segfault

[Marcel Ebmer]

anonymous report:

I get this when trying to run it:

* created ringbuffer of size 5880
* created thread for sound sample acquisition
Segmentation fault

Any ideas?

Thanks

[Marcel Ebmer (idefixs)]

The segfault must be caused by either the sample acquisition thread - which is unlikeliy, as it includes extensive error checking - or by the main thread in vectorfield.c, which would be called next after the ringbuffer. No ideas yet.

[Marcel Ebmer (idefixs)]

anonymous details:

I can't do a launchpad report as you need a login. I just run the program like:

./fische

to get the seg fault. No other messages are shown except what I posted above. Any kiind of options result in the same error.

If I add strace before ./fische I get the last lines:

write(1, "* created ringbuffer of size 588"..., 34* created ringbuffer of size 5880
) = 34
mmap(NULL, 8392704, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = 0x7f9f0df31000
mprotect(0x7f9f0df31000, 4096, PROT_NONE) = 0
clone(* created thread for sound sample acquisition

+++ killed by SIGSEGV +++

Hope this helps.

[Andrea Veri (av)]

hi Marcel,

before packaging the latest release into Debian I would like to see this bug fixed to prevent any crash reports and to have a fully working application. What do you think?

[Marcel Ebmer (idefixs)]

Yep, you're right. Unfortunately, I am (so far) unable to reproduce this bug, and noone else has reported the same problem. Do you think it might be related to the reporter using debian unstable?

[Marcel Ebmer (idefixs)]

Problem found. When the ALSA "default" device is somehow not readable (which is to be classified as mis-configuration), a segfault killed fische. This error is now "recovered" into a clean exit().

[Andrea Veri (av)]

is this fix available on latest fische release?

[Andrea Veri (av)]

as marcel told me by email, this fix has been included into latest 3.0.3 release already uploaded.