diff options
author | James Smart <jsmart2021@gmail.com> | 2019-08-14 16:56:44 -0700 |
---|---|---|
committer | Martin K. Petersen <martin.petersen@oracle.com> | 2019-08-19 22:41:09 -0400 |
commit | 6ede2ddd8bdd28bdaf43d15e70dc0ef38c5448db (patch) | |
tree | 79721f79d6e557de8f6c043282f843b3fb2054be /drivers/scsi/lpfc | |
parent | 3ad348d94452d6ab3aa0316105fb9f34f8ed3140 (diff) |
scsi: lpfc: Fix FLOGI handling across multiple link up/down conditions
It's possible for the driver to initiate an FLOGI and before it completes,
another link down/up transition occurs requiring a new FLOGI. Currently,
nothing is done to abort/noop the older FLOGI request to the adapter, so if
this transition occurs and the FLOGI completion is received after the link
down/up transition, the driver may erroneously act on the older FLOGI. In
most cases, the adapter properly terminates/fails the FLOGI, but there is a
timing condition where the FLOGI may complete on the wire prior to the
transition, but the response may not be seen/processed by the driver before
the driver sees the link transition.
Fix by having the link down handler in the driver run through any
outstanding ELS's and change the completion handler of the ELS so that it
will be no-op'd and released.
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Diffstat (limited to 'drivers/scsi/lpfc')
-rw-r--r-- | drivers/scsi/lpfc/lpfc_els.c | 75 |
1 files changed, 72 insertions, 3 deletions
diff --git a/drivers/scsi/lpfc/lpfc_els.c b/drivers/scsi/lpfc/lpfc_els.c index f12780f4cfbb..8103635adc38 100644 --- a/drivers/scsi/lpfc/lpfc_els.c +++ b/drivers/scsi/lpfc/lpfc_els.c @@ -1207,6 +1207,39 @@ out: } /** + * lpfc_cmpl_els_link_down - Completion callback function for ELS command + * aborted during a link down + * @phba: pointer to lpfc hba data structure. + * @cmdiocb: pointer to lpfc command iocb data structure. + * @rspiocb: pointer to lpfc response iocb data structure. + * + */ +static void +lpfc_cmpl_els_link_down(struct lpfc_hba *phba, struct lpfc_iocbq *cmdiocb, + struct lpfc_iocbq *rspiocb) +{ + IOCB_t *irsp; + uint32_t *pcmd; + uint32_t cmd; + + pcmd = (uint32_t *)(((struct lpfc_dmabuf *)cmdiocb->context2)->virt); + cmd = *pcmd; + irsp = &rspiocb->iocb; + + lpfc_printf_log(phba, KERN_INFO, LOG_ELS, + "6445 ELS completes after LINK_DOWN: " + " Status %x/%x cmd x%x flg x%x\n", + irsp->ulpStatus, irsp->un.ulpWord[4], cmd, + cmdiocb->iocb_flag); + + if (cmdiocb->iocb_flag & LPFC_IO_FABRIC) { + cmdiocb->iocb_flag &= ~LPFC_IO_FABRIC; + atomic_dec(&phba->fabric_iocb_count); + } + lpfc_els_free_iocb(phba, cmdiocb); +} + +/** * lpfc_issue_els_flogi - Issue an flogi iocb command for a vport * @vport: pointer to a host virtual N_Port data structure. * @ndlp: pointer to a node-list data structure. @@ -7960,18 +7993,40 @@ lpfc_els_flush_cmd(struct lpfc_vport *vport) if (phba->sli_rev == LPFC_SLI_REV4) spin_lock(&pring->ring_lock); + /* First we need to issue aborts to outstanding cmds on txcmpl */ list_for_each_entry_safe(piocb, tmp_iocb, &pring->txcmplq, list) { if (piocb->iocb_flag & LPFC_IO_LIBDFC) continue; if (piocb->vport != vport) continue; - list_add_tail(&piocb->dlist, &abort_list); + + /* On the ELS ring we can have ELS_REQUESTs or + * GEN_REQUESTs waiting for a response. + */ + cmd = &piocb->iocb; + if (cmd->ulpCommand == CMD_ELS_REQUEST64_CR) { + list_add_tail(&piocb->dlist, &abort_list); + + /* If the link is down when flushing ELS commands + * the firmware will not complete them till after + * the link comes back up. This may confuse + * discovery for the new link up, so we need to + * change the compl routine to just clean up the iocb + * and avoid any retry logic. + */ + if (phba->link_state == LPFC_LINK_DOWN) + piocb->iocb_cmpl = lpfc_cmpl_els_link_down; + } + if (cmd->ulpCommand == CMD_GEN_REQUEST64_CR) + list_add_tail(&piocb->dlist, &abort_list); } + if (phba->sli_rev == LPFC_SLI_REV4) spin_unlock(&pring->ring_lock); spin_unlock_irq(&phba->hbalock); - /* Abort each iocb on the aborted list and remove the dlist links. */ + + /* Abort each txcmpl iocb on aborted list and remove the dlist links. */ list_for_each_entry_safe(piocb, tmp_iocb, &abort_list, dlist) { spin_lock_irq(&phba->hbalock); list_del_init(&piocb->dlist); @@ -7987,6 +8042,9 @@ lpfc_els_flush_cmd(struct lpfc_vport *vport) if (phba->sli_rev == LPFC_SLI_REV4) spin_lock(&pring->ring_lock); + /* No need to abort the txq list, + * just queue them up for lpfc_sli_cancel_iocbs + */ list_for_each_entry_safe(piocb, tmp_iocb, &pring->txq, list) { cmd = &piocb->iocb; @@ -8007,11 +8065,22 @@ lpfc_els_flush_cmd(struct lpfc_vport *vport) list_del_init(&piocb->list); list_add_tail(&piocb->list, &abort_list); } + + /* The same holds true for any FLOGI/FDISC on the fabric_iocb_list */ + if (vport == phba->pport) { + list_for_each_entry_safe(piocb, tmp_iocb, + &phba->fabric_iocb_list, list) { + cmd = &piocb->iocb; + list_del_init(&piocb->list); + list_add_tail(&piocb->list, &abort_list); + } + } + if (phba->sli_rev == LPFC_SLI_REV4) spin_unlock(&pring->ring_lock); spin_unlock_irq(&phba->hbalock); - /* Cancell all the IOCBs from the completions list */ + /* Cancel all the IOCBs from the completions list */ lpfc_sli_cancel_iocbs(phba, &abort_list, IOSTAT_LOCAL_REJECT, IOERR_SLI_ABORTED); |