* [PATCH] jfs: fix null log dereference from sb during readonly mode
@ 2026-03-31 2:20 rafad900
0 siblings, 0 replies; only message in thread
From: rafad900 @ 2026-03-31 2:20 UTC (permalink / raw)
To: shaggy, jfs-discussion, linux-kernel, syzbot+1d096d31de6a0491b55e,
skhan
#syz test
Clearing the inode pointer is necessary to ensure no memory leaks
after txBegin fails to initialize the superblock during read only
mode.
- This error was found by syzkaller
- After ialloc() is called within jfs_create(), the inode pointer
is created but never freed when txBegin() fails.
- The patch addresses the need to stop commiting a file
transaction after txBegin() fails and the clean up of pointers.
- The patch was tested on QEMU with the repro provided by syzbot
Signed-off-by: rafad900 <19312533+rafad900@users.noreply.github.com>
---
fs/jfs/namei.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/fs/jfs/namei.c b/fs/jfs/namei.c
index 60c4a0e0fca5..3a5f45cdeae0 100644
--- a/fs/jfs/namei.c
+++ b/fs/jfs/namei.c
@@ -97,7 +97,13 @@ static int jfs_create(struct mnt_idmap *idmap, struct inode *dip,
}
tid = txBegin(dip->i_sb, 0);
-
+ if (tid == 0) {
+ jfs_err("jfs_create: unable to create tblk due to read only filesystem");
+ free_ea_wmap(ip);
+ clear_nlink(ip);
+ discard_new_inode(ip);
+ return -EROFS;
+ }
mutex_lock_nested(&JFS_IP(dip)->commit_mutex, COMMIT_MUTEX_PARENT);
mutex_lock_nested(&JFS_IP(ip)->commit_mutex, COMMIT_MUTEX_CHILD);
--
2.43.0
^ permalink raw reply related [flat|nested] only message in thread
only message in thread, other threads:[~2026-03-31 2:20 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-03-31 2:20 [PATCH] jfs: fix null log dereference from sb during readonly mode rafad900
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox