Java + ALSA - OSS = rage

[McLink]

As a result of PaulBredbury's posts in the "Flash: no sound" thread, I decided it might be best to drop OSS emulation altogether. Lo and behold, I can have a flash video open and Amarok will still work regardless. Unfortunately, nothing is that simple; the royal screwup known as Sun Java fails to play nice. I play a game called RuneScape, which is Java-based; it stretches Java's capabilities to its limits, and as such anything except Sun's VM tends to go nards-up with it.

So, the problem now is that when I play RuneScape, its sound starts, plays for about five seconds, then dies for about five seconds, then starts playing again - and so on. In the mean time, other applications using ALSA stop producing sound as well. Mind, this is using pure ALSA, with no OSS support in the kernel whatsoever. This happens with both sun-jdk-1.6.0.20 and sun-jdk-1.7.0.0_alpha89 from java-experimental. icedtea6-bin-1.7.2 also appears to work with RuneScape, but using that, I get no sound whatsoever (although other ALSA-using programs continue to produce sound, unlike with sun-jdk-*). I have read a tonne about Java and ALSA not playing nice, but any solution that possibly works for Java so far screwed up the sound of normal ALSA applications.

Some information:

Code: Select all

$ eix -I* --format '<installedversions:EQNAMEVERSION>' alsa
=media-libs/alsa-lib-1.0.22-r1
=media-plugins/alsa-plugins-1.0.22
=media-plugins/gst-plugins-alsa-0.10.25
=media-sound/alsa-headers-1.0.22
=media-sound/alsa-tools-1.0.22
=media-sound/alsa-utils-1.0.22-r1

