md/raid10: fix that replacement cannot complete recovery after reassemble

[ Upstream commit bda3153998f3eb2cafa4a6311971143628eacdbc ]

During assemble, the spare marked for replacement is not checked.
conf->fullsync cannot be updated to be 1. As a result, recovery will
treat it as a clean array. All recovering sectors are skipped. Original
device is replaced with the not-recovered spare.

mdadm -C /dev/md0 -l10 -n4 -pn2 /dev/loop[0123]
mdadm /dev/md0 -a /dev/loop4
mdadm /dev/md0 --replace /dev/loop0
mdadm -S /dev/md0 # stop array during recovery

mdadm -A /dev/md0 /dev/loop[01234]

After reassemble, you can see recovery go on, but it completes
immediately. In fact, recovery is not actually processed.

To solve this problem, we just add the missing logics for replacment
spares. (In raid1.c or raid5.c, they have already been checked.)

Reported-by: Alex Chen <alexchen@synology.com>
Reviewed-by: Alex Wu <alexwu@synology.com>
Reviewed-by: Chung-Chiang Cheng <cccheng@synology.com>
Signed-off-by: BingJing Chang <bingjingc@synology.com>
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index b138b5cba286ab6a3d7f09fc7ef7b3f34a4e22ed..6da66c3acd4638533c05bca0d295f724925516e6 100644 (file)
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -3734,6 +3734,13 @@ static int raid10_run(struct mddev *mddev)
                            disk->rdev->saved_raid_disk < 0)
                                conf->fullsync = 1;
                }
+
+               if (disk->replacement &&
+                   !test_bit(In_sync, &disk->replacement->flags) &&
+                   disk->replacement->saved_raid_disk < 0) {
+                       conf->fullsync = 1;
+               }
+
                disk->recovery_disabled = mddev->recovery_disabled - 1;
        }