On 04/18/18 12:30, Sakari Ailus wrote:
> On Wed, Apr 18, 2018 at 01:46:08AM -0700, Matt Ranostay wrote:
>
> ...
>
>> On Wed, Apr 18, 2018 at 1:03 AM, Sakari Ailus <[email protected]> wrote:
>>>> + if (vid_cap_buf) {
>>>> + struct vb2_buffer *vb2_buf =
>>>> &vid_cap_buf->vb.vb2_buf;
>>>> + void *vbuf = vb2_plane_vaddr(vb2_buf, 0);
>>>> + int ret = data->chip->xfer(data, vbuf);
>>>
>>> As the assignment in variable declaration does more than just initialise a
>>> variable, it'd be nice to make the assignment separately from the variable
>>> declaration.
>>
>> Guessing you mean it is that initialization here is getting pushed and
>> popped off the stack if the data isn't in a register?
>
> No, just that functionality is placed where variables are declared. The
> code is easier to read if you separate the two. I.e.
>
> int ret;
>
> ret = ...->xfer();
Matt, I'm making a pull request for this v8. I've split up this line myself,
so no need to post a v9.
Regards,
Hans
>
>>
>>>
>>>> +
>>>> + vb2_buf->timestamp = ktime_get_ns();
>>>> + vid_cap_buf->vb.sequence = data->sequence++;
>>>> + vb2_buffer_done(vb2_buf, ret ?
>>>> + VB2_BUF_STATE_ERROR : VB2_BUF_STATE_DONE);
>>>> + }
>>>> +
>>>> + schedule_delay = delay - (jiffies - start_jiffies);
>>>> +
>>>> + if (time_after(jiffies, start_jiffies + delay))
>>>> + schedule_delay = delay;
>>>> +
>>>> + schedule_timeout_interruptible(schedule_delay);
>>>> + } while (!kthread_should_stop());
>>>> +
>>>> + return 0;
>>>> +}
>
