[0];
+ phba->hbq_put = NULL;
+ }
+
+ /* mask off BAR0's flag bits 0 - 3 */
+ phba->pcb->hgpAddrLow = (bar_low & PCI_BASE_ADDRESS_MEM_MASK) +
+ (void __iomem *)phba->host_gp -
+ (void __iomem *)phba->MBslimaddr;
+ if (bar_low & PCI_BASE_ADDRESS_MEM_TYPE_64)
+ phba->pcb->hgpAddrHigh = bar_high;
+ else
+ phba->pcb->hgpAddrHigh = 0;
+ /* write HGP data to SLIM at the required longword offset */
+ memset(&hgp, 0, sizeof(struct lpfc_hgp));
+
+ for (i = 0; i < phba->sli.num_rings; i++) {
+ lpfc_memcpy_to_slim(phba->host_gp + i, &hgp,
+ sizeof(*phba->host_gp));
+ }
+ }
+
+ /* Setup Port Group offset */
+ if (phba->sli_rev == 3)
+ pgp_offset = offsetof(struct lpfc_sli2_slim,
+ mbx.us.s3_pgp.port);
+ else
+ pgp_offset = offsetof(struct lpfc_sli2_slim, mbx.us.s2.port);
+ pdma_addr = phba->slim2p.phys + pgp_offset;
+ phba->pcb->pgpAddrHigh = putPaddrHigh(pdma_addr);
+ phba->pcb->pgpAddrLow = putPaddrLow(pdma_addr);
+
+ /* Use callback routine to setp rings in the pcb */
+ lpfc_config_pcb_setup(phba);
+
+ /* special handling for LC HBAs */
+ if (lpfc_is_LC_HBA(phba->pcidev->device)) {
+ uint32_t hbainit[5];
+
+ lpfc_hba_init(phba, hbainit);
+
+ memcpy(&mb->un.varCfgPort.hbainit, hbainit, 20);
+ }
+
+ /* Swap PCB if needed */
+ lpfc_sli_pcimem_bcopy(phba->pcb, phba->pcb, sizeof(PCB_t));
+}
+
+/**
+ * lpfc_kill_board - Prepare a mailbox command for killing board
+ * @phba: pointer to lpfc hba data structure.
+ * @pmb: pointer to the driver internal queue element for mailbox command.
+ *
+ * The kill board mailbox command is used to tell firmware to perform a
+ * graceful shutdown of a channel on a specified board to prepare for reset.
+ * When the kill board mailbox command is received, the ER3 bit is set to 1
+ * in the Host Status register and the ER Attention bit is set to 1 in the
+ * Host Attention register of the HBA function that received the kill board
+ * command.
+ *
+ * This routine prepares the mailbox command for killing the board in
+ * preparation for a graceful shutdown.
+ **/
+void
+lpfc_kill_board(struct lpfc_hba * phba, LPFC_MBOXQ_t * pmb)
+{
+ MAILBOX_t *mb = &pmb->u.mb;
+
+ memset(pmb, 0, sizeof(LPFC_MBOXQ_t));
+ mb->mbxCommand = MBX_KILL_BOARD;
+ mb->mbxOwner = OWN_HOST;
+ return;
+}
+
+/**
+ * lpfc_mbox_put - Put a mailbox cmd into the tail of driver's mailbox queue
+ * @phba: pointer to lpfc hba data structure.
+ * @mbq: pointer to the driver internal queue element for mailbox command.
+ *
+ * Driver maintains a internal mailbox command queue implemented as a linked
+ * list. When a mailbox command is issued, it shall be put into the mailbox
+ * command queue such that they shall be processed orderly as HBA can process
+ * one mailbox command at a time.
+ **/
+void
+lpfc_mbox_put(struct lpfc_hba * phba, LPFC_MBOXQ_t * mbq)
+{
+ struct lpfc_sli *psli;
+
+ psli = &phba->sli;
+
+ list_add_tail(&mbq->list, &psli->mboxq);
+
+ psli->mboxq_cnt++;
+
+ return;
+}
+
+/**
+ * lpfc_mbox_get - Remove a mailbox cmd from the head of driver's mailbox queue
+ * @phba: pointer to lpfc hba data structure.
+ *
+ * Driver maintains a internal mailbox command queue implemented as a linked
+ * list. When a mailbox command is issued, it shall be put into the mailbox
+ * command queue such that they shall be processed orderly as HBA can process
+ * one mailbox command at a time. After HBA finished processing a mailbox
+ * command, the driver will remove a pending mailbox command from the head of
+ * the mailbox command queue and send to the HBA for processing.
+ *
+ * Return codes
+ * pointer to the driver internal queue element for mailbox command.
+ **/
+LPFC_MBOXQ_t *
+lpfc_mbox_get(struct lpfc_hba * phba)
+{
+ LPFC_MBOXQ_t *mbq = NULL;
+ struct lpfc_sli *psli = &phba->sli;
+
+ list_remove_head((&psli->mboxq), mbq, LPFC_MBOXQ_t, list);
+ if (mbq)
+ psli->mboxq_cnt--;
+
+ return mbq;
+}
+
+/**
+ * __lpfc_mbox_cmpl_put - Put mailbox cmd into mailbox cmd complete list
+ * @phba: pointer to lpfc hba data structure.
+ * @mbq: pointer to the driver internal queue element for mailbox command.
+ *
+ * This routine put the completed mailbox command into the mailbox command
+ * complete list. This is the unlocked version of the routine. The mailbox
+ * complete list is used by the driver worker thread to process mailbox
+ * complete callback functions outside the driver interrupt handler.
+ **/
+void
+__lpfc_mbox_cmpl_put(struct lpfc_hba *phba, LPFC_MBOXQ_t *mbq)
+{
+ list_add_tail(&mbq->list, &phba->sli.mboxq_cmpl);
+}
+
+/**
+ * lpfc_mbox_cmpl_put - Put mailbox command into mailbox command complete list
+ * @phba: pointer to lpfc hba data structure.
+ * @mbq: pointer to the driver internal queue element for mailbox command.
+ *
+ * This routine put the completed mailbox command into the mailbox command
+ * complete list. This is the locked version of the routine. The mailbox
+ * complete list is used by the driver worker thread to process mailbox
+ * complete callback functions outside the driver interrupt handler.
+ **/
+void
+lpfc_mbox_cmpl_put(struct lpfc_hba *phba, LPFC_MBOXQ_t *mbq)
+{
+ unsigned long iflag;
+
+ /* This function expects to be called from interrupt context */
+ spin_lock_irqsave(&phba->hbalock, iflag);
+ __lpfc_mbox_cmpl_put(phba, mbq);
+ spin_unlock_irqrestore(&phba->hbalock, iflag);
+ return;
+}
+
+/**
+ * lpfc_mbox_cmd_check - Check the validality of a mailbox command
+ * @phba: pointer to lpfc hba data structure.
+ * @mboxq: pointer to the driver internal queue element for mailbox command.
+ *
+ * This routine is to check whether a mailbox command is valid to be issued.
+ * This check will be performed by both the mailbox issue API when a client
+ * is to issue a mailbox command to the mailbox transport.
+ *
+ * Return 0 - pass the check, -ENODEV - fail the check
+ **/
+int
+lpfc_mbox_cmd_check(struct lpfc_hba *phba, LPFC_MBOXQ_t *mboxq)
+{
+ /* Mailbox command that have a completion handler must also have a
+ * vport specified.
