2018-06-17 08:53:27

by Stefan Agner

Subject: [PATCH] drm/amdgpu: use first uvd instance to avoid clang build error

Explicitly use the first uvd instance to avoid a build error when
using clang 6:
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1148:52: error: expected ')'
container_of(work, struct amdgpu_device, uvd.inst->idle_work.work);
^
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1148:3: note: to match this '('
container_of(work, struct amdgpu_device, uvd.inst->idle_work.work);
^
./include/linux/kernel.h:967:21: note: expanded from macro 'container_of'
((type *)(__mptr - offsetof(type, member))); })
^
./include/linux/stddef.h:17:32: note: expanded from macro 'offsetof'
^
./include/linux/compiler-gcc.h:170:20: note: expanded from macro
'__compiler_offsetof'
__builtin_offsetof(a, b)
^
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1147:24: error: initializing
'struct amdgpu_device *' with an expression of incompatible type 'void'
struct amdgpu_device *adev =
^
2 errors generated.

Fixes: 10dd74eac4db ("drm/amdgpu/vg20:Restruct uvd.inst to support multiple instances")
Cc: James Zhu <[email protected]>
Signed-off-by: Stefan Agner <[email protected]>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index bcf68f80bbf0..6666a5888c44 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -1145,7 +1145,7 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
{
struct amdgpu_device *adev =
- container_of(work, struct amdgpu_device, uvd.inst->idle_work.work);
+ container_of(work, struct amdgpu_device, uvd.inst[0].idle_work.work);
unsigned fences = 0, i, j;

for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
--
2.17.1

2018-06-18 02:06:20

by James Zhu

Subject: Re: [PATCH] drm/amdgpu: use first uvd instance to avoid clang build error

On 2018-06-17 04:52 AM, Stefan Agner wrote:
> Explicitly use the first uvd instance to avoid a build error when
> using clang 6:
> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1148:52: error: expected ')'
> container_of(work, struct amdgpu_device, uvd.inst->idle_work.work);
> ^
> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1148:3: note: to match this '('
> container_of(work, struct amdgpu_device, uvd.inst->idle_work.work);
> ^
> ./include/linux/kernel.h:967:21: note: expanded from macro 'container_of'
> ((type *)(__mptr - offsetof(type, member))); })
> ^
> ./include/linux/stddef.h:17:32: note: expanded from macro 'offsetof'
> ^
> ./include/linux/compiler-gcc.h:170:20: note: expanded from macro
> '__compiler_offsetof'
> __builtin_offsetof(a, b)
> ^
> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1147:24: error: initializing
> 'struct amdgpu_device *' with an expression of incompatible type 'void'
> struct amdgpu_device *adev =
> ^
> 2 errors generated.
>
> Fixes: 10dd74eac4db ("drm/amdgpu/vg20:Restruct uvd.inst to support multiple instances")
> Cc: James Zhu <[email protected]>
> Signed-off-by: Stefan Agner <[email protected]>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
> index bcf68f80bbf0..6666a5888c44 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
> @@ -1145,7 +1145,7 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
> static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
> {
> struct amdgpu_device *adev =
> - container_of(work, struct amdgpu_device, uvd.inst->idle_work.work);
> + container_of(work, struct amdgpu_device, uvd.inst[0].idle_work.work);
Hi Alex,
If all instances share one idle work from hardware view currently and in
the future ,
should we move struct delayed_work idle_work from struct amdgpu_uvd_inst
to struct amdgpu_uvd?
James
> unsigned fences = 0, i, j;
>
> for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {

2018-06-18 14:38:47

by Alex Deucher

Subject: Re: [PATCH] drm/amdgpu: use first uvd instance to avoid clang build error

On Sun, Jun 17, 2018 at 10:03 PM, James Zhu <[email protected]> wrote:
>
>
> On 2018-06-17 04:52 AM, Stefan Agner wrote:
>>
>> Explicitly use the first uvd instance to avoid a build error when
>> using clang 6:
>> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1148:52: error: expected ')'
>> container_of(work, struct amdgpu_device,
>> uvd.inst->idle_work.work);
>> ^
>> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1148:3: note: to match this '('
>> container_of(work, struct amdgpu_device,
>> uvd.inst->idle_work.work);
>> ^
>> ./include/linux/kernel.h:967:21: note: expanded from macro 'container_of'
>> ((type *)(__mptr - offsetof(type, member))); })
>> ^
>> ./include/linux/stddef.h:17:32: note: expanded from macro 'offsetof'
>> ^
>> ./include/linux/compiler-gcc.h:170:20: note: expanded from macro
>> '__compiler_offsetof'
>> __builtin_offsetof(a, b)
>> ^
>> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c:1147:24: error: initializing
>> 'struct amdgpu_device *' with an expression of incompatible type
>> 'void'
>> struct amdgpu_device *adev =
>> ^
>> 2 errors generated.
>>
>> Fixes: 10dd74eac4db ("drm/amdgpu/vg20:Restruct uvd.inst to support
>> multiple instances")
>> Cc: James Zhu <[email protected]>
>> Signed-off-by: Stefan Agner <[email protected]>
>> ---
>> drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>> index bcf68f80bbf0..6666a5888c44 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
>> @@ -1145,7 +1145,7 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring
>> *ring, uint32_t handle,
>> static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
>> {
>> struct amdgpu_device *adev =
>> - container_of(work, struct amdgpu_device,
>> uvd.inst->idle_work.work);
>> + container_of(work, struct amdgpu_device,
>> uvd.inst[0].idle_work.work);
>
> Hi Alex,
> If all instances share one idle work from hardware view currently and in the
> future ,
> should we move struct delayed_work idle_work from struct amdgpu_uvd_inst to
> struct amdgpu_uvd?

Yes, please do.

Alex

> James
>>
>> unsigned fences = 0, i, j;
>> for (i = 0; i < adev->uvd.num_uvd_inst; ++i) {
>
>
> _______________________________________________
> dri-devel mailing list
> [email protected]
> https://lists.freedesktop.org/mailman/listinfo/dri-devel