OSDN Git Service

dm mirror: do not degrade the mirror on discard error
authorMikulas Patocka <mpatocka@redhat.com>
Thu, 12 Feb 2015 15:09:20 +0000 (10:09 -0500)
committerMike Snitzer <snitzer@redhat.com>
Sat, 14 Feb 2015 00:50:46 +0000 (19:50 -0500)
It may be possible that a device claims discard support but it rejects
discards with -EOPNOTSUPP.  It happens when using loopback on ext2/ext3
filesystem driven by the ext4 driver.  It may also happen if the
underlying devices are moved from one disk on another.

If discard error happens, we reject the bio with -EOPNOTSUPP, but we do
not degrade the array.

This patch fixes failed test shell/lvconvert-repair-transient.sh in the
lvm2 testsuite if the testsuite is extracted on an ext2 or ext3
filesystem and it is being driven by the ext4 driver.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
drivers/md/dm-raid1.c

index 7dfdb5c..089d627 100644 (file)
@@ -604,6 +604,15 @@ static void write_callback(unsigned long error, void *context)
                return;
        }
 
+       /*
+        * If the bio is discard, return an error, but do not
+        * degrade the array.
+        */
+       if (bio->bi_rw & REQ_DISCARD) {
+               bio_endio(bio, -EOPNOTSUPP);
+               return;
+       }
+
        for (i = 0; i < ms->nr_mirrors; i++)
                if (test_bit(i, &error))
                        fail_mirror(ms->mirror + i, DM_RAID1_WRITE_ERROR);