+ */
+ if (mboxq->mbox_cmpl && mboxq->mbox_cmpl != lpfc_sli_def_mbox_cmpl &&
+ mboxq->mbox_cmpl != lpfc_sli_wake_mbox_wait) {
+ if (!mboxq->vport) {
+ lpfc_printf_log(phba, KERN_ERR, LOG_MBOX | LOG_VPORT,
+ "1814 Mbox x%x failed, no vport\n",
+ mboxq->u.mb.mbxCommand);
+ dump_stack();
+ return -ENODEV;
+ }
+ }
+ return 0;
+}
+
+/**
+ * lpfc_mbox_dev_check - Check the device state for issuing a mailbox command
+ * @phba: pointer to lpfc hba data structure.
+ *
+ * This routine is to check whether the HBA device is ready for posting a
+ * mailbox command. It is used by the mailbox transport API at the time the
+ * to post a mailbox command to the device.
+ *
+ * Return 0 - pass the check, -ENODEV - fail the check
+ **/
+int
+lpfc_mbox_dev_check(struct lpfc_hba *phba)
+{
+ /* If the PCI channel is in offline state, do not issue mbox */
+ if (unlikely(pci_channel_offline(phba->pcidev)))
+ return -ENODEV;
+
+ /* If the HBA is in error state, do not issue mbox */
+ if (phba->link_state == LPFC_HBA_ERROR)
+ return -ENODEV;
+
+ return 0;
+}
+
+/**
+ * lpfc_mbox_tmo_val - Retrieve mailbox command timeout value
+ * @phba: pointer to lpfc hba data structure.
+ * @mboxq: pointer to the driver internal queue element for mailbox command.
+ *
+ * This routine retrieves the proper timeout value according to the mailbox
+ * command code.
+ *
+ * Return codes
+ * Timeout value to be used for the given mailbox command
+ **/
+int
+lpfc_mbox_tmo_val(struct lpfc_hba *phba, LPFC_MBOXQ_t *mboxq)
+{
+ MAILBOX_t *mbox = &mboxq->u.mb;
+ uint8_t subsys, opcode;
+
+ switch (mbox->mbxCommand) {
+ case MBX_WRITE_NV: /* 0x03 */
+ case MBX_DUMP_MEMORY: /* 0x17 */
+ case MBX_UPDATE_CFG: /* 0x1B */
+ case MBX_DOWN_LOAD: /* 0x1C */
+ case MBX_DEL_LD_ENTRY: /* 0x1D */
+ case MBX_WRITE_VPARMS: /* 0x32 */
+ case MBX_LOAD_AREA: /* 0x81 */
+ case MBX_WRITE_WWN: /* 0x98 */
+ case MBX_LOAD_EXP_ROM: /* 0x9C */
+ case MBX_ACCESS_VDATA: /* 0xA5 */
+ return LPFC_MBOX_TMO_FLASH_CMD;
+ case MBX_SLI4_CONFIG: /* 0x9b */
+ subsys = lpfc_sli_config_mbox_subsys_get(phba, mboxq);
+ opcode = lpfc_sli_config_mbox_opcode_get(phba, mboxq);
+ if (subsys == LPFC_MBOX_SUBSYSTEM_COMMON) {
+ switch (opcode) {
+ case LPFC_MBOX_OPCODE_READ_OBJECT:
+ case LPFC_MBOX_OPCODE_WRITE_OBJECT:
+ case LPFC_MBOX_OPCODE_READ_OBJECT_LIST:
+ case LPFC_MBOX_OPCODE_DELETE_OBJECT:
+ case LPFC_MBOX_OPCODE_GET_PROFILE_LIST:
+ case LPFC_MBOX_OPCODE_SET_ACT_PROFILE:
+ case LPFC_MBOX_OPCODE_GET_PROFILE_CONFIG:
+ case LPFC_MBOX_OPCODE_SET_PROFILE_CONFIG:
+ case LPFC_MBOX_OPCODE_GET_FACTORY_PROFILE_CONFIG:
+ case LPFC_MBOX_OPCODE_GET_PROFILE_CAPACITIES:
+ case LPFC_MBOX_OPCODE_SEND_ACTIVATION:
+ case LPFC_MBOX_OPCODE_RESET_LICENSES:
+ case LPFC_MBOX_OPCODE_SET_BOOT_CONFIG:
+ case LPFC_MBOX_OPCODE_GET_VPD_DATA:
+ case LPFC_MBOX_OPCODE_SET_PHYSICAL_LINK_CONFIG:
+ return LPFC_MBOX_SLI4_CONFIG_EXTENDED_TMO;
+ }
+ }
+ if (subsys == LPFC_MBOX_SUBSYSTEM_FCOE) {
+ switch (opcode) {
+ case LPFC_MBOX_OPCODE_FCOE_SET_FCLINK_SETTINGS:
+ return LPFC_MBOX_SLI4_CONFIG_EXTENDED_TMO;
+ }
+ }
+ return LPFC_MBOX_SLI4_CONFIG_TMO;
+ }
+ return LPFC_MBOX_TMO;
+}
+
+/**
+ * lpfc_sli4_mbx_sge_set - Set a sge entry in non-embedded mailbox command
+ * @mbox: pointer to lpfc mbox command.
+ * @sgentry: sge entry index.
+ * @phyaddr: physical address for the sge
+ * @length: Length of the sge.
+ *
+ * This routine sets up an entry in the non-embedded mailbox command at the sge
+ * index location.
+ **/
+void
+lpfc_sli4_mbx_sge_set(struct lpfcMboxq *mbox, uint32_t sgentry,
+ dma_addr_t phyaddr, uint32_t length)
+{
+ struct lpfc_mbx_nembed_cmd *nembed_sge;
+
+ nembed_sge = (struct lpfc_mbx_nembed_cmd *)
+ &mbox->u.mqe.un.nembed_cmd;
+ nembed_sge->sge[sgentry].pa_lo = putPaddrLow(phyaddr);
+ nembed_sge->sge[sgentry].pa_hi = putPaddrHigh(phyaddr);
+ nembed_sge->sge[sgentry].length = length;
+}
+
+/**
+ * lpfc_sli4_mbx_sge_get - Get a sge entry from non-embedded mailbox command
+ * @mbox: pointer to lpfc mbox command.
+ * @sgentry: sge entry index.
+ * @sge: pointer to lpfc mailbox sge to load into.
+ *
+ * This routine gets an entry from the non-embedded mailbox command at the sge
+ * index location.
+ **/
+void
+lpfc_sli4_mbx_sge_get(struct lpfcMboxq *mbox, uint32_t sgentry,
+ struct lpfc_mbx_sge *sge)
+{
+ struct lpfc_mbx_nembed_cmd *nembed_sge;
+
+ nembed_sge = (struct lpfc_mbx_nembed_cmd *)
+ &mbox->u.mqe.un.nembed_cmd;
+ sge->pa_lo = nembed_sge->sge[sgentry].pa_lo;
+ sge->pa_hi = nembed_sge->sge[sgentry].pa_hi;
+ sge->length = nembed_sge->sge[sgentry].length;
+}
+
+/**
+ * lpfc_sli4_mbox_cmd_free - Free a sli4 mailbox command
+ * @phba: pointer to lpfc hba data structure.
+ * @mbox: pointer to lpfc mbox command.
+ *
+ * This routine frees SLI4 specific mailbox command for sending IOCTL command.
