[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20250424143322.18830-1-aha310510@gmail.com>
Date: Thu, 24 Apr 2025 23:33:22 +0900
From: Jeongjun Park <aha310510@...il.com>
To: akpm@...ux-foundation.org,
Liam.Howlett@...cle.com
Cc: lorenzo.stoakes@...cle.com,
willy@...radead.org,
linux-kernel@...r.kernel.org,
syzbot+a2b84e569d06ca3a949c@...kaller.appspotmail.com,
Jeongjun Park <aha310510@...il.com>
Subject: [PATCH v3] ipc: fix to protect IPCS lookups using RCU
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().
Reported-by: syzbot+a2b84e569d06ca3a949c@...kaller.appspotmail.com
Fixes: b34a6b1da371 ("ipc: introduce shm_rmid_forced sysctl")
Signed-off-by: Jeongjun Park <aha310510@...il.com>
---
v3: Add necessary descriptions and links as advised by Lorenzo Stoakes
- Link to v2: https://lore.kernel.org/all/20250422124843.17188-1-aha310510@gmail.com/
v2: Change to use RCU read critical section only when it is certain that idr_for_each() is called
- Link to v1: https://lore.kernel.org/all/20250219132905.8214-1-aha310510@gmail.com/
---
ipc/shm.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/ipc/shm.c b/ipc/shm.c
index 99564c870084..492fcc699985 100644
--- 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);
}
--
Powered by blists - more mailing lists