<div class="gmail_quote">在 2012年3月30日 下午12:48,Arun Raghavan <span dir="ltr"><<a href="mailto:arun.raghavan@collabora.co.uk">arun.raghavan@collabora.co.uk</a>></span>写道:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
<div class="im"><br>
On Thu, 2012-03-29 at 21:31 +0800, Deng Zhenrong wrote:<br>
> According to the description below:<br>
><br>
> <a href="http://lkml.org/lkml/2009/7/17/258" target="_blank">http://lkml.org/lkml/2009/7/17/258</a><br>
><br>
> There are two benefits:<br>
> a) it doesn't need to access the hardware.<br>
> b) avoid the syscall by using vdso clock_gettime().<br>
><br>
> The cons is the CLOCK_MONOTONIC_COARSE returns the time at the last<br>
> tick.<br>
<br>
</div>As Pierre mentions, we actually _want_ high resolution time sources for<br>
our uses. We use these timers to predict when the audio buffer will be<br>
nearing being empty so that we can wake up and fill the buffer again.<br>
Having a coarse timer for this purpose would mean that we can be woken<br>
up late, causing the device to underrun and users to hear glitches.<br>
<br></blockquote><div><br></div><div>Maybe I'm testing this patch in an idle state, so there's no underrun seen, I'll take that into account. Thanks for your clarification here. It helps me understand pulseaudio better.</div>
<div><br></div></div>