| Message ID | 20200508085242.23406-1-chandrakanth.patil@broadcom.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
| Series | None | expand |
On 5/8/20 10:52 AM, Chandrakanth Patil wrote: > Issue: When TM command times-out driver invokes the controller > reset. Post reset, driver re-fires pended TM commands which leads > to firmware crash. > > Fix: Post controller reset, return pended TM commands back to OS. > > Cc: stable@vger.kernel.org > Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com> > Signed-off-by: Chandrakanth Patil <chandrakanth.patil@broadcom.com> > --- > drivers/scsi/megaraid/megaraid_sas_fusion.c | 7 ++++++- > 1 file changed, 6 insertions(+), 1 deletion(-) > > diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c > index 87f91a38..319f241 100644 > --- a/drivers/scsi/megaraid/megaraid_sas_fusion.c > +++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c > @@ -4180,6 +4180,7 @@ static void megasas_refire_mgmt_cmd(struct megasas_instance *instance, > struct fusion_context *fusion; > struct megasas_cmd *cmd_mfi; > union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc; > + struct MPI2_RAID_SCSI_IO_REQUEST *scsi_io_req; > u16 smid; > bool refire_cmd = false; > u8 result; > @@ -4247,6 +4248,11 @@ static void megasas_refire_mgmt_cmd(struct megasas_instance *instance, > result = COMPLETE_CMD; > } > > + scsi_io_req = (struct MPI2_RAID_SCSI_IO_REQUEST *) > + cmd_fusion->io_request; > + if (scsi_io_req->Function == MPI2_FUNCTION_SCSI_TASK_MGMT) > + result = RETURN_CMD; > + > switch (result) { > case REFIRE_CMD: > megasas_fire_cmd_fusion(instance, req_desc); > @@ -4475,7 +4481,6 @@ megasas_issue_tm(struct megasas_instance *instance, u16 device_handle, > if (!timeleft) { > dev_err(&instance->pdev->dev, > "task mgmt type 0x%x timed out\n", type); > - cmd_mfi->flags |= DRV_DCMD_SKIP_REFIRE; > mutex_unlock(&instance->reset_mutex); > rc = megasas_reset_fusion(instance->host, MFI_IO_TIMEOUT_OCR); > mutex_lock(&instance->reset_mutex); > Why didn't the 'DRV_DCMD_SKIP_REFIRE' work? And if it doesn't work, can't it be removed completely? Cheers, Hannes
> > Issue: When TM command times-out driver invokes the controller reset. > > Post reset, driver re-fires pended TM commands which leads to firmware > > crash. > > > > Fix: Post controller reset, return pended TM commands back to OS. > > > > Cc: stable@vger.kernel.org > > Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com> > > Signed-off-by: Chandrakanth Patil <chandrakanth.patil@broadcom.com> > > --- > > drivers/scsi/megaraid/megaraid_sas_fusion.c | 7 ++++++- > > 1 file changed, 6 insertions(+), 1 deletion(-) > > > > diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c > > b/drivers/scsi/megaraid/megaraid_sas_fusion.c > > index 87f91a38..319f241 100644 > > --- a/drivers/scsi/megaraid/megaraid_sas_fusion.c > > +++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c > > @@ -4180,6 +4180,7 @@ static void megasas_refire_mgmt_cmd(struct > megasas_instance *instance, > > struct fusion_context *fusion; > > struct megasas_cmd *cmd_mfi; > > union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc; > > + struct MPI2_RAID_SCSI_IO_REQUEST *scsi_io_req; > > u16 smid; > > bool refire_cmd = false; > > u8 result; > > @@ -4247,6 +4248,11 @@ static void megasas_refire_mgmt_cmd(struct > megasas_instance *instance, > > result = COMPLETE_CMD; > > } > > > > + scsi_io_req = (struct MPI2_RAID_SCSI_IO_REQUEST *) > > + cmd_fusion->io_request; > > + if (scsi_io_req->Function == MPI2_FUNCTION_SCSI_TASK_MGMT) > > + result = RETURN_CMD; > > + > > switch (result) { > > case REFIRE_CMD: > > megasas_fire_cmd_fusion(instance, req_desc); @@ -4475,7 > +4481,6 > > @@ megasas_issue_tm(struct megasas_instance *instance, u16 > > device_handle, > > if (!timeleft) { > > dev_err(&instance->pdev->dev, > > "task mgmt type 0x%x timed out\n", type); > > - cmd_mfi->flags |= DRV_DCMD_SKIP_REFIRE; > > mutex_unlock(&instance->reset_mutex); > > rc = megasas_reset_fusion(instance->host, MFI_IO_TIMEOUT_OCR); > > mutex_lock(&instance->reset_mutex); > > > Why didn't the 'DRV_DCMD_SKIP_REFIRE' work? > And if it doesn't work, can't it be removed completely? Re-fire logic doesn't check 'DRV_DCMD_SKIP_REFIRE' flag for TM commands but it will check the flag for DCMDs Hence, 'DRV_DCMD_SKIP_REFIRE' flag is only removed for TM commands. -Chandrakanth Patil
diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c b/drivers/scsi/megaraid/megaraid_sas_fusion.c index 87f91a38..319f241 100644 --- a/drivers/scsi/megaraid/megaraid_sas_fusion.c +++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c @@ -4180,6 +4180,7 @@ static void megasas_refire_mgmt_cmd(struct megasas_instance *instance, struct fusion_context *fusion; struct megasas_cmd *cmd_mfi; union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc; + struct MPI2_RAID_SCSI_IO_REQUEST *scsi_io_req; u16 smid; bool refire_cmd = false; u8 result; @@ -4247,6 +4248,11 @@ static void megasas_refire_mgmt_cmd(struct megasas_instance *instance, result = COMPLETE_CMD; } + scsi_io_req = (struct MPI2_RAID_SCSI_IO_REQUEST *) + cmd_fusion->io_request; + if (scsi_io_req->Function == MPI2_FUNCTION_SCSI_TASK_MGMT) + result = RETURN_CMD; + switch (result) { case REFIRE_CMD: megasas_fire_cmd_fusion(instance, req_desc); @@ -4475,7 +4481,6 @@ megasas_issue_tm(struct megasas_instance *instance, u16 device_handle, if (!timeleft) { dev_err(&instance->pdev->dev, "task mgmt type 0x%x timed out\n", type); - cmd_mfi->flags |= DRV_DCMD_SKIP_REFIRE; mutex_unlock(&instance->reset_mutex); rc = megasas_reset_fusion(instance->host, MFI_IO_TIMEOUT_OCR); mutex_lock(&instance->reset_mutex);