No stack trace printed nor does process exit.

[jspavlick]

I was having some trouble with my Heroku app exiting with status 139. I installed segfault-handler, but now this is all I get. No stack trace.

Apr 11 13:31:14 my-app app/worker.2: PID 3 received SIGSEGV for address: 0x0

The process doesn't exit anymore.

Please advise.

[jspavlick]

Attempts to reproduce the segfault on my laptop (OS X) are proving unsuccessful.

[No9]

@jspavlick Can you checkout the repo and run the tests?

[jspavlick]

Thanks for the quick response.

Running example.js on my laptop (OS X) provides a wonderfully nice stack trace.

Laptop:

~> node example.js
NodeSegfaultHandlerNative: about to dereference NULL (will cause a SIGSEGV)
PID 44292 received SIGSEGV for address: 0x1
0   segfault-handler.node               0x00000001008de47a _ZL16segfault_handleriP9__siginfoPv + 282
1   libsystem_platform.dylib            0x00007fff92ddbf1a _sigtramp + 26
2   ???                                 0x00007fff5fbfef10 0x0 + 140734799802128
3   segfault-handler.node               0x00000001008de246 _Z22segfault_stack_frame_2v + 22
4   segfault-handler.node               0x00000001008de276 _Z13CauseSegfaultRKN2v89ArgumentsE + 38
5   node                                0x0000000100145b9a _ZN2v88internalL21Builtin_HandleApiCallENS0_12_GLOBAL__N_116BuiltinArgumentsILNS0_21BuiltinExtraArgumentsE1EEEPNS0_7IsolateE + 459
6   ???                                 0x00002f747c00618e 0x0 + 52177343111566
7   ???                                 0x00002f747c063df3 0x0 + 52177343495667

Heroku:

~> heroku run crash --app my-app
Running `crash` attached to terminal... up, run.6659
Detected 512 MB available memory, 512 MB limit per process (WEB_MEMORY)
Recommending WEB_CONCURRENCY=1
NodeSegfaultHandlerNative: about to dereference NULL (will cause a SIGSEGV)
PID 3 received SIGSEGV for address: 0x1
/app/node_modules/segfault-handler/build/Release/segfault-handler.node(+0x10e5)[0x7f33b06960e5]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x10340)[0x7f33b0c6d340]
/app/node_modules/segfault-handler/build/Release/segfault-handler.node(_Z22segfault_stack_frame_1v+0x10)[0x7f33b06961d0]
/app/node_modules/segfault-handler/build/Release/segfault-handler.node(_Z13CauseSegfaultRKN2v89ArgumentsE+0x1c)[0x7f33b069620c]
node[0x70fdc7]
[0x25ebbdd06362]
stty: standard input: unable to perform all requested operations

However, I receive no such stack dump when not running the example code.

Thanks.

[jspavlick]

I was just able to get my laptop node process to segfault. Notice node-segfault-handler did not help me. I'm not sure what to do next.

node(43906,0x7fff7536a300) malloc: *** error for object 0x10298d008: incorrect checksum for freed object - object was probably modified after being freed.
*** set a breakpoint in malloc_error_break to debug
/Library/Ruby/Gems/2.0.0/gems/foreman-0.77.0/lib/foreman/cli.rb:103:in `exit': no implicit conversion from nil to integer (TypeError)
    from /Library/Ruby/Gems/2.0.0/gems/foreman-0.77.0/lib/foreman/cli.rb:103:in `run'
    from /Library/Ruby/Gems/2.0.0/gems/thor-0.19.1/lib/thor/command.rb:27:in `run'
    from /Library/Ruby/Gems/2.0.0/gems/thor-0.19.1/lib/thor/invocation.rb:126:in `invoke_command'
    from /Library/Ruby/Gems/2.0.0/gems/thor-0.19.1/lib/thor.rb:359:in `dispatch'
    from /Library/Ruby/Gems/2.0.0/gems/thor-0.19.1/lib/thor/base.rb:440:in `start'
    from /Library/Ruby/Gems/2.0.0/gems/foreman-0.77.0/bin/foreman:7:in `<top (required)>'
    from /usr/bin/foreman:23:in `load'
    from /usr/bin/foreman:23:in `<main>'

[No9]

I haven't worked with heroku so you will have to bear with me here
but it looks like heroku are using foreman as the process manager?

So heroku run crash --app my-app
Is really a wrapper around forman

If that's the case can you run the node process directly on the server as you would on your local machine?
If you want a process manager then https://www.npmjs.com/package/forever might help.

Other than that you could ask the forman folks what might be preventing the stacktrace from surfacing up

[No9]

Hey @jspavlick I am going to close this as it's been around for a while.
If you have more input or further questions please feel free to spin up another issue.

[relovution]

I'm having a similar issue, not with not exiting but I have no stack trace:

root@beaglebone:~/test/node_modules/segfault-handler# node example.js 
NodeSegfaultHandlerNative: about to dereference NULL (will cause a SIGSEGV)
PID 1907 received SIGSEGV for address: 0x2
root@beaglebone:~/test/node_modules/segfault-handler# uname -a
Linux beaglebone 4.1.19-bone-rt-r20 #1 Sun Mar 13 04:30:34 UTC 2016 armv7l GNU/Linux

Any ideas?

[relovution]

I found a solution: adding -funwind-tables to CFLAGS in binding.gyp fixes this problem (see http://stackoverflow.com/questions/24700150/on-raspberry-pi-backtrace-returns-0-frames). Test after doing node-gyp rebuild:

root@beaglebone:~/test/node_modules/segfault-handler# node example.js 
NodeSegfaultHandlerNative: about to dereference NULL (will cause a SIGSEGV)
PID 2268 received SIGSEGV for address: 0x2
/root/test/node_modules/segfault-handler/build/Release/segfault-handler.node(+0x12a0)[0xb68ba2a0]
/lib/arm-linux-gnueabihf/libc.so.6(+0x26b00)[0xb68f1b00]
/root/test/node_modules/segfault-handler/build/Release/segfault-handler.node(_Z22segfault_stack_frame_1v+0x1f)[0xb68ba5c4]
/root/test/node_modules/segfault-handler/build/Release/segfault-handler.node(+0x1434)[0xb68ba434]

I can create a pull request, but I haven't tested it on x86/x86_64, any idea if this will have any repercussions?