Problem with ALSA and Flashplayer on Slackware 13.0

edorig · 02-20-2014, 04:57 PM

I have the following problem with ALSA on Slackware 13.0.
When I am watching a Flash video from a web site, sometimes
I start hearing a clicking sound while the video freezes,
and I have to close firefox to stop that clicking sound.

After that, the sound card (ATI IXP 400) seems to be in a strange state. Trying to play a .au file or a .wav file with aplay
produces a periodic noise, with a period of about 1/10th of a second.
Attempting to play the same .au file with cat file.au > /dev/audio
also gives a cyclic sound, but the period is about 2s. Sometimes
the playback jumps to a later part of the sound file, and start looping there, again with a 2s period.

with xmms file.mp3 no sound is played. If I seek to a different position, I hear sound for less than 1s then playback stops again.

Quote:

The output of speaker-test:

speaker-test 1.0.18

Playback device is default
Stream parameters are 48000Hz, S16_LE, 1 channels
Using 16 octaves of pink noise
Rate set to 48000Hz (requested 48000Hz)
Buffer size range from 2048 to 8192
Period size range from 1024 to 1024
Using max buffer size 8192
Periods = 4
was set period_size = 1024
was set buffer_size = 8192
0 - Front Left

Quote:

When doing cat file.au >/dev/audio,

/proc/asound/card0/pcm0p/sub0/*params

access: RW_INTERLEAVED
format: S16_LE
subformat: STD
channels: 2
rate: 8000 (8000/1)
period_size: 4096
buffer_size: 16384
OSS format: MU_LAW
OSS channels: 1
OSS rate: 8000
OSS period bytes: 4096
OSS periods: 4
OSS period frames: 4096
tstamp_mode: NONE
period_step: 1
avail_min: 1
start_threshold: 1
stop_threshold: 16384
silence_threshold: 4112
silence_size: 4112
boundary: 1073741824

It is still possible to adjust sound level using alsamixer, but this
does not help, and neither does using alsactl. Also using modprobe to
remove snd_atiixp and reinserting snd_atiixp does not solve the problem. I know that rebooting the computer solves the problem,
but is there a way to reinitialize the sound driver by echo 'something' into a special file under /proc or /sys ?

For now, I am trying to collect data from /sys/ that might help to
identify the problem.

aus9 · 02-21-2014, 02:08 AM

simple look at this site
http://www.adobe.com/software/flash/about/

if you have FF you have an old flash so consider trying google chrome

2) reboot to clear errors

aplay is not the best player around....YMMV I prefer vlc

edorig · 02-22-2014, 06:28 AM

I know that flashplayer 11 is now unsupported on Linux. But I have never read about the bug
I have encountered after various searches on Google, so it is unlikely to be known to Adobe.
I encounter that bug rather infrequently, generally as a result of watching videos on web sites
of TV channels (once with the french Canal Plus and twice with NBC). The bug is non-reproducible:
watching again the video after a reboot does not trigger the clicking sound anymore.
My guess is that it is caused by some latency in the flux of data between the server and my computer
that results in a buffer getting empty or a pointer containing some incorrect address. The flashplayer
plugin is not detecting the problem, and instead of freezing temporarily the video playback, it sends
incorrect data to the sound card, creating the clicking sound, and then disabling it.
For me, the problem is not so much the flashplayer bug but the fragility of the ALSA sound architecture.
Once the bug has disabled the sound card, I have no solution to reinitialize it properly except rebooting the computer. I consider that kind of flaw unacceptable in a Unix type operating system. There ought to be
a way to make the kernel reinitialize the sound card either by some ioctl or by writing to a /proc or /sys file. Maybe this problem has been fixed in more recent versions of the Linux kernel (I am still using a 2.6
kernel) and the solution is to compile a LTS 3.x kernel compatible with Slackware 13.0 or to recompile the kernel with OSSv4 instead of ALSA. But if that it the case, I would like some pointers to which kernels/OSS modules are fixing the problem.