$ emerge --info
Portage 2.2_rc67 (default/linux/amd64/10.0, gcc-4.3.4, glibc-2.10.1-r1, 2.6.33-gentoo-r1 x86_64)
=================================================================
System uname: Linux-2.6.33-gentoo-r1-x86_64-AMD_Phenom-tm-_9850_Quad-Core_Processor-with-gentoo-2.0.1
Timestamp of tree: Tue, 27 Apr 2010 14:45:01 +0000
ccache version 2.4 [enabled]
app-shells/bash:     4.0_p37
dev-java/java-config: 2.1.10
dev-lang/python:     2.4.6, 2.5.4-r4, 2.6.4-r1, 3.1.1-r1
dev-python/pycrypto: 2.1.0_beta1
dev-util/ccache:     2.4-r7
dev-util/cmake:      2.6.4-r3
sys-apps/baselayout: 2.0.1
sys-apps/openrc:     0.6.0-r1
sys-apps/sandbox:    2.2
sys-devel/autoconf:  2.13, 2.64
sys-devel/automake:  1.4_p6-r1, 1.7.9-r2, 1.8.5-r4, 1.9.6-r3, 1.10.3, 1.11.1
sys-devel/binutils:  2.18-r3
sys-devel/gcc:       3.4.6-r2, 4.2.4-r1, 4.3.4
sys-devel/gcc-config: 1.4.1
sys-devel/libtool:   2.2.6b
virtual/os-headers:  2.6.30-r1
ACCEPT_KEYWORDS="amd64"
ACCEPT_LICENSE="*"
CBUILD="x86_64-pc-linux-gnu"
CFLAGS="-O2 -march=native -pipe -ggdb"
CHOST="x86_64-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/share/X11/xkb /usr/share/config /var/lib/hsqldb"
CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d /etc/env.d/java/ /etc/eselect/postgresql /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/sandbox.d /etc/splash /etc/terminfo /etc/texmf/language.dat.d /etc/texmf/language.def.d /etc/texmf/updmap.d /etc/texmf/web2c"
CXXFLAGS="-O2 -march=native -pipe -ggdb"
DISTDIR="/usr/portage/distfiles"
EMERGE_DEFAULT_OPTS="--keep-going"
FEATURES="assume-digests ccache distlocks fixpackages news nostrip parallel-fetch preserve-libs protect-owned sandbox sfperms splitdebug strict unmerge-logs unmerge-orphans userfetch"
GENTOO_MIRRORS="http://ftp.heanet.ie/pub/gentoo/ http://mirror.ovh.net/gentoo-distfiles/ http://ftp.snt.utwente.nl/pub/os/linux/gentoo http://de-mirror.org/distro/gentoo/ http://mirror.qubenet.net/mirror/gentoo/ http://mirror.muntinternet.net/pub/gentoo/ ftp://gentoo.tiscali.nl/pub/mirror/gentoo/ http://linux.rz.ruhr-uni-bochum.de/download/gentoo-mirror/ http://gentoo.tiscali.nl/ ftp://mirror.muntinternet.net/pub/gentoo/"
LANG="en_GB.UTF-8"
LDFLAGS="-Wl,-O1"
LINGUAS="en_GB en nl de fr es it pl sv el la fi ru"
MAKEOPTS="-j5"
PKGDIR="/usr/portage/packages"
PORTAGE_CONFIGROOT="/"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/usr/local/crossdev-overlay /var/lib/layman/science /var/lib/layman/sunrise /var/lib/layman/pengdev /var/lib/layman/kde-testing /var/lib/layman/java-overlay /var/lib/layman/gnome /usr/local/portage"
SYNC="rsync://rsync.europe.gentoo.org/gentoo-portage"
USE="3ds 7zip X Xaw3d aac aalib accessibility acl acpi addressbook adns allegro alsa amd64 amrnb amrwb apache2 applet ares artworkextra aspell audiofile avahi bash-completion bcp bdf berkdb big-tables binfilter bjam blender-game bookmarks branding bugzilla bzip2 cairo cal3d calendar caps captury ccache cdaudio cdda cddb cdparanoia cdr cegui cgi chardet chipcard chroot cjk cli cpudetection cracklib crypt cscope css ctype cups curl curlwrappers cursors cvs cxx darcs dbus dedicated designer-plugin device-mapper devil dhcp directfb djvu dmi dmx doc double-precision dri dssi dts dv dvd dvdr dvdread dvi elf encode epydoc esd espeak examples exif expat export extensions extraengine extrafilters extras fat fax fbcon fbcondecor festival ffmpeg filter flac flash flatfile fontconfig fontforge foomaticdb force-cgi-redirect fortran freetts ftp fts3 fuse gadu gcj gcrypt gd gdbm gdm gdu gedit geoip geolocation ggi gif gimp git glib glitz glut gmail gmp gnome gnome-keyring gnuplot gnutls gold gpg gphoto2 gpm grammar graphviz groupwise gs gstreamer gtk gtkhtml gucharmap guile gzip hash hbci hddtemp hdf5 hfs howl-compat htmlhandbook i18n icons iconv icu id3tag ieee1394 ifp image imagemagick imap imlib inifile inkjar iostats ipod ipv6 jabber jai java java5 java6 javamail javascript jce jfs jikes jingle jms joystick jpeg jpeg2k json justify kde kdm kdrive kig-scripting kpathsea kqemu latex lcms libcaca libedit libnotify libsamplerate lightning lilypond lm_sensors logitech-mouse lua lua-cairo lua-imlib lzma lzo mad mailwrapper maps math mbox md5sum mdnsresponder-compat meanwhile menubar mhash midi mikmod mjpeg mmx mng modplug modules motif mozdevelop mozdom mp3 mp3tunes mp4 mpeg mplayer msn mtp mudflap multilib multiuser musepack musicbrainz mysql mysqli nas nautilus ncurses netboot network networking new-login nextaw nfs njb nls nptl nptlonly nsplugin ntfs nvidia objc objc++ objc-gc ode odk offensive ofx ogg openal openexr opengl openmp openntpd ots overlays pam pango paste64 pch pcre pdf pdo perl phonon php player plotutils png pop posix postgres postscript povray ppds pppd prediction pyste python qq qt3 qt3support qt4 quicktime quotas quotes radio rar raw rdesktop readline reflection regex reiser4 reiserfs replytolist rpc rss ruby sametime scanner sdl seamonkey secure-delete semantic-desktop session silc skey skins slang slp smartcard smi smp sndfile snmp sockets socks5 solver sound source sox speex spell spl sql sqlite srt sse sse2 ssl stars startup-notification stream subversion svg svgz swat symlink sysfs syslog system-sqlite talkfilters tcl tcpd templates tetex tga theora thesaurus threads threadsafe thumbnail tidy tiff timezone timidity tk tokenizer toolbar tools truetype tweak-mode unicode urandom usb v4l v4l2 vcd vdr verse vim vim-pager vim-syntax vim-with-x visualization vlm vorbis wavpack wcwidth weather-metar weather-xoap webkit wma wxwidgets wxwindows x264 xattr xchatdccserver xcomposite xerces-c xfs xft xine xinerama xinetd xml xmlreader xmlrpc xmlwriter xorg xosd xpm xprint xrandr xscreensaver xsl xterm-color xulrunner xv xvid xvmc zephyr zeroconf zip zlib zsh-completion" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mmap_emul mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions alias auth_basic auth_digest authn_anon authn_dbd authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache dav dav_fs dav_lock dbd deflate dir disk_cache env expires ext_filter file_cache filter headers ident imagemap include info log_config logio mem_cache mime mime_magic negotiation proxy proxy_ajp proxy_balancer proxy_connect proxy_http rewrite setenvif so speling status unique_id userdir usertrack vhost_alias" CAMERAS="ptp ptp2 canon" ELIBC="glibc" INPUT_DEVICES="keyboard mouse evdev wacom" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="en_GB en nl de fr es it pl sv el la fi ru" RUBY_TARGETS="ruby18" USERLAND="GNU" VIDEO_CARDS="nvidia" 
Unset:  CPPFLAGS, CTARGET, FFLAGS, INSTALL_MASK, LC_ALL, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS

