linux/drivers/md
Coly Li 3fd47bfe55 bcache: stop dc->writeback_rate_update properly
struct delayed_work writeback_rate_update in struct cache_dev is a delayed
worker to call function update_writeback_rate() in period (the interval is
defined by dc->writeback_rate_update_seconds).

When a metadate I/O error happens on cache device, bcache error handling
routine bch_cache_set_error() will call bch_cache_set_unregister() to
retire whole cache set. On the unregister code path, this delayed work is
stopped by calling cancel_delayed_work_sync(&dc->writeback_rate_update).

dc->writeback_rate_update is a special delayed work from others in bcache.
In its routine update_writeback_rate(), this delayed work is re-armed
itself. That means when cancel_delayed_work_sync() returns, this delayed
work can still be executed after several seconds defined by
dc->writeback_rate_update_seconds.

The problem is, after cancel_delayed_work_sync() returns, the cache set
unregister code path will continue and release memory of struct cache set.
Then the delayed work is scheduled to run, __update_writeback_rate()
will reference the already released cache_set memory, and trigger a NULL
pointer deference fault.

This patch introduces two more bcache device flags,
- BCACHE_DEV_WB_RUNNING
  bit set:  bcache device is in writeback mode and running, it is OK for
            dc->writeback_rate_update to re-arm itself.
  bit clear:bcache device is trying to stop dc->writeback_rate_update,
            this delayed work should not re-arm itself and quit.
- BCACHE_DEV_RATE_DW_RUNNING
  bit set:  routine update_writeback_rate() is executing.
  bit clear: routine update_writeback_rate() quits.

This patch also adds a function cancel_writeback_rate_update_dwork() to
wait for dc->writeback_rate_update quits before cancel it by calling
cancel_delayed_work_sync(). In order to avoid a deadlock by unexpected
quit dc->writeback_rate_update, after time_out seconds this function will
give up and continue to call cancel_delayed_work_sync().

And here I explain how this patch stops self re-armed delayed work properly
with the above stuffs.

update_writeback_rate() sets BCACHE_DEV_RATE_DW_RUNNING at its beginning
and clears BCACHE_DEV_RATE_DW_RUNNING at its end. Before calling
cancel_writeback_rate_update_dwork() clear flag BCACHE_DEV_WB_RUNNING.

Before calling cancel_delayed_work_sync() wait utill flag
BCACHE_DEV_RATE_DW_RUNNING is clear. So when calling
cancel_delayed_work_sync(), dc->writeback_rate_update must be already re-
armed, or quite by seeing BCACHE_DEV_WB_RUNNING cleared. In both cases
delayed work routine update_writeback_rate() won't be executed after
cancel_delayed_work_sync() returns.

Inside update_writeback_rate() before calling schedule_delayed_work(), flag
BCACHE_DEV_WB_RUNNING is checked before. If this flag is cleared, it means
someone is about to stop the delayed work. Because flag
BCACHE_DEV_RATE_DW_RUNNING is set already and cancel_delayed_work_sync()
has to wait for this flag to be cleared, we don't need to worry about race
condition here.

If update_writeback_rate() is scheduled to run after checking
BCACHE_DEV_RATE_DW_RUNNING and before calling cancel_delayed_work_sync()
in cancel_writeback_rate_update_dwork(), it is also safe. Because at this
moment BCACHE_DEV_WB_RUNNING is cleared with memory barrier. As I mentioned
previously, update_writeback_rate() will see BCACHE_DEV_WB_RUNNING is clear
and quit immediately.

Because there are more dependences inside update_writeback_rate() to struct
cache_set memory, dc->writeback_rate_update is not a simple self re-arm
delayed work. After trying many different methods (e.g. hold dc->count, or
use locks), this is the only way I can find which works to properly stop
dc->writeback_rate_update delayed work.

Changelog:
v3: change values of BCACHE_DEV_WB_RUNNING and BCACHE_DEV_RATE_DW_RUNNING
    to bit index, for test_bit().
v2: Try to fix the race issue which is pointed out by Junhui.
v1: The initial version for review