+ **/
+void
+lpfc_sli4_mbox_cmd_free(struct lpfc_hba *phba, struct lpfcMboxq *mbox)
+{
+ struct lpfc_mbx_sli4_config *sli4_cfg;
+ struct lpfc_mbx_sge sge;
+ dma_addr_t phyaddr;
+ uint32_t sgecount, sgentry;
+
+ sli4_cfg = &mbox->u.mqe.un.sli4_config;
+
+ /* For embedded mbox command, just free the mbox command */
+ if (bf_get(lpfc_mbox_hdr_emb, &sli4_cfg->header.cfg_mhdr)) {
+ mempool_free(mbox, phba->mbox_mem_pool);
+ return;
+ }
+
+ /* For non-embedded mbox command, we need to free the pages first */
+ sgecount = bf_get(lpfc_mbox_hdr_sge_cnt, &sli4_cfg->header.cfg_mhdr);
+ /* There is nothing we can do if there is no sge address array */
+ if (unlikely(!mbox->sge_array)) {
+ mempool_free(mbox, phba->mbox_mem_pool);
+ return;
+ }
+ /* Each non-embedded DMA memory was allocated in the length of a page */
+ for (sgentry = 0; sgentry < sgecount; sgentry++) {
+ lpfc_sli4_mbx_sge_get(mbox, sgentry, &sge);
+ phyaddr = getPaddr(sge.pa_hi, sge.pa_lo);
+ dma_free_coherent(&phba->pcidev->dev, SLI4_PAGE_SIZE,
+ mbox->sge_array->addr[sgentry], phyaddr);
+ }
+ /* Free the sge address array memory */
+ kfree(mbox->sge_array);
+ /* Finally, free the mailbox command itself */
+ mempool_free(mbox, phba->mbox_mem_pool);
+}
+
+/**
+ * lpfc_sli4_config - Initialize the SLI4 Config Mailbox command
+ * @phba: pointer to lpfc hba data structure.
+ * @mbox: pointer to lpfc mbox command.
+ * @subsystem: The sli4 config sub mailbox subsystem.
+ * @opcode: The sli4 config sub mailbox command opcode.
+ * @length: Length of the sli4 config mailbox command (including sub-header).
+ * @emb: True if embedded mbox command should be setup.
+ *
+ * This routine sets up the header fields of SLI4 specific mailbox command
+ * for sending IOCTL command.
+ *
+ * Return: the actual length of the mbox command allocated (mostly useful
+ * for none embedded mailbox command).
+ **/
+int
+lpfc_sli4_config(struct lpfc_hba *phba, struct lpfcMboxq *mbox,
+ uint8_t subsystem, uint8_t opcode, uint32_t length, bool emb)
+{
+ struct lpfc_mbx_sli4_config *sli4_config;
+ union lpfc_sli4_cfg_shdr *cfg_shdr = NULL;
+ uint32_t alloc_len;
+ uint32_t resid_len;
+ uint32_t pagen, pcount;
+ void *viraddr;
+ dma_addr_t phyaddr;
+
+ /* Set up SLI4 mailbox command header fields */
+ memset(mbox, 0, sizeof(*mbox));
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_SLI4_CONFIG);
+
+ /* Set up SLI4 ioctl command header fields */
+ sli4_config = &mbox->u.mqe.un.sli4_config;
+
+ /* Setup for the embedded mbox command */
+ if (emb) {
+ /* Set up main header fields */
+ bf_set(lpfc_mbox_hdr_emb, &sli4_config->header.cfg_mhdr, 1);
+ sli4_config->header.cfg_mhdr.payload_length = length;
+ /* Set up sub-header fields following main header */
+ bf_set(lpfc_mbox_hdr_opcode,
+ &sli4_config->header.cfg_shdr.request, opcode);
+ bf_set(lpfc_mbox_hdr_subsystem,
+ &sli4_config->header.cfg_shdr.request, subsystem);
+ sli4_config->header.cfg_shdr.request.request_length =
+ length - LPFC_MBX_CMD_HDR_LENGTH;
+ return length;
+ }
+
+ /* Setup for the non-embedded mbox command */
+ pcount = (SLI4_PAGE_ALIGN(length))/SLI4_PAGE_SIZE;
+ pcount = (pcount > LPFC_SLI4_MBX_SGE_MAX_PAGES) ?
+ LPFC_SLI4_MBX_SGE_MAX_PAGES : pcount;
+ /* Allocate record for keeping SGE virtual addresses */
+ mbox->sge_array = kzalloc(sizeof(struct lpfc_mbx_nembed_sge_virt),
+ GFP_KERNEL);
+ if (!mbox->sge_array) {
+ lpfc_printf_log(phba, KERN_ERR, LOG_MBOX,
+ "2527 Failed to allocate non-embedded SGE "
+ "array.\n");
+ return 0;
+ }
+ for (pagen = 0, alloc_len = 0; pagen < pcount; pagen++) {
+ /* The DMA memory is always allocated in the length of a
+ * page even though the last SGE might not fill up to a
+ * page, this is used as a priori size of SLI4_PAGE_SIZE for
+ * the later DMA memory free.
+ */
+ viraddr = dma_alloc_coherent(&phba->pcidev->dev,
+ SLI4_PAGE_SIZE, &phyaddr,
+ GFP_KERNEL);
+ /* In case of malloc fails, proceed with whatever we have */
+ if (!viraddr)
+ break;
+ mbox->sge_array->addr[pagen] = viraddr;
+ /* Keep the first page for later sub-header construction */
+ if (pagen == 0)
+ cfg_shdr = (union lpfc_sli4_cfg_shdr *)viraddr;
+ resid_len = length - alloc_len;
+ if (resid_len > SLI4_PAGE_SIZE) {
+ lpfc_sli4_mbx_sge_set(mbox, pagen, phyaddr,
+ SLI4_PAGE_SIZE);
+ alloc_len += SLI4_PAGE_SIZE;
+ } else {
+ lpfc_sli4_mbx_sge_set(mbox, pagen, phyaddr,
+ resid_len);
+ alloc_len = length;
+ }
+ }
+
+ /* Set up main header fields in mailbox command */
+ sli4_config->header.cfg_mhdr.payload_length = alloc_len;
+ bf_set(lpfc_mbox_hdr_sge_cnt, &sli4_config->header.cfg_mhdr, pagen);
+
+ /* Set up sub-header fields into the first page */
+ if (pagen > 0) {
+ bf_set(lpfc_mbox_hdr_opcode, &cfg_shdr->request, opcode);
+ bf_set(lpfc_mbox_hdr_subsystem, &cfg_shdr->request, subsystem);
+ cfg_shdr->request.request_length =
+ alloc_len - sizeof(union lpfc_sli4_cfg_shdr);
+ }
+ /* The sub-header is in DMA memory, which needs endian converstion */
+ if (cfg_shdr)
+ lpfc_sli_pcimem_bcopy(cfg_shdr, cfg_shdr,
+ sizeof(union lpfc_sli4_cfg_shdr));
+ return alloc_len;
+}
+
+/**
+ * lpfc_sli4_mbox_rsrc_extent - Initialize the opcode resource extent.
+ * @phba: pointer to lpfc hba data structure.
+ * @mbox: pointer to an allocated lpfc mbox resource.
+ * @exts_count: the number of extents, if required, to allocate.
+ * @rsrc_type: the resource extent type.
+ * @emb: true if LPFC_SLI4_MBX_EMBED. false if LPFC_SLI4_MBX_NEMBED.
+ *
+ * This routine completes the subcommand header for SLI4 resource extent
+ * mailbox commands. It is called after lpfc_sli4_config. The caller must
+ * pass an allocated mailbox and the attributes required to initialize the
+ * mailbox correctly.
+ *
+ * Return: the actual length of the mbox command allocated.
