[PATCH libdrm 3/4] amdgpu: add a function to find bo by cpu mapping

[Christian König]

Am 07.08.2018 um 11:52 schrieb Zhang, Jerry (Junwei):
> On 08/07/2018 05:33 PM, Christian König wrote:
>> Am 07.08.2018 um 10:28 schrieb Zhang, Jerry (Junwei):
>>> On 08/07/2018 04:20 PM, Christian König wrote:
>>>> Well NAK, that wasn't the intention of putting all BOs into the 
>>>> handle table.
>>>>
>>>> You should still use the kernel implementation.
>>>
>>> I thought we have discussed that in below mail thread. any gap?
>>>
>>> [PATCH 1/2] drm/amdgpu: return bo itself if userptr is cpu addr of 
>>> bo (v3)
>>> {{{
>>> >> Well we could just completely drop the kernel implementation and use
>>> >> an userspace implementation.
>>> >
>>> > Do you mean to implement finding bo by cpu address in libdrm 
>>> completely?
>>>
>>> Yes, exactly.
>>> }}}
>>
>> Ok, well that is a misunderstanding. I was just speculating about that.
>>
>> On the other hand if searching all BOs for the right one is not to 
>> much overhead for that workaround I'm fine with it.
>>
>> Just drop patch #2 in this series. All unused entries should be 
>> initialized to zero and walking the extra 512 isn't that much overhead.
>
> Yes, reserve it locally for now.
> May I get your RB?

I've got a few minor comments on the last patch as well.

Give me a moment to figure out how to push to libdrm master repostory 
now (they migrated to github and my old account doesn't work any more).

After that I can comment on your patches.

Thanks,
Christian.

>
> Regards,
> Jerry
>
>>
>> Regards,
>> Christian.
>>
>>>
>>> Regards,
>>> Jerry
>>>
>>>>
>>>> Christian.
>>>>
>>>> Am 07.08.2018 um 09:26 schrieb Junwei Zhang:
>>>>> Userspace needs to know if the user memory is from BO or malloc.
>>>>>
>>>>> Signed-off-by: Junwei Zhang <Jerry.Zhang at amd.com>
>>>>> ---
>>>>>   amdgpu/amdgpu.h    | 23 +++++++++++++++++++++++
>>>>>   amdgpu/amdgpu_bo.c | 34 ++++++++++++++++++++++++++++++++++
>>>>>   2 files changed, 57 insertions(+)
>>>>>
>>>>> diff --git a/amdgpu/amdgpu.h b/amdgpu/amdgpu.h
>>>>> index be83b45..a8c353c 100644
>>>>> --- a/amdgpu/amdgpu.h
>>>>> +++ b/amdgpu/amdgpu.h
>>>>> @@ -678,6 +678,29 @@ int 
>>>>> amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
>>>>>                       amdgpu_bo_handle *buf_handle);
>>>>>   /**
>>>>> + * Validate if the user memory comes from BO
>>>>> + *
>>>>> + * \param dev - [in] Device handle. See #amdgpu_device_initialize()
>>>>> + * \param cpu - [in] CPU address of user allocated memory which we
>>>>> + * want to map to GPU address space (make GPU accessible)
>>>>> + * (This address must be correctly aligned).
>>>>> + * \param size - [in] Size of allocation (must be correctly aligned)
>>>>> + * \param buf_handle - [out] Buffer handle for the userptr memory
>>>>> + * if the user memory is not from BO, the buf_handle will be NULL.
>>>>> + * \param offset_in_bo - [out] offset in this BO for this user 
>>>>> memory
>>>>> + *
>>>>> + *
>>>>> + * \return   0 on success\n
>>>>> + *          <0 - Negative POSIX Error code
>>>>> + *
>>>>> +*/
>>>>> +int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
>>>>> +                  void *cpu,
>>>>> +                  uint64_t size,
>>>>> +                  amdgpu_bo_handle *buf_handle,
>>>>> +                  uint64_t *offset_in_bo);
>>>>> +
>>>>> +/**
>>>>>    * Free previosuly allocated memory
>>>>>    *
>>>>>    * \param   dev           - \c [in] Device handle. See 
>>>>> #amdgpu_device_initialize()
>>>>> diff --git a/amdgpu/amdgpu_bo.c b/amdgpu/amdgpu_bo.c
>>>>> index b24e698..a631050 100644
>>>>> --- a/amdgpu/amdgpu_bo.c
>>>>> +++ b/amdgpu/amdgpu_bo.c
>>>>> @@ -529,6 +529,40 @@ int amdgpu_bo_wait_for_idle(amdgpu_bo_handle bo,
>>>>>       }
>>>>>   }
>>>>> +int amdgpu_find_bo_by_cpu_mapping(amdgpu_device_handle dev,
>>>>> +                  void *cpu,
>>>>> +                  uint64_t size,
>>>>> +                  amdgpu_bo_handle *buf_handle,
>>>>> +                  uint64_t *offset_in_bo)
>>>>> +{
>>>>> +    int i;
>>>>> +    struct amdgpu_bo *bo;
>>>>> +
>>>>> +    if (cpu == NULL || size == 0)
>>>>> +        return -EINVAL;
>>>>> +
>>>>> +    pthread_mutex_lock(&dev->bo_table_mutex);
>>>>> +    for (i = 0; i <= dev->bo_handles.max_key; i++) {
>>>>> +        bo = handle_table_lookup(&dev->bo_handles, i);
>>>>> +        if (!bo || !bo->cpu_ptr || size > bo->alloc_size)
>>>>> +            continue;
>>>>> +        if (cpu >= bo->cpu_ptr && cpu < (bo->cpu_ptr + 
>>>>> bo->alloc_size))
>>>>> +            break;
>>>>> +    }
>>>>> +    pthread_mutex_unlock(&dev->bo_table_mutex);
>>>>> +
>>>>> +    if (i <= dev->bo_handles.max_key) {
>>>>> +        atomic_inc(&bo->refcount);
>>>>> +        *buf_handle = bo;
>>>>> +        *offset_in_bo = cpu - bo->cpu_ptr;
>>>>> +    } else {
>>>>> +        *buf_handle = NULL;
>>>>> +        *offset_in_bo = 0;
>>>>> +    }
>>>>> +
>>>>> +    return 0;
>>>>> +}
>>>>> +
>>>>>   int amdgpu_create_bo_from_user_mem(amdgpu_device_handle dev,
>>>>>                       void *cpu,
>>>>>                       uint64_t size,
>>>>
>>>> _______________________________________________
>>>> amd-gfx mailing list
>>>> amd-gfx at lists.freedesktop.org
>>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>>> _______________________________________________
>>> amd-gfx mailing list
>>> amd-gfx at lists.freedesktop.org
>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>>