diff options
| author |   Dmitry Fomichev <dmitry.fomichev@wdc.com> | 2019-11-06 14:34:35 -0800 |
|---|---|---|
| committer |   Mike Snitzer <snitzer@redhat.com> | 2019-11-07 10:08:36 -0500 |
| commit | e7fad909b68aa37470d9f2d2731b5bec355ee5d6 (patch) | |
| tree | 46aa9d420dd63b0767cab11267db4a56b99435c0 /drivers/md/dm-zoned-reclaim.c | |
| parent | dm dust: add limited write failure mode (diff) | |
| download | linux-dev-e7fad909b68aa37470d9f2d2731b5bec355ee5d6.tar.xz linux-dev-e7fad909b68aa37470d9f2d2731b5bec355ee5d6.zip | |
dm zoned: reduce overhead of backing device checks
Commit 75d66ffb48efb3 added backing device health checks and as a part
of these checks, check_events() block ops template call is invoked in
dm-zoned mapping path as well as in reclaim and flush path. Calling
check_events() with ATA or SCSI backing devices introduces a blocking
scsi_test_unit_ready() call being made in sd_check_events(). Even though
the overhead of calling scsi_test_unit_ready() is small for ATA zoned
devices, it is much larger for SCSI and it affects performance in a very
negative way.
Fix this performance regression by executing check_events() only in case
of any I/O errors. The function dmz_bdev_is_dying() is modified to call
only blk_queue_dying(), while calls to check_events() are made in a new
helper function, dmz_check_bdev().
Reported-by: zhangxiaoxu <zhangxiaoxu5@huawei.com>
Fixes: 75d66ffb48efb3 ("dm zoned: properly handle backing device failure")
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Fomichev <dmitry.fomichev@wdc.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Diffstat (limited to 'drivers/md/dm-zoned-reclaim.c')
| -rw-r--r-- | drivers/md/dm-zoned-reclaim.c | 8 |
1 files changed, 2 insertions, 6 deletions
diff --git a/drivers/md/dm-zoned-reclaim.c b/drivers/md/dm-zoned-reclaim.c index d240d7ca8a8a..e7ace908a9b7 100644 --- a/drivers/md/dm-zoned-reclaim.c +++ b/drivers/md/dm-zoned-reclaim.c @@ -82,6 +82,7 @@ static int dmz_reclaim_align_wp(struct dmz_reclaim *zrc, struct dm_zone *zone, "Align zone %u wp %llu to %llu (wp+%u) blocks failed %d", dmz_id(zmd, zone), (unsigned long long)wp_block, (unsigned long long)block, nr_blocks, ret); + dmz_check_bdev(zrc->dev); return ret; } @@ -489,12 +490,7 @@ static void dmz_reclaim_work(struct work_struct *work) ret = dmz_do_reclaim(zrc); if (ret) { dmz_dev_debug(zrc->dev, "Reclaim error %d\n", ret); - if (ret == -EIO) - /* - * LLD might be performing some error handling sequence - * at the underlying device. To not interfere, do not - * attempt to schedule the next reclaim run immediately. - */ + if (!dmz_check_bdev(zrc->dev)) return; } |