From: James Smart <jsmart2021@gmail.com>
Date: Wed, 7 Jul 2021 11:43:46 -0700
Subject: scsi: lpfc: Delay unregistering from transport until GIDFT or ADISC
completes
Patch-mainline: v5.15-rc1
Git-commit: 0614568361b0c1827f999b1fff21223a496c740b
References: bsc#1189385
On an RSCN event, the nodes specified in RSCN payload and in MAPPED state
are moved to NPR state in order to revalidate the login. This triggers an
immediate unregister from SCSI/NVMe backend. The assumption is that the
node may be missing. The re-registration with the backend happens after
either relogin (PLOGI/PRLI; if ADISC is disabled or login truly lost) or
when ADISC completes successfully (rediscover with ADISC enabled).
However, the NVMe-FC standard provides for an RSCN to be triggered when
the remote port supports a discovery controller and there was a change
of discovery log content. As the remote port typically also supports
storage subsystems, this unregister causes all storage controller
connections to fail and require reconnect.
Correct by reworking the code to ensure that the unregistration only occurs
when a login state is truly terminated, thereby leaving the NVMe storage
controllers in place.
The changes made are:
- Retain node state in ADISC_ISSUE when scheduling ADISC ELS retry.
- Do not clear wwpn/wwnn values upon ADISC failure.
- Move MAPPED nodes to NPR during RSCN processing, but do not unregister
with transport. On GIDFT completion, identify missing nodes (not marked
NLP_NPR_2B_DISC) and unregister them.
- Perform unregistration for nodes that will go through ADISC processing
if ADISC completion fails.
- Successful ADISC completion will move node back to MAPPED state.
Link: https://lore.kernel.org/r/20210707184351.67872-16-jsmart2021@gmail.com
Co-developed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[dwagner: - updated fc4_xpt_flags defines due to missing context
- use fallthrough comment instead of statement]
Acked-by: Daniel Wagner <dwagner@suse.de>
---
drivers/scsi/lpfc/lpfc_crtn.h | 2
drivers/scsi/lpfc/lpfc_disc.h | 9 -
drivers/scsi/lpfc/lpfc_els.c | 66 +++++++-----
drivers/scsi/lpfc/lpfc_hbadisc.c | 197 +++++++++++++++++++++++++++----------
drivers/scsi/lpfc/lpfc_nportdisc.c | 9 -
drivers/scsi/lpfc/lpfc_nvme.c | 10 -
drivers/scsi/lpfc/lpfc_nvme.h | 4
7 files changed, 207 insertions(+), 90 deletions(-)
--- a/drivers/scsi/lpfc/lpfc_crtn.h
+++ b/drivers/scsi/lpfc/lpfc_crtn.h
@@ -88,6 +88,8 @@ void lpfc_unregister_vfi_cmpl(struct lpf
void lpfc_enqueue_node(struct lpfc_vport *, struct lpfc_nodelist *);
void lpfc_dequeue_node(struct lpfc_vport *, struct lpfc_nodelist *);
void lpfc_nlp_set_state(struct lpfc_vport *, struct lpfc_nodelist *, int);
+void lpfc_nlp_reg_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp);
+void lpfc_nlp_unreg_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp);
void lpfc_drop_node(struct lpfc_vport *, struct lpfc_nodelist *);
void lpfc_set_disctmo(struct lpfc_vport *);
int lpfc_can_disctmo(struct lpfc_vport *);
