| Message ID | 1601268657-940-9-git-send-email-muneendra.kumar@broadcom.com |
|---|---|
| State | New |
| Headers | show |
| Series | scsi: Support to handle Intermittent errors | expand |
On 9/28/20 6:50 AM, Muneendra wrote: > Added additional check to set SCMD_NORETRIES_ABORT bit in scmd->state > if port state is marginal prior to initiating io to the port. > > Signed-off-by: Muneendra <muneendra.kumar@broadcom.com> > > --- > v2: > New patch > --- > drivers/scsi/lpfc/lpfc_scsi.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c > index 5e802c8b22a9..1198351d34a8 100644 > --- a/drivers/scsi/lpfc/lpfc_scsi.c > +++ b/drivers/scsi/lpfc/lpfc_scsi.c > @@ -4526,6 +4526,12 @@ lpfc_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *cmnd) > cmnd->result = err; > goto out_fail_command; > } > + > + /* > + * If port state is marginal > + * Set the SCMD_NORETRIES_ABORT bit in scmd->state > + */ > + fc_rport_chkmarginal_set_noretries(rport, cmnd); > ndlp = rdata->pnode; > > if ((scsi_get_prot_op(cmnd) != SCSI_PROT_NORMAL) && > This really should be moved into the transport class; fc_block_rport() would be an ideal place for it. Cheers, Hannes -- Dr. Hannes Reinecke Kernel Storage Architect hare@suse.de +49 911 74053 688 SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
Hi Hannes, Thanks for the input. >> +++ b/drivers/scsi/lpfc/lpfc_scsi.c >> @@ -4526,6 +4526,12 @@ lpfc_queuecommand(struct Scsi_Host *shost, struct >> scsi_cmnd *cmnd) >> cmnd->result = err; >> goto out_fail_command; >> } >> + >> + /* >> + * If port state is marginal >> + * Set the SCMD_NORETRIES_ABORT bit in scmd->state >> + */ >> + fc_rport_chkmarginal_set_noretries(rport, cmnd); >> ndlp = rdata->pnode; >> >> if ((scsi_get_prot_op(cmnd) != SCSI_PROT_NORMAL) && >> >This really should be moved into the transport class; fc_block_rport() >would be an ideal place for it. [Muneendra]Correct me if I didn't understand correctly. As fc_block_rport cannot take arg of scsi_cmd can we add it as part of fc_block_scsi_eh ? Regards, Muneendra.
On 9/30/20 1:33 PM, Muneendra Kumar M wrote: > Hi Hannes, > Thanks for the input. > >>> +++ b/drivers/scsi/lpfc/lpfc_scsi.c >>> @@ -4526,6 +4526,12 @@ lpfc_queuecommand(struct Scsi_Host *shost, struct >>> scsi_cmnd *cmnd) >>> cmnd->result = err; >>> goto out_fail_command; >>> } >>> + >>> + /* >>> + * If port state is marginal >>> + * Set the SCMD_NORETRIES_ABORT bit in scmd->state >>> + */ >>> + fc_rport_chkmarginal_set_noretries(rport, cmnd); >>> ndlp = rdata->pnode; >>> >>> if ((scsi_get_prot_op(cmnd) != SCSI_PROT_NORMAL) && >>> >> This really should be moved into the transport class; fc_block_rport() >> would be an ideal place for it. > > [Muneendra]Correct me if I didn't understand correctly. > As fc_block_rport cannot take arg of scsi_cmd can we add it as part of > fc_block_scsi_eh ? > Ah, right. Actually I meant 'fc_remote_port_chkready()'. That doesnt' take an scmd as argument, but I don't see why we can't modify it to have an additonal 'sdev' parameter... Cheers, Hannes -- Dr. Hannes Reinecke Kernel Storage Architect hare@suse.de +49 911 74053 688 SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
Hi Hannes, >>> + * If port state is marginal >>> + * Set the SCMD_NORETRIES_ABORT bit in scmd->state >>> + */ >>> + fc_rport_chkmarginal_set_noretries(rport, cmnd); >>> ndlp = rdata->pnode; >>> >>> if ((scsi_get_prot_op(cmnd) != SCSI_PROT_NORMAL) && >>> >> This really should be moved into the transport class; >> fc_block_rport() would be an ideal place for it. > >> [Muneendra]Correct me if I didn't understand correctly. >> As fc_block_rport cannot take arg of scsi_cmd can we add it as part of >> fc_block_scsi_eh ? > > >>Ah, right. >>Actually I meant 'fc_remote_port_chkready()'. >>That doesnt' take an scmd as argument, but I don't see why we can't modify >>it to have an additonal 'sdev' parameter... fc_remote_port_chkready function is even getting called in slave alloc (lpfc_slave_alloc, qla2xxx_slave_alloc,...) and few ELS initiative functions bnx2fc_initiate_els, qedf_initiate_els.. where scsi_cmnd cannot be passed. In these cases we need to pass NULL for scsi_cmnd ? Regards, Muneendra.
diff --git a/drivers/scsi/lpfc/lpfc_scsi.c b/drivers/scsi/lpfc/lpfc_scsi.c index 5e802c8b22a9..1198351d34a8 100644 --- a/drivers/scsi/lpfc/lpfc_scsi.c +++ b/drivers/scsi/lpfc/lpfc_scsi.c @@ -4526,6 +4526,12 @@ lpfc_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *cmnd) cmnd->result = err; goto out_fail_command; } + + /* + * If port state is marginal + * Set the SCMD_NORETRIES_ABORT bit in scmd->state + */ + fc_rport_chkmarginal_set_noretries(rport, cmnd); ndlp = rdata->pnode; if ((scsi_get_prot_op(cmnd) != SCSI_PROT_NORMAL) &&
Added additional check to set SCMD_NORETRIES_ABORT bit in scmd->state if port state is marginal prior to initiating io to the port. Signed-off-by: Muneendra <muneendra.kumar@broadcom.com> --- v2: New patch --- drivers/scsi/lpfc/lpfc_scsi.c | 6 ++++++ 1 file changed, 6 insertions(+)