Description of Problem: During the recording test audio fails, system has a segmentation fault. /usr/bin/rec: line 205: 25833 Segmentation fault sox $volume $arch_defines $fopts $device $fopts2 "$filename" $effects Begin Digital Audio Playback Test Digital Audio Playback Test Complete Begin Recording Test /etc/init.d/rhr-audio: line 109: 25730 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME /etc/init.d/rhr-audio: line 109: 25743 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME sox: Unable to set audio speed to 11000 (set to 11025) /etc/init.d/rhr-audio: line 109: 25756 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME sox: Unable to set audio speed to 11000 (set to 11025) /etc/init.d/rhr-audio: line 109: 25769 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME sox: Unable to set audio speed to 22000 (set to 22050) /etc/init.d/rhr-audio: line 109: 25782 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME sox: Unable to set audio speed to 22000 (set to 22050) /etc/init.d/rhr-audio: line 109: 25795 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME sox: Unable to set audio speed to 44000 (set to 44100) /etc/init.d/rhr-audio: line 109: 25808 Killed rec -d /dev/ $DEVICE -r $x`echo 000` -c $y $TESTFILENAME sox: Unable to set audio speed to 44000 (set to 44100) /usr/bin/rec: line 205: 25833 Segmentation fault sox $volume $arch_defines $fopts $device $fopts2 "$filename" $effects sox: Premature EOF on .wav input file sox: Premature EOF on .wav input file sox: Premature EOF on .wav input file sox: Premature EOF on .wav input file sox: Premature EOF on .wav input file sox: Premature EOF on .wav input file sox: Premature EOF on .wav input file sox: Can't open input file '44_2-test.out.wav': No such file or directory Recording Test Completed System also fails Midi test same message as bug 65876 on-board audio i810. Midi files play fine using kmidi. Begin Midi Audio Playback Test open /dev/sequencer: No such device Error playing midi sound. Midi Audio Playback Test Complete Version-Release number of selected component (if applicable): How Reproducible:100% Steps to Reproduce: 1. Run RHR7.3 Cert suite on WS530 with SBLive audio card. 2. 3. Actual Results: Expected Results: Pass without errors Additional Information:
Those segmentation faults aren't oopses I hope ?
No here is the error from the Digital Audio Playback Test. /usr/bin/rec: line 205: 25833 Segmentation fault sox $volume $arch_defines $fopts $device $fopts2 "$filename" $effects
We can get this to work by editing the rhr-audio file. In the record section, if you remark out "(sleep 5; killall -w -9 rec sox) >/dev/null 1>/dev/null 2>/dev/null", and then manually kill each record process (at each frequency/channel) with CTRL-C, the test will pass.
The EMU10K doesn't have hardware midi; kmidi is translating the midi files from midi to digial output using TiMidity++. rhr-midi requires /dev/sequencer to exist to run properly (ie hardware midi support). Because the EMU10K doesn't in hardware support midi it is not required to be tested. This doesn't solve the digital audio portion though. Would you like to run through this as if it were an rhr-interact script problem or leave it as kernel driver problem?
Does the i810 audio not have hardware midi support either (bugzilla #65876)? We get the same error on both: /dev/sequencer device does not exist.
Not as far as I know; most PCI audio adapters use software to emulate MIDI but don't actually support it in hardware.
No, there is no hardware midi support on any i810 hardware variant.
Can I asume that you're not getting the segmentation faults with the i810 or does that mirror as well?
The only i810 problem we have is the midi /dev/sequencer issue.
One other issue with i810 the cert test records and plays back all the files, but still reports the test as failing see below is this a bug in the script? sox: Sound card appears to only support 2 channels. Overriding format [FAILED]
Per Chris K, this is not a failure. The sound card only supports two channels and the cert test writes this into the log.
This is a different issue from the two-channel problem. This failure is actually a seg fault.
Do we know if this fails (ie seg faults); if record is run by hand instead of within the scripts? Also; do we know if other record apps fail in a similar way? The point here it to determine if it's a driver bug, a record bug or if something has changed within bash that is causing the scripts to fail.
Chris says this is because audio is only 2 channels. Cert test writes this in log.... This is not a failure but only a limitation of the hardware. ---- This Bugzilla should be closed as limitation of this specific hardware... not an error/failure.
A segmentation fault is a limitation of the hardware??
Closing. Does not occur with Milan.