Bug #5867: Segfault with (likely) the new GC on OSX

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 2.0.0dev (2012-01-08 trunk 34231) [x86_64-darwin10.8.0]

[Eregon (Benoit Daloze)]

Hello,

I tried the new GC yesterday, and experienced a few segfaults.
I tried again with r34221 (the only significant revisions between this and r34231 are GC-related) and did not see any problem in ~2000 runs. I tried on Linux and saw no problem in ~30000 tries.

Unfortunately, it happens rather rarely, about once in 200 times at me.
It seems to always happen at the same spot.

Here the way I used to produce them:

git clone git://github.com/eregon/epath.git
cd epath
gem i rspec

path/to/ruby-r34231 -rrake -e "Rake.application.run"

or

ruby -e 'i=0; while system("path/to/ruby-r34231 -rrake -e "Rake.application.run""); p i+=1; end'

I attach a few crashlogs and a gdb backtrace. I can also provide a core dump.
I tried with GC.stress=true once but it worked well. I saw once what was likely memory-corruption effects.
I tried on another project and did not see anything abnormal, make test-all also passed.

[Eregon (Benoit Daloze)]

It seems to already be solved as of r34253 !
You can close this issue.

[naruse (Yui NARUSE)]

• Status changed from Open to Closed