[alsa-devel] Problem with PPH plugin
Hi,
Just wanted to highlight a bug over on Mozilla relating to a crash in the PPH plugin.
https://bugzilla.mozilla.org/show_bug.cgi?id=540460
It was initially though to be in PulseAudio, but on closer inspection this doesn't involve PA at all and in fact seems to relate to the PPH plugin from what I can gather.
The backtrace is here: https://bug540460.bugzilla.mozilla.org/attachment.cgi?id=422264
The user did submit a bug report to the bugtracker (https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4878) but given previous experience there (see other mail!) I figured I'd post a message here to highlight it seeing as the reporter has already done a lot to try and work out what is going on and posting a message on his behalf seemed like a good idea :)
Cheers
All the best.
At Tue, 19 Jan 2010 17:49:16 +0000, Colin Guthrie wrote:
Hi,
Just wanted to highlight a bug over on Mozilla relating to a crash in the PPH plugin.
https://bugzilla.mozilla.org/show_bug.cgi?id=540460
It was initially though to be in PulseAudio, but on closer inspection this doesn't involve PA at all and in fact seems to relate to the PPH plugin from what I can gather.
The backtrace is here: https://bug540460.bugzilla.mozilla.org/attachment.cgi?id=422264
The user did submit a bug report to the bugtracker (https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4878) but given previous experience there (see other mail!) I figured I'd post a message here to highlight it seeing as the reporter has already done a lot to try and work out what is going on and posting a message on his behalf seemed like a good idea :)
Hm, this is a horrible combination for debugging: a binary-only app plugin from another app.... The backtrace shows the function: that's good. But, it shows the very first line of the function. So, it can't tell where actually hits. It makes hard to guess...
And, I can't reproduce the bug.
The reporter can try the following, at least: - uninstall alsa-plugins package to see whether it's a speexrate plugin issue - update packages from OBS multimedia:audio:libs repo, to be sure, whether the problem got fixed in the later versions
Takashi
Takashi Iwai <tiwai <at> suse.de> writes:
Hm, this is a horrible combination for debugging: a binary-only app plugin from another app.... The backtrace shows the function: that's good. But, it shows the very first line of the function. So, it can't tell where actually hits. It makes hard to guess...
And, I can't reproduce the bug.
The reporter can try the following, at least:
- uninstall alsa-plugins package to see whether it's a speexrate plugin issue
- update packages from OBS multimedia:audio:libs repo, to be sure, whether the problem got fixed in the later versions
Takashi
Hi and thanks for the advice. Yes the issue is present only if a package "alsa-plugins" installed. The problem also occurs after the upgrade packages http://download.opensuse.org/repositories/multimedia:/libs/openSUSE_11.2/ . I also added information to the alsa bugtracker. https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4878
Anton
At Wed, 20 Jan 2010 22:24:46 +0000 (UTC), Anton Yashin wrote:
Takashi Iwai <tiwai <at> suse.de> writes:
Hm, this is a horrible combination for debugging: a binary-only app plugin from another app.... The backtrace shows the function: that's good. But, it shows the very first line of the function. So, it can't tell where actually hits. It makes hard to guess...
And, I can't reproduce the bug.
The reporter can try the following, at least:
- uninstall alsa-plugins package to see whether it's a speexrate plugin issue
- update packages from OBS multimedia:audio:libs repo, to be sure, whether the problem got fixed in the later versions
Takashi
Hi and thanks for the advice. Yes the issue is present only if a package "alsa-plugins" installed. The problem also occurs after the upgrade packages http://download.opensuse.org/repositories/multimedia:/libs/openSUSE_11.2/ . I also added information to the alsa bugtracker. https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4878
OK, just one more thing to be sure -- does this happen also with samplerate plugin? I suppose it's PPH specific, but it's not clear whether this comes from the PCM core side or just in the SRC side.
Also, did you see any similar error with other apps?
thanks,
Takashi
Takashi Iwai <tiwai <at> suse.de> writes:
OK, just one more thing to be sure -- does this happen also with samplerate plugin? I suppose it's PPH specific, but it's not clear whether this comes from the PCM core side or just in the SRC side.
You mean alsa-plugins-samplerate ? No this issue exists if alsa-plugins is installed. Installing or uninstalling alsa-plugins-samplerate does not change the situation.
Also, did you see any similar error with other apps?
thanks,
Takashi
There is unique to firefox. Is the only browser that makes sounds. The rest multithreaded audio output is working properly. For example i can watch two videos at once on youtube.
One more thing. Raymond says:
is flashsupport.c the source code of libflashplayer.so
http://labs.adobe.com/wiki/index.php/Flash_Player:Additional_Interface_Suppo...
[^]
It seem that there are some bugs in flashsupport.c
#14 0xb5ae6379 in snd_pcm_mmap_writei (pcm=0xb0c96ac0, buffer=0xaa1ae000,
size=940) at pcm_mmap.c:186
#15 0xb5ad415e in _snd_pcm_writei (size=<value optimized out>, buffer=<value
optimized out>, pcm=<value optimized out>) at pcm_local.h:516
#16 snd_pcm_writei (size=<value optimized out>, buffer=<value optimized out>,
pcm=<value optimized out>) at pcm.c:1247
#17 0xab01acd4 in ?? () from /usr/lib/browser-plugins/libflashplayer.so #18 0xab01af15 in ?? () from /usr/lib/browser-plugins/libflashplayer.so #19 0xb7e876e5 in start_thread (arg=0x0) at pthread_create.c:297 #20 0xb7e87600 in ?? () at pthread_create.c:216 from /lib/libpthread.so.0
at https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4878
Later I will send a bug in the adobe.
