On 10/25/23 00:25, Damien Le Moal wrote:
On 10/25/23 02:22, Bart Van Assche wrote:
On 10/23/23 17:13, Damien Le Moal wrote:
On 10/24/23 06:54, Bart Van Assche wrote:
case ILLEGAL_REQUEST:
+ /*
+ * Unaligned write command. This may indicate that zoned writes
+ * have been received by the device in the wrong order. If zone
+ * write locking is disabled, retry after all pending commands
+ * have completed.
+ */
+ if (sshdr.asc == 0x21 && sshdr.ascq == 0x04 &&
+ !req->q->limits.use_zone_write_lock &&
+ blk_rq_is_seq_zoned_write(req) &&
+ scmd->retries <= scmd->allowed) {
+ sdev_printk(KERN_INFO, scmd->device,
+ "Retrying unaligned write at LBA %#llx.\n",
+ scsi_get_lba(scmd));
KERN_INFO ? Did you perhaps mean KERN_DEBUG ? An info message for this will be
way too noisy.
Hi Damien,
Are you sure that KERN_INFO will be too noisy? On our test setups we see
this message less than once a day. Anyway, I will change the severity level.
I am not sure. But better safe than sorry :)
So given that we should not scare the user with errors that are not errors (as
the next tries will succeed), we should be silent and log a message only if the
retry count is exhausted and we still see a failure.
Hi Damien,
I will wrap SCSI_LOG_ERROR_RECOVERY(1, ...) around the above
sdev_printk() call. As you probably know SCSI logging is disabled by
default.
Thanks,
Bart.
