Ryan Ding
2015-Aug-24 07:23 UTC
[Ocfs2-devel] [PATCH] ocfs2: direct write will call ocfs2_rw_unlock() twice when doing aio+dio
Orabug: 21612107
Use wrong return value in ocfs2_file_write_iter(). This will cause
ocfs2_rw_unlock() be called both in write_iter & end_io, and trigger a
BUG_ON.
This issue exist since commit 7da839c475894ea872ec909a5d2e83dddccff5be.
Signed-off-by: Ryan Ding <ryan.ding at oracle.com>
---
fs/ocfs2/file.c | 28 ++++++++++++++--------------
1 files changed, 14 insertions(+), 14 deletions(-)
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 719f7f4..33efa33 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -2372,6 +2372,20 @@ relock:
/* buffered aio wouldn't have proper lock coverage today */
BUG_ON(written == -EIOCBQUEUED && !(iocb->ki_flags &
IOCB_DIRECT));
+ /*
+ * deep in g_f_a_w_n()->ocfs2_direct_IO we pass in a ocfs2_dio_end_io
+ * function pointer which is called when o_direct io completes so that
+ * it can unlock our rw lock.
+ * Unfortunately there are error cases which call end_io and others
+ * that don't. so we don't have to unlock the rw_lock if either an
+ * async dio is going to do it in the future or an end_io after an
+ * error has already done it.
+ */
+ if ((written == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) {
+ rw_level = -1;
+ unaligned_dio = 0;
+ }
+
if (unlikely(written <= 0))
goto no_sync;
@@ -2396,20 +2410,6 @@ relock:
}
no_sync:
- /*
- * deep in g_f_a_w_n()->ocfs2_direct_IO we pass in a ocfs2_dio_end_io
- * function pointer which is called when o_direct io completes so that
- * it can unlock our rw lock.
- * Unfortunately there are error cases which call end_io and others
- * that don't. so we don't have to unlock the rw_lock if either an
- * async dio is going to do it in the future or an end_io after an
- * error has already done it.
- */
- if ((ret == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) {
- rw_level = -1;
- unaligned_dio = 0;
- }
-
if (unaligned_dio) {
ocfs2_iocb_clear_unaligned_aio(iocb);
mutex_unlock(&OCFS2_I(inode)->ip_unaligned_aio);
--
1.7.1
Junxiao Bi
2015-Aug-24 07:39 UTC
[Ocfs2-devel] [PATCH] ocfs2: direct write will call ocfs2_rw_unlock() twice when doing aio+dio
On 08/24/2015 03:23 PM, Ryan Ding wrote:> Orabug: 21612107 > > Use wrong return value in ocfs2_file_write_iter(). This will cause > ocfs2_rw_unlock() be called both in write_iter & end_io, and trigger a BUG_ON. > > This issue exist since commit 7da839c475894ea872ec909a5d2e83dddccff5be.Better say: This issue is introduced by commit 7da839c47589 ("ocfs2: use __generic_file_write_iter()") , or checkpatch will report a style error. Other looks good. Reviewed-by: Junxiao Bi <junxiao.bi at oracle.com> Thanks, Junxiao.> > Signed-off-by: Ryan Ding <ryan.ding at oracle.com> > --- > fs/ocfs2/file.c | 28 ++++++++++++++-------------- > 1 files changed, 14 insertions(+), 14 deletions(-) > > diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c > index 719f7f4..33efa33 100644 > --- a/fs/ocfs2/file.c > +++ b/fs/ocfs2/file.c > @@ -2372,6 +2372,20 @@ relock: > /* buffered aio wouldn't have proper lock coverage today */ > BUG_ON(written == -EIOCBQUEUED && !(iocb->ki_flags & IOCB_DIRECT)); > > + /* > + * deep in g_f_a_w_n()->ocfs2_direct_IO we pass in a ocfs2_dio_end_io > + * function pointer which is called when o_direct io completes so that > + * it can unlock our rw lock. > + * Unfortunately there are error cases which call end_io and others > + * that don't. so we don't have to unlock the rw_lock if either an > + * async dio is going to do it in the future or an end_io after an > + * error has already done it. > + */ > + if ((written == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) { > + rw_level = -1; > + unaligned_dio = 0; > + } > + > if (unlikely(written <= 0)) > goto no_sync; > > @@ -2396,20 +2410,6 @@ relock: > } > > no_sync: > - /* > - * deep in g_f_a_w_n()->ocfs2_direct_IO we pass in a ocfs2_dio_end_io > - * function pointer which is called when o_direct io completes so that > - * it can unlock our rw lock. > - * Unfortunately there are error cases which call end_io and others > - * that don't. so we don't have to unlock the rw_lock if either an > - * async dio is going to do it in the future or an end_io after an > - * error has already done it. > - */ > - if ((ret == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) { > - rw_level = -1; > - unaligned_dio = 0; > - } > - > if (unaligned_dio) { > ocfs2_iocb_clear_unaligned_aio(iocb); > mutex_unlock(&OCFS2_I(inode)->ip_unaligned_aio); >