[SCSI] lpfc: bug fixes
Following the NPIV support, the following changes have been accumulated
in the testing and qualification of the driver:
- Fix affinity of ELS ring to slow/deferred event processing
- Fix Ring attention masks
- Defer dev_loss_tmo timeout handling to worker thread
- Consolidate link down error classification for better error checking
- Remove unused/deprecated nlp_initiator_tmr timer
- Fix for async scan - move adapter init code back into pci_probe_one
context. Fix async scan interfaces.
- Expand validation of ability to create vports
- Extract VPI resource cnt from firmware
- Tuning of Login/Reject policies to better deal with overwhelmned targets
- Misc ELS and discovery fixes
- Export the npiv_enable attribute to sysfs
- Mailbox handling fix
- Add debugfs support
- A few other small misc fixes:
- wrong return values, double-frees, bad locking
- Added adapter failure heartbeat
Signed-off-by: James Smart <James.Smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
diff --git a/drivers/scsi/lpfc/lpfc_hbadisc.c b/drivers/scsi/lpfc/lpfc_hbadisc.c
index 94ee967..f2f4639 100644
--- a/drivers/scsi/lpfc/lpfc_hbadisc.c
+++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
@@ -37,6 +37,7 @@
#include "lpfc_logmsg.h"
#include "lpfc_crtn.h"
#include "lpfc_vport.h"
+#include "lpfc_debugfs.h"
/* AlpaArray for assignment of scsid for scan-down and bind_method */
static uint8_t lpfcAlpaArray[] = {
@@ -77,6 +78,10 @@
phba = ndlp->vport->phba;
+ lpfc_debugfs_disc_trc(ndlp->vport, LPFC_DISC_TRC_RPORT,
+ "rport terminate: sid:x%x did:x%x flg:x%x",
+ ndlp->nlp_sid, ndlp->nlp_DID, ndlp->nlp_flag);
+
if (ndlp->nlp_sid != NLP_NO_SID) {
lpfc_sli_abort_iocb(phba, &phba->sli.ring[phba->sli.fcp_ring],
ndlp->nlp_sid, 0, 0, LPFC_CTX_TGT);
@@ -93,12 +98,10 @@
{
struct lpfc_rport_data *rdata;
struct lpfc_nodelist * ndlp;
- uint8_t *name;
- int warn_on = 0;
- struct lpfc_hba *phba;
struct lpfc_vport *vport;
- int put_node;
- int put_rport;
+ struct lpfc_hba *phba;
+ struct completion devloss_compl;
+ struct lpfc_work_evt *evtp;
rdata = rport->dd_data;
ndlp = rdata->pnode;
@@ -112,7 +115,70 @@
return;
}
+ vport = ndlp->vport;
+ phba = vport->phba;
+
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_RPORT,
+ "rport devlosscb: sid:x%x did:x%x flg:x%x",
+ ndlp->nlp_sid, ndlp->nlp_DID, ndlp->nlp_flag);
+
+ init_completion(&devloss_compl);
+ evtp = &ndlp->dev_loss_evt;
+
+ if (!list_empty(&evtp->evt_listp))
+ return;
+
+ spin_lock_irq(&phba->hbalock);
+ evtp->evt_arg1 = ndlp;
+ evtp->evt_arg2 = &devloss_compl;
+ evtp->evt = LPFC_EVT_DEV_LOSS;
+ list_add_tail(&evtp->evt_listp, &phba->work_list);
+ if (phba->work_wait)
+ wake_up(phba->work_wait);
+
+ spin_unlock_irq(&phba->hbalock);
+
+ wait_for_completion(&devloss_compl);
+
+ return;
+}
+
+/*
+ * This function is called from the worker thread when dev_loss_tmo
+ * expire.