+ **/
+int
+lpfc_sli4_mbox_rsrc_extent(struct lpfc_hba *phba, struct lpfcMboxq *mbox,
+ uint16_t exts_count, uint16_t rsrc_type, bool emb)
+{
+ uint8_t opcode = 0;
+ struct lpfc_mbx_nembed_rsrc_extent *n_rsrc_extnt = NULL;
+ void *virtaddr = NULL;
+
+ /* Set up SLI4 ioctl command header fields */
+ if (emb == LPFC_SLI4_MBX_NEMBED) {
+ /* Get the first SGE entry from the non-embedded DMA memory */
+ virtaddr = mbox->sge_array->addr[0];
+ if (virtaddr == NULL)
+ return 1;
+ n_rsrc_extnt = (struct lpfc_mbx_nembed_rsrc_extent *) virtaddr;
+ }
+
+ /*
+ * The resource type is common to all extent Opcodes and resides in the
+ * same position.
+ */
+ if (emb == LPFC_SLI4_MBX_EMBED)
+ bf_set(lpfc_mbx_alloc_rsrc_extents_type,
+ &mbox->u.mqe.un.alloc_rsrc_extents.u.req,
+ rsrc_type);
+ else {
+ /* This is DMA data. Byteswap is required. */
+ bf_set(lpfc_mbx_alloc_rsrc_extents_type,
+ n_rsrc_extnt, rsrc_type);
+ lpfc_sli_pcimem_bcopy(&n_rsrc_extnt->word4,
+ &n_rsrc_extnt->word4,
+ sizeof(uint32_t));
+ }
+
+ /* Complete the initialization for the particular Opcode. */
+ opcode = lpfc_sli_config_mbox_opcode_get(phba, mbox);
+ switch (opcode) {
+ case LPFC_MBOX_OPCODE_ALLOC_RSRC_EXTENT:
+ if (emb == LPFC_SLI4_MBX_EMBED)
+ bf_set(lpfc_mbx_alloc_rsrc_extents_cnt,
+ &mbox->u.mqe.un.alloc_rsrc_extents.u.req,
+ exts_count);
+ else
+ bf_set(lpfc_mbx_alloc_rsrc_extents_cnt,
+ n_rsrc_extnt, exts_count);
+ break;
+ case LPFC_MBOX_OPCODE_GET_ALLOC_RSRC_EXTENT:
+ case LPFC_MBOX_OPCODE_GET_RSRC_EXTENT_INFO:
+ case LPFC_MBOX_OPCODE_DEALLOC_RSRC_EXTENT:
+ /* Initialization is complete.*/
+ break;
+ default:
+ lpfc_printf_log(phba, KERN_ERR, LOG_MBOX,
+ "2929 Resource Extent Opcode x%x is "
+ "unsupported\n", opcode);
+ return 1;
+ }
+
+ return 0;
+}
+
+/**
+ * lpfc_sli_config_mbox_subsys_get - Get subsystem from a sli_config mbox cmd
+ * @phba: pointer to lpfc hba data structure.
+ * @mbox: pointer to lpfc mbox command queue entry.
+ *
+ * This routine gets the subsystem from a SLI4 specific SLI_CONFIG mailbox
+ * command. If the mailbox command is not MBX_SLI4_CONFIG (0x9B) or if the
+ * sub-header is not present, subsystem LPFC_MBOX_SUBSYSTEM_NA (0x0) shall
+ * be returned.
+ **/
+uint8_t
+lpfc_sli_config_mbox_subsys_get(struct lpfc_hba *phba, LPFC_MBOXQ_t *mbox)
+{
+ struct lpfc_mbx_sli4_config *sli4_cfg;
+ union lpfc_sli4_cfg_shdr *cfg_shdr;
+
+ if (mbox->u.mb.mbxCommand != MBX_SLI4_CONFIG)
+ return LPFC_MBOX_SUBSYSTEM_NA;
+ sli4_cfg = &mbox->u.mqe.un.sli4_config;
+
+ /* For embedded mbox command, get opcode from embedded sub-header*/
+ if (bf_get(lpfc_mbox_hdr_emb, &sli4_cfg->header.cfg_mhdr)) {
+ cfg_shdr = &mbox->u.mqe.un.sli4_config.header.cfg_shdr;
+ return bf_get(lpfc_mbox_hdr_subsystem, &cfg_shdr->request);
+ }
+
+ /* For non-embedded mbox command, get opcode from first dma page */
+ if (unlikely(!mbox->sge_array))
+ return LPFC_MBOX_SUBSYSTEM_NA;
+ cfg_shdr = (union lpfc_sli4_cfg_shdr *)mbox->sge_array->addr[0];
+ return bf_get(lpfc_mbox_hdr_subsystem, &cfg_shdr->request);
+}
+
+/**
+ * lpfc_sli_config_mbox_opcode_get - Get opcode from a sli_config mbox cmd
+ * @phba: pointer to lpfc hba data structure.
+ * @mbox: pointer to lpfc mbox command queue entry.
+ *
+ * This routine gets the opcode from a SLI4 specific SLI_CONFIG mailbox
+ * command. If the mailbox command is not MBX_SLI4_CONFIG (0x9B) or if
+ * the sub-header is not present, opcode LPFC_MBOX_OPCODE_NA (0x0) be
+ * returned.
+ **/
+uint8_t
+lpfc_sli_config_mbox_opcode_get(struct lpfc_hba *phba, LPFC_MBOXQ_t *mbox)
+{
+ struct lpfc_mbx_sli4_config *sli4_cfg;
+ union lpfc_sli4_cfg_shdr *cfg_shdr;
+
+ if (mbox->u.mb.mbxCommand != MBX_SLI4_CONFIG)
+ return LPFC_MBOX_OPCODE_NA;
+ sli4_cfg = &mbox->u.mqe.un.sli4_config;
+
+ /* For embedded mbox command, get opcode from embedded sub-header*/
+ if (bf_get(lpfc_mbox_hdr_emb, &sli4_cfg->header.cfg_mhdr)) {
+ cfg_shdr = &mbox->u.mqe.un.sli4_config.header.cfg_shdr;
+ return bf_get(lpfc_mbox_hdr_opcode, &cfg_shdr->request);
+ }
+
+ /* For non-embedded mbox command, get opcode from first dma page */
+ if (unlikely(!mbox->sge_array))
+ return LPFC_MBOX_OPCODE_NA;
+ cfg_shdr = (union lpfc_sli4_cfg_shdr *)mbox->sge_array->addr[0];
+ return bf_get(lpfc_mbox_hdr_opcode, &cfg_shdr->request);
+}
+
+/**
+ * lpfc_sli4_mbx_read_fcf_rec - Allocate and construct read fcf mbox cmd
+ * @phba: pointer to lpfc hba data structure.
+ * @mboxq: pointer to lpfc mbox command.
+ * @fcf_index: index to fcf table.
+ *
+ * This routine routine allocates and constructs non-embedded mailbox command
+ * for reading a FCF table entry referred by @fcf_index.
+ *
+ * Return: pointer to the mailbox command constructed if successful, otherwise
+ * NULL.
+ **/
+int
+lpfc_sli4_mbx_read_fcf_rec(struct lpfc_hba *phba,
+ struct lpfcMboxq *mboxq,
+ uint16_t fcf_index)
+{
+ void *virt_addr;
+ uint8_t *bytep;
+ struct lpfc_mbx_sge sge;
+ uint32_t alloc_len, req_len;
+ struct lpfc_mbx_read_fcf_tbl *read_fcf;
+
+ if (!mboxq)
+ return -ENOMEM;
+
+ req_len = sizeof(struct fcf_record) +
+ sizeof(union lpfc_sli4_cfg_shdr) + 2 * sizeof(uint32_t);
+
+ /* Set up READ_FCF SLI4_CONFIG mailbox-ioctl command */
+ alloc_len = lpfc_sli4_config(phba, mboxq, LPFC_MBOX_SUBSYSTEM_FCOE,
+ LPFC_MBOX_OPCODE_FCOE_READ_FCF_TABLE, req_len,
+ LPFC_SLI4_MBX_NEMBED);
+
+ if (alloc_len < req_len) {
+ lpfc_printf_log(phba, KERN_ERR, LOG_MBOX,
+ "0291 Allocated DMA memory size (x%x) is "
+ "less than the requested DMA memory "
+ "size (x%x)\n", alloc_len, req_len);
+ return -ENOMEM;
+ }
+
+ /* Get the first SGE entry from the non-embedded DMA memory. This
+ * routine only uses a single SGE.
