[PATCH 4/4] drm/amdgpu: Init the cp MQD if it's not be initialized before

[Alex Deucher]

On Thu, Feb 18, 2021 at 8:20 PM shaoyunl <shaoyun.liu at amd.com> wrote:
>
> The MQD might not be initialized duirng first init period if the device need to be reset
> druing probe. Driver need to proper init them in gpu recovery period
>
> Signed-off-by: shaoyunl <shaoyun.liu at amd.com>
> Change-Id: Iad58a050939af2afa46d1c74a90866c47ba9efd2
> ---
>  drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 20 +++++++++++++++++---
>  1 file changed, 17 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> index 65db88bb6cbc..8fc2fd518a1b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> @@ -3696,11 +3696,18 @@ static int gfx_v9_0_kiq_init_queue(struct amdgpu_ring *ring)
>         struct amdgpu_device *adev = ring->adev;
>         struct v9_mqd *mqd = ring->mqd_ptr;
>         int mqd_idx = AMDGPU_MAX_COMPUTE_RINGS;
> +       struct v9_mqd *tmp_mqd;
>
>         gfx_v9_0_kiq_setting(ring);
>
> -       if (amdgpu_in_reset(adev)) { /* for GPU_RESET case */
> -               /* reset MQD to a clean status */
> +       /* GPU could be in bad state during probe, driver trigger the reset
> +        * after load the SMU, in this case , the mqd is not be initialized.
> +        * driver need to re-init the mqd in this case.
> +        * check mqd->cp_hqd_pq_control since this value should not be 0
> +        */
> +       tmp_mqd = (struct v9_mqd *)adev->gfx.mec.mqd_backup[mqd_idx];
> +       if (amdgpu_in_reset(adev) && tmp_mqd->cp_hqd_pq_control){
> +               /* for GPU_RESET case , reset MQD to a clean status */
>                 if (adev->gfx.mec.mqd_backup[mqd_idx])
>                         memcpy(mqd, adev->gfx.mec.mqd_backup[mqd_idx], sizeof(struct v9_mqd_allocation));
>
> @@ -3736,8 +3743,15 @@ static int gfx_v9_0_kcq_init_queue(struct amdgpu_ring *ring)
>         struct amdgpu_device *adev = ring->adev;
>         struct v9_mqd *mqd = ring->mqd_ptr;
>         int mqd_idx = ring - &adev->gfx.compute_ring[0];
> +       struct v9_mqd *tmp_mqd;
>
> -       if (!amdgpu_in_reset(adev) && !adev->in_suspend) {
> +       /* Samw as above kiq init, driver need to re-init the mqd if mqd->cp_hqd_pq_control

Samw -> Same

I think this also needs to come before patch 1.  With these comments fixed:
Acked-by: Alex Deucher <alexander.deucher at amd.com>

> +        * is not be initialized before
> +        */
> +       tmp_mqd = (struct v9_mqd *)adev->gfx.mec.mqd_backup[mqd_idx];
> +
> +       if (!tmp_mqd->cp_hqd_pq_control ||
> +           (!amdgpu_in_reset(adev) && !adev->in_suspend)) {
>                 memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation));
>                 ((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF;
>                 ((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF;
> --
> 2.17.1
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx