Re: [PATCH 10/21] drm/amdgpu: wire up emit_wreg for UVD v6

["Christian König" <christian.koenig-5C7GfCeVMHo@public.gmane.org>]

[-- Attachment #1.1: Type: text/plain, Size: 3725 bytes --]

Am 17.01.2018 um 15:22 schrieb Deucher, Alexander:
>
>
> ------------------------------------------------------------------------
> *From:* amd-gfx <amd-gfx-bounces-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org> on behalf of 
> Christian König <ckoenig.leichtzumerken-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> *Sent:* Wednesday, January 17, 2018 7:42 AM
> *To:* Alex Deucher
> *Cc:* amd-gfx list
> *Subject:* Re: [PATCH 10/21] drm/amdgpu: wire up emit_wreg for UVD v6
> Am 16.01.2018 um 21:01 schrieb Alex Deucher:
> > On Tue, Jan 16, 2018 at 2:40 PM, Christian König
> > <ckoenig.leichtzumerken-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> >> Needed for vm_flush unification.
> >>
> >> Signed-off-by: Christian König <christian.koenig-5C7GfCeVMHo@public.gmane.org>
> >> ---
> >>   drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c | 12 ++++++++++++
> >>   1 file changed, 12 insertions(+)
> >>
> >> diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c 
> b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
> >> index c6e22eff89e7..b48b843c66ae 100644
> >> --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
> >> +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
> >> @@ -1059,6 +1059,17 @@ static void uvd_v6_0_enc_ring_emit_ib(struct 
> amdgpu_ring *ring,
> >>          amdgpu_ring_write(ring, ib->length_dw);
> >>   }
> >>
> >> +static void uvd_v6_0_ring_emit_wreg(struct amdgpu_ring *ring,
> >> +                                   uint32_t reg, uint32_t val)
> >> +{
> >> +       amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
> >> +       amdgpu_ring_write(ring, reg << 2);
> >> +       amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
> >> +       amdgpu_ring_write(ring, val);
> >> +       amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
> >> +       amdgpu_ring_write(ring, 0x8);
> > Is this valid on UVD6?
>
> This is valid on the decode ring of UVD6, only the HEVC UVD6 encode and
> VCE3 rings use different commands.
>
> >    I think this may be a UVD7 thing.  Looking at
> > the other UVD code, I think we just want:
> >          amdgpu_ring_write(ring, PACKET0(reg, 0));
> >          amdgpu_ring_write(ring, val);
>
> IIRC that won't work. The UVD RBC can only write to UVD registers using
> the PACKET0 command.
>
> We probably need to fix a bunch of functions in uvd4,5,6,7 in that 
> case which use packet0 directly to write non-UVD registers.

Yeah, I see. I briefly remember that I questioned if this works or not 
two years ago when we added the code.

Going to wire things up correctly with the new wreg callback.

Christian.

>
> Alex
>
>
> Regards,
> Christian.
>
> >
> > Alex
> >
> >
> >> +}
> >> +
> >>   static void uvd_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
> >> unsigned vmid, unsigned pasid,
> >> uint64_t pd_addr)
> >> @@ -1595,6 +1606,7 @@ static const struct amdgpu_ring_funcs 
> uvd_v6_0_ring_vm_funcs = {
> >>          .pad_ib = amdgpu_ring_generic_pad_ib,
> >>          .begin_use = amdgpu_uvd_ring_begin_use,
> >>          .end_use = amdgpu_uvd_ring_end_use,
> >> +       .emit_wreg = uvd_v6_0_ring_emit_wreg,
> >>   };
> >>
> >>   static const struct amdgpu_ring_funcs uvd_v6_0_enc_ring_vm_funcs = {
> >> --
> >> 2.14.1
> >>
> >> _______________________________________________
> >> amd-gfx mailing list
> >> amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
> >> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx


[-- Attachment #1.2: Type: text/html, Size: 9851 bytes --]

[-- Attachment #2: Type: text/plain, Size: 154 bytes --]

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx