Re: [PATCH] media: venus: add support for selection rectangles

From: mgottam
Date: Fri Nov 09 2018 - 02:38:32 EST


On 2018-11-02 03:16, Tomasz Figa wrote:
On Fri, Nov 2, 2018 at 12:02 AM Stanimir Varbanov
<stanimir.varbanov@xxxxxxxxxx> wrote:

Hi Malathi,

On 11/1/18 3:10 PM, mgottam@xxxxxxxxxxxxxx wrote:
> On 2018-10-16 15:11, Stanimir Varbanov wrote:
>> Hi Malathi,
>>
>> On 10/09/2018 10:53 AM, Malathi Gottam wrote:
>>> Handles target type crop by setting the new active rectangle
>>> to hardware. The new rectangle should be within YUV size.
>>>
>>> Signed-off-by: Malathi Gottam <mgottam@xxxxxxxxxxxxxx>
>>> ---
>>> drivers/media/platform/qcom/venus/venc.c | 19 +++++++++++++++++--
>>> 1 file changed, 17 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/media/platform/qcom/venus/venc.c
>>> b/drivers/media/platform/qcom/venus/venc.c
>>> index 3f50cd0..754c19a 100644
>>> --- a/drivers/media/platform/qcom/venus/venc.c
>>> +++ b/drivers/media/platform/qcom/venus/venc.c
>>> @@ -478,16 +478,31 @@ static int venc_g_fmt(struct file *file, void
>>> *fh, struct v4l2_format *f)
>>> venc_s_selection(struct file *file, void *fh, struct v4l2_selection *s)
>>> {
>>> struct venus_inst *inst = to_inst(file);
>>> + int ret;
>>> + u32 buftype;
>>>
>>> if (s->type != V4L2_BUF_TYPE_VIDEO_OUTPUT)
>>> return -EINVAL;
>>>
>>> switch (s->target) {
>>> case V4L2_SEL_TGT_CROP:
>>> - if (s->r.width != inst->out_width ||
>>> - s->r.height != inst->out_height ||
>>> + if (s->r.width > inst->out_width ||
>>> + s->r.height > inst->out_height ||
>>> s->r.top != 0 || s->r.left != 0)
>>> return -EINVAL;

Note that returning -EINVAL is not what VIDIOC_S_SELECTION should do here.

As per the general V4L2 spec [1], VIDIOC_S_SELECTION is expected to
adjust the crop rectangle "as close as possible to the requested one".
In this case that would likely mean something like this:

if (s->r.left != 0) {
s->r.width += s->r.left;
s->r.left = 0;
}

if (s->r.top != 0) {
s->r.height += s->r.top;
s->r.top = 0;
}

if (s->r.width > inst->out_width)
s->r.width = inst->out_width;

if (s->r.height > inst->out_height)
s->r.height = inst->out_height;

[1]
https://www.kernel.org/doc/html/latest/media/uapi/v4l/vidioc-g-selection.html


Yes Tomasz, I overlooked the spec that we cannot return EINVAL in this case.

I will make the necessary changes and update the new patch.
>>> + if (s->r.width != inst->width ||
>>> + s->r.height != inst->height) {
>>> + buftype = HFI_BUFFER_OUTPUT;
>>> + ret = venus_helper_set_output_resolution(inst,
>>> + s->r.width,
>>> + s->r.height,
>>> + buftype);
>>
>> I'm afraid that set_output_resolution cannot be called at any time. Do
>> you think we can set it after start_session?
>
> Yes Stan, we can set output_resolution after the session has been
> initialization.
> As per the spec, this call s_selection is an optional step under
> Initialization
> procedure of encoder even before we request buffers.

Setting this output resolution here I feel it as a repetition as we
have this width and height set to firmware as a part of VIDIOC_REQBUFS().
So updating the selection rectangle would be sufficient in this case.


What spec you are referring to? The spec for the encoders [1] or
something else.

For our convenience, Hans was nice enough to host a compiled version at:
https://hverkuil.home.xs4all.nl/request-api/uapi/v4l/dev-encoder.html

Best regards,
Tomasz