drbd: fix race between drbdadm invalidate/verify and finishing resync

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>

diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index 308beb8c0c1e4130dff271259f9b036031be341b..867bf1d829888bb498e06977c5d1208bbcc59040 100644 (file)
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -1963,6 +1963,7 @@ static int drbd_nl_invalidate(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl
 
        /* 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);
@@ -1981,6 +1982,7 @@ static int drbd_nl_invalidate(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl
 
                retcode = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_T));
        }
+       drbd_resume_io(mdev);
 
        reply->ret_code = retcode;
        return 0;
@@ -2002,6 +2004,7 @@ static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_re
 
        /* 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);
@@ -2020,6 +2023,7 @@ static int drbd_nl_invalidate_peer(struct drbd_conf *mdev, struct drbd_nl_cfg_re
                } else
                        retcode = drbd_request_state(mdev, NS(conn, C_STARTING_SYNC_S));
        }
+       drbd_resume_io(mdev);
 
        reply->ret_code = retcode;
        return 0;
@@ -2192,11 +2196,13 @@ static int drbd_nl_start_ov(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp,
 
        /* 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));
 
        /* w_make_ov_request expects position to be aligned */
        mdev->ov_start_sector = args.start_sector & ~BM_SECT_PER_BIT;
        reply->ret_code = drbd_request_state(mdev,NS(conn,C_VERIFY_S));
+       drbd_resume_io(mdev);
        return 0;
 }