diff options
| author | NeilBrown <neilb@suse.com> | 2016-11-18 16:16:11 +1100 |
|---|---|---|
| committer | Shaohua Li <shli@fb.com> | 2016-11-22 09:11:33 -0800 |
| commit | 46533ff7fefb7e9e3539494f5873b00091caa8eb (patch) | |
| tree | c594fb27cb08069d7ce63bcaa5439d5e7313a6aa /drivers/md/bitmap.c | |
| parent | 688834e6ae6b21e3d98b5cf2586aa4a9b515c3a0 (diff) | |
| download | linux-46533ff7fefb7e9e3539494f5873b00091caa8eb.tar.bz2 | |
md: Use REQ_FAILFAST_* on metadata writes where appropriate
This can only be supported on personalities which ensure
that md_error() never causes an array to enter the 'failed'
state. i.e. if marking a device Faulty would cause some
data to be inaccessible, the device is status is left as
non-Faulty. This is true for RAID1 and RAID10.
If we get a failure writing metadata but the device doesn't
fail, it must be the last device so we re-write without
FAILFAST to improve chance of success. We also flag the
device as LastDev so that future metadata updates don't
waste time on failfast writes.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
Diffstat (limited to 'drivers/md/bitmap.c')
| -rw-r--r-- | drivers/md/bitmap.c | 15 |
1 files changed, 12 insertions, 3 deletions
diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c index cf77cbf9ed22..c4621571b718 100644 --- a/drivers/md/bitmap.c +++ b/drivers/md/bitmap.c @@ -209,11 +209,13 @@ static struct md_rdev *next_active_rdev(struct md_rdev *rdev, struct mddev *mdde static int write_sb_page(struct bitmap *bitmap, struct page *page, int wait) { - struct md_rdev *rdev = NULL; + struct md_rdev *rdev; struct block_device *bdev; struct mddev *mddev = bitmap->mddev; struct bitmap_storage *store = &bitmap->storage; +restart: + rdev = NULL; while ((rdev = next_active_rdev(rdev, mddev)) != NULL) { int size = PAGE_SIZE; loff_t offset = mddev->bitmap_info.offset; @@ -269,8 +271,8 @@ static int write_sb_page(struct bitmap *bitmap, struct page *page, int wait) page); } - if (wait) - md_super_wait(mddev); + if (wait && md_super_wait(mddev) < 0) + goto restart; return 0; bad_alignment: @@ -428,6 +430,13 @@ static void bitmap_wait_writes(struct bitmap *bitmap) wait_event(bitmap->write_wait, atomic_read(&bitmap->pending_writes)==0); else + /* Note that we ignore the return value. The writes + * might have failed, but that would just mean that + * some bits which should be cleared haven't been, + * which is safe. The relevant bitmap blocks will + * probably get written again, but there is no great + * loss if they aren't. + */ md_super_wait(bitmap->mddev); } |