6.6-stable review patch. If anyone has any objections, please let me know.
------------------
From: Justin Tee justin.tee@broadcom.com
[ Upstream commit 1f0f7679ad8942f810b0f19ee9cf098c3502d66a ]
A kref imbalance occurs when handling an unsolicited PRLO in direct attached topology.
Rework PRLO rcv handling when in MAPPED state. Save the state that we were handling a PRLO by setting nlp_last_elscmd to ELS_CMD_PRLO. Then in the lpfc_cmpl_els_logo_acc() completion routine, manually restart discovery. By issuing the PLOGI, which nlp_gets, before nlp_put at the end of the lpfc_cmpl_els_logo_acc() routine, we are saving us from a final nlp_put. And, we are still allowing the unreg_rpi to happen.
Signed-off-by: Justin Tee justin.tee@broadcom.com Link: https://lore.kernel.org/r/20240726231512.92867-7-justintee8345@gmail.com Signed-off-by: Martin K. Petersen martin.petersen@oracle.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/scsi/lpfc/lpfc_els.c | 27 ++++++++++++++++----------- drivers/scsi/lpfc/lpfc_nportdisc.c | 22 ++++++++++++++++++++-- 2 files changed, 36 insertions(+), 13 deletions(-)
diff --git a/drivers/scsi/lpfc/lpfc_els.c b/drivers/scsi/lpfc/lpfc_els.c index 44d3ada9fbbcb..f67d72160d36e 100644 --- a/drivers/scsi/lpfc/lpfc_els.c +++ b/drivers/scsi/lpfc/lpfc_els.c @@ -5228,9 +5228,10 @@ lpfc_cmpl_els_logo_acc(struct lpfc_hba *phba, struct lpfc_iocbq *cmdiocb, /* ACC to LOGO completes to NPort <nlp_DID> */ lpfc_printf_vlog(vport, KERN_INFO, LOG_ELS, "0109 ACC to LOGO completes to NPort x%x refcnt %d " - "Data: x%x x%x x%x\n", - ndlp->nlp_DID, kref_read(&ndlp->kref), ndlp->nlp_flag, - ndlp->nlp_state, ndlp->nlp_rpi); + "last els x%x Data: x%x x%x x%x\n", + ndlp->nlp_DID, kref_read(&ndlp->kref), + ndlp->nlp_last_elscmd, ndlp->nlp_flag, ndlp->nlp_state, + ndlp->nlp_rpi);
/* This clause allows the LOGO ACC to complete and free resources * for the Fabric Domain Controller. It does deliberately skip @@ -5242,18 +5243,22 @@ lpfc_cmpl_els_logo_acc(struct lpfc_hba *phba, struct lpfc_iocbq *cmdiocb, goto out;
if (ndlp->nlp_state == NLP_STE_NPR_NODE) { - /* If PLOGI is being retried, PLOGI completion will cleanup the - * node. The NLP_NPR_2B_DISC flag needs to be retained to make - * progress on nodes discovered from last RSCN. - */ - if ((ndlp->nlp_flag & NLP_DELAY_TMO) && - (ndlp->nlp_last_elscmd == ELS_CMD_PLOGI)) - goto out; - if (ndlp->nlp_flag & NLP_RPI_REGISTERED) lpfc_unreg_rpi(vport, ndlp);
+ /* If came from PRLO, then PRLO_ACC is done. + * Start rediscovery now. + */ + if (ndlp->nlp_last_elscmd == ELS_CMD_PRLO) { + spin_lock_irq(&ndlp->lock); + ndlp->nlp_flag |= NLP_NPR_2B_DISC; + spin_unlock_irq(&ndlp->lock); + ndlp->nlp_prev_state = ndlp->nlp_state; + lpfc_nlp_set_state(vport, ndlp, NLP_STE_PLOGI_ISSUE); + lpfc_issue_els_plogi(vport, ndlp->nlp_DID, 0); + } } + out: /* * The driver received a LOGO from the rport and has ACK'd it. diff --git a/drivers/scsi/lpfc/lpfc_nportdisc.c b/drivers/scsi/lpfc/lpfc_nportdisc.c index 3ed211d093dd1..fe174062e4946 100644 --- a/drivers/scsi/lpfc/lpfc_nportdisc.c +++ b/drivers/scsi/lpfc/lpfc_nportdisc.c @@ -2635,8 +2635,26 @@ lpfc_rcv_prlo_mapped_node(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp, /* flush the target */ lpfc_sli_abort_iocb(vport, ndlp->nlp_sid, 0, LPFC_CTX_TGT);
- /* Treat like rcv logo */ - lpfc_rcv_logo(vport, ndlp, cmdiocb, ELS_CMD_PRLO); + /* Send PRLO_ACC */ + spin_lock_irq(&ndlp->lock); + ndlp->nlp_flag |= NLP_LOGO_ACC; + spin_unlock_irq(&ndlp->lock); + lpfc_els_rsp_acc(vport, ELS_CMD_PRLO, cmdiocb, ndlp, NULL); + + /* Save ELS_CMD_PRLO as the last elscmd and then set to NPR. + * lpfc_cmpl_els_logo_acc is expected to restart discovery. + */ + ndlp->nlp_last_elscmd = ELS_CMD_PRLO; + ndlp->nlp_prev_state = ndlp->nlp_state; + + lpfc_printf_vlog(vport, KERN_INFO, LOG_NODE | LOG_ELS | LOG_DISCOVERY, + "3422 DID x%06x nflag x%x lastels x%x ref cnt %u\n", + ndlp->nlp_DID, ndlp->nlp_flag, + ndlp->nlp_last_elscmd, + kref_read(&ndlp->kref)); + + lpfc_nlp_set_state(vport, ndlp, NLP_STE_NPR_NODE); + return ndlp->nlp_state; }