On Fri, 2009-08-07 at 01:45 +0300, Maxim Levitsky wrote:
> Finally, here is the fix,
> it turns out that gstreamer were asking for timeout that sometimes
> already happened. (maybe due to wakeups from pulse to update volume,
> anyway there is a check for such condition, but it is broken.
>
> pa_usec_t is unsigned, so a result of a subtraction will always be non
> negative, so, the following fixes this bug for me.
> or it is possible to make this a one liner patch by changing the type of
> 'usec' from pa_usec_t to int64_t
>
>
> ---
>
>
> Correctly deal with events in the past in calc_next_timeout
>
> From: Maxim Levitsky <maximlevitsky at gmail.com>
>
> pa_usec_t is unsigned, thus it will always be >= 0
> This makes gstreamer pulse mixer work again
> ---
>
> src/pulse/mainloop.c | 10 +++++-----
> 1 files changed, 5 insertions(+), 5 deletions(-)
>
>
> diff --git a/src/pulse/mainloop.c b/src/pulse/mainloop.c
> index c418d10..fb5f148 100644
> --- a/src/pulse/mainloop.c
> +++ b/src/pulse/mainloop.c
> @@ -765,7 +765,7 @@ static pa_time_event*
> find_next_time_event(pa_mainloop *m) {
>
> static int calc_next_timeout(pa_mainloop *m) {
> pa_time_event *t;
> - pa_usec_t usec;
> + pa_usec_t clock_now;
>
> if (!m->n_enabled_time_events)
> return -1;
> @@ -776,12 +776,12 @@ static int calc_next_timeout(pa_mainloop *m) {
> if (t->time == 0)
> return 0;
>
> - usec = t->time - pa_rtclock_now();
> + clock_now = pa_rtclock_now();
>
> - if (usec <= 0)
> - return 0;
> + if (t->time < clock_now)
> + return 0;
>
> - return (int) (usec / 1000); /* in milliseconds */
> + return (int) ((t->time - clock_now) / 1000); /* in milliseconds */
> }
>
> static int dispatch_timeout(pa_mainloop *m) {
>
Lennart Poettering, what do you think about this?
Regards,
Maxim Levitsky
