rbspy: Crash: thread '' has overflowed its stack
I saw this while running the latest rbspy v0.3.13.
Command:
sudo rbspy record -p $(pgrep -n -f 'puma ') --subprocesses --duration 10
Crash:
thread '<unknown>' has overflowed its stack
fatal runtime error: stack overflow
Aborted (core dumped)
gdb on the core dump gave me:
(gdb) bt
#0 0x000000000070fe1f in __restore_sigs ()
#1 0x000000000070fe60 in raise ()
#2 0x00007f821ffdc4b0 in ?? ()
#3 0x00007f82203edac8 in ?? ()
#4 0x00000000006e0242 in std::io::Write::write_fmt::h9ff815313448b677 () at library/std/src/io/mod.rs:1518
#5 0x00007f8449189af0 in ?? ()
#6 0x00007f82203edac8 in ?? ()
#7 0x000000000070ddb2 in abort ()
#8 0x0000000000a3df28 in ?? ()
#9 0x0000000000000002 in ?? ()
#10 0x0000000000000000 in ?? ()
It’s happening consistently. Could be related to something new.
About this issue
- Original URL
- State: closed
- Created 3 years ago
- Comments: 16 (6 by maintainers)
The fix for this will be included in 0.3.14 when it’s published. Thanks again for the detailed feedback, @igorwwwwwwwwwwwwwwwwwwww and @jarthod.
Okay, I found a way to reliably repro this crash. I should have time over the next few days to dig into it.
@acj Fix confirmed, thank you so much!
The application in question is gitlab-org/gitlab.
Here’s a profile and flamegraph that I captured with stackprof. To give an idea of the size of symbol names.
stacks.sorted.out.gz flamegraph.svg.gz
Sample of longest strings:
@acj it’s puma
5.1.1.