--- a/drivers/scsi/lpfc/lpfc_disc.h
+++ b/drivers/scsi/lpfc/lpfc_disc.h
@@ -78,10 +78,11 @@ struct lpfc_node_rrqs {
};
enum lpfc_fc4_xpt_flags {
- NLP_WAIT_FOR_UNREG = 0x1,
- SCSI_XPT_REGD = 0x2,
- NVME_XPT_REGD = 0x4,
- NLP_XPT_HAS_HH = 0x8,
+ NLP_XPT_REGD = 0x1,
+ SCSI_XPT_REGD = 0x2,
+ NVME_XPT_REGD = 0x4,
+ NVME_XPT_UNREG_WAIT = 0x8,
+ NLP_XPT_HAS_HH = 0x10
};
struct lpfc_nodelist {
--- a/drivers/scsi/lpfc/lpfc_els.c
+++ b/drivers/scsi/lpfc/lpfc_els.c
@@ -1641,6 +1641,12 @@ lpfc_plogi_confirm_nport(struct lpfc_hba
if (!new_ndlp || (new_ndlp == ndlp))
return ndlp;
+ /*
+ * Unregister from backend if not done yet. Could have been skipped
+ * due to ADISC
+ */
+ lpfc_nlp_unreg_node(vport, new_ndlp);
+
if (phba->sli_rev == LPFC_SLI_REV4) {
active_rrqs_xri_bitmap = mempool_alloc(phba->active_rrq_pool,
GFP_KERNEL);
@@ -4316,7 +4322,7 @@ lpfc_els_retry(struct lpfc_hba *phba, st
(cmd == ELS_CMD_NVMEPRLI))
lpfc_nlp_set_state(vport, ndlp,
NLP_STE_PRLI_ISSUE);
- else
+ else if (cmd != ELS_CMD_ADISC)
lpfc_nlp_set_state(vport, ndlp,
NLP_STE_NPR_NODE);
ndlp->nlp_last_elscmd = cmd;
@@ -5604,25 +5610,40 @@ lpfc_els_disc_adisc(struct lpfc_vport *v
/* go thru NPR nodes and issue any remaining ELS ADISCs */
list_for_each_entry_safe(ndlp, next_ndlp, &vport->fc_nodes, nlp_listp) {
- if (ndlp->nlp_state == NLP_STE_NPR_NODE &&
- (ndlp->nlp_flag & NLP_NPR_2B_DISC) != 0 &&
- (ndlp->nlp_flag & NLP_NPR_ADISC) != 0) {
- spin_lock_irq(&ndlp->lock);
- ndlp->nlp_flag &= ~NLP_NPR_ADISC;
- spin_unlock_irq(&ndlp->lock);
- ndlp->nlp_prev_state = ndlp->nlp_state;
- lpfc_nlp_set_state(vport, ndlp, NLP_STE_ADISC_ISSUE);
- lpfc_issue_els_adisc(vport, ndlp, 0);
- sentadisc++;
- vport->num_disc_nodes++;
- if (vport->num_disc_nodes >=
- vport->cfg_discovery_threads) {
- spin_lock_irq(shost->host_lock);
- vport->fc_flag |= FC_NLP_MORE;
- spin_unlock_irq(shost->host_lock);
- break;
- }
+
+ if (ndlp->nlp_state != NLP_STE_NPR_NODE ||
+ !(ndlp->nlp_flag & NLP_NPR_ADISC))
+ continue;
+
+ spin_lock_irq(&ndlp->lock);
+ ndlp->nlp_flag &= ~NLP_NPR_ADISC;
+ spin_unlock_irq(&ndlp->lock);
+
+ if (!(ndlp->nlp_flag & NLP_NPR_2B_DISC)) {
+ /* This node was marked for ADISC but was not picked
+ * for discovery. This is possible if the node was
+ * missing in gidft response.
+ *
+ * At time of marking node for ADISC, we skipped unreg
+ * from backend
+ */
+ lpfc_nlp_unreg_node(vport, ndlp);
+ continue;
}
+
+ ndlp->nlp_prev_state = ndlp->nlp_state;
+ lpfc_nlp_set_state(vport, ndlp, NLP_STE_ADISC_ISSUE);
+ lpfc_issue_els_adisc(vport, ndlp, 0);
+ sentadisc++;
+ vport->num_disc_nodes++;
+ if (vport->num_disc_nodes >=
+ vport->cfg_discovery_threads) {
+ spin_lock_irq(shost->host_lock);
+ vport->fc_flag |= FC_NLP_MORE;
+ spin_unlock_irq(shost->host_lock);
+ break;
+ }
+
}
if (sentadisc == 0) {
spin_lock_irq(shost->host_lock);
@@ -6833,13 +6854,6 @@ lpfc_rscn_recovery_check(struct lpfc_vpo
continue;
}
- /* Check to see if we need to NVME rescan this target
- * remoteport.
