Bug#763985: pa log

[Adam Borowski]

On Tue, Jul 18, 2017 at 07:14:04PM -0400, Felipe Sateler wrote:
> On Tue, Jul 4, 2017 at 9:39 PM, Adam Borowski <kilobyte at angband.pl> wrote:
> 
> > > > Is there any more debugging info I should attach?  I have !clue about
> > > > pulseaudio hacking.
> > .
> > > You could attach a verbose log of pulseaudio;
> > >
> > > $ pulseaudio --kill ; pulseaudio --log-target=file:pa.log --log-time &
> > > # suspend, resume and check the problem happens
> > > $ pulseaudio --kill ; pulseaudio --start # to get a new pulse daemon
> > >
> > > That would most certainly help.
> > ---
> > > Oops, I forgot to add -vvvv to the pa command line.
> >
> > Ouchie, somehow this bug mail was lost -- and in the meantime I was happy
> > with ALSA.  Well, I'm still happy, but Firefox is currently w/o audio...
> > Sorry for this.  Thus, with an, ahem, small delay:
> >
> > Here's the log.  The bug still reliably reproduces, even on 4.12 kernel
> > fresh from Linus, and so it does on our official 4.9.
> >
> 
> Is this machine particularly slow? 4s to probe alsa looks a bit slow.

A regular amd64 desktop, somewhat long in the tooth now but was quite decent
when bought 5 years ago.  No non-negligible load at the time of the test.

> Anyway, when (relative to the log) did you suspend the machine, and when
> did you experience the problem?

Started the sample sound using mplayer (visible in the log connecting at
16.667), suspended right after -- at human reaction speed, without special
haste, so it should be ~3 seconds, waited around 10 seconds before resuming.
The start of errors at 29.281 apparently matches resume.

> I see a few Wakeups from alsa, plus some underruns. Possibly tsched does
> not work very well on your driver, could you try passing tsched=0 to
> module-udev-detect in default.pa ?

Will test this when I can reboot with pulse.


Meow!
-- 
⢀⣴⠾⠻⢶⣦⠀ 
⣾⠁⢠⠒⠀⣿⡁ A dumb species has no way to open a tuna can.
⢿⡄⠘⠷⠚⠋⠀ A smart species invents a can opener.
⠈⠳⣄⠀⠀⠀⠀ A master species delegates.