+ */
+ lpfc_sli4_mbx_sge_get(mboxq, 0, &sge);
+ virt_addr = mboxq->sge_array->addr[0];
+ read_fcf = (struct lpfc_mbx_read_fcf_tbl *)virt_addr;
+
+ /* Set up command fields */
+ bf_set(lpfc_mbx_read_fcf_tbl_indx, &read_fcf->u.request, fcf_index);
+ /* Perform necessary endian conversion */
+ bytep = virt_addr + sizeof(union lpfc_sli4_cfg_shdr);
+ lpfc_sli_pcimem_bcopy(bytep, bytep, sizeof(uint32_t));
+
+ return 0;
+}
+
+/**
+ * lpfc_request_features: Configure SLI4 REQUEST_FEATURES mailbox
+ * @phba: pointer to lpfc hba data structure.
+ * @mboxq: pointer to lpfc mbox command.
+ *
+ * This routine sets up the mailbox for an SLI4 REQUEST_FEATURES
+ * mailbox command.
+ **/
+void
+lpfc_request_features(struct lpfc_hba *phba, struct lpfcMboxq *mboxq)
+{
+ /* Set up SLI4 mailbox command header fields */
+ memset(mboxq, 0, sizeof(LPFC_MBOXQ_t));
+ bf_set(lpfc_mqe_command, &mboxq->u.mqe, MBX_SLI4_REQ_FTRS);
+
+ /* Set up host requested features. */
+ bf_set(lpfc_mbx_rq_ftr_rq_fcpi, &mboxq->u.mqe.un.req_ftrs, 1);
+ bf_set(lpfc_mbx_rq_ftr_rq_perfh, &mboxq->u.mqe.un.req_ftrs, 1);
+
+ /* Enable DIF (block guard) only if configured to do so. */
+ if (phba->cfg_enable_bg)
+ bf_set(lpfc_mbx_rq_ftr_rq_dif, &mboxq->u.mqe.un.req_ftrs, 1);
+
+ /* Enable NPIV only if configured to do so. */
+ if (phba->max_vpi && phba->cfg_enable_npiv)
+ bf_set(lpfc_mbx_rq_ftr_rq_npiv, &mboxq->u.mqe.un.req_ftrs, 1);
+
+ if (phba->nvmet_support) {
+ bf_set(lpfc_mbx_rq_ftr_rq_mrqp, &mboxq->u.mqe.un.req_ftrs, 1);
+ /* iaab/iaar NOT set for now */
+ bf_set(lpfc_mbx_rq_ftr_rq_iaab, &mboxq->u.mqe.un.req_ftrs, 0);
+ bf_set(lpfc_mbx_rq_ftr_rq_iaar, &mboxq->u.mqe.un.req_ftrs, 0);
+ }
+ return;
+}
+
+/**
+ * lpfc_init_vfi - Initialize the INIT_VFI mailbox command
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @vport: Vport associated with the VF.
+ *
+ * This routine initializes @mbox to all zeros and then fills in the mailbox
+ * fields from @vport. INIT_VFI configures virtual fabrics identified by VFI
+ * in the context of an FCF. The driver issues this command to setup a VFI
+ * before issuing a FLOGI to login to the VSAN. The driver should also issue a
+ * REG_VFI after a successful VSAN login.
+ **/
+void
+lpfc_init_vfi(struct lpfcMboxq *mbox, struct lpfc_vport *vport)
+{
+ struct lpfc_mbx_init_vfi *init_vfi;
+
+ memset(mbox, 0, sizeof(*mbox));
+ mbox->vport = vport;
+ init_vfi = &mbox->u.mqe.un.init_vfi;
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_INIT_VFI);
+ bf_set(lpfc_init_vfi_vr, init_vfi, 1);
+ bf_set(lpfc_init_vfi_vt, init_vfi, 1);
+ bf_set(lpfc_init_vfi_vp, init_vfi, 1);
+ bf_set(lpfc_init_vfi_vfi, init_vfi,
+ vport->phba->sli4_hba.vfi_ids[vport->vfi]);
+ bf_set(lpfc_init_vfi_vpi, init_vfi,
+ vport->phba->vpi_ids[vport->vpi]);
+ bf_set(lpfc_init_vfi_fcfi, init_vfi,
+ vport->phba->fcf.fcfi);
+}
+
+/**
+ * lpfc_reg_vfi - Initialize the REG_VFI mailbox command
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @vport: vport associated with the VF.
+ * @phys: BDE DMA bus address used to send the service parameters to the HBA.
+ *
+ * This routine initializes @mbox to all zeros and then fills in the mailbox
+ * fields from @vport, and uses @buf as a DMAable buffer to send the vport's
+ * fc service parameters to the HBA for this VFI. REG_VFI configures virtual
+ * fabrics identified by VFI in the context of an FCF.
+ **/
+void
+lpfc_reg_vfi(struct lpfcMboxq *mbox, struct lpfc_vport *vport, dma_addr_t phys)
+{
+ struct lpfc_mbx_reg_vfi *reg_vfi;
+ struct lpfc_hba *phba = vport->phba;
+ uint8_t bbscn_fabric = 0, bbscn_max = 0, bbscn_def = 0;
+
+ memset(mbox, 0, sizeof(*mbox));
+ reg_vfi = &mbox->u.mqe.un.reg_vfi;
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_REG_VFI);
+ bf_set(lpfc_reg_vfi_vp, reg_vfi, 1);
+ bf_set(lpfc_reg_vfi_vfi, reg_vfi,
+ phba->sli4_hba.vfi_ids[vport->vfi]);
+ bf_set(lpfc_reg_vfi_fcfi, reg_vfi, phba->fcf.fcfi);
+ bf_set(lpfc_reg_vfi_vpi, reg_vfi, phba->vpi_ids[vport->vpi]);
+ memcpy(reg_vfi->wwn, &vport->fc_portname, sizeof(struct lpfc_name));
+ reg_vfi->wwn[0] = cpu_to_le32(reg_vfi->wwn[0]);
+ reg_vfi->wwn[1] = cpu_to_le32(reg_vfi->wwn[1]);
+ reg_vfi->e_d_tov = phba->fc_edtov;
+ reg_vfi->r_a_tov = phba->fc_ratov;
+ if (phys) {
+ reg_vfi->bde.addrHigh = putPaddrHigh(phys);
+ reg_vfi->bde.addrLow = putPaddrLow(phys);
+ reg_vfi->bde.tus.f.bdeSize = sizeof(vport->fc_sparam);
+ reg_vfi->bde.tus.f.bdeFlags = BUFF_TYPE_BDE_64;
+ }
+ bf_set(lpfc_reg_vfi_nport_id, reg_vfi, vport->fc_myDID);
+
+ /* Only FC supports upd bit */
+ if ((phba->sli4_hba.lnk_info.lnk_tp == LPFC_LNK_TYPE_FC) &&
+ (vport->fc_flag & FC_VFI_REGISTERED) &&
+ (!phba->fc_topology_changed))
+ bf_set(lpfc_reg_vfi_upd, reg_vfi, 1);
+
+ bf_set(lpfc_reg_vfi_bbcr, reg_vfi, 0);
+ bf_set(lpfc_reg_vfi_bbscn, reg_vfi, 0);
+ bbscn_fabric = (phba->fc_fabparam.cmn.bbRcvSizeMsb >> 4) & 0xF;
+
+ if (phba->bbcredit_support && phba->cfg_enable_bbcr &&
+ bbscn_fabric != 0) {
+ bbscn_max = bf_get(lpfc_bbscn_max,
+ &phba->sli4_hba.bbscn_params);
+ if (bbscn_fabric <= bbscn_max) {
+ bbscn_def = bf_get(lpfc_bbscn_def,
+ &phba->sli4_hba.bbscn_params);
+
+ if (bbscn_fabric > bbscn_def)
+ bf_set(lpfc_reg_vfi_bbscn, reg_vfi,
+ bbscn_fabric);
+ else
+ bf_set(lpfc_reg_vfi_bbscn, reg_vfi, bbscn_def);
+
+ bf_set(lpfc_reg_vfi_bbcr, reg_vfi, 1);
+ }
+ }
+ lpfc_printf_vlog(vport, KERN_INFO, LOG_MBOX,
+ "3134 Register VFI, mydid:x%x, fcfi:%d, "
+ " vfi:%d, vpi:%d, fc_pname:%x%x fc_flag:x%x"
+ " port_state:x%x topology chg:%d bbscn_fabric :%d\n",
+ vport->fc_myDID,
+ phba->fcf.fcfi,
+ phba->sli4_hba.vfi_ids[vport->vfi],
+ phba->vpi_ids[vport->vpi],
+ reg_vfi->wwn[0], reg_vfi->wwn[1], vport->fc_flag,
+ vport->port_state, phba->fc_topology_changed,
+ bbscn_fabric);
+}
+
+/**
+ * lpfc_init_vpi - Initialize the INIT_VPI mailbox command
+ * @phba: pointer to the hba structure to init the VPI for.