+ */
+void
+lpfc_dev_loss_tmo_handler(struct lpfc_nodelist *ndlp)
+{
+ struct lpfc_rport_data *rdata;
+ struct fc_rport *rport;
+ struct lpfc_vport *vport;
+ struct lpfc_hba *phba;
+ uint8_t *name;
+ int warn_on = 0;
+
+ rport = ndlp->rport;
+
+ if (!rport)
+ return;
+
+ rdata = rport->dd_data;
+ name = (uint8_t *) &ndlp->nlp_portname;
+ vport = ndlp->vport;
+ phba = vport->phba;
+
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_RPORT,
+ "rport devlosstmo:did:x%x type:x%x id:x%x",
+ ndlp->nlp_DID, ndlp->nlp_type, rport->scsi_target_id);
+
+ if (!(vport->load_flag & FC_UNLOADING) &&
+ ndlp->nlp_state == NLP_STE_MAPPED_NODE)
+ return;
+
if (ndlp->nlp_type & NLP_FABRIC) {
+ int put_node;
+ int put_rport;
+
/* We will clean up these Nodes in linkup */
put_node = rdata->pnode != NULL;
put_rport = ndlp->rport != NULL;
@@ -125,15 +191,6 @@
return;
}
- name = (uint8_t *)&ndlp->nlp_portname;
- vport = ndlp->vport;
- phba = vport->phba;
-
- if (!(vport->load_flag & FC_UNLOADING) &&
- ndlp->nlp_state == NLP_STE_MAPPED_NODE)
- return;
-
-
if (ndlp->nlp_sid != NLP_NO_SID) {
warn_on = 1;
/* flush the target */
@@ -171,6 +228,9 @@
(ndlp->nlp_state != NLP_STE_UNMAPPED_NODE))
lpfc_disc_state_machine(vport, ndlp, NULL, NLP_EVT_DEVICE_RM);
else {
+ int put_node;
+ int put_rport;
+
put_node = rdata->pnode != NULL;
put_rport = ndlp->rport != NULL;
rdata->pnode = NULL;
@@ -180,7 +240,6 @@
if (put_rport)
put_device(&rport->dev);
}
- return;
}
@@ -206,12 +265,17 @@
spin_unlock_irq(&phba->hbalock);
free_evt = 1;
switch (evtp->evt) {
- case LPFC_EVT_DEV_LOSS:
+ case LPFC_EVT_DEV_LOSS_DELAY:
free_evt = 0; /* evt is part of ndlp */
ndlp = (struct lpfc_nodelist *) (evtp->evt_arg1);
vport = ndlp->vport;
if (!vport)
break;
+
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_RPORT,
+ "rport devlossdly:did:x%x flg:x%x",
+ ndlp->nlp_DID, ndlp->nlp_flag, 0);
+
if (!(vport->load_flag & FC_UNLOADING) &&
!(ndlp->nlp_flag & NLP_DELAY_TMO) &&
!(ndlp->nlp_flag & NLP_NPR_2B_DISC)) {
@@ -224,6 +288,14 @@
lpfc_els_retry_delay_handler(ndlp);
free_evt = 0; /* evt is part of ndlp */
break;
+ case LPFC_EVT_DEV_LOSS:
+ ndlp = (struct lpfc_nodelist *)(evtp->evt_arg1);
+ lpfc_nlp_get(ndlp);
+ lpfc_dev_loss_tmo_handler(ndlp);
+ free_evt = 0;
+ complete((struct completion *)(evtp->evt_arg2));
+ lpfc_nlp_put(ndlp);
+ break;
case LPFC_EVT_ONLINE:
if (phba->link_state < LPFC_LINK_DOWN)
*(int *) (evtp->evt_arg1) = lpfc_online(phba);
@@ -272,13 +344,12 @@
}
-static void
+void
lpfc_work_done(struct lpfc_hba *phba)
{
struct lpfc_sli_ring *pring;
- uint32_t ha_copy, control, work_port_events;
+ uint32_t ha_copy, status, control, work_port_events;
struct lpfc_vport *vport;
- int i;
spin_lock_irq(&phba->hbalock);
ha_copy = phba->work_ha;
@@ -310,6 +381,9 @@
if (work_port_events & WORKER_ELS_TMO)
lpfc_els_timeout_handler(vport);
+ if (work_port_events & WORKER_HB_TMO)
+ lpfc_hb_timeout_handler(phba);
+
if (work_port_events & WORKER_MBOX_TMO)
lpfc_mbox_timeout_handler(phba);
@@ -333,30 +407,31 @@
}
spin_unlock_irq(&phba->hbalock);
- for (i = 0; i < phba->sli.num_rings; i++, ha_copy >>= 4) {
- pring = &phba->sli.ring[i];
- if ((ha_copy & HA_RXATT)
- || (pring->flag & LPFC_DEFERRED_RING_EVENT)) {
