Strange error - bitcoind terminated #8196

Hi all,

I just had this error - new to me:

*** stack smashing detected ***: bitcoind terminated

..this is the latest repo running on Ubuntu 64bit - any ideas?

Can you post the debug.log and maybe the time when the "stack smashing" was detected?

The "stack smashing" error message was only in my console, there's nothing mentioned in my logs though. Here are the logs from when the client Bitcoin version v0.12.99.0-3e4cf8f terminated:

2016-06-12 23:44:56 socket send error Connection timed out (110) 2016-06-12 23:45:00 socket send error Connection timed out (110) 2016-06-12 23:45:07 socket recv error Connection timed out (110) 2016-06-12 23:45:18 socket recv error Connection timed out (110) 2016-06-12 23:45:18 UpdateTip: new best=0000000000000000033927d833b8c8b037a643d904b4924c1bbb83149944e3af height=416033 version=0x20000001 log2_work=84.820649 tx=135389797 date='2016-06-12 23:44:51' progress=1.000000 cache=53.2MiB(7183tx) warning='4 of last 100 blocks have unexpected version' 2016-06-12 23:45:23 connect() to 216.155.129.35:8333 failed after select(): Connection refused (111) 2016-06-12 23:45:27 receive version message: /bitcoinj:0.14.1/: version 70001, blocks=0, us=127.0.0.1:8333, peer=98 2016-06-12 23:45:31 ProcessMessages: advertising address 92..__.114:8333 [redacted] 2016-06-12 23:45:31 receive version message: /Satoshi:0.12.1/: version 70012, blocks=416033, us=92.3.33.114:39254, peer=99 2016-06-12 23:45:31 AdvertiseLocal: advertising address 92...114:8333 [redacted] 2016-06-12 23:45:34 receive version message: /bitcoinj:0.14.1/: version 70001, blocks=0, us=127.0.0.1:8333, peer=100 2016-06-12 23:45:37 connect() to 198.46.152.2:8333 failed after select(): Connection refused (111) 2016-06-12 23:45:37 ProcessMessages: advertising address 92.*..114:8333 [redacted] 2016-06-12 23:45:37 receive version message: /Satoshi:0.12.1/: version 70012, blocks=416033, us=92.3.33.114:38186, peer=101 2016-06-12 23:45:37 AdvertiseLocal: advertising address 92..__.114:8333 [redacted] 2016-06-12 23:45:52 receive version message: /bitnodes.21.co:0.1/: version 70002, blocks=416030, us=92.3.33.114:8333, peer=102 2016-06-12 23:45:52 AdvertiseLocal: advertising address 92..**.114:8333 [redacted] 2016-06-12 23:46:01 tor: Thread interrupt 2016-06-12 23:46:01 torcontrol thread exit 2016-06-12 23:46:01 msghand thread interrupt 2016-06-12 23:46:01 opencon thread interrupt 2016-06-12 23:46:01 scheduler thread interrupt 2016-06-12 23:46:01 addcon thread interrupt 2016-06-12 23:46:01 net thread interrupt 2016-06-12 23:46:01 Shutdown: In progress... 2016-06-12 23:46:01 StopNode() 2016-06-12 23:46:01 Shutdown: done 2016-06-12 23:47:07

No idea why part of the log is in bold - ignore it.

Looks like it happens during shutdown, if you got that crash only after sending the stop command yourself. Is that the case?

Can you reproduce this?

I didn't issue any stop command, it terminated itself with the "stack" error in my screen console. I've since reverted back to the stable version I'm afraid, so am unable to reproduce.

There's nothing in that log to indicate that something else than RPC stop triggered the shutdown (it looks like a normal shutdown, even the final message Shutdown: done). Are you sure you quoted the right log?

Hmmmm. I done a search for "stack" in the debug log but it showed no results, so I posted the shutdown log closest to when I received the "*** stack smashing detected ***: bitcoind terminated" error.

I just done another search for "stack", "smashing" & "terminated" in the debug log - but nothing turns up, it's very strange.

Edit: What does "stack smashing" mean actually?

That "stack smashing detected" will not be written to debug.log, but printed to the console.

Right, that's where I saw it. Never seen or heard of that message before, that's why I thought I should report it. Maybe if that message was reported in the log it would be easier to locate the cause?

It can't be. It's not bitcoind's code that is doing this detection, but the built-in stack smashing detector, added by the compiler.

We're lacking information to troubleshoot this.

If this is an intermittent problem that happens randomly, I'd suggest starting bitcoind in gdb, then generating a backtrace if this happens and posting it here.

To run bitcoind in gdb transparently, as well as possibly get a mail when it crashes, see my scripts here: https://gist.github.com/laanwj/29bc141fb8d10608651c