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @vpi: VPI to be initialized.
+ *
+ * The INIT_VPI mailbox command supports virtual N_Ports. The driver uses the
+ * command to activate a virtual N_Port. The HBA assigns a MAC address to use
+ * with the virtual N Port. The SLI Host issues this command before issuing a
+ * FDISC to connect to the Fabric. The SLI Host should issue a REG_VPI after a
+ * successful virtual NPort login.
+ **/
+void
+lpfc_init_vpi(struct lpfc_hba *phba, struct lpfcMboxq *mbox, uint16_t vpi)
+{
+ memset(mbox, 0, sizeof(*mbox));
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_INIT_VPI);
+ bf_set(lpfc_init_vpi_vpi, &mbox->u.mqe.un.init_vpi,
+ phba->vpi_ids[vpi]);
+ bf_set(lpfc_init_vpi_vfi, &mbox->u.mqe.un.init_vpi,
+ phba->sli4_hba.vfi_ids[phba->pport->vfi]);
+}
+
+/**
+ * lpfc_unreg_vfi - Initialize the UNREG_VFI mailbox command
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @vport: vport associated with the VF.
+ *
+ * The UNREG_VFI mailbox command causes the SLI Host to put a virtual fabric
+ * (logical NPort) into the inactive state. The SLI Host must have logged out
+ * and unregistered all remote N_Ports to abort any activity on the virtual
+ * fabric. The SLI Port posts the mailbox response after marking the virtual
+ * fabric inactive.
+ **/
+void
+lpfc_unreg_vfi(struct lpfcMboxq *mbox, struct lpfc_vport *vport)
+{
+ memset(mbox, 0, sizeof(*mbox));
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_UNREG_VFI);
+ bf_set(lpfc_unreg_vfi_vfi, &mbox->u.mqe.un.unreg_vfi,
+ vport->phba->sli4_hba.vfi_ids[vport->vfi]);
+}
+
+/**
+ * lpfc_sli4_dump_cfg_rg23 - Dump sli4 port config region 23
+ * @phba: pointer to the hba structure containing.
+ * @mbox: pointer to lpfc mbox command to initialize.
+ *
+ * This function create a SLI4 dump mailbox command to dump configure
+ * region 23.
+ **/
+int
+lpfc_sli4_dump_cfg_rg23(struct lpfc_hba *phba, struct lpfcMboxq *mbox)
+{
+ struct lpfc_dmabuf *mp = NULL;
+ MAILBOX_t *mb;
+
+ memset(mbox, 0, sizeof(*mbox));
+ mb = &mbox->u.mb;
+
+ mp = kmalloc(sizeof(struct lpfc_dmabuf), GFP_KERNEL);
+ if (mp)
+ mp->virt = lpfc_mbuf_alloc(phba, 0, &mp->phys);
+
+ if (!mp || !mp->virt) {
+ kfree(mp);
+ /* dump config region 23 failed to allocate memory */
+ lpfc_printf_log(phba, KERN_WARNING, LOG_MBOX,
+ "2569 lpfc dump config region 23: memory"
+ " allocation failed\n");
+ return 1;
+ }
+
+ memset(mp->virt, 0, LPFC_BPL_SIZE);
+ INIT_LIST_HEAD(&mp->list);
+
+ /* save address for completion */
+ mbox->ctx_buf = (uint8_t *)mp;
+
+ mb->mbxCommand = MBX_DUMP_MEMORY;
+ mb->un.varDmp.type = DMP_NV_PARAMS;
+ mb->un.varDmp.region_id = DMP_REGION_23;
+ mb->un.varDmp.sli4_length = DMP_RGN23_SIZE;
+ mb->un.varWords[3] = putPaddrLow(mp->phys);
+ mb->un.varWords[4] = putPaddrHigh(mp->phys);
+ return 0;
+}
+
+static void
+lpfc_mbx_cmpl_rdp_link_stat(struct lpfc_hba *phba, LPFC_MBOXQ_t *mboxq)
+{
+ MAILBOX_t *mb;
+ int rc = FAILURE;
+ struct lpfc_rdp_context *rdp_context =
+ (struct lpfc_rdp_context *)(mboxq->ctx_ndlp);
+
+ mb = &mboxq->u.mb;
+ if (mb->mbxStatus)
+ goto mbx_failed;
+
+ memcpy(&rdp_context->link_stat, &mb->un.varRdLnk, sizeof(READ_LNK_VAR));
+
+ rc = SUCCESS;
+
+mbx_failed:
+ lpfc_sli4_mbox_cmd_free(phba, mboxq);
+ rdp_context->cmpl(phba, rdp_context, rc);
+}
+
+static void
+lpfc_mbx_cmpl_rdp_page_a2(struct lpfc_hba *phba, LPFC_MBOXQ_t *mbox)
+{
+ struct lpfc_dmabuf *mp = (struct lpfc_dmabuf *)mbox->ctx_buf;
+ struct lpfc_rdp_context *rdp_context =
+ (struct lpfc_rdp_context *)(mbox->ctx_ndlp);
+
+ if (bf_get(lpfc_mqe_status, &mbox->u.mqe))
+ goto error_mbuf_free;
+
+ lpfc_sli_bemem_bcopy(mp->virt, &rdp_context->page_a2,
+ DMP_SFF_PAGE_A2_SIZE);
+
+ /* We don't need dma buffer for link stat. */
+ lpfc_mbuf_free(phba, mp->virt, mp->phys);
+ kfree(mp);
+
+ memset(mbox, 0, sizeof(*mbox));
+ lpfc_read_lnk_stat(phba, mbox);
+ mbox->vport = rdp_context->ndlp->vport;
+ mbox->mbox_cmpl = lpfc_mbx_cmpl_rdp_link_stat;
+ mbox->ctx_ndlp = (struct lpfc_rdp_context *)rdp_context;