Any help is very much appreciated! Thanks in advance!

[chithanh]

Maybe you can start it with the aoss wrapper?

[PaulBredbury]

drop OSS emulation altogether. Lo and behold

So obviously, an app was using OSS emulation. Did you actually try to figure out which one?

fuser -v /dev/snd/* /dev/dsp*

I keep posting that command, that *shows* you which apps are using ALSA or OSS. So, did you try, you know, actually running it, with the differing audio-using apps open and playing sound?

I keep seeing people have audio problems, who then either give up or have the problem "magically" fix itself - this is ridiculous.

[Iguana8649]

Just checking, you don't have a pulseaudio flag screwing around your system do you? I don't remember if pulseaudio is enabled on the desktop profile by default and you don't have any "-pulseaudio" there.

TL;DR: using ALSA on a system using pulseaudio caused sound to fail every now and then on mplayer (on Ubuntu at least).

[McLink]

PaulBredbury, I ran that command back when I still had OSS emulation, but I ran it for diagnosing Flash, which did use the DSP. Then I got rid of OSS support and Flash started playing nice, but Java didn't. I actually got rid of OSS emulation to get Flash working.

For Java, I always had to use aoss to get it working, but aoss is kludgy, and has caused RuneScape to crash on numerous occasions. For reasons beyond me, aoss chromium now even gives an error stating libaoss.so can't be preloaded. The chromium binary, as well as the libraries it pulls in dynamically are all 64-bit, so that probably isn't the cause of that either. In any case, for some reason, aoss has always caused instability in the programs that used it.

Pulseaudio isn't installed and the USE-flag isn't active, and I'm not using the desktop profile anyway.

When RuneScape is running in chromium and giving the on-off-on-off effect, the fuser command shows nothing (other than that /dev/dsp* doesn't exist). Starting an application I know uses ALSA will then produce the classic "device or resource busy", because Java locks ALSA's hw:0,0.

[PaulBredbury]

Then I got rid of OSS support and Flash started playing nice

Well, that doesn't make any sense, because FLASH USES ALSA. So, you are barking up the wrong tree, and blaming the wrong app.

Java locks ALSA's hw:0,0.

I suggest you file a bug report with Sun about that. Hint, hint

[McLink]

I know it does, or at least can do - but it still locked /dev/dsp for me. How, or why, I don't know, but what I do know is that e.g. a YouTube video and Amarok were mutually exclusive before I removed OSS support, and aren't any more with it removed. I'll see what happens if I reenable OSS support in the form of kernel modules - but that means it'll be a while before I can get any results, because I don't want to reboot right now.

