[PATCH] ocfs2: reset folio to NULL when get folio fails

Lizhi Xu posted 1 patch 3 months, 3 weeks ago
fs/ocfs2/aops.c | 1 +
1 file changed, 1 insertion(+)
[PATCH] ocfs2: reset folio to NULL when get folio fails
Posted by Lizhi Xu 3 months, 3 weeks ago
The reproducer uses FAULT_INJECTION to make memory allocation fail, which
causes __filemap_get_folio() to fail, when initializing w_folios[i] in
ocfs2_grab_folios_for_write(), it only returns an error code and the value
of w_folios[i] is the error code, which causes ocfs2_unlock_and_free_folios()
to recycle the invalid w_folios[i] when releasing folios.

Reported-by: syzbot+c2ea94ae47cd7e3881ec@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=c2ea94ae47cd7e3881ec
Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com>
---
 fs/ocfs2/aops.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 40b6bce12951..89aadc6cdd87 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -1071,6 +1071,7 @@ static int ocfs2_grab_folios_for_write(struct address_space *mapping,
 			if (IS_ERR(wc->w_folios[i])) {
 				ret = PTR_ERR(wc->w_folios[i]);
 				mlog_errno(ret);
+				wc->w_folios[i] = NULL;
 				goto out;
 			}
 		}
-- 
2.43.0
Re: [PATCH] ocfs2: reset folio to NULL when get folio fails
Posted by Joseph Qi 3 months, 3 weeks ago

On 2025/6/16 09:31, Lizhi Xu wrote:
> The reproducer uses FAULT_INJECTION to make memory allocation fail, which
> causes __filemap_get_folio() to fail, when initializing w_folios[i] in
> ocfs2_grab_folios_for_write(), it only returns an error code and the value
> of w_folios[i] is the error code, which causes ocfs2_unlock_and_free_folios()
> to recycle the invalid w_folios[i] when releasing folios.
> 
> Reported-by: syzbot+c2ea94ae47cd7e3881ec@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=c2ea94ae47cd7e3881ec
> Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com>

Looks good.
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>

> ---
>  fs/ocfs2/aops.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
> index 40b6bce12951..89aadc6cdd87 100644
> --- a/fs/ocfs2/aops.c
> +++ b/fs/ocfs2/aops.c
> @@ -1071,6 +1071,7 @@ static int ocfs2_grab_folios_for_write(struct address_space *mapping,
>  			if (IS_ERR(wc->w_folios[i])) {
>  				ret = PTR_ERR(wc->w_folios[i]);
>  				mlog_errno(ret);
> +				wc->w_folios[i] = NULL;
>  				goto out;
>  			}
>  		}