+ if (lpfc_sli_issue_mbox(phba, mbox, MBX_NOWAIT) == MBX_NOT_FINISHED)
+ goto error_cmd_free;
+
+ return;
+
+error_mbuf_free:
+ lpfc_mbuf_free(phba, mp->virt, mp->phys);
+ kfree(mp);
+error_cmd_free:
+ lpfc_sli4_mbox_cmd_free(phba, mbox);
+ rdp_context->cmpl(phba, rdp_context, FAILURE);
+}
+
+void
+lpfc_mbx_cmpl_rdp_page_a0(struct lpfc_hba *phba, LPFC_MBOXQ_t *mbox)
+{
+ int rc;
+ struct lpfc_dmabuf *mp = (struct lpfc_dmabuf *)(mbox->ctx_buf);
+ struct lpfc_rdp_context *rdp_context =
+ (struct lpfc_rdp_context *)(mbox->ctx_ndlp);
+
+ if (bf_get(lpfc_mqe_status, &mbox->u.mqe))
+ goto error;
+
+ lpfc_sli_bemem_bcopy(mp->virt, &rdp_context->page_a0,
+ DMP_SFF_PAGE_A0_SIZE);
+
+ memset(mbox, 0, sizeof(*mbox));
+
+ memset(mp->virt, 0, DMP_SFF_PAGE_A2_SIZE);
+ INIT_LIST_HEAD(&mp->list);
+
+ /* save address for completion */
+ mbox->ctx_buf = mp;
+ mbox->vport = rdp_context->ndlp->vport;
+
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_DUMP_MEMORY);
+ bf_set(lpfc_mbx_memory_dump_type3_type,
+ &mbox->u.mqe.un.mem_dump_type3, DMP_LMSD);
+ bf_set(lpfc_mbx_memory_dump_type3_link,
+ &mbox->u.mqe.un.mem_dump_type3, phba->sli4_hba.physical_port);
+ bf_set(lpfc_mbx_memory_dump_type3_page_no,
+ &mbox->u.mqe.un.mem_dump_type3, DMP_PAGE_A2);
+ bf_set(lpfc_mbx_memory_dump_type3_length,
+ &mbox->u.mqe.un.mem_dump_type3, DMP_SFF_PAGE_A2_SIZE);
+ mbox->u.mqe.un.mem_dump_type3.addr_lo = putPaddrLow(mp->phys);
+ mbox->u.mqe.un.mem_dump_type3.addr_hi = putPaddrHigh(mp->phys);
+
+ mbox->mbox_cmpl = lpfc_mbx_cmpl_rdp_page_a2;
+ mbox->ctx_ndlp = (struct lpfc_rdp_context *)rdp_context;
+ rc = lpfc_sli_issue_mbox(phba, mbox, MBX_NOWAIT);
+ if (rc == MBX_NOT_FINISHED)
+ goto error;
+
+ return;
+
+error:
+ lpfc_mbuf_free(phba, mp->virt, mp->phys);
+ kfree(mp);
+ lpfc_sli4_mbox_cmd_free(phba, mbox);
+ rdp_context->cmpl(phba, rdp_context, FAILURE);
+}
+
+
+/*
+ * lpfc_sli4_dump_sfp_pagea0 - Dump sli4 read SFP Diagnostic.
+ * @phba: pointer to the hba structure containing.
+ * @mbox: pointer to lpfc mbox command to initialize.
+ *
+ * This function create a SLI4 dump mailbox command to dump configure
+ * type 3 page 0xA0.
+ */
+int
+lpfc_sli4_dump_page_a0(struct lpfc_hba *phba, struct lpfcMboxq *mbox)
+{
+ struct lpfc_dmabuf *mp = NULL;
+
+ memset(mbox, 0, sizeof(*mbox));
+
+ mp = kmalloc(sizeof(struct lpfc_dmabuf), GFP_KERNEL);
+ if (mp)
+ mp->virt = lpfc_mbuf_alloc(phba, 0, &mp->phys);
+ if (!mp || !mp->virt) {
+ kfree(mp);
+ lpfc_printf_log(phba, KERN_WARNING, LOG_MBOX,
+ "3569 dump type 3 page 0xA0 allocation failed\n");
+ return 1;
+ }
+
+ memset(mp->virt, 0, LPFC_BPL_SIZE);
+ INIT_LIST_HEAD(&mp->list);
+
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_DUMP_MEMORY);
+ /* save address for completion */
+ mbox->ctx_buf = mp;
+
+ bf_set(lpfc_mbx_memory_dump_type3_type,
+ &mbox->u.mqe.un.mem_dump_type3, DMP_LMSD);
+ bf_set(lpfc_mbx_memory_dump_type3_link,
+ &mbox->u.mqe.un.mem_dump_type3, phba->sli4_hba.physical_port);
+ bf_set(lpfc_mbx_memory_dump_type3_page_no,
+ &mbox->u.mqe.un.mem_dump_type3, DMP_PAGE_A0);
+ bf_set(lpfc_mbx_memory_dump_type3_length,
+ &mbox->u.mqe.un.mem_dump_type3, DMP_SFF_PAGE_A0_SIZE);
+ mbox->u.mqe.un.mem_dump_type3.addr_lo = putPaddrLow(mp->phys);
+ mbox->u.mqe.un.mem_dump_type3.addr_hi = putPaddrHigh(mp->phys);
+
+ return 0;
+}
+
+/**
+ * lpfc_reg_fcfi - Initialize the REG_FCFI mailbox command
+ * @phba: pointer to the hba structure containing the FCF index and RQ ID.
+ * @mbox: pointer to lpfc mbox command to initialize.
+ *
+ * The REG_FCFI mailbox command supports Fibre Channel Forwarders (FCFs). The
+ * SLI Host uses the command to activate an FCF after it has acquired FCF
+ * information via a READ_FCF mailbox command. This mailbox command also is used
+ * to indicate where received unsolicited frames from this FCF will be sent. By
+ * default this routine will set up the FCF to forward all unsolicited frames
+ * the the RQ ID passed in the @phba. This can be overridden by the caller for
+ * more complicated setups.
