iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

btrfs: scrub: Introduce full stripe lock for RAID56

Browse Source 
Unlike mirror based profiles, RAID5/6 recovery needs to read out the
whole full stripe.

And if we don't do proper protection, it can easily cause race condition.

Introduce 2 new functions: lock_full_stripe() and unlock_full_stripe()
for RAID5/6.
Which store a rb_tree of mutexes for full stripes, so scrub callers can
use them to lock a full stripe to avoid race.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ minor comment adjustments ]
Signed-off-by: David Sterba <dsterba@suse.com>
This commit is contained in:
Qu Wenruo
2017-04-14 08:35:54 +08:00
committed by David Sterba
parent fa7aede2ab
commit 0966a7b130
3 changed files with 251 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
fs/btrfs/ctree.h
View File

@ -539,6 +539,14 @@ struct btrfs_io_ctl {
unsigned check_crcs:1;
};
/*
* Tree to record all locked full stripes of a RAID5/6 block group
*/
struct btrfs_full_stripe_locks_tree {
struct rb_root root;
struct mutex lock;
};
struct btrfs_block_group_cache {
struct btrfs_key key;
struct btrfs_block_group_item item;
@ -649,6 +657,9 @@ struct btrfs_block_group_cache {
* Protected by free_space_lock.
*/
int needs_free_space;
/* Record locked full stripes for RAID5/6 block group */
struct btrfs_full_stripe_locks_tree full_stripe_locks_root;
};
/* delayed seq elem */
@ -3653,6 +3664,12 @@ int btrfs_scrub_cancel_dev(struct btrfs_fs_info *info,
struct btrfs_device *dev);
int btrfs_scrub_progress(struct btrfs_fs_info *fs_info, u64 devid,
struct btrfs_scrub_progress *progress);
static inline void btrfs_init_full_stripe_locks_tree(
struct btrfs_full_stripe_locks_tree *locks_root)
{
locks_root->root = RB_ROOT;
mutex_init(&locks_root->lock);
}
/* dev-replace.c */
void btrfs_bio_counter_inc_blocked(struct btrfs_fs_info *fs_info);