io_rsrc_ref_quiesce will unlock the uring while it waits for references to
the io_rsrc_data to be killed.
There are other places to the data that might add references to data via
calls to io_rsrc_node_switch.
There is a race condition where this reference can be added after the
completion has been signalled. At this point the io_rsrc_ref_quiesce call
will wake up and relock the uring, assuming the data is unused and can be
freed - although it is actually being used.
To fix this check in io_rsrc_ref_quiesce if a resource has been revived.
Reported-by: syzbot+ca8bf833622a1662745b@xxxxxxxxxxxxxxxxxxxxxxxxx
Fixes: b36a2050040b ("io_uring: fix bug in slow unregistering of nodes")
Signed-off-by: Dylan Yudaken <dylany@xxxxxx>
---
fs/io_uring.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 77b9c7e4793b..02086e8e0dec 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -7924,7 +7924,15 @@ static __cold int io_rsrc_ref_quiesce(struct io_rsrc_data *data,
ret = wait_for_completion_interruptible(&data->done);
if (!ret) {
mutex_lock(&ctx->uring_lock);
- break;
+ if (atomic_read(&data->refs) > 0) {
+ /*
+ * it has been revived by another thread while
+ * we were unlocked
+ */
+ mutex_unlock(&ctx->uring_lock);
+ } else {
+ break;
+ }
}
atomic_inc(&data->refs);
base-commit: cfb92440ee71adcc2105b0890bb01ac3cddb8507
--
2.30.2
