In the Linux kernel, the following vulnerability has been resolved:

io_uring: lock overflowing for IOPOLL

syzbot reports an issue with overflow filling for IOPOLL:

WARNING: CPU: 0 PID: 28 at io_uring/io_uring.c:734 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734
CPU: 0 PID: 28 Comm: kworker/u4:1 Not tainted 6.2.0-rc3-syzkaller-16369-g358a161a6a9e #0
Workqueue: events_unbound io_ring_exit_work
Call trace:
 io_cqring_event_overflow+0x1c0/0x230 io_uring/io_uring.c:734
 io_req_cqe_overflow+0x5c/0x70 io_uring/io_uring.c:773
 io_fill_cqe_req io_uring/io_uring.h:168 [inline]
 io_do_iopoll+0x474/0x62c io_uring/rw.c:1065
 io_iopoll_try_reap_events+0x6c/0x108 io_uring/io_uring.c:1513
 io_uring_try_cancel_requests+0x13c/0x258 io_uring/io_uring.c:3056
 io_ring_exit_work+0xec/0x390 io_uring/io_uring.c:2869
 process_one_work+0x2d8/0x504 kernel/workqueue.c:2289
 worker_thread+0x340/0x610 kernel/workqueue.c:2436
 kthread+0x12c/0x158 kernel/kthread.c:376
 ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:863

There is no real problem for normal IOPOLL as flush is also called with
uring_lock taken, but it's getting more complicated for IOPOLL|SQPOLL,
for which __io_cqring_overflow_flush() happens from the CQ waiting path.

References

• https://nvd.nist.gov/vuln/detail/CVE-2023-52903
• https://git.kernel.org/stable/c/544d163d659d45a206d8929370d5a2984e546cb7
• https://git.kernel.org/stable/c/7fc3990dad04a677606337ebc61964094d6cb41b
• https://git.kernel.org/stable/c/de77faee280163ff03b7ab64af6c9d779a43d4c4
• https://git.kernel.org/stable/c/ed4629d1e968359fbb91d0a3780b1e86a2c08845