Hi all, a quick follow-up to my earlier post. I now rootcaused what exactly happens and wanted to share the results. On Fri, 31 May 2019, Kai Vehmanen wrote: > I've been testing pulse with the new Intel SOF [1] audio drivers (that are > going to 5.2 kernel). > > In my tests, e.g. system suspend/stress tests work much better on latest > master of pulseaudio, compared to test with latest released Pulseaudio. This turned out to be a bug in how PA handled the first iteration after ALSA SUSPEND state. See the description in (and link to a PA12.2 patch to fix): https://github.com/thesofproject/linux/issues/1000#issuecomment-508128140 https://github.com/kv2019i/pulseaudio/commit/0b3c52341d1d8d098f315302d6d3e1c40965493f The affected code was reworked by Georg Chini's patch "alsa: Improve resume logic after alsa suspend" in PA master in March this year, so issue is already fixed. With 13.x release looming (and apparently bunch of other issues with S3 handling if streams are active), probably better for people to update to PA13 rather than take my point-fix to PA12.2. Br, Kai _______________________________________________ pulseaudio-discuss mailing list pulseaudio-discuss@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/pulseaudio-discuss
