]> git.ipfire.org Git - thirdparty/kernel/linux.git/commitdiff
ipc: fix to protect IPCS lookups using RCU
authorJeongjun Park <aha310510@gmail.com>
Thu, 24 Apr 2025 14:33:22 +0000 (23:33 +0900)
committerAndrew Morton <akpm@linux-foundation.org>
Mon, 12 May 2025 00:54:11 +0000 (17:54 -0700)
syzbot reported that it discovered a use-after-free vulnerability, [0]

[0]: https://lore.kernel.org/all/67af13f8.050a0220.21dd3.0038.GAE@google.com/

idr_for_each() is protected by rwsem, but this is not enough.  If it is
not protected by RCU read-critical region, when idr_for_each() calls
radix_tree_node_free() through call_rcu() to free the radix_tree_node
structure, the node will be freed immediately, and when reading the next
node in radix_tree_for_each_slot(), the already freed memory may be read.

Therefore, we need to add code to make sure that idr_for_each() is
protected within the RCU read-critical region when we call it in
shm_destroy_orphaned().

Link: https://lkml.kernel.org/r/20250424143322.18830-1-aha310510@gmail.com
Fixes: b34a6b1da371 ("ipc: introduce shm_rmid_forced sysctl")
Signed-off-by: Jeongjun Park <aha310510@gmail.com>
Reported-by: syzbot+a2b84e569d06ca3a949c@syzkaller.appspotmail.com
Cc: Jeongjun Park <aha310510@gmail.com>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vasiliy Kulikov <segoon@openwall.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
ipc/shm.c

index 99564c87008408b649f1f7f6733d30d2913765d2..492fcc6999857a30b8c2e1c98f550ba9b9e81e5e 100644 (file)
--- a/ipc/shm.c
+++ b/ipc/shm.c
@@ -431,8 +431,11 @@ static int shm_try_destroy_orphaned(int id, void *p, void *data)
 void shm_destroy_orphaned(struct ipc_namespace *ns)
 {
        down_write(&shm_ids(ns).rwsem);
-       if (shm_ids(ns).in_use)
+       if (shm_ids(ns).in_use) {
+               rcu_read_lock();
                idr_for_each(&shm_ids(ns).ipcs_idr, &shm_try_destroy_orphaned, ns);
+               rcu_read_unlock();
+       }
        up_write(&shm_ids(ns).rwsem);
 }