+ **/
+void
+lpfc_reg_fcfi(struct lpfc_hba *phba, struct lpfcMboxq *mbox)
+{
+ struct lpfc_mbx_reg_fcfi *reg_fcfi;
+
+ memset(mbox, 0, sizeof(*mbox));
+ reg_fcfi = &mbox->u.mqe.un.reg_fcfi;
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_REG_FCFI);
+ if (phba->nvmet_support == 0) {
+ bf_set(lpfc_reg_fcfi_rq_id0, reg_fcfi,
+ phba->sli4_hba.hdr_rq->queue_id);
+ /* Match everything - rq_id0 */
+ bf_set(lpfc_reg_fcfi_type_match0, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_type_mask0, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_rctl_match0, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_rctl_mask0, reg_fcfi, 0);
+
+ bf_set(lpfc_reg_fcfi_rq_id1, reg_fcfi, REG_FCF_INVALID_QID);
+
+ /* addr mode is bit wise inverted value of fcf addr_mode */
+ bf_set(lpfc_reg_fcfi_mam, reg_fcfi,
+ (~phba->fcf.addr_mode) & 0x3);
+ } else {
+ /* This is ONLY for NVMET MRQ == 1 */
+ if (phba->cfg_nvmet_mrq != 1)
+ return;
+
+ bf_set(lpfc_reg_fcfi_rq_id0, reg_fcfi,
+ phba->sli4_hba.nvmet_mrq_hdr[0]->queue_id);
+ /* Match type FCP - rq_id0 */
+ bf_set(lpfc_reg_fcfi_type_match0, reg_fcfi, FC_TYPE_FCP);
+ bf_set(lpfc_reg_fcfi_type_mask0, reg_fcfi, 0xff);
+ bf_set(lpfc_reg_fcfi_rctl_match0, reg_fcfi,
+ FC_RCTL_DD_UNSOL_CMD);
+
+ bf_set(lpfc_reg_fcfi_rq_id1, reg_fcfi,
+ phba->sli4_hba.hdr_rq->queue_id);
+ /* Match everything else - rq_id1 */
+ bf_set(lpfc_reg_fcfi_type_match1, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_type_mask1, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_rctl_match1, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_rctl_mask1, reg_fcfi, 0);
+ }
+ bf_set(lpfc_reg_fcfi_rq_id2, reg_fcfi, REG_FCF_INVALID_QID);
+ bf_set(lpfc_reg_fcfi_rq_id3, reg_fcfi, REG_FCF_INVALID_QID);
+ bf_set(lpfc_reg_fcfi_info_index, reg_fcfi,
+ phba->fcf.current_rec.fcf_indx);
+ if (phba->fcf.current_rec.vlan_id != LPFC_FCOE_NULL_VID) {
+ bf_set(lpfc_reg_fcfi_vv, reg_fcfi, 1);
+ bf_set(lpfc_reg_fcfi_vlan_tag, reg_fcfi,
+ phba->fcf.current_rec.vlan_id);
+ }
+}
+
+/**
+ * lpfc_reg_fcfi_mrq - Initialize the REG_FCFI_MRQ mailbox command
+ * @phba: pointer to the hba structure containing the FCF index and RQ ID.
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @mode: 0 to register FCFI, 1 to register MRQs
+ *
+ * The REG_FCFI_MRQ mailbox command supports Fibre Channel Forwarders (FCFs).
+ * The SLI Host uses the command to activate an FCF after it has acquired FCF
+ * information via a READ_FCF mailbox command. This mailbox command also is used
+ * to indicate where received unsolicited frames from this FCF will be sent. By
+ * default this routine will set up the FCF to forward all unsolicited frames
+ * the the RQ ID passed in the @phba. This can be overridden by the caller for
+ * more complicated setups.
+ **/
+void
+lpfc_reg_fcfi_mrq(struct lpfc_hba *phba, struct lpfcMboxq *mbox, int mode)
+{
+ struct lpfc_mbx_reg_fcfi_mrq *reg_fcfi;
+
+ /* This is ONLY for MRQ */
+ if (phba->cfg_nvmet_mrq <= 1)
+ return;
+
+ memset(mbox, 0, sizeof(*mbox));
+ reg_fcfi = &mbox->u.mqe.un.reg_fcfi_mrq;
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_REG_FCFI_MRQ);
+ if (mode == 0) {
+ bf_set(lpfc_reg_fcfi_mrq_info_index, reg_fcfi,
+ phba->fcf.current_rec.fcf_indx);
+ if (phba->fcf.current_rec.vlan_id != LPFC_FCOE_NULL_VID) {
+ bf_set(lpfc_reg_fcfi_mrq_vv, reg_fcfi, 1);
+ bf_set(lpfc_reg_fcfi_mrq_vlan_tag, reg_fcfi,
+ phba->fcf.current_rec.vlan_id);
+ }
+ return;
+ }
+
+ bf_set(lpfc_reg_fcfi_mrq_rq_id0, reg_fcfi,
+ phba->sli4_hba.nvmet_mrq_hdr[0]->queue_id);
+ /* Match NVME frames of type FCP (protocol NVME) - rq_id0 */
+ bf_set(lpfc_reg_fcfi_mrq_type_match0, reg_fcfi, FC_TYPE_FCP);
+ bf_set(lpfc_reg_fcfi_mrq_type_mask0, reg_fcfi, 0xff);
+ bf_set(lpfc_reg_fcfi_mrq_rctl_match0, reg_fcfi, FC_RCTL_DD_UNSOL_CMD);
+ bf_set(lpfc_reg_fcfi_mrq_rctl_mask0, reg_fcfi, 0xff);
+ bf_set(lpfc_reg_fcfi_mrq_ptc0, reg_fcfi, 1);
+ bf_set(lpfc_reg_fcfi_mrq_pt0, reg_fcfi, 1);
+
+ bf_set(lpfc_reg_fcfi_mrq_policy, reg_fcfi, 3); /* NVME connection id */
+ bf_set(lpfc_reg_fcfi_mrq_mode, reg_fcfi, 1);
+ bf_set(lpfc_reg_fcfi_mrq_filter, reg_fcfi, 1); /* rq_id0 */
+ bf_set(lpfc_reg_fcfi_mrq_npairs, reg_fcfi, phba->cfg_nvmet_mrq);
+
+ bf_set(lpfc_reg_fcfi_mrq_rq_id1, reg_fcfi,
+ phba->sli4_hba.hdr_rq->queue_id);
+ /* Match everything - rq_id1 */
+ bf_set(lpfc_reg_fcfi_mrq_type_match1, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_mrq_type_mask1, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_mrq_rctl_match1, reg_fcfi, 0);
+ bf_set(lpfc_reg_fcfi_mrq_rctl_mask1, reg_fcfi, 0);
+
+ bf_set(lpfc_reg_fcfi_mrq_rq_id2, reg_fcfi, REG_FCF_INVALID_QID);
+ bf_set(lpfc_reg_fcfi_mrq_rq_id3, reg_fcfi, REG_FCF_INVALID_QID);
+}
+
+/**
+ * lpfc_unreg_fcfi - Initialize the UNREG_FCFI mailbox command
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @fcfi: FCFI to be unregistered.
+ *
+ * The UNREG_FCFI mailbox command supports Fibre Channel Forwarders (FCFs).
+ * The SLI Host uses the command to inactivate an FCFI.
+ **/
+void
+lpfc_unreg_fcfi(struct lpfcMboxq *mbox, uint16_t fcfi)
+{
+ memset(mbox, 0, sizeof(*mbox));
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_UNREG_FCFI);
+ bf_set(lpfc_unreg_fcfi, &mbox->u.mqe.un.unreg_fcfi, fcfi);
+}
+
+/**
+ * lpfc_resume_rpi - Initialize the RESUME_RPI mailbox command
+ * @mbox: pointer to lpfc mbox command to initialize.
+ * @ndlp: The nodelist structure that describes the RPI to resume.
+ *
+ * The RESUME_RPI mailbox command is used to restart I/O to an RPI after a
+ * link event.
+ **/
+void
+lpfc_resume_rpi(struct lpfcMboxq *mbox, struct lpfc_nodelist *ndlp)
+{
+ struct lpfc_hba *phba = ndlp->phba;
+ struct lpfc_mbx_resume_rpi *resume_rpi;
+
+ memset(mbox, 0, sizeof(*mbox));
+ resume_rpi = &mbox->u.mqe.un.resume_rpi;
+ bf_set(lpfc_mqe_command, &mbox->u.mqe, MBX_RESUME_RPI);
+ bf_set(lpfc_resume_rpi_index, resume_rpi,
+ phba->sli4_hba.rpi_ids[ndlp->nlp_rpi]);
+ bf_set(lpfc_resume_rpi_ii, resume_rpi, RESUME_INDEX_RPI);
+ resume_rpi->event_tag = ndlp->phba->fc_eventTag;
+}
+