- if (pring->flag & LPFC_STOP_IOCB_MASK) {
- pring->flag |= LPFC_DEFERRED_RING_EVENT;
- } else {
- lpfc_sli_handle_slow_ring_event(phba, pring,
- (ha_copy &
- HA_RXMASK));
- pring->flag &= ~LPFC_DEFERRED_RING_EVENT;
- }
- /*
- * Turn on Ring interrupts
- */
- spin_lock_irq(&phba->hbalock);
- control = readl(phba->HCregaddr);
- control |= (HC_R0INT_ENA << i);
+ pring = &phba->sli.ring[LPFC_ELS_RING];
+ status = (ha_copy & (HA_RXMASK << (4*LPFC_ELS_RING)));
+ status >>= (4*LPFC_ELS_RING);
+ if ((status & HA_RXMASK)
+ || (pring->flag & LPFC_DEFERRED_RING_EVENT)) {
+ if (pring->flag & LPFC_STOP_IOCB_MASK) {
+ pring->flag |= LPFC_DEFERRED_RING_EVENT;
+ } else {
+ lpfc_sli_handle_slow_ring_event(phba, pring,
+ (status &
+ HA_RXMASK));
+ pring->flag &= ~LPFC_DEFERRED_RING_EVENT;
+ }
+ /*
+ * Turn on Ring interrupts
+ */
+ spin_lock_irq(&phba->hbalock);
+ control = readl(phba->HCregaddr);
+ if (!(control & (HC_R0INT_ENA << LPFC_ELS_RING))) {
+ control |= (HC_R0INT_ENA << LPFC_ELS_RING);
writel(control, phba->HCregaddr);
readl(phba->HCregaddr); /* flush */
- spin_unlock_irq(&phba->hbalock);
}
+ spin_unlock_irq(&phba->hbalock);
}
-
lpfc_work_list_done(phba);
}
@@ -365,7 +440,7 @@
{
struct lpfc_vport *vport;
struct lpfc_sli_ring *pring;
- int i, rc = 0;
+ int rc = 0;
spin_lock_irq(&phba->hbalock);
list_for_each_entry(vport, &phba->port_list, listentry) {
@@ -380,13 +455,10 @@
rc = 1;
goto exit;
}
- for (i = 0; i < phba->sli.num_rings; i++) {
- pring = &phba->sli.ring[i];
- if (pring->flag & LPFC_DEFERRED_RING_EVENT) {
- rc = 1;
- goto exit;
- }
- }
+
+ pring = &phba->sli.ring[LPFC_ELS_RING];
+ if (pring->flag & LPFC_DEFERRED_RING_EVENT)
+ rc = 1;
exit:
if (rc)
phba->work_found++;
@@ -506,6 +578,10 @@
fc_host_post_event(shost, fc_get_event_number(), FCH_EVT_LINKDOWN, 0);
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_ELS_CMD,
+ "Link Down: state:x%x rtry:x%x flg:x%x",
+ vport->port_state, vport->fc_ns_retry, vport->fc_flag);
+
/* Cleanup any outstanding RSCN activity */
lpfc_els_flush_rscn(vport);
@@ -617,6 +693,10 @@
if ((vport->load_flag & FC_UNLOADING) != 0)
return;
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_ELS_CMD,
+ "Link Up: top:x%x speed:x%x flg:x%x",
+ phba->fc_topology, phba->fc_linkspeed, phba->link_flag);
+
/* If NPIV is not enabled, only bring the physical port up */
if (!(phba->sli3_options & LPFC_SLI3_NPIV_ENABLED) &&
(vport != phba->pport))
@@ -935,7 +1015,7 @@
}
} else {
if (!(phba->sli3_options & LPFC_SLI3_NPIV_ENABLED)) {
- if (phba->max_vpi && lpfc_npiv_enable &&
+ if (phba->max_vpi && phba->cfg_npiv_enable &&
(phba->sli_rev == 3))
phba->sli3_options |= LPFC_SLI3_NPIV_ENABLED;
}
@@ -1124,8 +1204,6 @@
"mb status = 0x%x\n",
phba->brd_no, vport->vpi, mb->mbxStatus);
break;
- default:
- phba->vpi_cnt--;
}
vport->unreg_vpi_cmpl = VPORT_OK;
mempool_free(pmb, phba->mbox_mem_pool);
@@ -1182,7 +1260,6 @@
vport->fc_myDID = 0;
goto out;
}
- phba->vpi_cnt++;
vport->num_disc_nodes = 0;
/* go thru NPR list and issue ELS PLOGIs */
@@ -1257,16 +1334,13 @@
if (phba->link_flag & LS_NPIV_FAB_SUPPORTED)
lpfc_initial_fdisc(next_vport);
- else {
- if (phba->sli3_options &
- LPFC_SLI3_NPIV_ENABLED) {
- lpfc_vport_set_state(vport,
- FC_VPORT_NO_FABRIC_SUPP);
