diff mbox series

ext4: set h_journal if there is a failure starting a reserved handle

Message ID 20180418154946.32006-1-tytso@mit.edu
State Accepted, archived
Headers show
Series ext4: set h_journal if there is a failure starting a reserved handle | expand

Commit Message

Theodore Ts'o April 18, 2018, 3:49 p.m. UTC
If ext4 tries to start a reserved handle via
jbd2_journal_start_reserved(), and the journal has been aborted, this
can result in a NULL pointer dereference.  This is because the fields
h_journal and h_transaction in the handle structure share the same
meory, via a union, so jbd2_journal_start_reserved() will clear
h_journal before calling start_this_handle().  If this function fails
due to an aborted handle, h_journal will still be NULL, and the call
to jbd2_journal_free_reserved() will pass a NULL journal to
sub_reserve_credits().

This can be reproduced by running "kvm-xfstests -c dioread_nolock
generic/475".

Fixes: 8f7d89f36829b ("jbd2: transaction reservation support")
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: Jan Kara <jack@suse.cz>
Cc: stable@kernel.org # 3.11
---
 fs/jbd2/transaction.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Andreas Dilger April 18, 2018, 4:07 p.m. UTC | #1
> On Apr 18, 2018, at 9:49 AM, Theodore Ts'o <tytso@mit.edu> wrote:
> 
> If ext4 tries to start a reserved handle via
> jbd2_journal_start_reserved(), and the journal has been aborted, this
> can result in a NULL pointer dereference.  This is because the fields
> h_journal and h_transaction in the handle structure share the same
> meory, via a union, so jbd2_journal_start_reserved() will clear

(typo) s/meory/memory/

Otherwise it looks fine.  I was wondering if it would be safer to
check if (journal == NULL) in jbd2_journal_free_reserved(), but that
adds overhead for each caller, while your patch only adds the extra
code in the error case.

Reviewed-by: Andreas Dilger <adilger@dilger.ca>

> h_journal before calling start_this_handle().  If this function fails
> due to an aborted handle, h_journal will still be NULL, and the call
> to jbd2_journal_free_reserved() will pass a NULL journal to
> sub_reserve_credits().
> 
> This can be reproduced by running "kvm-xfstests -c dioread_nolock
> generic/475".
> 
> Fixes: 8f7d89f36829b ("jbd2: transaction reservation support")
> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
> Cc: Jan Kara <jack@suse.cz>
> Cc: stable@kernel.org # 3.11
> ---
> fs/jbd2/transaction.c | 1 +
> 1 file changed, 1 insertion(+)
> 
> diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
> index ac311037d7a5..8aa453784402 100644
> --- a/fs/jbd2/transaction.c
> +++ b/fs/jbd2/transaction.c
> @@ -532,6 +532,7 @@ int jbd2_journal_start_reserved(handle_t *handle, unsigned int type,
> 	 */
> 	ret = start_this_handle(journal, handle, GFP_NOFS);
> 	if (ret < 0) {
> +		handle->h_journal = journal;
> 		jbd2_journal_free_reserved(handle);
> 		return ret;
> 	}
> --
> 2.16.1.72.g5be1f00a9a
> 


Cheers, Andreas
Jan Kara April 19, 2018, 9:32 a.m. UTC | #2
On Wed 18-04-18 11:49:46, Theodore Ts'o wrote:
> If ext4 tries to start a reserved handle via
> jbd2_journal_start_reserved(), and the journal has been aborted, this
> can result in a NULL pointer dereference.  This is because the fields
> h_journal and h_transaction in the handle structure share the same
> meory, via a union, so jbd2_journal_start_reserved() will clear
> h_journal before calling start_this_handle().  If this function fails
> due to an aborted handle, h_journal will still be NULL, and the call
> to jbd2_journal_free_reserved() will pass a NULL journal to
> sub_reserve_credits().
> 
> This can be reproduced by running "kvm-xfstests -c dioread_nolock
> generic/475".
> 
> Fixes: 8f7d89f36829b ("jbd2: transaction reservation support")
> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
> Cc: Jan Kara <jack@suse.cz>
> Cc: stable@kernel.org # 3.11

Thanks for fixing this! The patch looks good to me. You can add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
>  fs/jbd2/transaction.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
> index ac311037d7a5..8aa453784402 100644
> --- a/fs/jbd2/transaction.c
> +++ b/fs/jbd2/transaction.c
> @@ -532,6 +532,7 @@ int jbd2_journal_start_reserved(handle_t *handle, unsigned int type,
>  	 */
>  	ret = start_this_handle(journal, handle, GFP_NOFS);
>  	if (ret < 0) {
> +		handle->h_journal = journal;
>  		jbd2_journal_free_reserved(handle);
>  		return ret;
>  	}
> -- 
> 2.16.1.72.g5be1f00a9a
>
Theodore Ts'o April 19, 2018, 12:41 p.m. UTC | #3
On Wed, Apr 18, 2018 at 10:07:43AM -0600, Andreas Dilger wrote:
> 
> > On Apr 18, 2018, at 9:49 AM, Theodore Ts'o <tytso@mit.edu> wrote:
> > 
> > If ext4 tries to start a reserved handle via
> > jbd2_journal_start_reserved(), and the journal has been aborted, this
> > can result in a NULL pointer dereference.  This is because the fields
> > h_journal and h_transaction in the handle structure share the same
> > meory, via a union, so jbd2_journal_start_reserved() will clear
> 
> (typo) s/meory/memory/

Thanks, fixed.

						- Ted
diff mbox series

Patch

diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index ac311037d7a5..8aa453784402 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -532,6 +532,7 @@  int jbd2_journal_start_reserved(handle_t *handle, unsigned int type,
 	 */
 	ret = start_this_handle(journal, handle, GFP_NOFS);
 	if (ret < 0) {
+		handle->h_journal = journal;
 		jbd2_journal_free_reserved(handle);
 		return ret;
 	}