| Summary: | Client freezes intermittently playing over network | ||
|---|---|---|---|
| Product: | PulseAudio | Reporter: | Tom Hunt <tomdicksonhunt> |
| Component: | daemon | Assignee: | pulseaudio-bugs |
| Status: | RESOLVED MOVED | QA Contact: | pulseaudio-bugs |
| Severity: | normal | ||
| Priority: | medium | CC: | bgamari, carl, copycat, flyser42, james, lennart |
| Version: | unspecified | ||
| Hardware: | x86-64 (AMD64) | ||
| OS: | Linux (All) | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: |
Client configuration
Server configuration |
||
|
Description
Tom Hunt
2015-04-03 03:16:46 UTC
http://cgit.freedesktop.org/pulseaudio/pulseaudio/commit/src/pulsecore/protocol-native.c?id=22827a5e1e627115ae44a297be2ba5b35561700f : [pulseaudio] protocol-native.c: Client sent non-aligned memblock: index 0, length 2841, frame size: 4 seem client do not send in wholw frames , server don't expect client send .25, .5 or .75 frame is this compressed format ? Compressed format? What's this? It is, as far as I can tell, a fairly vanilla pulse install; the only change I've made is to enable TCP audio and set up cookies for that. Just for fun, I just compiled a version of pulse 6.0 with the change at that link reversed. The bug indeed disappears at this point. This is almost certainly not the best way to do it. I would venture that the proper fix is more likely in libpulse somewhere, but I couldn't begin to localize further. if you look at http://cgit.freedesktop.org/pulseaudio/pulseaudio/tree/src/utils/pacat.c static void do_stream_write(size_t length) { size_t l; pa_assert(length); if (!buffer || !buffer_length) return; l = length; if (l > buffer_length) l = buffer_length; if (pa_stream_write(stream, (uint8_t*) buffer + buffer_index, l, NULL, 0, PA_SEEK_RELATIVE) < 0) { pa_log(_("pa_stream_write() failed: %s"), pa_strerror(pa_context_errno(context))); quit(1); return; } buffer_length -= l; buffer_index += l; if (!buffer_length) { pa_xfree(buffer); buffer = NULL; buffer_index = buffer_length = 0; } } either length is no multiple of frame size or index which is not multiple of frame size are invalid when call pa_stream_write() : 2015-04-01T19:50:39.905427-06:00 localhost pulseaudio[28649]: W: [pulseaudio] protocol-native.c: Client sent non-aligned memblock: index 2841, length 1255, frame size: 4 If you look there, you see that it seems to get the invalid memblocks in clusters, and the sum of the lengths in each cluster is indeed valid. I would guess the network is doing something strange. Same problem here. Configuration: Server - Debian sid / Pulseaudio v.6.0-5 from repo / headless, launched via "dbus-launch pulseaudio" from non-root user via ssh / config attached Client - Calculate (Gentoo) / Pulseaudio v 6.0 from repo (sources) / MATE 1.8.1 / config and USE flags attached Steps to reproduce: - Turn on pulseaudio server (dbus-launch pulseaudio in my case) - Choose remote device in mate-volume-control, "Output" tab. - Change volume/balance via mate-volume-control Sometimes bug appeared suddenly, without volume/balance changes, e.g when I watch video using smplayer with sound output to pulseaudio. Log at server side attached. USE flags at client side: Installed versions: 6.0(00:14:45 10.09.2015)(X alsa alsa-plugin asyncns bluetooth caps dbus gdbm glib gnome gtk ipv6 native-headset orc ssl tcpd udev webrtc-aec zeroconf -doc -equalizer -jack -libsamplerate -lirc -neon -ofono-headset -oss -qt4 -realtime -selinux -system-wide -systemd -test -xen ABI_MIPS="-n32 -n64 -o32" ABI_PPC="-32 -64" ABI_S390="-32 -64" ABI_X86="32 64 -x32") Created attachment 118287 [details]
Client configuration
Created attachment 118288 [details]
Server configuration
I think I am suffering from the same issue, using pulseaudio 9.0. On the client, I get timeout and "stream killed" (module-tunnel.c), on the server the same message, "non-aligned memblock" (protocol-native.c). Is there a fix or workaround or some misconfiguration that I might have to look at? -- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/issues/521. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.