md/raid10: remove use-after-free bug.

We always need to be careful when calling generic_make_request, as it
can start a chain of events which might free something that we are
using.

Here is one place I wasn't careful enough.  If the wbio2 is not in
use, then it might get freed at the first generic_make_request call.
So perform all necessary tests first.

This bug was introduced in 3.3-rc3 (24afd80d99) and can cause an
oops, so fix is suitable for any -stable since then.

Cc: stable@vger.kernel.org (3.3+)
Signed-off-by: NeilBrown <neilb@suse.de>

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 957a719..df7b0a0 100644 (file)
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -2290,12 +2290,18 @@ static void recovery_request_write(struct mddev *mddev, struct r10bio *r10_bio)
        d = r10_bio->devs[1].devnum;
        wbio = r10_bio->devs[1].bio;
        wbio2 = r10_bio->devs[1].repl_bio;
+       /* Need to test wbio2->bi_end_io before we call
+        * generic_make_request as if the former is NULL,
+        * the latter is free to free wbio2.
+        */
+       if (wbio2 && !wbio2->bi_end_io)
+               wbio2 = NULL;
        if (wbio->bi_end_io) {
                atomic_inc(&conf->mirrors[d].rdev->nr_pending);
                md_sync_acct(conf->mirrors[d].rdev->bdev, bio_sectors(wbio));
                generic_make_request(wbio);
        }
-       if (wbio2 && wbio2->bi_end_io) {
+       if (wbio2) {
                atomic_inc(&conf->mirrors[d].replacement->nr_pending);
                md_sync_acct(conf->mirrors[d].replacement->bdev,
                             bio_sectors(wbio2));