Ocfs2 devel - Jan 2013 - [PATCH 0/4 v2] Fix possible use after free with AIO

Hi,

  since nobody seem to have picked up any of the patches (at least
nobody told me so), I've added acks I received to the patches and I'm
sending them to Al for merging. Al, can you please merge these? Thanks.

								Honza

Running AIO is pinning inode in memory using file reference. Once AIO
is completed using aio_complete(), file reference is put and inode can
be freed from memory. So we have to be sure that calling aio_complete()
is the last thing we do with the inode.

CC: linux-ext4 at vger.kernel.org
CC: "Theodore Ts'o" <tytso at mit.edu>
CC: stable at vger.kernel.org
Reviewed-by: Carlos Maiolino <cmaiolino at redhat.com>
Acked-by: Jeff Moyer <jmoyer at redhat.com>
Signed-off-by: Jan Kara <jack at suse.cz>
---
 fs/ext4/inode.c   |    2 +-
 fs/ext4/page-io.c |    9 ++++-----
 2 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index cbfe13b..ba06638 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -2977,9 +2977,9 @@ static void ext4_end_io_dio(struct kiocb *iocb, loff_t
offset,
 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
 		ext4_free_io_end(io_end);
 out:
+		inode_dio_done(inode);
 		if (is_async)
 			aio_complete(iocb, ret, 0);
-		inode_dio_done(inode);
 		return;
 	}
 
diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c
index 0016fbc..b42d04f 100644
--- a/fs/ext4/page-io.c
+++ b/fs/ext4/page-io.c
@@ -103,14 +103,13 @@ static int ext4_end_io(ext4_io_end_t *io)
 			 "(inode %lu, offset %llu, size %zd, error %d)",
 			 inode->i_ino, offset, size, ret);
 	}
-	if (io->iocb)
-		aio_complete(io->iocb, io->result, 0);
-
-	if (io->flag & EXT4_IO_END_DIRECT)
-		inode_dio_done(inode);
 	/* Wake up anyone waiting on unwritten extent conversion */
 	if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
 		wake_up_all(ext4_ioend_wq(inode));
+	if (io->flag & EXT4_IO_END_DIRECT)
+		inode_dio_done(inode);
+	if (io->iocb)
+		aio_complete(io->iocb, io->result, 0);
 	return ret;
 }
 
-- 
1.7.1

Running AIO is pinning inode in memory using file reference. Once AIO
is completed using aio_complete(), file reference is put and inode can
be freed from memory. So we have to be sure that calling aio_complete()
is the last thing we do with the inode.

CC: xfs at oss.sgi.com
CC: Ben Myers <bpm at sgi.com>
CC: stable at vger.kernel.org
Reviewed-by: Ben Myers <bpm at sgi.com>
Acked-by: Jeff Moyer <jmoyer at redhat.com>
Signed-off-by: Jan Kara <jack at suse.cz>
---
 fs/xfs/xfs_aops.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index 4111a40..5f707e5 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -86,11 +86,11 @@ xfs_destroy_ioend(
 	}
 
 	if (ioend->io_iocb) {
+		inode_dio_done(ioend->io_inode);
 		if (ioend->io_isasync) {
 			aio_complete(ioend->io_iocb, ioend->io_error ?
 					ioend->io_error : ioend->io_result, 0);
 		}
-		inode_dio_done(ioend->io_inode);
 	}
 
 	mempool_free(ioend, xfs_ioend_pool);
-- 
1.7.1

Running AIO is pinning inode in memory using file reference. Once AIO
is completed using aio_complete(), file reference is put and inode can
be freed from memory. So we have to be sure that calling aio_complete()
is the last thing we do with the inode.

CC: Joel Becker <jlbec at evilplan.org>
CC: ocfs2-devel at oss.oracle.com
CC: stable at vger.kernel.org
Acked-by: Jeff Moyer <jmoyer at redhat.com>
Signed-off-by: Jan Kara <jack at suse.cz>
---
 fs/ocfs2/aops.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 6577432..340bd02 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -593,9 +593,9 @@ static void ocfs2_dio_end_io(struct kiocb *iocb,
 	level = ocfs2_iocb_rw_locked_level(iocb);
 	ocfs2_rw_unlock(inode, level);
 
+	inode_dio_done(inode);
 	if (is_async)
 		aio_complete(iocb, ret, 0);
-	inode_dio_done(inode);
 }
 
 /*
-- 
1.7.1

Running AIO is pinning inode in memory using file reference. Once AIO
is completed using aio_complete(), file reference is put and inode can
be freed from memory. So we have to be sure that calling aio_complete()
is the last thing we do with the inode.

CC: Christoph Hellwig <hch at infradead.org>
CC: Jens Axboe <axboe at kernel.dk>
CC: Jeff Moyer <jmoyer at redhat.com>
CC: stable at vger.kernel.org
Acked-by: Jeff Moyer <jmoyer at redhat.com>
Signed-off-by: Jan Kara <jack at suse.cz>
---
 fs/direct-io.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/direct-io.c b/fs/direct-io.c
index cf5b44b..f853263 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -261,9 +261,9 @@ static ssize_t dio_complete(struct dio *dio, loff_t offset,
ssize_t ret, bool is
 		dio->end_io(dio->iocb, offset, transferred,
 			    dio->private, ret, is_async);
 	} else {
+		inode_dio_done(dio->inode);
 		if (is_async)
 			aio_complete(dio->iocb, ret, 0);
-		inode_dio_done(dio->inode);
 	}
 
 	return ret;
-- 
1.7.1

On Wed, Jan 30, 2013 at 12:27:57AM +0100, Jan Kara
wrote:
> 
>   Hi,
> 
>   since nobody seem to have picked up any of the patches (at least
> nobody told me so), I've added acks I received to the patches and
I'm
> sending them to Al for merging. Al, can you please merge these? Thanks.
VFS part (4/4) picked; IMO at least ext* and xfs should go through the
filesystem trees.  If ocfs2 folks don't pick fs/ocfs2 part, I'll grab
it as well - not sure how active their tree is these days...