提交 e0a33270 编写于 作者: N NeilBrown 提交者: Linus Torvalds

[PATCH] md: Fixed refcounting/locking when attempting read error correction in raid10

We need to hold a reference to rdevs while reading and writing to attempt to
correct read errors.  This reference must be taken under an rcu lock.
Signed-off-by: NNeil Brown <neilb@suse.de>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>
上级 df30d0f4
...@@ -1407,36 +1407,45 @@ static void raid10d(mddev_t *mddev) ...@@ -1407,36 +1407,45 @@ static void raid10d(mddev_t *mddev)
if (s > (PAGE_SIZE>>9)) if (s > (PAGE_SIZE>>9))
s = PAGE_SIZE >> 9; s = PAGE_SIZE >> 9;
rcu_read_lock();
do { do {
int d = r10_bio->devs[sl].devnum; int d = r10_bio->devs[sl].devnum;
rdev = conf->mirrors[d].rdev; rdev = rcu_dereference(conf->mirrors[d].rdev);
if (rdev && if (rdev &&
test_bit(In_sync, &rdev->flags) && test_bit(In_sync, &rdev->flags)) {
sync_page_io(rdev->bdev, atomic_inc(&rdev->nr_pending);
r10_bio->devs[sl].addr + rcu_read_unlock();
sect + rdev->data_offset, success = sync_page_io(rdev->bdev,
s<<9, r10_bio->devs[sl].addr +
conf->tmppage, READ)) sect + rdev->data_offset,
success = 1; s<<9,
else { conf->tmppage, READ);
sl++; rdev_dec_pending(rdev, mddev);
if (sl == conf->copies) rcu_read_lock();
sl = 0; if (success)
break;
} }
sl++;
if (sl == conf->copies)
sl = 0;
} while (!success && sl != r10_bio->read_slot); } while (!success && sl != r10_bio->read_slot);
rcu_read_unlock();
if (success) { if (success) {
int start = sl; int start = sl;
/* write it back and re-read */ /* write it back and re-read */
rcu_read_lock();
while (sl != r10_bio->read_slot) { while (sl != r10_bio->read_slot) {
int d; int d;
if (sl==0) if (sl==0)
sl = conf->copies; sl = conf->copies;
sl--; sl--;
d = r10_bio->devs[sl].devnum; d = r10_bio->devs[sl].devnum;
rdev = conf->mirrors[d].rdev; rdev = rcu_dereference(conf->mirrors[d].rdev);
if (rdev && if (rdev &&
test_bit(In_sync, &rdev->flags)) { test_bit(In_sync, &rdev->flags)) {
atomic_inc(&rdev->nr_pending);
rcu_read_unlock();
atomic_add(s, &rdev->corrected_errors); atomic_add(s, &rdev->corrected_errors);
if (sync_page_io(rdev->bdev, if (sync_page_io(rdev->bdev,
r10_bio->devs[sl].addr + r10_bio->devs[sl].addr +
...@@ -1444,6 +1453,8 @@ static void raid10d(mddev_t *mddev) ...@@ -1444,6 +1453,8 @@ static void raid10d(mddev_t *mddev)
s<<9, conf->tmppage, WRITE) == 0) s<<9, conf->tmppage, WRITE) == 0)
/* Well, this device is dead */ /* Well, this device is dead */
md_error(mddev, rdev); md_error(mddev, rdev);
rdev_dec_pending(rdev, mddev);
rcu_read_lock();
} }
} }
sl = start; sl = start;
...@@ -1453,17 +1464,22 @@ static void raid10d(mddev_t *mddev) ...@@ -1453,17 +1464,22 @@ static void raid10d(mddev_t *mddev)
sl = conf->copies; sl = conf->copies;
sl--; sl--;
d = r10_bio->devs[sl].devnum; d = r10_bio->devs[sl].devnum;
rdev = conf->mirrors[d].rdev; rdev = rcu_dereference(conf->mirrors[d].rdev);
if (rdev && if (rdev &&
test_bit(In_sync, &rdev->flags)) { test_bit(In_sync, &rdev->flags)) {
atomic_inc(&rdev->nr_pending);
rcu_read_unlock();
if (sync_page_io(rdev->bdev, if (sync_page_io(rdev->bdev,
r10_bio->devs[sl].addr + r10_bio->devs[sl].addr +
sect + rdev->data_offset, sect + rdev->data_offset,
s<<9, conf->tmppage, READ) == 0) s<<9, conf->tmppage, READ) == 0)
/* Well, this device is dead */ /* Well, this device is dead */
md_error(mddev, rdev); md_error(mddev, rdev);
rdev_dec_pending(rdev, mddev);
rcu_read_lock();
} }
} }
rcu_read_unlock();
} else { } else {
/* Cannot read from anywhere -- bye bye array */ /* Cannot read from anywhere -- bye bye array */
md_error(mddev, conf->mirrors[r10_bio->devs[r10_bio->read_slot].devnum].rdev); md_error(mddev, conf->mirrors[r10_bio->devs[r10_bio->read_slot].devnum].rdev);
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册