Signed-off-by: Coly Li <colyli@suse.de>
Reviewed-by: Junhui Tang <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Cc: Michael Lyle <mlyle@lyle.org>
Cc: Hannes Reinecke <hare@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-03-18 20:15:20 -06:00
..
bcache bcache: stop dc->writeback_rate_update properly 2018-03-18 20:15:20 -06:00
persistent-data dm btree: fix serious bug in btree_split_beneath() 2018-01-17 09:07:55 -05:00
dm-bio-prison-v1.c dm bio prison: use rb_entry() rather than container_of() 2017-06-19 11:03:50 -04:00
dm-bio-prison-v1.h block: switch bios to blk_status_t 2017-06-09 09:27:32 -06:00
dm-bio-prison-v2.c dm bio prison: use rb_entry() rather than container_of() 2017-06-19 11:03:50 -04:00
dm-bio-prison-v2.h dm bio prison v2: new interface for the bio prison 2017-03-07 11:30:16 -05:00
dm-bio-record.h block: replace bi_bdev with a gendisk pointer and partitions index 2017-08-23 12:49:55 -06:00
dm-bufio.c dm bufio: eliminate unnecessary labels in dm_bufio_client_create() 2018-01-17 09:16:04 -05:00
dm-bufio.h dm integrity: optimize writing dm-bufio buffers that are partially changed 2017-08-28 11:47:17 -04:00
dm-builtin.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
dm-cache-background-tracker.c dm cache background tracker: limit amount of background work that may be issued at once 2017-11-10 15:45:03 -05:00
dm-cache-background-tracker.h dm cache: significant rework to leverage dm-bio-prison-v2 2017-03-07 13:28:31 -05:00
dm-cache-block-types.h linux: drop __bitwise__ everywhere 2016-12-16 00:13:41 +02:00
dm-cache-metadata.c dm cache: convert dm_cache_metadata.ref_count from atomic_t to refcount_t 2017-10-24 15:09:51 -04:00
dm-cache-metadata.h dm cache: significant rework to leverage dm-bio-prison-v2 2017-03-07 13:28:31 -05:00
dm-cache-policy-internal.h dm cache: significant rework to leverage dm-bio-prison-v2 2017-03-07 13:28:31 -05:00
dm-cache-policy-smq.c dm cache policy smq: allocate cache blocks in order 2017-11-10 15:45:05 -05:00
dm-cache-policy.c
dm-cache-policy.h dm cache: significant rework to leverage dm-bio-prison-v2 2017-03-07 13:28:31 -05:00
dm-cache-target.c dm: fix various targets to dm_register_target after module __init resources created 2017-12-04 10:23:10 -05:00
dm-core.h dm: various cleanups to md->queue initialization code 2018-01-29 13:44:55 -05:00
dm-crypt.c - DM core fixes to ensure that bio submission follows a depth-first tree 2018-01-31 11:05:47 -08:00
dm-delay.c dm: backfill missing calls to mutex_destroy() 2018-01-17 09:16:15 -05:00
dm-era-target.c dm: do not set 'discards_supported' in targets that do not need it 2017-11-16 16:33:54 -05:00
dm-exception-store.c
dm-exception-store.h
dm-flakey.c dm flakey: check for null arg_name in parse_features() 2018-01-17 09:16:13 -05:00
dm-integrity.c dm integrity: don't store cipher request on the stack 2018-01-17 09:08:57 -05:00
dm-io.c dm io: remove BIOSET_NEED_RESCUER flag from bios bioset 2017-12-13 12:15:56 -05:00
dm-ioctl.c vfs: do bulk POLL* -> EPOLL* replacement 2018-02-11 14:34:03 -08:00
dm-kcopyd.c dm: backfill missing calls to mutex_destroy() 2018-01-17 09:16:15 -05:00
dm-linear.c - Some request-based DM core and DM multipath fixes and cleanups 2017-09-14 13:43:16 -07:00
dm-log-userspace-base.c
dm-log-userspace-transfer.c
dm-log-userspace-transfer.h
dm-log-writes.c dm log writes: fix max length used for kstrndup 2018-01-17 09:16:16 -05:00
dm-log.c block,fs: use REQ_* flags directly 2016-11-01 09:43:26 -06:00
dm-mpath.c - DM core fixes to ensure that bio submission follows a depth-first tree 2018-01-31 11:05:47 -08:00
dm-mpath.h
dm-path-selector.c
dm-path-selector.h
dm-queue-length.c dm mpath selector: more evenly distribute ties 2018-01-29 13:44:58 -05:00
dm-raid1.c md: Convert timers to use timer_setup() 2017-11-14 20:11:57 -07:00
dm-raid.c - DM core fixes to ensure that bio submission follows a depth-first tree 2018-01-31 11:05:47 -08:00
dm-region-hash.c block: rename bio bi_rw to bi_opf 2016-08-07 14:41:02 -06:00
dm-round-robin.c dm round robin: revert "use percpu 'repeat_count' and 'current_path'" 2017-02-17 00:54:09 -05:00
dm-rq.c for-linus-20180204 2018-02-04 11:16:35 -08:00
dm-rq.h dm rq: do not update rq partially in each ending bio 2017-08-28 10:23:28 -04:00
