RE: [PATCH] drm/amd/amdgpu: Add pcie indirect support to amdgpu_mm_wreg_mmio_rlc()

["Zhang, Hawking" <Hawking.Zhang@amd.com>]

[AMD Official Use Only]

RE - Actually, for older asics, we shouldn't we be using mmINDEX/mmDATA rather than the pcie indirect registers?  Or is that handled already somehow?

I remember we checked this with hw team before (might two years ago when make the change in amdgpu_device_rreg/wreg helper). The answer was it is safe to retire mmINDEX/DATA approach for SI and onwards. PCIE_INDEX/DATA should be good enough for indirect access in amdgpu driver. For radeon driver, mmINDEX/mmDATA shall still be kept.

Regards,
Hawking
-----Original Message-----
From: amd-gfx <amd-gfx-bounces@lists.freedesktop.org> On Behalf Of Alex Deucher
Sent: Sunday, January 9, 2022 05:13
To: StDenis, Tom <Tom.StDenis@amd.com>
Cc: amd-gfx list <amd-gfx@lists.freedesktop.org>
Subject: Re: [PATCH] drm/amd/amdgpu: Add pcie indirect support to amdgpu_mm_wreg_mmio_rlc()

On Fri, Jan 7, 2022 at 7:07 AM Tom St Denis <tom.stdenis@amd.com> wrote:
>
> The function amdgpu_mm_wreg_mmio_rlc() is used by debugfs to write to 
> MMIO registers.  It didn't support registers beyond the BAR mapped 
> MMIO space.  This adds pcie indirect write support.
>
> Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c 
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index c38e0e87090b..53a04095a6db 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -552,7 +552,7 @@ void amdgpu_device_wreg(struct amdgpu_device 
> *adev,  }
>
>  /**
> - * amdgpu_mm_wreg_mmio_rlc -  write register either with mmio or with 
> RLC path if in range
> + * amdgpu_mm_wreg_mmio_rlc -  write register either with 
> + direct/indirect mmio or with RLC path if in range
>   *
>   * this function is invoked only the debugfs register access
>   */
> @@ -567,6 +567,8 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
>             adev->gfx.rlc.funcs->is_rlcg_access_range) {
>                 if (adev->gfx.rlc.funcs->is_rlcg_access_range(adev, reg))
>                         return adev->gfx.rlc.funcs->sriov_wreg(adev, 
> reg, v, 0, 0);
> +       } else if ((reg * 4) >= adev->rmmio_size) {
> +               adev->pcie_wreg(adev, reg * 4, v);

Actually, for older asics, we shouldn't we be using mmINDEX/mmDATA rather than the pcie indirect registers?  Or is that handled already somehow?

Alex

>         } else {
>                 writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
>         }
> --
> 2.32.0
>