core dumps from bitcoin-qt 0.8.2

[lano1106]

It happens every five/ten minutes. Here is what is in my journal log file. I will recompile it without striping out symbols to have a nice call stack but in the meantime, if it is a known issue, please point me to possible workaround:

Jun 10 22:51:13 Wailaba2 kernel: bitcoin-net[5370]: segfault at 10 ip 00007f21b1cb5d31 sp 00007f216d3d39a0 error 6 in libc-2.17.so[7f21b1c3c000+1a3000]
Jun 11 00:07:09 Wailaba2 kernel: bitcoin-net[5980]: segfault at 10 ip 00007f7e25852d31 sp 00007f7dd7feea80 error 6 in libc-2.17.so[7f7e257d9000+1a3000]
Jun 11 00:36:17 Wailaba2 kernel: bitcoin-net[6674]: segfault at 10 ip 00007f66675a3d31 sp 00007f6631393a80 error 6 in libc-2.17.so[7f666752a000+1a3000]
Jun 11 00:46:53 Wailaba2 kernel: bitcoin-net[6815]: segfault at 10 ip 00007f5ec29aad31 sp 00007f5e79ddf9a0 error 6 in libc-2.17.so[7f5ec2931000+1a3000]
Jun 11 00:53:49 Wailaba2 kernel: bitcoin-net[6929]: segfault at 12 ip 00007fed00515d31 sp 00007fecb27eb9a0 error 6 in libc-2.17.so[7fed0049c000+1a3000]
Jun 11 00:56:00 Wailaba2 kernel: bitcoin-net[7000]: segfault at 10 ip 00007fdd80d3cd31 sp 00007fdd45feaa50 error 6 in libc-2.17.so[7fdd80cc3000+1a3000]

[kodo]

Update your bitcoinQT client.

[lano1106]

interesting suggestion.

I was under impression that 0.8.2 was the latest and greatest. Anyway, my investigation has advanced a little bit.:

#0  0x00007f11d50df3d1 in _int_malloc () from /usr/lib/libc.so.6
(gdb) where
#0  0x00007f11d50df3d1 in _int_malloc () from /usr/lib/libc.so.6
#1  0x00007f11d50e261a in calloc () from /usr/lib/libc.so.6
#2  0x00007f11d8a2d340 in ?? () from /usr/lib/trickle/trickle-overload.so
#3  0x00007f11d8a2d922 in select () from /usr/lib/trickle/trickle-overload.so
#4  0x00000000004e3926 in ThreadSocketHandler() ()
#5  0x00000000004e4a3d in void TraceThread<void (*)()>(char const*, void (*)()) ()
#6  0x00007f11d7337a09 in ?? () from /usr/lib/libboost_thread.so.1.53.0
#7  0x00007f11d7115b52 in start_thread () from /usr/lib/libpthread.so.0
#8  0x00007f11d515893d in clone () from /usr/lib/libc.so.6

If you core dump deep inside libc while allocating memory, that can only mean 1 thing:

Heap corruption

Will run valgrind on bitcoin-qt to see what it finds. If valgrind finds something easily, I'll be a bit surprised as this should be a mandatory check before bitcoin dev team release a new version.

[lano1106]

I fixed my core dumps.

The problem was in trickle that I am using to stop bitcoin-qt to upload 20-30 GB of data daily!

http://monkey.org/~marius/pages/?page=trickle
https://wiki.archlinux.org/index.php/Trickle

If there are other trickle users here, you are likely to experience the same core dumps with bitcoin-qt 0.8.2, you can find a description of the problem and a patch file to fix the core dumps here:

https://bugs.archlinux.org/task/35872

If this helps you, BTC donations are welcome