2019-10-19 08:41:26

by Daniel Wagner

Subject: [PATCH] scsi: lpfc: Check queue pointer before use

The queue pointer might not be valid. The rest of the code checks the
pointer before accessing it. lpfc_sli4_process_missed_mbox_completions
is the only place where the check is missing.

Fixes: 657add4e5e15 ("scsi: lpfc: Fix poor use of hardware queues if fewer irq vectors")
Cc: James Smart <[email protected]>
Signed-off-by: Daniel Wagner <[email protected]>
---
Hi,

Not entirely sure if this correct. I tried to understand the logic of
the mentioned patch but failed to grasps all the details. Anyway, we
observe a crash in lpfc_sli4_process_missed_mbox_completions() while
iterating the array. All but the last one entry has a valid pointer.

Thanks,
Daniel

drivers/scsi/lpfc/lpfc_sli.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/drivers/scsi/lpfc/lpfc_sli.c
index 379c37451645..149966ba8a17 100644
--- a/drivers/scsi/lpfc/lpfc_sli.c
+++ b/drivers/scsi/lpfc/lpfc_sli.c
@@ -7906,7 +7906,7 @@ lpfc_sli4_process_missed_mbox_completions(struct lpfc_hba *phba)
if (sli4_hba->hdwq) {
for (eqidx = 0; eqidx < phba->cfg_irq_chann; eqidx++) {
eq = phba->sli4_hba.hba_eq_hdl[eqidx].eq;
- if (eq->queue_id == sli4_hba->mbx_cq->assoc_qid) {
+ if (eq && eq->queue_id == sli4_hba->mbx_cq->assoc_qid) {
fpeq = eq;
break;
}
--
2.16.4

2019-10-19 09:21:49

by James Smart

Subject: Re: [PATCH] scsi: lpfc: Check queue pointer before use

On 10/18/2019 9:21 AM, Daniel Wagner wrote:
> The queue pointer might not be valid. The rest of the code checks the
> pointer before accessing it. lpfc_sli4_process_missed_mbox_completions
> is the only place where the check is missing.
>
> Fixes: 657add4e5e15 ("scsi: lpfc: Fix poor use of hardware queues if fewer irq vectors")
> Cc: James Smart <[email protected]>
> Signed-off-by: Daniel Wagner <[email protected]>
> ---
> Hi,
>
> Not entirely sure if this correct. I tried to understand the logic of
> the mentioned patch but failed to grasps all the details. Anyway, we
> observe a crash in lpfc_sli4_process_missed_mbox_completions() while
> iterating the array. All but the last one entry has a valid pointer.
>
> Thanks,
> Daniel
>
> drivers/scsi/lpfc/lpfc_sli.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/drivers/scsi/lpfc/lpfc_sli.c
> index 379c37451645..149966ba8a17 100644
> --- a/drivers/scsi/lpfc/lpfc_sli.c
> +++ b/drivers/scsi/lpfc/lpfc_sli.c
> @@ -7906,7 +7906,7 @@ lpfc_sli4_process_missed_mbox_completions(struct lpfc_hba *phba)
> if (sli4_hba->hdwq) {
> for (eqidx = 0; eqidx < phba->cfg_irq_chann; eqidx++) {
> eq = phba->sli4_hba.hba_eq_hdl[eqidx].eq;
> - if (eq->queue_id == sli4_hba->mbx_cq->assoc_qid) {
> + if (eq && eq->queue_id == sli4_hba->mbx_cq->assoc_qid) {
> fpeq = eq;
> break;
> }

looks fine. Thanks!

Reviewed by: James Smart <[email protected]>

-- james

2019-10-19 09:23:11

by Martin K. Petersen

Subject: Re: [PATCH] scsi: lpfc: Check queue pointer before use

Daniel,

> The queue pointer might not be valid. The rest of the code checks the
> pointer before accessing it. lpfc_sli4_process_missed_mbox_completions
> is the only place where the check is missing.

Applied to 5.4/scsi-fixes, thanks!

--
Martin K. Petersen Oracle Linux Engineering