[PATCH 14/21] drm/amdgpu: add new amdgpu_gmc_emit_flush_gpu_tlb callback

[Christian König]

Am 18.01.2018 um 08:08 schrieb Chunming Zhou:
>
>
> On 2018年01月17日 20:19, Christian König wrote:
>> Am 17.01.2018 um 07:24 schrieb Chunming Zhou:
>>> I'm not sure if this is a good ongoing, as I know, our per IP 
>>> topology is intended for every generation asic bringup.
>>
>> Actually the intention goes beyond that, e.g. we want to reuse IPs 
>> for newer ASICs as well when they haven't changed.
>>
>>>
>>> I think there are many similar logics which we can abstract, but we 
>>> don't do that.
>>>
>>> For emit_wreg above patches, they only can be used by vm flush part, 
>>> but they are common func fallback.
>>>
>>> Can we only make emit_wreg as static in their files, and replace 
>>> every where they are used?
>>
>> Not sure what you are suggesting here?
> my mean is like the attached, only abstract them in file per ip.

That is not a good idea. The functions should be independent from the 
hardware version.

Additional to that we want to avoid using structures for ring buffer 
content.

Regards,
Christian.

>
>
> Regards,
> David Zhou
>>
>> Regards,
>> Christian.
>>
>>>
>>> Regards,
>>> David Zhou
>>> On 2018年01月17日 03:40, Christian König wrote:
>>>> Add a new GMC function to unify vm flushing.
>>>>
>>>> Signed-off-by: Christian König <christian.koenig at amd.com>
>>>> ---
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu.h     | 1 +
>>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h | 3 +++
>>>>   2 files changed, 4 insertions(+)
>>>>
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>>> index 93eae393b08d..230826718c98 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>>> @@ -1779,6 +1779,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring 
>>>> *ring)
>>>>   #define amdgpu_asic_flush_hdp(adev) 
>>>> (adev)->asic_funcs->flush_hdp((adev))
>>>>   #define amdgpu_asic_invalidate_hdp(adev) 
>>>> (adev)->asic_funcs->invalidate_hdp((adev))
>>>>   #define amdgpu_gmc_flush_gpu_tlb(adev, vmid) 
>>>> (adev)->gmc.gmc_funcs->flush_gpu_tlb((adev), (vmid))
>>>> +#define amdgpu_gmc_emit_flush_gpu_tlb(r, vmid, pasid, addr) 
>>>> (r)->adev->gmc.gmc_funcs->emit_flush_gpu_tlb((r), (vmid), (pasid), 
>>>> (addr))
>>>>   #define amdgpu_gmc_set_pte_pde(adev, pt, idx, addr, flags) 
>>>> (adev)->gmc.gmc_funcs->set_pte_pde((adev), (pt), (idx), (addr), 
>>>> (flags))
>>>>   #define amdgpu_gmc_get_vm_pde(adev, level, dst, flags) 
>>>> (adev)->gmc.gmc_funcs->get_vm_pde((adev), (level), (dst), (flags))
>>>>   #define amdgpu_gmc_get_pte_flags(adev, flags) 
>>>> (adev)->gmc.gmc_funcs->get_vm_pte_flags((adev),(flags))
>>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h 
>>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
>>>> index a4a8374f7f3f..114350a4693f 100644
>>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
>>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
>>>> @@ -52,6 +52,9 @@ struct amdgpu_gmc_funcs {
>>>>       /* flush the vm tlb via mmio */
>>>>       void (*flush_gpu_tlb)(struct amdgpu_device *adev,
>>>>                     uint32_t vmid);
>>>> +    /* flush the vm tlb via ring */
>>>> +    uint64_t (*emit_flush_gpu_tlb)(struct amdgpu_ring *ring, 
>>>> unsigned vmid,
>>>> +                       unsigned pasid, uint64_t pd_addr);
>>>>       /* write pte/pde updates using the cpu */
>>>>       int (*set_pte_pde)(struct amdgpu_device *adev,
>>>>                  void *cpu_pt_addr, /* cpu addr of page table */
>>>
>>
>