[PATCH 2/2] drm/amdgpu/gfx9: put queue resets behind a debug option

[Alex Deucher]

On Wed, Aug 21, 2024 at 5:59 PM Felix Kuehling <felix.kuehling at amd.com> wrote:
>
>
> On 2024-08-20 16:25, Alex Deucher wrote:
> > Pending extended validation.
> >
> > Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
> > ---
> >   drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c | 4 ++++
> >   drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c             | 4 ++++
> >   drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c           | 6 ++++++
> >   3 files changed, 14 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
> > index c63528a4e8941..1254a43ec96b6 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
> > @@ -1151,6 +1151,10 @@ uint64_t kgd_gfx_v9_hqd_get_pq_addr(struct amdgpu_device *adev,
> >       uint32_t low, high;
> >       uint64_t queue_addr = 0;
> >
> > +     if (!adev->debug_exp_resets &&
> > +         !adev->gfx.num_gfx_rings)
> > +             return 0;
> > +
>
> Did you put this in the HW-specific code path intentionally? If you want
> this check to apply to all ASICs, you should put it into
> detect_queue_hang in kfd_device_queue_manager.c. But maybe the extended
> validation is HW-specific.

I only want to apply it to MI parts at this point.  We will likely
have a different default on other parts.

Alex

>
> Either way, the patch is
>
> Acked-by: Felix Kuehling <felix.kuehling at amd.com>
>
>
> >       kgd_gfx_v9_acquire_queue(adev, pipe_id, queue_id, inst);
> >       amdgpu_gfx_rlc_enter_safe_mode(adev, inst);
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> > index 21089aadbb7b4..8cf5d7925b51c 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> > @@ -7233,6 +7233,10 @@ static int gfx_v9_0_reset_kcq(struct amdgpu_ring *ring,
> >       unsigned long flags;
> >       int i, r;
> >
> > +     if (!adev->debug_exp_resets &&
> > +         !adev->gfx.num_gfx_rings)
> > +             return -EINVAL;
> > +
> >       if (amdgpu_sriov_vf(adev))
> >               return -EINVAL;
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
> > index 2067f26d3a9d8..f8649546b9c4c 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
> > @@ -3052,6 +3052,9 @@ static void gfx_v9_4_3_ring_soft_recovery(struct amdgpu_ring *ring,
> >       struct amdgpu_device *adev = ring->adev;
> >       uint32_t value = 0;
> >
> > +     if (!adev->debug_exp_resets)
> > +             return;
> > +
> >       value = REG_SET_FIELD(value, SQ_CMD, CMD, 0x03);
> >       value = REG_SET_FIELD(value, SQ_CMD, MODE, 0x01);
> >       value = REG_SET_FIELD(value, SQ_CMD, CHECK_VMID, 1);
> > @@ -3475,6 +3478,9 @@ static int gfx_v9_4_3_reset_kcq(struct amdgpu_ring *ring,
> >       unsigned long flags;
> >       int r, i;
> >
> > +     if (!adev->debug_exp_resets)
> > +             return -EINVAL;
> > +
> >       if (amdgpu_sriov_vf(adev))
> >               return -EINVAL;
> >