- */
- if (ndlp->nlp_fc4_type & NLP_FC4_NVME &&
- ndlp->nlp_type & (NLP_NVME_TARGET | NLP_NVME_DISCOVERY))
- lpfc_nvme_rescan_port(vport, ndlp);
-
lpfc_disc_state_machine(vport, ndlp, NULL,
NLP_EVT_DEVICE_RECOVERY);
lpfc_cancel_retry_delay_tmo(vport, ndlp);
--- a/drivers/scsi/lpfc/lpfc_hbadisc.c
+++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
@@ -4349,10 +4349,152 @@ lpfc_nlp_counters(struct lpfc_vport *vpo
spin_unlock_irqrestore(shost->host_lock, iflags);
}
+/* Register a node with backend if not already done */
+void
+lpfc_nlp_reg_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
+{
+
+ unsigned long iflags;
+
+ spin_lock_irqsave(&ndlp->lock, iflags);
+ if (ndlp->fc4_xpt_flags & NLP_XPT_REGD) {
+ /* Already registered with backend, trigger rescan */
+ spin_unlock_irqrestore(&ndlp->lock, iflags);
+
+ if (ndlp->fc4_xpt_flags & NVME_XPT_REGD &&
+ ndlp->nlp_type & (NLP_NVME_TARGET | NLP_NVME_DISCOVERY)) {
+ lpfc_nvme_rescan_port(vport, ndlp);
+ }
+ return;
+ }
+
+ ndlp->fc4_xpt_flags |= NLP_XPT_REGD;
+ spin_unlock_irqrestore(&ndlp->lock, iflags);
+
+ if (lpfc_valid_xpt_node(ndlp)) {
+ 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
+ */
+ lpfc_register_remote_port(vport, ndlp);
+ }
+
+ /* We are done if we do not have any NVME remote node */
+ if (!(ndlp->nlp_fc4_type & NLP_FC4_NVME))
+ return;
+
+ /* Notify the NVME transport of this new rport. */
+ if (vport->phba->sli_rev >= LPFC_SLI_REV4 &&
+ ndlp->nlp_fc4_type & NLP_FC4_NVME) {
+ if (vport->phba->nvmet_support == 0) {
+ /* Register this rport with the transport.
+ * Only NVME Target Rports are registered with
+ * the transport.
+ */
+ if (ndlp->nlp_type & NLP_NVME_TARGET) {
+ vport->phba->nport_event_cnt++;
+ lpfc_nvme_register_port(vport, ndlp);
+ }
+ } else {
+ /* Just take an NDLP ref count since the
+ * target does not register rports.
+ */
+ lpfc_nlp_get(ndlp);
+ }
+ }
+}
+
+/* Unregister a node with backend if not already done */
+void
+lpfc_nlp_unreg_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
+{
+ unsigned long iflags;
+
+ spin_lock_irqsave(&ndlp->lock, iflags);
+ if (!(ndlp->fc4_xpt_flags & NLP_XPT_REGD)) {
+ spin_unlock_irqrestore(&ndlp->lock, iflags);
+ return;
+ }
+
+ ndlp->fc4_xpt_flags &= ~NLP_XPT_REGD;
+ spin_unlock_irqrestore(&ndlp->lock, iflags);
+
+ if (ndlp->rport &&
+ ndlp->fc4_xpt_flags & SCSI_XPT_REGD) {
+ vport->phba->nport_event_cnt++;
+ lpfc_unregister_remote_port(ndlp);
+ }
+
+ if (ndlp->fc4_xpt_flags & NVME_XPT_REGD) {
+ vport->phba->nport_event_cnt++;
+ if (vport->phba->nvmet_support == 0) {
+ /* Start devloss if target. */
+ if (ndlp->nlp_type & NLP_NVME_TARGET)
+ lpfc_nvme_unregister_port(vport, ndlp);
+ } else {
+ /* NVMET has no upcall. */
+ lpfc_nlp_put(ndlp);
+ }
+ }
+
+}
+
+/*
+ * Adisc state change handling
+ */
+static void
+lpfc_handle_adisc_state(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp,
+ int new_state)
+{
+ switch (new_state) {
+ /*
+ * Any state to ADISC_ISSUE
+ * Do nothing, adisc cmpl handling will trigger state changes
+ */
+ case NLP_STE_ADISC_ISSUE:
+ break;
+
+ /*
+ * ADISC_ISSUE to mapped states
+ * Trigger a registration with backend, it will be nop if
+ * already registered
+ */
+ case NLP_STE_UNMAPPED_NODE:
+ ndlp->nlp_type |= NLP_FC_NODE;
+ /* fallthrough */
+ case NLP_STE_MAPPED_NODE:
+ ndlp->nlp_flag &= ~NLP_NODEV_REMOVE;
+ lpfc_nlp_reg_node(vport, ndlp);
+ break;
+
+ /*
+ * ADISC_ISSUE to non-mapped states
+ * We are moving from ADISC_ISSUE to a non-mapped state because
+ * ADISC failed, we would have skipped unregistering with
+ * backend, attempt it now
+ */
+ case NLP_STE_NPR_NODE:
+ ndlp->nlp_flag &= ~NLP_RCV_PLOGI;
+ /* fallthrough */
+ default:
+ lpfc_nlp_unreg_node(vport, ndlp);
+ break;
+ }
+
+}
+
static void
lpfc_nlp_state_cleanup(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp,
int old_state, int new_state)
{
+ /* Trap ADISC changes here */
+ if (new_state == NLP_STE_ADISC_ISSUE ||
+ old_state == NLP_STE_ADISC_ISSUE) {
+ lpfc_handle_adisc_state(vport, ndlp, new_state);
+ return;
+ }
+
if (new_state == NLP_STE_UNMAPPED_NODE) {
ndlp->nlp_flag &= ~NLP_NODEV_REMOVE;
ndlp->nlp_type |= NLP_FC_NODE;
@@ -4362,60 +4504,17 @@ lpfc_nlp_state_cleanup(struct lpfc_vport
if (new_state == NLP_STE_NPR_NODE)
ndlp->nlp_flag &= ~NLP_RCV_PLOGI;
- /* FCP and NVME Transport interface */
+ /* Reg/Unreg for FCP and NVME Transport interface */
if ((old_state == NLP_STE_MAPPED_NODE ||
old_state == NLP_STE_UNMAPPED_NODE)) {
- if (ndlp->rport &&
- lpfc_valid_xpt_node(ndlp)) {
- vport->phba->nport_event_cnt++;
- lpfc_unregister_remote_port(ndlp);
- }
-
- if (ndlp->nlp_fc4_type & NLP_FC4_NVME) {
- vport->phba->nport_event_cnt++;
- if (vport->phba->nvmet_support == 0) {
- /* Start devloss if target. */
- if (ndlp->nlp_type & NLP_NVME_TARGET)
- lpfc_nvme_unregister_port(vport, ndlp);
- } else {
- /* NVMET has no upcall. */
- lpfc_nlp_put(ndlp);
- }
- }
+ /* For nodes marked for ADISC, Handle unreg in ADISC cmpl */
+ if (!(ndlp->nlp_flag & NLP_NPR_ADISC))
+ lpfc_nlp_unreg_node(vport, ndlp);
}
- /* FCP and NVME Transport interfaces */
-
if (new_state == NLP_STE_MAPPED_NODE ||
- new_state == NLP_STE_UNMAPPED_NODE) {
- if (lpfc_valid_xpt_node(ndlp)) {
- 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
- */
- lpfc_register_remote_port(vport, ndlp);
- }
- /* Notify the NVME transport of this new rport. */
- if (vport->phba->sli_rev >= LPFC_SLI_REV4 &&
- ndlp->nlp_fc4_type & NLP_FC4_NVME) {
- if (vport->phba->nvmet_support == 0) {
- /* Register this rport with the transport.
- * Only NVME Target Rports are registered with
- * the transport.
- */
- if (ndlp->nlp_type & NLP_NVME_TARGET) {
- vport->phba->nport_event_cnt++;
- lpfc_nvme_register_port(vport, ndlp);
- }
- } else {
- /* Just take an NDLP ref count since the
- * target does not register rports.
