[PATCH] drm/amdgpu: skip force complete fences when ring is uninitialized

Zhou, Bob Bob.Zhou at amd.com
Wed May 31 08:56:50 UTC 2023


[AMD Official Use Only - General]

Hi Christian,

Thanks for your suggestion.  I will apply it and send out the v2 patch for review.

Regards,
Bob

-----Original Message-----
From: Koenig, Christian <Christian.Koenig at amd.com>
Sent: Wednesday, May 31, 2023 4:33 PM
To: Zhou, Bob <Bob.Zhou at amd.com>; amd-gfx at lists.freedesktop.org; Min, Frank <Frank.Min at amd.com>
Cc: Chen, Guchun <Guchun.Chen at amd.com>
Subject: Re: [PATCH] drm/amdgpu: skip force complete fences when ring is uninitialized



Am 31.05.23 um 06:48 schrieb Bob Zhou:
> uvd ring in uvd_v7_0_sw_init only initializes ring in bare metal case,
> so when executing amdgpu_uvd_resume to restore fence seq in SRIOV, a
> null pointer dereference will occur. This patch correct this.

Ah! I was already wondering how this bug reports came to be.

In this case please fix the UVD code to not call
amdgpu_fence_driver_force_completion() in the first place instead of hacking around that in the fence code.

Regards,
Christian.

>
> Fixes: 043f2271e2d0a ("drm/amdgpu: mark force completed fences with
> -ECANCELED")
>
> BUG: kernel NULL pointer dereference, address: 0000000000000000
> Oops: 0000 [#1] PREEMPT SMP PTI
> RIP: 0010:amdgpu_fence_driver_set_error+0x3f/0xc0 [amdgpu] Call Trace:
>   <TASK>
>   amdgpu_fence_driver_force_completion+0x18/0x50 [amdgpu]
>   amdgpu_uvd_resume+0x1b2/0x380 [amdgpu]
>   ? amdgpu_ring_init+0x73c/0x820 [amdgpu]
>   uvd_v7_0_sw_init+0x358/0x450 [amdgpu]
>   amdgpu_device_init.cold+0x198c/0x1fdf [amdgpu]
>   ? pci_bus_read_config_byte+0x40/0x80
>   ? pci_read_config_byte+0x27/0x50
>   amdgpu_driver_load_kms+0x1a/0x160 [amdgpu]
>   amdgpu_pci_probe+0x186/0x3c0 [amdgpu]
>
> Signed-off-by: Bob Zhou <bob.zhou at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 2 ++
>   1 file changed, 2 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> index c694b41f6461..d2c5484309a6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> @@ -723,6 +723,8 @@ void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error)
>    */
>   void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring)
>   {
> +     if (ring->fence_drv.irq_src)
> +             return;
>       amdgpu_fence_driver_set_error(ring, -ECANCELED);
>       amdgpu_fence_write(ring, ring->fence_drv.sync_seq);
>       amdgpu_fence_process(ring);



More information about the amd-gfx mailing list