On 07.07.2018 12:07, Georg Chini wrote:
On 06.07.2018 08:43, Sathish Narasimman wrote:
Hi Luiz,
Yes, I am doing the polling. I an not changing that part of the code.
In which it is default in which NBS also uses.
Thanks,
Sathish N
The current code does not poll for POLLOUT in all situations,
see comment in the code. You should check if POLLOUT is
really set in pollfd->events. Actually a situation like you
describe is "normal" but should not impact sound quality,
again see comment in the code. You could try if it helps to
remove the "if (blocks_to_write > 0)" and set writable=false
unconditionally.
In fact, in all situations where you are woken up by POLLOUT but you
do not have a block to write, POLLOUT will not be polled on the next
iteration to prevent unnecessary looping. Therefore removing the if
condition may not help much. Does the write_block_size equal the
read_block_size? In theory, different sizes are supported, but I think
it could cause issues.
On Wed, Jul 4, 2018 at 9:29 PM, Luiz Augusto von Dentz
<[email protected] <mailto:[email protected]>> wrote:
Hi Sathish,
On Mon, Jul 2, 2018 at 4:25 PM, Sathish Narasimman
<[email protected] <mailto:[email protected]>> wrote:
> Hi,
>
> I am trying to make WBS with PulseAudio. I am almost successful in
> establishing the connection and making a WBS traffic.
>
> But out of 10 attempts or sometimes the TX packets are not sent
due to
> POLLOUT is not reflected in the pollfd
> /* Test if the stream is writable */
> if (pollfd->revents & POLLOUT)
> writable = true;
Are you sure you are polling with POLLOUT?
> I would like to know what causes the stream as not writable.
> What are the connecting modules to resolve this?
> This happening only with WBS as NBS works good without any problem.
>
>
> Thanks,
> Sathish N
>
_______________________________________________
pulseaudio-discuss mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
