[Intel-gfx] kernel 5.5.4: BUG: kernel NULL pointer dereference, address: 000000000000000

[Toralf Förster]

On 2/16/20 4:26 AM, Hillf Danton wrote:
> Looks like a stray lock counts for the above NULL dereference.
Hi, the patch applied on top of 5.5.4 breaks the internal display now even in the boot phase.
Gert just a black screen after few seconds, nothing in the logs except:

Feb 16 11:21:57 t44 kernel: elogind-daemon[1431]: Removed session c15.
Feb 16 11:21:57 t44 start-stop-daemon[6462]: Will stop PID 1431
Feb 16 11:21:57 t44 start-stop-daemon[6462]: Sending signal 15 to PID 1431
Feb 16 11:21:57 t44 kernel: elogind-daemon[1431]: Received signal 15 [TERM]
Feb 16 11:21:57 t44 kernel: elogind-daemon[1431]: segfault at 56264c000000 ip 00007fddfcf76882 sp 00007ffc98c721b0 error 4 in libc-2.29.so[7fddfcf0c000+15a000]
Feb 16 11:21:57 t44 kernel: Code: a8 02 75 4c 48 8b 15 05 e5 13 00 64 48 83 3a 00 0f 84 f2 00 00 00 48 8d 3d 2b f2 13 00 a8 04 74 0c 48 89 f0 48 25 00 00 00 fc <48> 8b 38 48 8b 44 24 18 64 48 33 04 25 28 00
00 00 0f 85 f8 00 00
Feb 16 11:21:57 t44 start-stop-daemon[6549]: Will stop /usr/sbin/dnsmasq
Feb 16 11:21:57 t44 start-stop-daemon[6549]: Will stop PID 2764
Feb 16 11:21:57 t44 start-stop-daemon[6549]: Sending signal 15 to PID 2764


>
> Btw, send pure text message please.

Ick, I do sned plain text to LKML, or?

>
> --- a/drivers/gpu/drm/i915/i915_request.c
> +++ b/drivers/gpu/drm/i915/i915_request.c
> @@ -254,8 +254,7 @@ bool i915_request_retire(struct i915_req
>
>  	spin_lock_irq(&rq->lock);
>  	i915_request_mark_complete(rq);
> -	if (!i915_request_signaled(rq))
> -		dma_fence_signal_locked(&rq->fence);
> +	dma_fence_signal(&rq->fence);
>  	if (test_bit(DMA_FENCE_FLAG_ENABLE_SIGNAL_BIT, &rq->fence.flags))
>  		i915_request_cancel_breadcrumb(rq);
>  	if (i915_request_has_waitboost(rq)) {
>


--
Toralf