Skip to content

Commit

Permalink
drbd: fix race between drbdadm invalidate/verify and finishing resync
Browse files Browse the repository at this point in the history
When a resync or online verify is finished or aborted,
drbd does a bulk write-out of changed bitmap pages.

If *in that very moment* a new verify or resync is triggered,
this can race:
 ASSERT( !test_bit(BITMAP_IO, &mdev->flags) ) in drbd_main.c
 FIXME going to queue 'set_n_write from StartingSync' but 'write from resync_finished' still pending?
and similar.

This can be observed with e.g. tight invalidate loops in test scripts,
and probably has no real-life implication.

Still, that race can be solved by first quiescen the device,
before starting a new resync or verify.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
  • Loading branch information
Lars Ellenberg authored and Philipp Reisner committed Nov 8, 2012
1 parent 07be15b commit 5016b82
Showing 1 changed file with 10 additions and 0 deletions.
10 changes: 10 additions & 0 deletions drivers/block/drbd/drbd_nl.c
Original file line number Diff line number Diff line change
Expand Up @@ -2372,6 +2372,7 @@ int drbd_adm_invalidate(struct sk_buff *skb, struct genl_info *info)

/* If there is still bitmap IO pending, probably because of a previous
* resync just being finished, wait for it before requesting a new resync. */
drbd_suspend_io(mdev);
wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));

retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_T), CS_ORDERED);
Expand All @@ -2390,6 +2391,7 @@ int drbd_adm_invalidate(struct sk_buff *skb, struct genl_info *info)

retcode = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_T));
}
drbd_resume_io(mdev);

out:
drbd_adm_finish(info, retcode);
Expand Down Expand Up @@ -2435,6 +2437,11 @@ int drbd_adm_invalidate_peer(struct sk_buff *skb, struct genl_info *info)

mdev = adm_ctx.mdev;

/* If there is still bitmap IO pending, probably because of a previous
* resync just being finished, wait for it before requesting a new resync. */
drbd_suspend_io(mdev);
wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));

retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S), CS_ORDERED);
if (retcode < SS_SUCCESS) {
if (retcode == SS_NEED_CONNECTION && mdev->state.role == R_PRIMARY) {
Expand All @@ -2450,6 +2457,7 @@ int drbd_adm_invalidate_peer(struct sk_buff *skb, struct genl_info *info)
} else
retcode = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S));
}
drbd_resume_io(mdev);

out:
drbd_adm_finish(info, retcode);
Expand Down Expand Up @@ -2903,8 +2911,10 @@ int drbd_adm_start_ov(struct sk_buff *skb, struct genl_info *info)
}
/* If there is still bitmap IO pending, e.g. previous resync or verify
* just being finished, wait for it before requesting a new resync. */
drbd_suspend_io(mdev);
wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));
retcode = drbd_request_state(mdev,NS(conn,C_VERIFY_S));
drbd_resume_io(mdev);
out:
drbd_adm_finish(info, retcode);
return 0;
Expand Down

0 comments on commit 5016b82

Please sign in to comment.