On 12/9/22 09:13, Mike Christie wrote:
>> @@ -595,6 +596,22 @@ enum scsi_disposition scsi_check_sense(struct scsi_cmnd *scmd)
>> if (sshdr.asc == 0x10) /* DIF */
>> return SUCCESS;
>>
>> + /*
>> + * Check aborts due to command duration limit policy:
>> + * ABORTED COMMAND additional sense code with the
>> + * COMMAND TIMEOUT BEFORE PROCESSING or
>> + * COMMAND TIMEOUT DURING PROCESSING or
>> + * COMMAND TIMEOUT DURING PROCESSING DUE TO ERROR RECOVERY
>> + * additional sense code qualifiers.
>> + */
>> + if (sshdr.asc == 0x2e &&
>> + sshdr.ascq >= 0x01 && sshdr.ascq <= 0x03) {
>> + set_scsi_ml_byte(scmd, SCSIML_STAT_DL_TIMEOUT);
>> + req->cmd_flags |= REQ_FAILFAST_DEV;
>
> Why are you setting the REQ_FAILFAST_DEV bit? Does libata check for it?
>
> I thought you might have set it because DID_TIME_OUT was set and you wanted
> to hit that check in scsi_noretry_cmd. However, I see that patch where you
> added the new flag so DID_TIME_OUT does not get set sometimes so you probably
> don't hit that path, and you have that check for SCSIML_STAT_DL_TIMEOUT in there
> below.
This is for the block layer (blk_noretry_request() helper) so that the
remainder of the request is not retried. Retrying a CDL command that
timedout goes against the goal of CDL.
--
Damien Le Moal
Western Digital Research
