[PATCH v2 01/10] drm/amdgpu: add skip_hw_access checks for sriov

[Christian König]

Am 28.05.24 um 19:23 schrieb Yunxiang Li:
> Accessing registers via host is missing the check for skip_hw_access and
> the lockdep check that comes with it.
>
> Signed-off-by: Yunxiang Li <Yunxiang.Li at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c | 9 +++++++++
>   1 file changed, 9 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> index 3d5f58e76f2d..3cf8416f8cb0 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> @@ -977,6 +977,9 @@ u32 amdgpu_virt_rlcg_reg_rw(struct amdgpu_device *adev, u32 offset, u32 v, u32 f
>   		return 0;
>   	}
>   
> +	if (amdgpu_device_skip_hw_access(adev))
> +		return 0;
> +

The only other caller of this function is amdgpu_device_xcc_rreg() and 
that one has the amdgpu_device_skip_hw_access() call already. So it 
could be that this one is duplicated.

On the other hand an extra check doesn't really hurt us.

So either way the patch is Reviewed-by: Christian König 
<christian.koenig at amd.com>

Regards,
Christian.


>   	reg_access_ctrl = &adev->gfx.rlc.reg_access_ctrl[xcc_id];
>   	scratch_reg0 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg0;
>   	scratch_reg1 = (void __iomem *)adev->rmmio + 4 * reg_access_ctrl->scratch_reg1;
> @@ -1047,6 +1050,9 @@ void amdgpu_sriov_wreg(struct amdgpu_device *adev,
>   {
>   	u32 rlcg_flag;
>   
> +	if (amdgpu_device_skip_hw_access(adev))
> +		return;
> +
>   	if (!amdgpu_sriov_runtime(adev) &&
>   		amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags, hwip, true, &rlcg_flag)) {
>   		amdgpu_virt_rlcg_reg_rw(adev, offset, value, rlcg_flag, xcc_id);
> @@ -1064,6 +1070,9 @@ u32 amdgpu_sriov_rreg(struct amdgpu_device *adev,
>   {
>   	u32 rlcg_flag;
>   
> +	if (amdgpu_device_skip_hw_access(adev))
> +		return 0;
> +
>   	if (!amdgpu_sriov_runtime(adev) &&
>   		amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags, hwip, false, &rlcg_flag))
>   		return amdgpu_virt_rlcg_reg_rw(adev, offset, 0, rlcg_flag, xcc_id);