Raid5 Bug Fixes
cover Raid5 Bug Fixes
Hey,
Please find two patches with fixes to the raid5 code.
The first patch fixes a bug in my recent commit that causes data corruption in very specific circumstances.
The second patch fixes a theoretical issue with nested waits in code that was recently cleaned up in the previous series (though the issue always existed).
Thanks,
Logan
--
Logan Gunthorpe (2):
md/raid5: Fix sectors_to_do bitmap overflow in raid5_make_request() md/raid5: Convert prepare_to_wait() to wait_woken() api
drivers/md/raid5.c | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-)
base-commit: ff4ec5f79108cf82fe7168547c76fe754c4ade0a --
2.30.2
PATCH 1/2 md/raid5: Fix sectors_to_do bitmap overflow in raid5_make_request()
For unaligned IO that have nearly maximum sectors, the number of stripes will end up being one greater than the size of the bitmap. When this happens, the last stripe in the IO will not be processed as it should be, resulting in data corruption.
However, this is not normally seen when the backing block devices have 4K physical block sizes seeing the block layer will split the request before that happens.
To fix this increase the bitmap size by one bit and ensure the full number of stripes are checked when calling find_first_bit().
--- drivers/md/raid5.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-)
drivers/md/raid5.c +11 -8
@@ -5872,8 +5872,11 @@ struct stripe_request_ctx {/* last sector in the request */ sector_t last_sector; -/* bitmap to track stripe sectors that have been added to stripes */ -DECLARE_BITMAP (sectors_to_do,RAID5_MAX_REQ_STRIPES ); +/* +* bitmap to track stripe sectors that have been added to stripes +* add one to account for unaligned requests +*/ +DECLARE_BITMAP (sectors_to_do,RAID5_MAX_REQ_STRIPES +1 );/* the request had REQ_PREFLUSH, cleared after the first stripe_head */ bool do_flush;
@@ -6046,7 +6049,7 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)const int rw =bio_data_dir (bi);enum stripe_result res;DEFINE_WAIT (w); -int s; +int s, stripe_cnt;if (unlikely (bi->bi_opf &REQ_PREFLUSH )) {int ret =log_handle_flush_request (conf, bi);
@@ -6090,9 +6093,9 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi) ctx.last_sector =bio_end_sector (bi); bi->bi_next =NULL ; -bitmap_set (ctx.sectors_to_do,0 , -DIV_ROUND_UP_SECTOR_T (ctx.last_sector - logical_sector, -RAID5_STRIPE_SECTORS (conf))); + stripe_cnt =DIV_ROUND_UP_SECTOR_T (ctx.last_sector - logical_sector, +RAID5_STRIPE_SECTORS (conf)); +bitmap_set (ctx.sectors_to_do,0 , stripe_cnt);pr_debug ("raid456: %s, logical %llu to %llu\n" , __func__, bi->bi_iter.bi_sector, ctx.last_sector);
@@ -6137,8 +6140,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)continue ; } - s =find_first_bit (ctx.sectors_to_do,RAID5_MAX_REQ_STRIPES ); -if (s ==RAID5_MAX_REQ_STRIPES ) + s =find_first_bit (ctx.sectors_to_do, stripe_cnt); +if (s == stripe_cnt)break ; logical_sector = ctx.first_sector +
Signature · 1 line
2.30.2
PATCH 2/2 md/raid5: Convert prepare_to_wait() to wait_woken() api
In reply to Logan Gunthorperaid5_get_active_stripe() can sleep in various situations and it is called by make_stripe_request() while inside the prepare_to_wait()/finish_wait() section. Nested waits like this are not supported.
This was noticed while making other changes that add different sleeps to raid5_get_active_stripe() that caused a WARNING with and CONFIG_DEBUG_ATOMIC_SLEEP.
No ill effects have been noticed with the code as is, but theoretically a nested and here could cause a dead lock so it should be fixed.
To fix this, convert the prepare_to_wait() call to use wake_woken() which supports nested sleeps.
--- drivers/md/raid5.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-)
drivers/md/raid5.c +6 -7
@@ -6043,12 +6043,12 @@ static enum stripe_result make_stripe_request(struct mddev *mddev,static bool raid5_make_request (struct mddev *mddev,struct bio * bi) { +DEFINE_WAIT_FUNC (wait, woken_wake_function);struct r5conf *conf = mddev->private;sector_t logical_sector;struct stripe_request_ctx ctx = {};const int rw =bio_data_dir (bi);enum stripe_result res; -DEFINE_WAIT (w);int s, stripe_cnt;if (unlikely (bi->bi_opf &REQ_PREFLUSH )) {
@@ -6111,7 +6111,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi)return true; }md_account_bio (mddev, &bi); -prepare_to_wait (&conf->wait_for_overlap, &w,TASK_UNINTERRUPTIBLE ); + +add_wait_queue (&conf->wait_for_overlap, &wait);while (1 ) { res =make_stripe_request (mddev, conf, &ctx, logical_sector, bi);
@@ -6134,9 +6135,8 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi) ctx.batch_last =NULL ; } -schedule (); -prepare_to_wait (&conf->wait_for_overlap, &w, -TASK_UNINTERRUPTIBLE ); +wait_woken (&wait,TASK_UNINTERRUPTIBLE , +MAX_SCHEDULE_TIMEOUT );continue ; }
@@ -6147,8 +6147,7 @@ static bool raid5_make_request(struct mddev *mddev, struct bio * bi) logical_sector = ctx.first_sector + (s <<RAID5_STRIPE_SHIFT (conf)); } - -finish_wait (&conf->wait_for_overlap, &w); +remove_wait_queue (&conf->wait_for_overlap, &wait);if (ctx.batch_last)raid5_release_stripe (ctx.batch_last);
Signature · 1 line
2.30.2
Quoted 5 lines · Logan Gunthorpe
Hey,
Please find two patches with fixes to the raid5 code.
Quoted 8 lines
The first patch fixes a bug in my recent commit that causes data corruption in very specific circumstances.
The second patch fixes a theoretical issue with nested waits in code that was recently cleaned up in the previous series (though the issue always existed).
Thanks,
Applied to md-next after fixing a couple typos.
Thanks!
Song
Quoted 16 lines
Logan
--
Logan Gunthorpe (2):
md/raid5: Fixsectors_to_dobitmap overflow inraid5_make_request()md/raid5: Convertprepare_to_wait()towait_woken()apidrivers/md/raid5.c | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-)base-commit:
ff4ec5f79108cf82fe7168547c76fe754c4ade0a--
2.30.2