| Message ID | 8fc77aa7ffbc61e7e55d57e8cfc7423642558b17.1625500974.git.johannes.thumshirn@wdc.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | btrfs: don't block if we can't acquire the reclaim lock | expand |
On 5.07.21 г. 19:32, Johannes Thumshirn wrote: > If we can acquire the reclaim_bgs_lock in on block group reclaim, we block nit: s/can/can't/ , also remove 'in' ? > until it is free. This can potentially stall for a long time. > > While reclaim of block groups is necessary for a good user experience on a > zoned file system, there still is no need to block as it is best effort > only, just like when we're deleting unused block groups. > > Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> > --- > fs/btrfs/block-group.c | 10 +++++++++- > 1 file changed, 9 insertions(+), 1 deletion(-) > > diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c > index c557327b4545..9e7d9d0c763d 100644 > --- a/fs/btrfs/block-group.c > +++ b/fs/btrfs/block-group.c > @@ -1498,7 +1498,15 @@ void btrfs_reclaim_bgs_work(struct work_struct *work) > if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) > return; > > - mutex_lock(&fs_info->reclaim_bgs_lock); > + /* > + * Long running balances can keep us blocked here for eternity, so > + * simply skip reclaim if we're unable to get the mutex. > + */ > + if (!mutex_trylock(&fs_info->reclaim_bgs_lock)) { > + btrfs_exclop_finish(fs_info); > + return; > + } > + > spin_lock(&fs_info->unused_bgs_lock); > while (!list_empty(&fs_info->reclaim_bgs)) { > u64 zone_unusable; >
On 06/07/2021 13:21, Nikolay Borisov wrote: > > > On 5.07.21 г. 19:32, Johannes Thumshirn wrote: >> If we can acquire the reclaim_bgs_lock in on block group reclaim, we block > nit: s/can/can't/ , also remove 'in' ? Oops, you're right. David can you fix this up on apply?
On Tue, Jul 06, 2021 at 01:32:38AM +0900, Johannes Thumshirn wrote: > If we can acquire the reclaim_bgs_lock in on block group reclaim, we block > until it is free. This can potentially stall for a long time. > > While reclaim of block groups is necessary for a good user experience on a > zoned file system, there still is no need to block as it is best effort > only, just like when we're deleting unused block groups. > > Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Added to misc-next with the fixup, thanks.
diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c index c557327b4545..9e7d9d0c763d 100644 --- a/fs/btrfs/block-group.c +++ b/fs/btrfs/block-group.c @@ -1498,7 +1498,15 @@ void btrfs_reclaim_bgs_work(struct work_struct *work) if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) return; - mutex_lock(&fs_info->reclaim_bgs_lock); + /* + * Long running balances can keep us blocked here for eternity, so + * simply skip reclaim if we're unable to get the mutex. + */ + if (!mutex_trylock(&fs_info->reclaim_bgs_lock)) { + btrfs_exclop_finish(fs_info); + return; + } + spin_lock(&fs_info->unused_bgs_lock); while (!list_empty(&fs_info->reclaim_bgs)) { u64 zone_unusable;
If we can acquire the reclaim_bgs_lock in on block group reclaim, we block until it is free. This can potentially stall for a long time. While reclaim of block groups is necessary for a good user experience on a zoned file system, there still is no need to block as it is best effort only, just like when we're deleting unused block groups. Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> --- fs/btrfs/block-group.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)