It's possible that an already dequeued put_async_work() will release the reference (*) that was put asynchronously after the dequeue happened. This leaves an async-put work pending, without any reference to release. A subsequent async-put may trigger the drm_WARN_ON(!queue_delayed_work(&power_domains->async_put_work)); warn due to async_put_work() still pending. To avoid the warn, cancel the pending async_put_work() when releasing the reference at (*) above. Fixes: https://gitlab.freedesktop.org/drm/intel/-/issues/3421 Fixes: https://gitlab.freedesktop.org/drm/intel/-/issues/2289 Signed-off-by: Imre Deak <imre.deak@xxxxxxxxx> --- drivers/gpu/drm/i915/display/intel_display_power.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/gpu/drm/i915/display/intel_display_power.c b/drivers/gpu/drm/i915/display/intel_display_power.c index 2f7d1664c4738..a95bbf23e6b7c 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power.c +++ b/drivers/gpu/drm/i915/display/intel_display_power.c @@ -2263,6 +2263,12 @@ intel_display_power_put_async_work(struct work_struct *work) fetch_and_zero(&power_domains->async_put_domains[1]); queue_async_put_domains_work(power_domains, fetch_and_zero(&new_work_wakeref)); + } else { + /* + * Cancel the work that got queued after this one got dequeued, + * since here we released the corresponding async-put reference. + */ + cancel_delayed_work(&power_domains->async_put_work); } out_verify: -- 2.27.0 _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
