diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2017-03-16 11:43:48 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2017-03-16 11:43:48 -0700 |
commit | 3009b303b0f9868295205d5f750a2d088e3965ce (patch) | |
tree | 2fa619ad6efb1dfdd53bb9ba54b29b4176edfa64 /drivers/md/raid10.c | |
parent | 69eea5a4ab9c705496e912b55a9d312325de19e6 (diff) | |
parent | 11353b9d10392e79e32603d2178e75feb25eaf0d (diff) |
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md
Pull MD fixes from Shaohua Li:
- fix a parity calculation bug of raid5 cache by Song
- fix a potential deadlock issue by me
- fix two endian issues by Jason
- fix a disk limitation issue by Neil
- other small fixes and cleanup
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md:
md/raid1: fix a trivial typo in comments
md/r5cache: fix set_syndrome_sources() for data in cache
md: fix incorrect use of lexx_to_cpu in does_sb_need_changing
md: fix super_offset endianness in super_1_rdev_size_change
md/raid1/10: fix potential deadlock
md: don't impose the MD_SB_DISKS limit on arrays without metadata.
md: move funcs from pers->resize to update_size
md-cluster: remove useless memset from gather_all_resync_info
md-cluster: free md_cluster_info if node leave cluster
md: delete dead code
md/raid10: submit bio directly to replacement disk
Diffstat (limited to 'drivers/md/raid10.c')
-rw-r--r-- | drivers/md/raid10.c | 41 |
1 files changed, 34 insertions, 7 deletions
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index 0536658c9d40..e89a8d78a9ed 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -1478,11 +1478,24 @@ retry_write: mbio->bi_bdev = (void*)rdev; atomic_inc(&r10_bio->remaining); + + cb = blk_check_plugged(raid10_unplug, mddev, + sizeof(*plug)); + if (cb) + plug = container_of(cb, struct raid10_plug_cb, + cb); + else + plug = NULL; spin_lock_irqsave(&conf->device_lock, flags); - bio_list_add(&conf->pending_bio_list, mbio); - conf->pending_count++; + if (plug) { + bio_list_add(&plug->pending, mbio); + plug->pending_cnt++; + } else { + bio_list_add(&conf->pending_bio_list, mbio); + conf->pending_count++; + } spin_unlock_irqrestore(&conf->device_lock, flags); - if (!mddev_check_plugged(mddev)) + if (!plug) md_wakeup_thread(mddev->thread); } } @@ -1572,7 +1585,25 @@ static void raid10_make_request(struct mddev *mddev, struct bio *bio) split = bio; } + /* + * If a bio is splitted, the first part of bio will pass + * barrier but the bio is queued in current->bio_list (see + * generic_make_request). If there is a raise_barrier() called + * here, the second part of bio can't pass barrier. But since + * the first part bio isn't dispatched to underlaying disks + * yet, the barrier is never released, hence raise_barrier will + * alays wait. We have a deadlock. + * Note, this only happens in read path. For write path, the + * first part of bio is dispatched in a schedule() call + * (because of blk plug) or offloaded to raid10d. + * Quitting from the function immediately can change the bio + * order queued in bio_list and avoid the deadlock. + */ __make_request(mddev, split); + if (split != bio && bio_data_dir(bio) == READ) { + generic_make_request(bio); + break; + } } while (split != bio); /* In case raid10d snuck in to freeze_array */ @@ -3944,10 +3975,6 @@ static int raid10_resize(struct mddev *mddev, sector_t sectors) return ret; } md_set_array_sectors(mddev, size); - if (mddev->queue) { - set_capacity(mddev->gendisk, mddev->array_sectors); - revalidate_disk(mddev->gendisk); - } if (sectors > mddev->dev_sectors && mddev->recovery_cp > oldsize) { mddev->recovery_cp = oldsize; |