Randomly occuring frequent crashes during playback

• Jul 7, 2017 - 02:19

Over the past couple of days, MuseScore has started crashing unpredictably but frequently during playback. When I restart and load my score after there's been a crash, playback works two or three times, then MuseScore crashes again. It doesn't matter where in the score I start the playback.

I run a stock Kubuntu 16.04 (Xenial) with KXstudio installed. The audio subsystem is pure ALSA. MuseScore is set to use JACK. I use an external sampler rather than the internal synthesizer. I haven't installed anything new or upgraded for two weeks or so.

grep "musescore" /var/log/syslog spits out the following:

Jul 6 14:58:25 aurelius kernel: [27314.953145] mscore[13202]: segfault at 20 ip 00000000007fa2b3 sp 00007ffffc6cf550 error 4 in mscore[400000+164c000]
Jul 6 15:03:52 aurelius kernel: [27642.329085] traps: mscore[13365] general protection ip:7f3e58184a73 sp:7ffe6ad65678 error:0 in libstdc++.so.6.0.21[7f3e580e0000+172000]
Jul 6 15:07:07 aurelius kernel: [27837.012730] traps: mscore[13505] general protection ip:7fdb155cea73 sp:7ffdcc3158b8 error:0 in libstdc++.so.6.0.21[7fdb1552a000+172000]
Jul 6 15:14:00 aurelius kernel: [28250.001626] traps: mscore[13580] general protection ip:7f433d404a73 sp:7fffcf9557a8 error:0 in libstdc++.so.6.0.21[7f433d360000+172000]
Jul 6 15:18:03 aurelius kernel: [28492.490803] mscore[13653]: segfault at 18 ip 00007fc4deee3a82 sp 00007ffc147bfc28 error 4 in libstdc++.so.6.0.21[7fc4dee3f000+172000]
Jul 6 15:19:56 aurelius kernel: [28605.663948] mscore[13777]: segfault at 7fb2d92e ip 00007fb2d7e86a73 sp 00007ffea327faf8 error 4 in libstdc++.so.6.0.21[7fb2d7de2000+172000]
Jul 6 15:21:45 aurelius kernel: [28714.584757] traps: mscore[13840] general protection ip:7ff0b5b72a73 sp:7ffe988e15a8 error:0 in libstdc++.so.6.0.21[7ff0b5ace000+172000]
Jul 6 15:22:52 aurelius kernel: [28781.966284] mscore[13881]: segfault at 28 ip 00007fe8fab23a73 sp 00007ffd4093da38 error 4 in libstdc++.so.6.0.21[7fe8faa7f000+172000]
Jul 6 17:21:40 aurelius kernel: [35910.010616] mscore[13900]: segfault at 1000000011 ip 00007f3bd9785a73 sp 00007ffc0fda2818 error 4 in libstdc++.so.6.0.21[7f3bd96e1000+172000]
Jul 6 17:22:43 aurelius kernel: [35972.541306] mscore[16228]: segfault at 18 ip 00007ff834cc5a82 sp 00007fffbfea78e8 error 4 in libstdc++.so.6.0.21[7ff834c21000+172000]
Jul 6 17:25:18 aurelius kernel: [36127.766314] mscore[16253]: segfault at 18 ip 00007f8c8f287a82 sp 00007ffe6fc19e98 error 4 in libstdc++.so.6.0.21[7f8c8f1e3000+172000]

Any idea what's causing this? My libstdc++ is the stock Kubuntu libstdc++6.


Comments

Which version of MuseScore? I had problems with MS 2.0.3 on xubuntu 16.04 following a software update but switched to the AppImage of MS 2.0.3 and it seems pretty solid.

In reply to by underquark

Not the sort of answer I need, but thanks.

The AppImage fails with:

noJACK ERROR: client linked with incompatible libjack version.
noJACK ERROR: could not attach to JACK server
JackAudio()::init(): failed, status 0x421
no JACK server found
no audio driver found
Cannot start I/O
sequencer init failed

It would be useful to include the correct JACK version on stderr since without it, the information is useless.

Meantime, 2.03 on my system crashed 40 or 50 times yesterday. Every second or third change I made to the score, I had to restart and reload in order to get playback.

Someone knowledgeable--please help me out. I can't get any work done. Is the playback crash likely coming from MuseScore, from libstdc++6, or from something KXstudio has mangled? I don't even know who to contact to file a bug report. I was hoping here, but that's clearly a no-starter.

In reply to by kuwitt

It's the distro package of 2.03 for Ubuntu 16.04 that started crashing during playback. After several months of working fine. I've been tussling with MuseScore non-stop for five years now, but this problem is completely beyond me, and, if I may whine, it's more than a PITA. It's a work killer. I can't get any composing done and I'm halfway into a project.

I see there's a snap of 2.1. I'm going to give it a shot. Will report back.

In reply to by Nicolas

I'm not sure I have any simple scores, however when I was working on the last few orchestral pieces, I experienced the same problem: hit the playback button, watch it crash. I can't say how to reproduce the bug because it occurs at different locations in the score, but never the same ones. And it doesn't happen on command. When the moon is in the seventh house and Jupiter aligns with Mars, I can go for several hours without crashing. Then, boom! Crashing, crashing, crashing. About the only helpful piece of information I can give is that I rarely experience the problem when starting a score. And endeavouring to reproduce it even in a problem score is impossible, since, as I say, when the gods are on my side, I can go for quite a while before it occurs.

In reply to by Peter Schaffter

I don't have really an idea and there are only a few ressources for the "libstdc++.so.6.0.21" error. Some of them indicate to a conflict with an installed Nvidia graphic driver. So, if you have a Nvidia graphic card and the Nvidia driver intalled, it's maybe useful to check if the same error still exist, if you change it to the Nouveau driver.

In reply to by Peter Schaffter

No joy. The problem persists with the nouveau drivers. When it first appeared, it seemed to coincide with systemd(1) starting the CUPS scheduler or cups-browsed(8). I don't need printing facilities, so I removed CUPS entirely. That, too, did nothing.

Another avenue I explored was installing libstdc++6 from the Ubuntu repo instead of using the version shipped with KXstudio. Again, no joy. I've also tried to run MuseScore with the --debug flag, but all that happens is it core dumps whenever I try to open a score.

In reply to by Nicolas

Already done. It's one of the first things I do with a fresh MuseScore install.

You might be onto something, though. I tried an experiment last night. After making a change in the score, no matter how trivial, I hit Ctrl-s, waited a few seconds, repeated the operation, then hit Esc several times. After another few seconds, I'd start playback. I was able to get through about an hour's worth of work without crashing.

It could just be coincidence. The crashes so far haven't been routinely reproducible and I may have just been lucky. If not, my suspicion is that they're related to how long I wait before starting playback after making a change. All those extra Ctrl-s and Esc were me taking out my frustration on the keyboard. The key thing, surely, is the time factor.

I'll pursue this today and report back.

Do you still have an unanswered question? Please log in first to post your question.