And yeah, maybe I am barking up the wrong tree - but I lack the knowledge on sound support in Linux to know for sure which tree is the right one, so I just go with whatever I observe.

All I know is that it would be great if all applications could just happily cooperate by using ALSA's default device instead of locking /dev/dsp or hw:0,0. And maybe they can, but I don't know how - hence this thread.

Java locks ALSA's hw:0,0.

I suggest you file a bug report with Sun about that. Hint, hint

...Yeah. Note how Sun conveniently closed that bug while the comments state it still exists.

[PaulBredbury]

How, or why, I don't know

No-one ever does, and the only ones who reckon that flash is to blame are the newbies, and they never provide a shred of meaningful evidence. I've seen many threads like this, especially on the Arch Linux forums of late. It's gotten to ridiculous proportions of hearsay vs facts.

it would be great if all applications could just happily cooperate

Agreed. See alternative dmix, which needs testing by people with a brain and spare time.

Note how Sun conveniently closed that bug

Then create a new bug - fight stupidity with perseverance

Sounds like a re-run of the Flash and Skype stupidity.

[McLink]

I'll check out the monster dmixer and (maybe) file a bug report later. In the mean time, here is proof that Flash does in fact lock /dev/dsp on my machine:

Code: Select all

# fuser -v /dev/snd* /dev/dsp*
                     USER        PID ACCESS COMMAND
/dev/dsp:            link       4456 F.... firefox

That's with a YouTube video running in the background (no libflashsupport or nspluginwrapper installed, mind you).

If sound is already playing through ALSA in the background, e.g. with sox's play, Flash just mixes it into the existing stream and _does_ play nice. It appears Flash will play nice so long as ALSA was holding a stream open when the first Flash object is loaded, and it will continue doing so even after the original ALSA stream has finished. In other words, Flash tries to be greedy first, and failing that (because the DSP is inaccessible), it does play nice.

Even more strangely, Java now continues its weird stuttering (on-off-on-off) even though the OSS modules are loaded. It doesn't lock /dev/dsp, but it does block sound from all other applications if it is run first. If sound is already playing, Java simply does not produce sound at all.

aoss chromium now produces the following error:

Code: Select all

ERROR: ld.so: object 'libaoss.so' from LD_PRELOAD cannot be preloaded: ignored.

Strangely, running any of several other programs (such as xclock, which doesn't produce sound, and sox's play, which uses ALSA) through aoss doesn't produce that error.

[PaulBredbury]

no libflashsupport or nspluginwrapper installed

Type about:plugins into your browser and take a look. 'Cos I don't believe you

That will also show the version of flash, which should show: Shockwave Flash 10.0 r45

Another possible explanation is that some other program was blocking ALSA when flash *started*, so flash had to use OSS, and then the other program closed. So run the fuser command immediately before, during, and after.

[McLink]

no libflashsupport or nspluginwrapper installed

Type about:plugins into your browser and take a look. 'Cos I don't believe you

That will also show the version of flash, which should show: Shockwave Flash 10.0 r45

Another possible explanation is that some other program was blocking ALSA when flash *started*, so flash had to use OSS, and then the other program closed. So run the fuser command immediately before, during, and after.

Quoth the raven, "Schockwave Flash 10.0 r45".

Before, during and after (scripted):

