]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
fs: fix lost error code in dio_complete
authorMaximilian Heyne <mheyne@amazon.de>
Fri, 30 Nov 2018 15:35:14 +0000 (08:35 -0700)
committerJens Axboe <axboe@kernel.dk>
Fri, 30 Nov 2018 15:35:14 +0000 (08:35 -0700)
commit e259221763a40403d5bb232209998e8c45804ab8 ("fs: simplify the
generic_write_sync prototype") reworked callers of generic_write_sync(),
and ended up dropping the error return for the directio path. Prior to
that commit, in dio_complete(), an error would be bubbled up the stack,
but after that commit, errors passed on to dio_complete were eaten up.

This was reported on the list earlier, and a fix was proposed in
https://lore.kernel.org/lkml/20160921141539.GA17898@infradead.org/, but
never followed up with.  We recently hit this bug in our testing where
fencing io errors, which were previously erroring out with EIO, were
being returned as success operations after this commit.

The fix proposed on the list earlier was a little short -- it would have
still called generic_write_sync() in case `ret` already contained an
error. This fix ensures generic_write_sync() is only called when there's
no pending error in the write. Additionally, transferred is replaced
with ret to bring this code in line with other callers.

Fixes: e259221763a4 ("fs: simplify the generic_write_sync prototype")
Reported-by: Ravi Nankani <rnankani@amazon.com>
Signed-off-by: Maximilian Heyne <mheyne@amazon.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
CC: Torsten Mehlan <tomeh@amazon.de>
CC: Uwe Dannowski <uwed@amazon.de>
CC: Amit Shah <aams@amazon.de>
CC: David Woodhouse <dwmw@amazon.co.uk>
CC: stable@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
fs/direct-io.c

index 722d17c88edb9016ac910072f4f758605089ec83..41a0e97252aed1f28a566a76ff7cc342c9606fae 100644 (file)
@@ -325,8 +325,8 @@ static ssize_t dio_complete(struct dio *dio, ssize_t ret, unsigned int flags)
                 */
                dio->iocb->ki_pos += transferred;
 
-               if (dio->op == REQ_OP_WRITE)
-                       ret = generic_write_sync(dio->iocb,  transferred);
+               if (ret > 0 && dio->op == REQ_OP_WRITE)
+                       ret = generic_write_sync(dio->iocb, ret);
                dio->iocb->ki_complete(dio->iocb, ret, 0);
        }