Skip to contentSkip to conversation

nilfs2: fix NULL i_assoc_inode dereference in nilfs_mdt_save_to_shadow_map

PATCH In review Ryusuke Konishi 2 participants 2 messages Mar 31 NILFS2 filesystem

lore
Share & tools
lore

Ryusuke Konishi

author maintainer
#0 lore raw

PATCH 1/1 nilfs2: fix NULL i_assoc_inode dereference in nilfs_mdt_save_to_shadow_map

From: Deepanshu Kartikey <[email protected]>

The DAT inode's btree node cache (i_assoc_inode) is initialized lazily during btree operations. However, nilfs_mdt_save_to_shadow_map() assumes i_assoc_inode is already initialized when copying dirty pages to the shadow map during GC.

If NILFS_IOCTL_CLEAN_SEGMENTS is called immediately after mount before any btree operation has occurred on the DAT inode, i_assoc_inode is NULL leading to a general protection fault.

Fix this by calling nilfs_attach_btree_node_cache() on the DAT inode in nilfs_dat_read() at mount time, ensuring i_assoc_inode is always initialized before any GC operation can use it.

Reported-by [email protected]
Fixes e897be17a441 ("nilfs2: fix lockdep warnings in page operations for btree nodes")

---
Hi Viacheslav,

Please queue this patch first.

It fixes a potential NULL pointer dereference defect that could be triggered (likely most cases are intentionally) via the GC ioctl.

Thanks,
Ryusuke Konishi

 fs/nilfs2/dat.c | 3 +++
 1 file changed, 3 insertions(+)
fs/nilfs2/dat.c +3 -0
index 674380837ab9..888dc1831c86 100644
@@ -524,6 +524,9 @@ int nilfs_dat_read(struct super_block *sb, size_t entry_size,
 	if (err)
 		goto failed;
 
+	err = nilfs_attach_btree_node_cache(dat);
+	if (err)
+		goto failed;
 	err = nilfs_read_inode_common(dat, raw_inode);
 	if (err)
 		goto failed;
Signature · 1 line
2.43.0
#1 lore raw

Hi Ryusuke,

Quoted 41 lines · Ryusuke Konishi

From: Deepanshu Kartikey <[email protected]>

The DAT inode's btree node cache (i_assoc_inode) is initialized lazily during btree operations. However, nilfs_mdt_save_to_shadow_map() assumes i_assoc_inode is already initialized when copying dirty pages to the shadow map during GC.

If NILFS_IOCTL_CLEAN_SEGMENTS is called immediately after mount before any btree operation has occurred on the DAT inode, i_assoc_inode is NULL leading to a general protection fault.

Fix this by calling nilfs_attach_btree_node_cache() on the DAT inode in nilfs_dat_read() at mount time, ensuring i_assoc_inode is always initialized before any GC operation can use it.

Reported-by [email protected]
Fixes e897be17a441 ("nilfs2: fix lockdep warnings in page operations for btree nodes")

---
Hi Viacheslav,

Please queue this patch first.

It fixes a potential NULL pointer dereference defect that could be triggered (likely most cases are intentionally) via the GC ioctl.

Thanks,
Ryusuke Konishi

 fs/nilfs2/dat.c | 3 +++
 1 file changed, 3 insertions(+)
fs/nilfs2/dat.c +0 -0
index 674380837ab9..888dc1831c86 100644

fs/nilfs2/dat.c int nilfs_dat_read(struct super_block *sb, size_t entry_size, · line 532

Quoted 9 lines
fs/nilfs2/dat.c +3 -0
@@ -524,6 +524,9 @@ int nilfs_dat_read(struct super_block *sb, size_t entry_size,
 	if (err)
 		goto failed;
 
+	err = nilfs_attach_btree_node_cache(dat);
+	if (err)
+		goto failed;
 	err = nilfs_read_inode_common(dat, raw_inode);
 	if (err)
 		goto failed;

Applied on for-next branch of NILFS2 git tree.

Thanks,
Slava.