Re: [RFC PATCH v6 6/9] media: tegra: Add Tegra210 Video input driver

[Sowjanya Komatineni <skomatineni@nvidia.com>]

On 4/6/20 1:56 PM, Dmitry Osipenko wrote:
> External email: Use caution opening links or attachments
>
>
> 06.04.2020 23:55, Sowjanya Komatineni пишет:
>> On 4/6/20 1:53 PM, Dmitry Osipenko wrote:
>>> External email: Use caution opening links or attachments
>>>
>>>
>>> 06.04.2020 23:50, Sowjanya Komatineni пишет:
>>>> On 4/6/20 1:45 PM, Dmitry Osipenko wrote:
>>>>> External email: Use caution opening links or attachments
>>>>>
>>>>>
>>>>> 04.04.2020 04:25, Sowjanya Komatineni пишет:
>>>>>> +static int chan_capture_kthread_start(void *data)
>>>>>> +{
>>>>>> +     struct tegra_vi_channel *chan = data;
>>>>>> +     struct tegra_channel_buffer *buf;
>>>>>> +     int err = 0;
>>>>>> +     int caps_inflight;
>>>>>> +
>>>>>> +     set_freezable();
>>>>>> +
>>>>>> +     while (1) {
>>>>>> +             try_to_freeze();
>>>>>> +
>>>>>> +             wait_event_interruptible(chan->start_wait,
>>>>>> +                                      !list_empty(&chan->capture) ||
>>>>>> +                                      kthread_should_stop());
>>>>>> +             /*
>>>>>> +              * Frame start and MW_ACK_DONE syncpoint condition
>>>>>> FIFOs are
>>>>>> +              * of max depth 2. So make sure max 2 capture
>>>>>> requests are
>>>>>> +              * in process by the hardware at a time.
>>>>>> +              */
>>>>>> +             while (!(kthread_should_stop() ||
>>>>>> list_empty(&chan->capture))) {
>>>>>> +                     caps_inflight = chan->capture_reqs -
>>>>>> chan->sequence;
>>>>>> +                     /*
>>>>>> +                      * Source is not streaming if error is non-zero.
>>>>>> +                      * So, do not dequeue buffers on capture error
>>>>>> or when
>>>>>> +                      * syncpoint requests in FIFO are full.
>>>>>> +                      */
>>>>>> +                     if (err || caps_inflight >= SYNCPT_FIFO_DEPTH)
>>>>>> +                             break;
>>>>> Am I understanding correctly that this thread will take 100% CPU,
>>>>> spinning here, if more than 2 frame-captures queued?
>>>> on more than 2 frames captures, it breaks thread and on next wakeup it
>>>> continues
>>> The wait_event() won't wait if condition is true.
>> condition is checked when waitqueue is woken up
> https://elixir.bootlin.com/linux/v5.6.2/source/include/linux/wait.h#L462

process is put to sleep until the condition evaluates to true or signal 
is received.

condition is checked each time the waitqueue head is woken up.

Also capture list may keep on getting updated with buffers from userspace.

but at a time we only limit 2 frames as VI supports double buffering and 
syncpt fifo's max depth is 2

Any more buffers waiting will be processing on subsequent iterations.

So basically thread run time is depending on buffers getting queued from 
userspace.