- */
- lpfc_nlp_get(ndlp);
- }
- }
- }
+ new_state == NLP_STE_UNMAPPED_NODE)
+ lpfc_nlp_reg_node(vport, ndlp);
if ((new_state == NLP_STE_MAPPED_NODE) &&
(vport->stat_data_enabled)) {
--- a/drivers/scsi/lpfc/lpfc_nportdisc.c
+++ b/drivers/scsi/lpfc/lpfc_nportdisc.c
@@ -855,6 +855,9 @@ lpfc_rcv_logo(struct lpfc_vport *vport,
ndlp->nlp_last_elscmd = ELS_CMD_PLOGI;
}
out:
+ /* Unregister from backend, could have been skipped due to ADISC */
+ lpfc_nlp_unreg_node(vport, ndlp);
+
ndlp->nlp_prev_state = ndlp->nlp_state;
lpfc_nlp_set_state(vport, ndlp, NLP_STE_NPR_NODE);
@@ -1669,9 +1672,6 @@ lpfc_cmpl_adisc_adisc_issue(struct lpfc_
spin_unlock_irq(&ndlp->lock);
ndlp->nlp_last_elscmd = ELS_CMD_PLOGI;
- memset(&ndlp->nlp_nodename, 0, sizeof(struct lpfc_name));
- memset(&ndlp->nlp_portname, 0, sizeof(struct lpfc_name));
-
ndlp->nlp_prev_state = NLP_STE_ADISC_ISSUE;
lpfc_nlp_set_state(vport, ndlp, NLP_STE_NPR_NODE);
lpfc_unreg_rpi(vport, ndlp);
@@ -2589,13 +2589,14 @@ lpfc_device_recov_mapped_node(struct lpf
void *arg,
uint32_t evt)
{
+ lpfc_disc_set_adisc(vport, ndlp);
+
ndlp->nlp_prev_state = NLP_STE_MAPPED_NODE;
lpfc_nlp_set_state(vport, ndlp, NLP_STE_NPR_NODE);
spin_lock_irq(&ndlp->lock);
ndlp->nlp_flag &= ~(NLP_NODEV_REMOVE | NLP_NPR_2B_DISC);
ndlp->nlp_fc4_type &= ~(NLP_FC4_FCP | NLP_FC4_NVME);
spin_unlock_irq(&ndlp->lock);
- lpfc_disc_set_adisc(vport, ndlp);
return ndlp->nlp_state;
}
--- a/drivers/scsi/lpfc/lpfc_nvme.c
+++ b/drivers/scsi/lpfc/lpfc_nvme.c
@@ -218,8 +218,8 @@ lpfc_nvme_remoteport_delete(struct nvme_
/* The register rebind might have occurred before the delete
* downcall. Guard against this race.
*/
- if (ndlp->fc4_xpt_flags & NLP_WAIT_FOR_UNREG)
- ndlp->fc4_xpt_flags &= ~(NLP_WAIT_FOR_UNREG | NVME_XPT_REGD);
+ if (ndlp->fc4_xpt_flags & NVME_XPT_UNREG_WAIT)
+ ndlp->fc4_xpt_flags &= ~(NVME_XPT_UNREG_WAIT | NVME_XPT_REGD);
spin_unlock_irq(&ndlp->lock);
@@ -2176,7 +2176,7 @@ lpfc_nvme_register_port(struct lpfc_vpor
* race that leaves the WAIT flag set.
*/
spin_lock_irq(&ndlp->lock);
- ndlp->fc4_xpt_flags &= ~NLP_WAIT_FOR_UNREG;
+ ndlp->fc4_xpt_flags &= ~NVME_XPT_UNREG_WAIT;
ndlp->fc4_xpt_flags |= NVME_XPT_REGD;
spin_unlock_irq(&ndlp->lock);
rport = remote_port->private;
@@ -2188,7 +2188,7 @@ lpfc_nvme_register_port(struct lpfc_vpor
*/
spin_lock_irq(&ndlp->lock);
ndlp->nrport = NULL;
- ndlp->fc4_xpt_flags &= ~NLP_WAIT_FOR_UNREG;
+ ndlp->fc4_xpt_flags &= ~NVME_XPT_UNREG_WAIT;
spin_unlock_irq(&ndlp->lock);
rport->ndlp = NULL;
rport->remoteport = NULL;
@@ -2340,7 +2340,7 @@ lpfc_nvme_unregister_port(struct lpfc_vp
* The transport will update it.
*/
spin_lock_irq(&vport->phba->hbalock);
- ndlp->fc4_xpt_flags |= NLP_WAIT_FOR_UNREG;
+ ndlp->fc4_xpt_flags |= NVME_XPT_UNREG_WAIT;
spin_unlock_irq(&vport->phba->hbalock);
/* Don't let the host nvme transport keep sending keep-alives
--- a/drivers/scsi/lpfc/lpfc_nvme.h
+++ b/drivers/scsi/lpfc/lpfc_nvme.h
@@ -33,8 +33,8 @@
#define LPFC_MAX_NVME_INFO_TMP_LEN 100
#define LPFC_NVME_INFO_MORE_STR "\nCould be more info...\n"
-#define lpfc_ndlp_get_nrport(ndlp) \
- ((!ndlp->nrport || (ndlp->fc4_xpt_flags & NLP_WAIT_FOR_UNREG)) \
+#define lpfc_ndlp_get_nrport(ndlp) \
+ ((!ndlp->nrport || (ndlp->fc4_xpt_flags & NVME_XPT_UNREG_WAIT))\
? NULL : ndlp->nrport)
struct lpfc_nvme_qhandle {