Re: [PATCH v4 2/2] radeon: use memcpy_to/fromio for UVD fw upload

2021-06-04 Thread Alex Deucher
Applied with the RB fixed.

Thanks!

Alex

On Fri, Jun 4, 2021 at 7:53 AM Chen Li  wrote:
>
>
> I met a gpu addr bug recently and the kernel log
> tells me the pc is memcpy/memset and link register is
> radeon_uvd_resume.
>
> As we know, in some architectures, optimized memcpy/memset
> may not work well on device memory. Trival memcpy_toio/memset_io
> can fix this problem.
>
> BTW, amdgpu has already done it in:
> commit ba0b2275a678 ("drm/amdgpu: use memcpy_to/fromio for UVD fw upload"),
> that's why it has no this issue on the same gpu and platform.
>
> Signed-off-by: Chen Li 
> Reviewed-by: Christian König
> ---
>  drivers/gpu/drm/radeon/radeon_uvd.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c 
> b/drivers/gpu/drm/radeon/radeon_uvd.c
> index 85a1f2c31749..753da95e6abb 100644
> --- a/drivers/gpu/drm/radeon/radeon_uvd.c
> +++ b/drivers/gpu/drm/radeon/radeon_uvd.c
> @@ -288,7 +288,7 @@ int radeon_uvd_resume(struct radeon_device *rdev)
> if (rdev->uvd.vcpu_bo == NULL)
> return -EINVAL;
>
> -   memcpy(rdev->uvd.cpu_addr, rdev->uvd_fw->data, rdev->uvd_fw->size);
> +   memcpy_toio((void __iomem *)rdev->uvd.cpu_addr, rdev->uvd_fw->data, 
> rdev->uvd_fw->size);
>
> size = radeon_bo_size(rdev->uvd.vcpu_bo);
> size -= rdev->uvd_fw->size;
> @@ -296,7 +296,7 @@ int radeon_uvd_resume(struct radeon_device *rdev)
> ptr = rdev->uvd.cpu_addr;
> ptr += rdev->uvd_fw->size;
>
> -   memset(ptr, 0, size);
> +   memset_io((void __iomem *)ptr, 0, size);
>
> return 0;
>  }
> --
> 2.31.1
>
>
>
>
> ___
> amd-gfx mailing list
> amd-...@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx


Re: [PATCH v4 2/2] radeon: use memcpy_to/fromio for UVD fw upload

2021-06-04 Thread Christian König

Am 04.06.21 um 10:43 schrieb Chen Li:

I met a gpu addr bug recently and the kernel log
tells me the pc is memcpy/memset and link register is
radeon_uvd_resume.

As we know, in some architectures, optimized memcpy/memset
may not work well on device memory. Trival memcpy_toio/memset_io
can fix this problem.

BTW, amdgpu has already done it in:
commit ba0b2275a678 ("drm/amdgpu: use memcpy_to/fromio for UVD fw upload"),
that's why it has no this issue on the same gpu and platform.

Signed-off-by: Chen Li 
Reviewed-by: Christian König


Reviewed-by: Christian König 

Alex will probably now pick them up for upstreaming.

Christian.


---
  drivers/gpu/drm/radeon/radeon_uvd.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c 
b/drivers/gpu/drm/radeon/radeon_uvd.c
index 85a1f2c31749..753da95e6abb 100644
--- a/drivers/gpu/drm/radeon/radeon_uvd.c
+++ b/drivers/gpu/drm/radeon/radeon_uvd.c
@@ -288,7 +288,7 @@ int radeon_uvd_resume(struct radeon_device *rdev)
if (rdev->uvd.vcpu_bo == NULL)
return -EINVAL;
  
-	memcpy(rdev->uvd.cpu_addr, rdev->uvd_fw->data, rdev->uvd_fw->size);

+   memcpy_toio((void __iomem *)rdev->uvd.cpu_addr, rdev->uvd_fw->data, 
rdev->uvd_fw->size);
  
  	size = radeon_bo_size(rdev->uvd.vcpu_bo);

size -= rdev->uvd_fw->size;
@@ -296,7 +296,7 @@ int radeon_uvd_resume(struct radeon_device *rdev)
ptr = rdev->uvd.cpu_addr;
ptr += rdev->uvd_fw->size;
  
-	memset(ptr, 0, size);

+   memset_io((void __iomem *)ptr, 0, size);
  
  	return 0;

  }