sd: tweak discard heuristics to work around QEMU SCSI issue

7985090aa0 changed the discard heuristics to give preference to the
WRITE SAME commands that (unlike UNMAP) guarantee deterministic results.

Ming Lei discovered that QEMU SCSI's WRITE SAME implementation
internally relied on limits that were only communicated for the UNMAP
case. And therefore discard commands backed by WRITE SAME would fail.

Tweak the heuristics so we still pick UNMAP in the LBPRZ=0 case and only
prefer the WRITE SAME variants if the device has the LBPRZ flag set.

Reported-by: Ming Lei <ming.lei@canonical.com>
Tested-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
This commit is contained in:
Martin K. Petersen 2014-12-05 08:58:03 -05:00 committed by Christoph Hellwig
parent b7392d2247
commit e461338b6c

View file

@ -2623,8 +2623,9 @@ static void sd_read_block_limits(struct scsi_disk *sdkp)
sd_config_discard(sdkp, SD_LBP_WS16);
} else { /* LBP VPD page tells us what to use */
if (sdkp->lbpws)
if (sdkp->lbpu && sdkp->max_unmap_blocks && !sdkp->lbprz)
sd_config_discard(sdkp, SD_LBP_UNMAP);
else if (sdkp->lbpws)
sd_config_discard(sdkp, SD_LBP_WS16);
else if (sdkp->lbpws10)
sd_config_discard(sdkp, SD_LBP_WS10);