dm-service-time.c dm mpath selector: more evenly distribute ties 2018-01-29 13:44:58 -05:00
dm-snap-persistent.c dm: make flush bios explicitly sync 2017-05-31 10:50:23 -04:00
dm-snap-transient.c
dm-snap.c dm snapshot: use mutex instead of rw_semaphore 2018-01-17 09:16:14 -05:00
dm-stats.c dm: backfill missing calls to mutex_destroy() 2018-01-17 09:16:15 -05:00
dm-stats.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
dm-stripe.c - Some request-based DM core and DM multipath fixes and cleanups 2017-09-14 13:43:16 -07:00
dm-switch.c locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() 2017-10-25 11:01:08 +02:00
dm-sysfs.c dm: move request-based code out to dm-rq.[hc] 2016-06-10 15:15:44 -04:00
dm-table.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
dm-target.c dm: don't return errnos from ->map 2017-06-09 09:27:32 -06:00
dm-thin-metadata.c dm thin metadata: THIN_MAX_CONCURRENT_LOCKS should be 6 2018-01-17 09:07:54 -05:00
dm-thin-metadata.h dm thin: fix a race condition between discarding and provisioning a block 2016-07-20 12:43:35 -04:00
dm-thin.c dm thin: fix trailing semicolon in __remap_and_issue_shared_cell 2018-01-29 13:44:57 -05:00
dm-uevent.c
dm-uevent.h
dm-unstripe.c dm unstripe: fix target length versus number of stripes size check 2018-01-29 13:44:58 -05:00
dm-verity-fec.c dm verity fec: fix GFP flags used with mempool_alloc() 2017-07-26 15:55:44 -04:00
dm-verity-fec.h dm verity fec: limit error correction recursion 2017-03-16 09:37:31 -04:00
dm-verity-target.c Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6 2017-11-14 10:52:09 -08:00
dm-verity.h dm: move dm-verity to generic async completion 2017-11-03 22:11:20 +08:00
dm-zero.c dm: don't return errnos from ->map 2017-06-09 09:27:32 -06:00
dm-zoned-metadata.c dm: backfill missing calls to mutex_destroy() 2018-01-17 09:16:15 -05:00
dm-zoned-reclaim.c dm zoned: use GFP_NOIO in I/O path 2017-07-26 15:55:43 -04:00
dm-zoned-target.c dm: backfill missing calls to mutex_destroy() 2018-01-17 09:16:15 -05:00
dm-zoned.h dm zoned: drive-managed zoned block device target 2017-06-19 11:05:20 -04:00
dm.c block: Add 'lock' as third argument to blk_alloc_queue_node() 2018-02-28 12:23:35 -07:00
dm.h dm: move dm_table_destroy() to same header as dm_table_create() 2018-01-17 09:16:06 -05:00
Kconfig dm: add unstriped target 2018-01-17 09:16:00 -05:00
Makefile dm: add unstriped target 2018-01-17 09:16:00 -05:00
md-bitmap.c Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md 2017-11-14 16:07:26 -08:00
md-bitmap.h Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md 2017-11-14 16:07:26 -08:00
md-cluster.c md-cluster: update document for raid10 2017-11-01 21:32:25 -07:00
md-cluster.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
md-faulty.c md: rename some drivers/md/ files to have an "md-" prefix 2017-10-16 19:06:36 -07:00
md-linear.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
md-linear.h Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md 2017-11-14 16:07:26 -08:00
md-multipath.c md: remove redundant variable q 2017-11-01 21:32:24 -07:00
md-multipath.h Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md 2017-11-14 16:07:26 -08:00
md.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
md.h raid5-ppl: PPL support for disks with write-back cache enabled 2018-01-15 14:29:42 -08:00
raid0.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
raid0.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
raid1-10.c License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
raid1.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
raid1.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
raid5-cache.c raid5-ppl: PPL support for disks with write-back cache enabled 2018-01-15 14:29:42 -08:00
raid5-log.h raid5-ppl: PPL support for disks with write-back cache enabled 2018-01-15 14:29:42 -08:00
raid5-ppl.c raid5-ppl: PPL support for disks with write-back cache enabled 2018-01-15 14:29:42 -08:00
raid5.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
raid5.h License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
raid10.c block: Use blk_queue_flag_*() in drivers instead of queue_flag_*() 2018-03-08 14:13:48 -07:00
raid10.h Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md 2017-11-14 16:07:26 -08:00