On 14.08.2023 21:06, Dikshita Agarwal wrote:
>
>
> On 7/28/2023 10:52 PM, Konrad Dybcio wrote:
>> On 28.07.2023 18:28, Bjorn Andersson wrote:
>>> On Fri, Jul 28, 2023 at 06:53:22PM +0530, Vikash Garodia wrote:
>>>> diff --git a/drivers/media/platform/qcom/iris/vidc/src/msm_vidc_memory.c b/drivers/media/platform/qcom/iris/vidc/src/msm_vidc_memory.c
>>> [..]
>>>> +static const struct msm_vidc_memory_ops msm_mem_ops = {
>>>> + .dma_buf_get = msm_vidc_dma_buf_get,
>>>> + .dma_buf_put = msm_vidc_dma_buf_put,
>>>> + .dma_buf_put_completely = msm_vidc_dma_buf_put_completely,
>>>> + .dma_buf_attach = msm_vidc_dma_buf_attach,
>>>> + .dma_buf_detach = msm_vidc_dma_buf_detach,
>>>> + .dma_buf_map_attachment = msm_vidc_dma_buf_map_attachment,
>>>> + .dma_buf_unmap_attachment = msm_vidc_dma_buf_unmap_attachment,
>>>> + .memory_alloc_map = msm_vidc_memory_alloc_map,
>>>> + .memory_unmap_free = msm_vidc_memory_unmap_free,
>>>> + .buffer_region = msm_vidc_buffer_region,
>>>
>>> Will there ever be more than one implementation of the
>>> msm_vidc_memory_ops?
>>>
>>> Unless there's a really strong reason, just call the functions directly
>>> without the function pointers and call_mem_op(), this will be slightly
>>> faster, but more importantly it allows for much faster navigation of the
>>> code base.
>> Same for HFI ops
> Hi Konrad,
> There are no HFI ops in this driver, are you referring to anything else
> here, could you please point me to it?
Sorry, I had that in my brain cache after reading through the downstream
driver..
Konrad
