[tip: core/urgent] signal: Prevent sigqueue caching after task got released

[tip-bot2 for Thomas Gleixner]

The following commit has been merged into the core/urgent branch of tip:

Commit-ID:     399f8dd9a866e107639eabd3c1979cd526ca3a98
Gitweb:        https://git.kernel.org/tip/399f8dd9a866e107639eabd3c1979cd526ca3a98
Author:        Thomas Gleixner <tglx@xxxxxxxxxxxxx>
AuthorDate:    Tue, 22 Jun 2021 01:08:30 +02:00
Committer:     Thomas Gleixner <tglx@xxxxxxxxxxxxx>
CommitterDate: Tue, 22 Jun 2021 15:55:41 +02:00

signal: Prevent sigqueue caching after task got released

syzbot reported a memory leak related to sigqueue caching.

The assumption that a task cannot cache a sigqueue after the signal handler
has been dropped and exit_task_sigqueue_cache() has been invoked turns out
to be wrong.

Such a task can still invoke release_task(other_task), which cleans up the
signals of 'other_task' and ends up in sigqueue_cache_or_free(), which in
turn will cache the signal because task->sigqueue_cache is NULL. That's
obviously bogus because nothing will free the cached signal of that task
anymore, so the cached item is leaked.

This happens when e.g. the last non-leader thread exits and reaps the
zombie leader.

Prevent this by setting tsk::sigqueue_cache to an error pointer value in
exit_task_sigqueue_cache() which forces any subsequent invocation of
sigqueue_cache_or_free() from that task to hand the sigqueue back to the
kmemcache.

Add comments to all relevant places.

Fixes: 4bad58ebc8bc ("signal: Allow tasks to cache one sigqueue struct")
Reported-by: syzbot+0bac5fec63d4f399ba98@xxxxxxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Reviewed-by: Oleg Nesterov <oleg@xxxxxxxxxx>
Acked-by: Christian Brauner <christian.brauner@xxxxxxxxxx>
Link: https://lore.kernel.org/r/878s32g6j5.ffs@xxxxxxxxxxxxxxxxxxxxxxx

---
 kernel/signal.c | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/kernel/signal.c b/kernel/signal.c
index f7c6ffc..f1ecd8f 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -435,6 +435,12 @@ __sigqueue_alloc(int sig, struct task_struct *t, gfp_t gfp_flags,
 		 * Preallocation does not hold sighand::siglock so it can't
 		 * use the cache. The lockless caching requires that only
 		 * one consumer and only one producer run at a time.
+		 *
+		 * For the regular allocation case it is sufficient to
+		 * check @q for NULL because this code can only be called
+		 * if the target task @t has not been reaped yet; which
+		 * means this code can never observe the error pointer which is
+		 * written to @t->sigqueue_cache in exit_task_sigqueue_cache().
 		 */
 		q = READ_ONCE(t->sigqueue_cache);
 		if (!q || sigqueue_flags)
@@ -463,13 +469,18 @@ void exit_task_sigqueue_cache(struct task_struct *tsk)
 	struct sigqueue *q = tsk->sigqueue_cache;
 
 	if (q) {
-		tsk->sigqueue_cache = NULL;
 		/*
 		 * Hand it back to the cache as the task might
 		 * be self reaping which would leak the object.
 		 */
 		 kmem_cache_free(sigqueue_cachep, q);
 	}
+
+	/*
+	 * Set an error pointer to ensure that @tsk will not cache a
+	 * sigqueue when it is reaping it's child tasks
+	 */
+	tsk->sigqueue_cache = ERR_PTR(-1);
 }
 
 static void sigqueue_cache_or_free(struct sigqueue *q)
@@ -481,6 +492,10 @@ static void sigqueue_cache_or_free(struct sigqueue *q)
 	 * is intentional when run without holding current->sighand->siglock,
 	 * which is fine as current obviously cannot run __sigqueue_free()
 	 * concurrently.
+	 *
+	 * The NULL check is safe even if current has been reaped already,
+	 * in which case exit_task_sigqueue_cache() wrote an error pointer
+	 * into current->sigqueue_cache.
 	 */
 	if (!READ_ONCE(current->sigqueue_cache))
 		WRITE_ONCE(current->sigqueue_cache, q);