- lpfc_printf_log(phba, KERN_ERR, LOG_ELS,
+ else if (phba->sli3_options & LPFC_SLI3_NPIV_ENABLED) {
+ lpfc_vport_set_state(vport,
+ FC_VPORT_NO_FABRIC_SUPP);
+ lpfc_printf_log(phba, KERN_ERR, LOG_ELS,
"%d (%d):0259 No NPIV Fabric "
"support\n",
phba->brd_no, vport->vpi);
- }
}
}
lpfc_do_scr_ns_plogi(phba, vport);
@@ -1377,6 +1451,11 @@
((struct lpfc_rport_data *) ndlp->rport->dd_data)->pnode == ndlp) {
lpfc_nlp_put(ndlp);
}
+
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_RPORT,
+ "rport add: did:x%x flg:x%x type x%x",
+ ndlp->nlp_DID, ndlp->nlp_flag, ndlp->nlp_type);
+
ndlp->rport = rport = fc_remote_port_add(shost, 0, &rport_ids);
if (!rport || !get_device(&rport->dev)) {
dev_printk(KERN_WARNING, &phba->pcidev->dev,
@@ -1394,7 +1473,6 @@
rport_ids.roles |= FC_RPORT_ROLE_FCP_TARGET;
if (ndlp->nlp_type & NLP_FCP_INITIATOR)
rport_ids.roles |= FC_RPORT_ROLE_FCP_INITIATOR;
- del_timer_sync(&ndlp->nlp_initiator_tmr);
if (rport_ids.roles != FC_RPORT_ROLE_UNKNOWN)
@@ -1412,6 +1490,10 @@
{
struct fc_rport *rport = ndlp->rport;
+ lpfc_debugfs_disc_trc(ndlp->vport, LPFC_DISC_TRC_RPORT,
+ "rport delete: did:x%x flg:x%x type x%x",
+ ndlp->nlp_DID, ndlp->nlp_flag, ndlp->nlp_type);
+
fc_remote_port_delete(rport);
return;
@@ -1478,20 +1560,19 @@
if (new_state == NLP_STE_MAPPED_NODE ||
new_state == NLP_STE_UNMAPPED_NODE) {
vport->phba->nport_event_cnt++;
- /*
- * Tell the fc transport about the port, if we haven't
- * already. If we have, and it's a scsi entity, be
- * sure to unblock any attached scsi devices
- */
- lpfc_register_remote_port(vport, ndlp);
+ /*
+ * Tell the fc transport about the port, if we haven't
+ * already. If we have, and it's a scsi entity, be
+ * sure to unblock any attached scsi devices
+ */
+ lpfc_register_remote_port(vport, ndlp);
}
-
- /*
- * if we added to Mapped list, but the remote port
- * registration failed or assigned a target id outside
- * our presentable range - move the node to the
- * Unmapped List
- */
+ /*
+ * if we added to Mapped list, but the remote port
+ * registration failed or assigned a target id outside
+ * our presentable range - move the node to the
+ * Unmapped List
+ */
if (new_state == NLP_STE_MAPPED_NODE &&
(!ndlp->rport ||
ndlp->rport->scsi_target_id == -1 ||
@@ -1533,11 +1614,16 @@
char name1[16], name2[16];
lpfc_printf_log(vport->phba, KERN_INFO, LOG_NODE,
- "%d:0904 NPort state transition x%06x, %s -> %s\n",
- vport->phba->brd_no,
+ "%d (%d):0904 NPort state transition x%06x, %s -> %s\n",
+ vport->phba->brd_no, vport->vpi,
ndlp->nlp_DID,
lpfc_nlp_state_name(name1, sizeof(name1), old_state),
lpfc_nlp_state_name(name2, sizeof(name2), state));
+
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_NODE,
+ "node statechg did:x%x old:%d ste:%d",
+ ndlp->nlp_DID, old_state, state);
+
if (old_state == NLP_STE_NPR_NODE &&
(ndlp->nlp_flag & NLP_DELAY_TMO) != 0 &&
state != NLP_STE_NPR_NODE)
@@ -1571,7 +1657,8 @@
spin_lock_irq(shost->host_lock);
list_del_init(&ndlp->nlp_listp);
spin_unlock_irq(shost->host_lock);
- lpfc_nlp_state_cleanup(vport, ndlp, ndlp->nlp_state, 0);
+ lpfc_nlp_state_cleanup(vport, ndlp, ndlp->nlp_state,
+ NLP_STE_UNUSED_NODE);
}
void
@@ -1585,6 +1672,7 @@
lpfc_nlp_counters(vport, ndlp->nlp_state, -1);