Thanks,
Anton
At Thu, 21 Jan 2010 11:19:24 +0000 (UTC), Anton Yashin wrote:
Takashi Iwai <tiwai <at> suse.de> writes:
OK, just one more thing to be sure -- does this happen also with samplerate plugin? I suppose it's PPH specific, but it's not clear whether this comes from the PCM core side or just in the SRC side.
You mean alsa-plugins-samplerate ? No this issue exists if alsa-plugins is installed. Installing or uninstalling alsa-plugins-samplerate does not change the situation.
You need to set up the configuration to use that plugin. Just installing doesn't enable the plugin setup. See README.
Also, did you see any similar error with other apps?
thanks,
Takashi
There is unique to firefox. Is the only browser that makes sounds. The rest multithreaded audio output is working properly. For example i can watch two videos at once on youtube.
How does firefox make sounds by itself? Since the bug is triggered when you enter some error texts in the search field, I guess this is causing a beep or whatever tone.
One more thing. Raymond says:
is flashsupport.c the source code of libflashplayer.so
http://labs.adobe.com/wiki/index.php/Flash_Player:Additional_Interface_Suppo...
[^]
It seem that there are some bugs in flashsupport.c
Quite possibly.
thanks,
Takashi
Takashi Iwai <tiwai <at> suse.de> writes:
You need to set up the configuration to use that plugin. Just installing doesn't enable the plugin setup. See README.
Thanks, after i added: defaults.pcm.rate_converter "samplerate" to ~/.asoundrc i got segmentation fault. I added crash-with-alsa-plugins-samplerate-enabled.txt to the alsa bugtracker.
How does firefox make sounds by itself? Since the bug is triggered when you enter some error texts in the search field, I guess this is causing a beep or whatever tone.
This is what I had in mind. Beep sound played by firefox. No other browser can not play sounds on events in my system.
thanks,
Anton.
'Twas brillig, and Anton Yashin at 21/01/10 11:19 did gyre and gimble:
One more thing. Raymond says:
is flashsupport.c the source code of libflashplayer.so
http://labs.adobe.com/wiki/index.php/Flash_Player:Additional_Interface_Suppo...
[^]
It seem that there are some bugs in flashsupport.c
#14 0xb5ae6379 in snd_pcm_mmap_writei (pcm=0xb0c96ac0, buffer=0xaa1ae000,
size=940) at pcm_mmap.c:186
#15 0xb5ad415e in _snd_pcm_writei (size=<value optimized out>, buffer=<value
optimized out>, pcm=<value optimized out>) at pcm_local.h:516
#16 snd_pcm_writei (size=<value optimized out>, buffer=<value optimized out>,
pcm=<value optimized out>) at pcm.c:1247
#17 0xab01acd4 in ?? () from /usr/lib/browser-plugins/libflashplayer.so #18 0xab01af15 in ?? () from /usr/lib/browser-plugins/libflashplayer.so #19 0xb7e876e5 in start_thread (arg=0x0) at pthread_create.c:297 #20 0xb7e87600 in ?? () at pthread_create.c:216 from /lib/libpthread.so.0
at https://bugtrack.alsa-project.org/alsa-bug/view.php?id=4878
Later I will send a bug in the adobe.
I'm not sure where Raymond sees flashsupport.c...
Could it be that you have libflashsupport.so (probably in /usr/lib/) on your system somewhere? If so you should try removing it (e.g. finding which package owns the file - rpm -qf <file> - and then removing the package) and see if the problem goes away.
It's used to extend Flash player and I guess it could be cocking up somewhere or other.
Col
Colin Guthrie <gmane <at> colin.guthr.ie> writes:
I'm not sure where Raymond sees flashsupport.c...
Could it be that you have libflashsupport.so (probably in /usr/lib/) on your system somewhere? If so you should try removing it (e.g. finding which package owns the file - rpm -qf <file> - and then removing the package) and see if the problem goes away.
It's used to extend Flash player and I guess it could be cocking up somewhere or other.
Col
Hi
#LANG=C find / | grep libflashsupport find: `/home/anton/.gvfs': Permission denied
May be this some internal features of flash player?
Thanks,
Anton.
'Twas brillig, and Anton Yashin at 21/01/10 15:04 did gyre and gimble:
Colin Guthrie <gmane <at> colin.guthr.ie> writes:
I'm not sure where Raymond sees flashsupport.c...
Could it be that you have libflashsupport.so (probably in /usr/lib/) on your system somewhere? If so you should try removing it (e.g. finding which package owns the file - rpm -qf <file> - and then removing the package) and see if the problem goes away.
It's used to extend Flash player and I guess it could be cocking up somewhere or other.
Col
Hi
#LANG=C find / | grep libflashsupport find: `/home/anton/.gvfs': Permission denied
May be this some internal features of flash player?
flashsupport is an external system. If your system has a libflashsupport.so file it will be loaded into the flashplayer. We used to use this some time ago when Flash's audio support was shocking (libflashsupport.so was used to provide pulseaudio support). Now flash works OK with alsa and PA via alsa, we no longer need this. Just wanted to ensure you weren't generating some odd code paths as a result of this file being present. It seems you are not and thus, I don't think flashsupport is an issue (tho as the internals of Flash are a black box I can't be 100% certain).
Cheers
Col
participants (3)
-
Anton Yashin -
Colin Guthrie -
Takashi Iwai