Re: [PATCH 1/2] lpfc: call lpfc_sli_validate_fcp_iocb() with the hbalock held

2016-07-20 Thread Martin K. Petersen
> "Johannes" == Johannes Thumshirn writes: Johannes> Call lpfc_sli_validate_fcp_iocb() with the hbalock held, as Johannes> the pointer to iocbq is not guaranteed to still be valid after Johannes> looking it up. Applied to 4.8/scsi-queue. -- Martin K. Petersen Oracle Linux Engineering

Re: [PATCH 1/2] lpfc: call lpfc_sli_validate_fcp_iocb() with the hbalock held

2016-07-19 Thread James Smart
The check is better and is fine to accept. However, in reviewing the code, there may still be some raciness as some of the fields validated aren't set under lock. We'll look deeper. Regardless, holding the lock while counting is a very good thing to do. Signed-off-by: James Smart -- james

Re: [PATCH 1/2] lpfc: call lpfc_sli_validate_fcp_iocb() with the hbalock held

2016-07-19 Thread Johannes Thumshirn
nnedy" > > Cc: "Linux SCSI Mailinglist" , "Johannes > > Thumshirn" > > Sent: Monday, July 18, 2016 10:06:03 AM > > Subject: [PATCH 1/2] lpfc: call lpfc_sli_validate_fcp_iocb() with the > > hbalock held > > > > Call lpfc_sli_va

Re: [PATCH 1/2] lpfc: call lpfc_sli_validate_fcp_iocb() with the hbalock held

2016-07-18 Thread Laurence Oberman
- Original Message - > From: "Johannes Thumshirn" > To: "Martin K . Petersen" , "James Bottomley" > , "James Smart" > , "Dick Kennedy" > Cc: "Linux SCSI Mailinglist" , "Johannes > Thum

[PATCH 1/2] lpfc: call lpfc_sli_validate_fcp_iocb() with the hbalock held

2016-07-18 Thread Johannes Thumshirn
Call lpfc_sli_validate_fcp_iocb() with the hbalock held, as the pointer to iocbq is not guaranteed to still be valid after looking it up. Signed-off-by: Johannes Thumshirn --- drivers/scsi/lpfc/lpfc_sli.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/driver