Code: Select all

# while true; do echo "At $(date):"; fuser -v /dev/snd* /dev/dsp*; pgrep firefox; echo "----------"; sleep 5; done
At Fri May  7 13:19:41 CEST 2010:
----------
At Fri May  7 13:19:46 CEST 2010:
18314
----------
At Fri May  7 13:19:51 CEST 2010:
18314
----------
At Fri May  7 13:19:56 CEST 2010:
18314
----------
At Fri May  7 13:20:01 CEST 2010:
18314
----------
At Fri May  7 13:20:06 CEST 2010:
18314
----------
At Fri May  7 13:20:11 CEST 2010:
18314
----------
At Fri May  7 13:20:16 CEST 2010:
18314
----------
At Fri May  7 13:20:21 CEST 2010:
18314
----------
At Fri May  7 13:20:26 CEST 2010:
                     USER        PID ACCESS COMMAND
/dev/dsp:            link      18314 F.... firefox
18314
----------
At Fri May  7 13:20:32 CEST 2010:
                     USER        PID ACCESS COMMAND
/dev/dsp:            link      18314 F.... firefox
18314
----------
At Fri May  7 13:20:37 CEST 2010:
                     USER        PID ACCESS COMMAND
/dev/dsp:            link      18314 F.... firefox
18314
----------
At Fri May  7 13:20:42 CEST 2010:
                     USER        PID ACCESS COMMAND
/dev/dsp:            link      18314 F.... firefox
18314
----------
At Fri May  7 13:20:47 CEST 2010:
----------
At Fri May  7 13:20:52 CEST 2010:
----------
At Fri May  7 13:20:57 CEST 2010:
----------
^C

[PaulBredbury]

Beats me But it's nice to see some decent investigative work, for a change

I suggest you try the latest beta. With it, about:plugins shows: Shockwave Flash 10.1 r53

See if that uses ALSA, like it's supposed to

[McLink]

Mhh, that's 32-bit only as far as I can tell.

In any case, it's Java that bugs me the most. Judging by what I now know about Flash, I can work around it by playing a looped null audio stream through ALSA in the background, which should be almost trivial to set up. Then, at least, I can keep OSS emulation for debugging and for those applications that really need it.

From what I figure, though, the only way to get Java to use ALSA reliably is to somehow make hw:0,0 and plughw:0,0 virtual sound cards that instead route to dmix. Is there any way to do that with a "simple" .asoundrc (as opposed to hacking ALSA or writing a kernel module)? None of the various workarounds I've found on Google (such as this one) seem to work.

[PaulBredbury]

looped null audio stream

LOL, as the kiddies say

I suggest you try contacting the Flash devs - their audio code to choose between ALSA and OSS must be buggy

that instead route to dmix

You would need to hack the ALSA source code.

[McLink]

looped null audio stream

LOL, as the kiddies say

Mhh, when I wrote that, I was thinking of writing a C program to stream '\0's to ALSA in an infinite loop - hence the terminology. In the end I decided to go with an easier way;

Code: Select all

#!/bin/bash
renice -n 18 -p $$ >/dev/null 2>&1
ionice -c 2 -n 7 -p $$ >/dev/null 2>&1

while true
do
	/usr/bin/play -V0 /home/link/stuff/silence.ogg >/dev/null 2>&1 &
	sleep 9
done

Same result, slightly different path to get there. No less kludgy, of course, but it actually works, which is fair enough.

that instead route to dmix

You would need to hack the ALSA source code.

Dang. Well then, I suppose I'll add contacting Sun and Adobe to my to-do list. Once more a reminder of why closed-source software sucks arse.

[PaulBredbury]

Do you have multiple soundcards, with an incomplete ALSA config in /etc/modprobe.d/ ?

I'm thinking maybe flash has a totally screwed-up method of choosing which soundcard to use, if ALSA gives it any choice whatsoever.

[McLink]

I have only one sound card - which technically isn't even a card; it's just the on-board audio controller of my motherboard.

The 'card':

Code: Select all

$ cat /proc/asound/cards 
 0 [SB             ]: HDA-Intel - HDA ATI SB
                      HDA ATI SB at 0xf9ef4000 irq 16

[PaulBredbury]

Grasping at straws, try playing with FLASH_ALSA_DEVICE, e.g.:

Code: Select all

export FLASH_ALSA_DEVICE=plug:dmix
firefox &

I found this from:

Code: Select all

strings libflashplayer.so | grep -i alsa

[PaulBredbury]

Flash is very particular about buffer_size it seems. Please try improved dmix.

[McLink]

I've given up on getting Flash to play nice. 64-bit Flash currently isn't supported by Adobe anymore (and old Flash versions don't work with new Firefox versions), so nspluginwrapper has to be used, which is shit.

Flash itself is shit, and has been for as long as I remember. Java applets are also shit, but at least they're not as common. Flash, on the other hand, is plastered over, oh, 80% of the interblag, or so it appears.