Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 7ee1fb93 authored by Lars Ellenberg's avatar Lars Ellenberg Committed by Philipp Reisner
Browse files

drbd: flush drbd work queue before invalidate/invalidate remote 



If you do back to back wait-sync/invalidate on a Primary in a tight loop,
during application IO load, you could trigger a race:
  kernel: block drbd6: FIXME going to queue 'set_n_write from StartingSync'
	but 'write from resync_finished' still pending?

Fix this by changing the order of the drbd_queue_work() and
the wake_up() in dec_ap_pending(), and adding the additional
drbd_flush_workqueue() before requesting the full sync.

Signed-off-by: default avatarPhilipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: default avatarLars Ellenberg <lars.ellenberg@linbit.com>
parent c12e9c89

drivers/block/drbd/drbd_int.h

+6 −4
Original line number Diff line number Diff line
@@ -2421,15 +2421,17 @@ static inline void dec_ap_bio(struct drbd_conf *mdev) 
	int ap_bio = atomic_dec_return(&mdev->ap_bio_cnt);



	D_ASSERT(ap_bio >= 0);



	if (ap_bio == 0 && test_bit(BITMAP_IO, &mdev->flags)) {

		if (!test_and_set_bit(BITMAP_IO_QUEUED, &mdev->flags))

			drbd_queue_work(&mdev->data.work, &mdev->bm_io_work.w);

	}



	/* this currently does wake_up for every dec_ap_bio!

	 * maybe rather introduce some type of hysteresis?

	 * e.g. (ap_bio == mxb/2 || ap_bio == 0) ? */

	if (ap_bio < mxb)

		wake_up(&mdev->misc_wait);

	if (ap_bio == 0 && test_bit(BITMAP_IO, &mdev->flags)) {

		if (!test_and_set_bit(BITMAP_IO_QUEUED, &mdev->flags))

			drbd_queue_work(&mdev->data.work, &mdev->bm_io_work.w);

	}

}



static inline int drbd_set_ed_uuid(struct drbd_conf *mdev, u64 val)

drivers/block/drbd/drbd_nl.c

+6 −2
Original line number Diff line number Diff line
@@ -1977,9 +1977,11 @@ static int drbd_nl_invalidate(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl 
	int retcode;



	/* If there is still bitmap IO pending, probably because of a previous

	 * resync just being finished, wait for it before requesting a new resync. */

	 * resync just being finished, wait for it before requesting a new resync.

	 * Also wait for it's after_state_ch(). */

	drbd_suspend_io(mdev);

	wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));

	drbd_flush_workqueue(mdev);



	retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_T), CS_ORDERED);
@@ -2018,9 +2020,11 @@ static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_re 
	int retcode;



	/* If there is still bitmap IO pending, probably because of a previous

	 * resync just being finished, wait for it before requesting a new resync. */

	 * resync just being finished, wait for it before requesting a new resync.

	 * Also wait for it's after_state_ch(). */

	drbd_suspend_io(mdev);

	wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));

	drbd_flush_workqueue(mdev);



	retcode = _drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S), CS_ORDERED);