Skip to contentSkip to conversation

trident driver bug

DISCUSSION Paul Davis 3 participants 4 messages Mar 19, 2003

lore
Share & tools
lore

Paul Davis

author
#0 lore raw

there still appears to be a bug in the trident driver. i have been running jack with a period size of 1024, and it works fine. when i change that to 4096 frames, the kernel does a hard oops, with the oops occuring in enable_hlt(). the rest of the trace doesn't make much sense, and i cannot feed it to ksymoops because the machine locks hard. what can i do to make it eas(y|ier) to debug?

-------------------------------------------------------
This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for?
http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en

#1 lore raw
Quoted 6 lines · Paul Davis

there still appears to be a bug in the trident driver. i have been running jack with a period size of 1024, and it works fine. when i change that to 4096 frames, the kernel does a hard oops, with the oops occuring in enable_hlt(). the rest of the trace doesn't make much sense, and i cannot feed it to ksymoops because the machine locks hard. what can i do to make it eas(y|ier) to debug?

I cannot reproduce it here. Could you try the same command?

test:/home/perex # aplay -v -F 92870 -B 185750 audio/audio_10.wav Playing WAVE 'audio/audio_10.wav' : Signed 16 bit Little Endian, Rate 44100 Hz, Stereo Plug PCM: Hardware PCM card 0 'Trident TRID4DWAVENX' device 0 subdevice 0

Its setup is:
stream : PLAYBACK
access : RW_INTERLEAVED
format : S16_LE
subformat : STD
channels : 2
rate : 44100
exact rate : 44100 (44100/1)
msbits : 16
buffer_size : 8192
period_size : 4096
period_time : 92879
tick_time : 10000
tstamp_mode : NONE
period_step : 1
sleep_min : 0
avail_min : 4096
xfer_align : 4096
start_threshold : 8192
stop_threshold : 8192
silence_threshold: 0
silence_size : 0
boundary : 1073741824
Aborted by signal Interrupt...

Jaroslav

-----
Jaroslav Kysela <[email protected]>
Linux Kernel Sound Maintainer
ALSA Project, SuSE Labs

-------------------------------------------------------
This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for?
http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en

#2 lore raw
In reply to Paul Davis
Quoted 7 lines · Paul Davis

there still appears to be a bug in the trident driver. i have been running jack with a period size of 1024, and it works fine. when i change that to 4096 frames, the kernel does a hard oops, with the oops occuring in enable_hlt(). the rest of the trace doesn't make much sense, and i cannot feed it to ksymoops because the machine locks hard. what can i do to make it eas(y|ier) to debug?

... just having a short view to the sourcecode, I saw that the maxbuffer for playback is bigger than the rest (spdif, capture,..). Maybe this is a bug, but I don't have either the hardware nor the documentation for the trident. So it's possible that I'm totally wrong here...

... it's just an idea to change that buffersize to the smaller buffersize like capture/spdif (128*1024) instead of original size (256*1024).

good luck!
martin

-------------------------------------------------------
This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for?
http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en

#3 lore raw
Quoted 9 lines · Martin Langer
Quoted 7 lines · Paul Davis

there still appears to be a bug in the trident driver. i have been running jack with a period size of 1024, and it works fine. when i change that to 4096 frames, the kernel does a hard oops, with the oops occuring in enable_hlt(). the rest of the trace doesn't make much sense, and i cannot feed it to ksymoops because the machine locks hard. what can i do to make it eas(y|ier) to debug?

Quoted 7 lines

... just having a short view to the sourcecode, I saw that the maxbuffer for playback is bigger than the rest (spdif, capture,..). Maybe this is a bug, but I don't have either the hardware nor the documentation for the trident. So it's possible that I'm totally wrong here...

... it's just an idea to change that buffersize to the smaller buffersize like capture/spdif (128*1024) instead of original size (256*1024).

Trident's NX playback hardware supports the 256kB ring buffer. It seems like some weird bug in the DMA allocation routines.

Jaroslav

-----
Jaroslav Kysela <[email protected]>
Linux Kernel Sound Maintainer
ALSA Project, SuSE Labs

-------------------------------------------------------
This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for?
http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en