On 4/7/25 08:03, Xuan Zhuo wrote:
On Fri, 4 Apr 2025 16:39:03 +0700, Bui Quang Minh <minhquangbui99@xxxxxxxxx> wrote:
When pausing rx (e.g. set up xdp, xsk pool, rx resize), we call
napi_disable() on the receive queue's napi. In delayed refill_work, it
also calls napi_disable() on the receive queue's napi. This can leads to
deadlock when napi_disable() is called on an already disabled napi. This
scenario can be reproducible by binding a XDP socket to virtio-net
interface without setting up the fill ring. As a result, try_fill_recv
will fail until the fill ring is set up and refill_work is scheduled.
So, what is the problem? The refill_work is waiting? As I know, that thread
will sleep some time, so the cpu can do other work.
When napi_disable is called on an already disabled napi, it will sleep
in napi_disable_locked while still holding the netdev_lock. As a result,
later napi_enable gets stuck too as it cannot acquire the netdev_lock.
This leads to refill_work and the pause-then-resume tx are stuck altogether.
Thanks,
Quang Minh.
