drbd: Fix a potential race that could case data inconsistency

When we have a write request and a state change C_WF_BITMAP_S -> C_SYNC_SOURCE
at the same time, and it happens that the line

    remote = remote && drbd_should_do_remote(s);

stills sees C_WF_BITMAP_S, and

     send_oos = rw == WRITE && drbd_should_send_oos(s);

already sees C_SYNC_SOURCE both are 0.

This causes the write to not be mirrored, but marked as out-of-sync on the
Sync_Source node.

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_req.c b/drivers/block/drbd/drbd_req.c
index 3e4dc070d3ff5070c7d8962d8810bf8795752f02..c3f99bde0e1125d920bf3a6cee3e04092ad5265e 100644 (file)
--- a/drivers/block/drbd/drbd_req.c
+++ b/drivers/block/drbd/drbd_req.c
@@ -822,6 +822,7 @@ int __drbd_make_request(struct drbd_conf *mdev, struct bio *bio, unsigned long s
        int local, remote, send_oos = 0;
        int err;
        int ret = 0;
+       union drbd_dev_state s;
 
        /* allocate outside of all locks; */
        req = drbd_req_new(mdev, bio);
@@ -884,8 +885,9 @@ int __drbd_make_request(struct drbd_conf *mdev, struct bio *bio, unsigned long s
                drbd_al_begin_io(mdev, &req->i);
        }
 
-       remote = remote && drbd_should_do_remote(mdev->state);
-       send_oos = rw == WRITE && drbd_should_send_out_of_sync(mdev->state);
+       s = mdev->state;
+       remote = remote && drbd_should_do_remote(s);
+       send_oos = rw == WRITE && drbd_should_send_out_of_sync(s);
        D_ASSERT(!(remote && send_oos));
 
        if (!(local || remote) && !drbd_suspended(mdev)) {