From: Justin Tee justin.tee@broadcom.com
[ Upstream commit 9914a3d033d3e1d836a43e93e9738e7dd44a096a ]
When NPIV ports are zoned to devices that support both initiator and target mode, a remote device's initiated PRLI results in unintended final kref clean up of the device's ndlp structure. This disrupts NPIV ports' discovery for target devices that support both initiator and target mode.
Modify the NPIV lpfc_drop_node clause such that we allow the ndlp to live so long as it was in NLP_STE_PLOGI_ISSUE, NLP_STE_REG_LOGIN_ISSUE, or NLP_STE_PRLI_ISSUE nlp_state. This allows lpfc's issued PRLI completion routine to determine if the final kref clean up should execute rather than a remote device's issued PRLI.
Fixes: db651ec22524 ("scsi: lpfc: Correct used_rpi count when devloss tmo fires with no recovery") Signed-off-by: Justin Tee justin.tee@broadcom.com Link: https://lore.kernel.org/r/20230523183206.7728-5-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 | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-)
diff --git a/drivers/scsi/lpfc/lpfc_els.c b/drivers/scsi/lpfc/lpfc_els.c index 62d2ca688cd14..e07242ac0f014 100644 --- a/drivers/scsi/lpfc/lpfc_els.c +++ b/drivers/scsi/lpfc/lpfc_els.c @@ -5466,9 +5466,19 @@ lpfc_cmpl_els_rsp(struct lpfc_hba *phba, struct lpfc_iocbq *cmdiocb, ndlp->nlp_flag &= ~NLP_RELEASE_RPI; spin_unlock_irq(&ndlp->lock); } + lpfc_drop_node(vport, ndlp); + } else if (ndlp->nlp_state != NLP_STE_PLOGI_ISSUE && + ndlp->nlp_state != NLP_STE_REG_LOGIN_ISSUE && + ndlp->nlp_state != NLP_STE_PRLI_ISSUE) { + /* Drop ndlp if there is no planned or outstanding + * issued PRLI. + * + * In cases when the ndlp is acting as both an initiator + * and target function, let our issued PRLI determine + * the final ndlp kref drop. + */ + lpfc_drop_node(vport, ndlp); } - - lpfc_drop_node(vport, ndlp); }
/* Release the originating I/O reference. */