diff options
author | NeilBrown <neilb@suse.de> | 2009-02-24 21:18:47 -0500 |
---|---|---|
committer | NeilBrown <neilb@suse.de> | 2009-02-24 21:18:47 -0500 |
commit | 73d5c38a9536142e062c35997b044e89166e063b (patch) | |
tree | 68fd385cbdee1fa75269974ef210d53a0ae5e311 /drivers/md/raid10.c | |
parent | 78200d45cde2a79c0d0ae0407883bb264caa3c18 (diff) |
md: avoid races when stopping resync.
There has been a race in raid10 and raid1 for a long time
which has only recently started showing up due to a scheduler changed.
When a sync_read request finishes, as soon as reschedule_retry
is called, another thread can mark the resync request as having
completed, so md_do_sync can finish, ->stop can be called, and
->conf can be freed. So using conf after reschedule_retry is not
safe.
Similarly, when finishing a sync_write, calling md_done_sync must be
the last thing we do, as it allows a chain of events which will free
conf and other data structures.
The first of these requires action in raid10.c
The second requires action in raid1.c and raid10.c
Cc: stable@kernel.org
Signed-off-by: NeilBrown <neilb@suse.de>
Diffstat (limited to 'drivers/md/raid10.c')
-rw-r--r-- | drivers/md/raid10.c | 7 |
1 files changed, 4 insertions, 3 deletions
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index e1feb87afc6c..7301631abe04 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c | |||
@@ -1236,6 +1236,7 @@ static void end_sync_read(struct bio *bio, int error) | |||
1236 | /* for reconstruct, we always reschedule after a read. | 1236 | /* for reconstruct, we always reschedule after a read. |
1237 | * for resync, only after all reads | 1237 | * for resync, only after all reads |
1238 | */ | 1238 | */ |
1239 | rdev_dec_pending(conf->mirrors[d].rdev, conf->mddev); | ||
1239 | if (test_bit(R10BIO_IsRecover, &r10_bio->state) || | 1240 | if (test_bit(R10BIO_IsRecover, &r10_bio->state) || |
1240 | atomic_dec_and_test(&r10_bio->remaining)) { | 1241 | atomic_dec_and_test(&r10_bio->remaining)) { |
1241 | /* we have read all the blocks, | 1242 | /* we have read all the blocks, |
@@ -1243,7 +1244,6 @@ static void end_sync_read(struct bio *bio, int error) | |||
1243 | */ | 1244 | */ |
1244 | reschedule_retry(r10_bio); | 1245 | reschedule_retry(r10_bio); |
1245 | } | 1246 | } |
1246 | rdev_dec_pending(conf->mirrors[d].rdev, conf->mddev); | ||
1247 | } | 1247 | } |
1248 | 1248 | ||
1249 | static void end_sync_write(struct bio *bio, int error) | 1249 | static void end_sync_write(struct bio *bio, int error) |
@@ -1264,11 +1264,13 @@ static void end_sync_write(struct bio *bio, int error) | |||
1264 | 1264 | ||
1265 | update_head_pos(i, r10_bio); | 1265 | update_head_pos(i, r10_bio); |
1266 | 1266 | ||
1267 | rdev_dec_pending(conf->mirrors[d].rdev, mddev); | ||
1267 | while (atomic_dec_and_test(&r10_bio->remaining)) { | 1268 | while (atomic_dec_and_test(&r10_bio->remaining)) { |
1268 | if (r10_bio->master_bio == NULL) { | 1269 | if (r10_bio->master_bio == NULL) { |
1269 | /* the primary of several recovery bios */ | 1270 | /* the primary of several recovery bios */ |
1270 | md_done_sync(mddev, r10_bio->sectors, 1); | 1271 | sector_t s = r10_bio->sectors; |
1271 | put_buf(r10_bio); | 1272 | put_buf(r10_bio); |
1273 | md_done_sync(mddev, s, 1); | ||
1272 | break; | 1274 | break; |
1273 | } else { | 1275 | } else { |
1274 | r10bio_t *r10_bio2 = (r10bio_t *)r10_bio->master_bio; | 1276 | r10bio_t *r10_bio2 = (r10bio_t *)r10_bio->master_bio; |
@@ -1276,7 +1278,6 @@ static void end_sync_write(struct bio *bio, int error) | |||
1276 | r10_bio = r10_bio2; | 1278 | r10_bio = r10_bio2; |
1277 | } | 1279 | } |
1278 | } | 1280 | } |
1279 | rdev_dec_pending(conf->mirrors[d].rdev, mddev); | ||
1280 | } | 1281 | } |
1281 | 1282 | ||
1282 | /* | 1283 | /* |