spin_lock_irq(shost->host_lock);
list_del_init(&ndlp->nlp_listp);
+ ndlp->nlp_flag &= ~NLP_TARGET_REMOVE;
spin_unlock_irq(shost->host_lock);
lpfc_nlp_put(ndlp);
}
@@ -1609,6 +1697,13 @@
tmo = ((phba->fc_ratov * 3) + 3);
}
+
+ if (!timer_pending(&vport->fc_disctmo)) {
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_ELS_CMD,
+ "set disc timer: tmo:x%x state:x%x flg:x%x",
+ tmo, vport->port_state, vport->fc_flag);
+ }
+
mod_timer(&vport->fc_disctmo, jiffies + HZ * tmo);
spin_lock_irq(shost->host_lock);
vport->fc_flag |= FC_DISC_TMO;
@@ -1635,6 +1730,10 @@
struct lpfc_hba *phba = vport->phba;
unsigned long iflags;
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_ELS_CMD,
+ "can disc timer: state:x%x rtry:x%x flg:x%x",
+ vport->port_state, vport->fc_ns_retry, vport->fc_flag);
+
/* Turn off discovery timer if its running */
if (vport->fc_flag & FC_DISC_TMO) {
spin_lock_irqsave(shost->host_lock, iflags);
@@ -1898,13 +1997,17 @@
ndlp->nlp_last_elscmd = 0;
del_timer_sync(&ndlp->nlp_delayfunc);
- del_timer_sync(&ndlp->nlp_initiator_tmr);
if (!list_empty(&ndlp->els_retry_evt.evt_listp))
list_del_init(&ndlp->els_retry_evt.evt_listp);
if (!list_empty(&ndlp->dev_loss_evt.evt_listp))
list_del_init(&ndlp->dev_loss_evt.evt_listp);
+ if (!list_empty(&ndlp->dev_loss_evt.evt_listp)) {
+ list_del_init(&ndlp->dev_loss_evt.evt_listp);
+ complete((struct completion *)(ndlp->dev_loss_evt.evt_arg2));
+ }
+
lpfc_unreg_rpi(vport, ndlp);
return 0;
@@ -2418,6 +2521,10 @@
vport->fc_flag &= ~FC_DISC_TMO;
spin_unlock_irq(shost->host_lock);
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_ELS_CMD,
+ "disc timeout: state:x%x rtry:x%x flg:x%x",
+ vport->port_state, vport->fc_ns_retry, vport->fc_flag);
+
switch (vport->port_state) {
case LPFC_LOCAL_CFG_LINK:
@@ -2743,7 +2850,7 @@
spin_lock_irq(shost->host_lock);
ndlp = __lpfc_find_node(vport, lpfc_filter_by_wwpn, wwpn);
spin_unlock_irq(shost->host_lock);
- return NULL;
+ return ndlp;
}
void
@@ -2764,7 +2871,7 @@
}
evtp->evt_arg1 = ndlp;
- evtp->evt = LPFC_EVT_DEV_LOSS;
+ evtp->evt = LPFC_EVT_DEV_LOSS_DELAY;
list_add_tail(&evtp->evt_listp, &phba->work_list);
if (phba->work_wait)
lpfc_worker_wake_up(phba);
@@ -2779,9 +2886,6 @@
memset(ndlp, 0, sizeof (struct lpfc_nodelist));
INIT_LIST_HEAD(&ndlp->els_retry_evt.evt_listp);
INIT_LIST_HEAD(&ndlp->dev_loss_evt.evt_listp);
- init_timer(&ndlp->nlp_initiator_tmr);
- ndlp->nlp_initiator_tmr.function = lpfc_dev_loss_delay;
- ndlp->nlp_initiator_tmr.data = (unsigned long)ndlp;
init_timer(&ndlp->nlp_delayfunc);
ndlp->nlp_delayfunc.function = lpfc_els_retry_delay;
ndlp->nlp_delayfunc.data = (unsigned long)ndlp;
@@ -2790,6 +2894,11 @@
ndlp->nlp_sid = NLP_NO_SID;
INIT_LIST_HEAD(&ndlp->nlp_listp);
kref_init(&ndlp->kref);
+
+ lpfc_debugfs_disc_trc(vport, LPFC_DISC_TRC_NODE,
+ "node init: did:x%x",
+ ndlp->nlp_DID, 0, 0);
+
return;
}
@@ -2798,6 +2907,11 @@
{
struct lpfc_nodelist *ndlp = container_of(kref, struct lpfc_nodelist,
kref);
+
+ lpfc_debugfs_disc_trc(ndlp->vport, LPFC_DISC_TRC_NODE,
+ "node release: did:x%x flg:x%x type:x%x",
+ ndlp->nlp_DID, ndlp->nlp_flag, ndlp->nlp_type);
+
lpfc_nlp_remove(ndlp->vport, ndlp);
mempool_free(ndlp, ndlp->vport->phba->nlp_mem_pool);
}