Vit Ondruch <v.ondruch / tiscali.cz> wrote:
> I gave it try with ruby 2.0.0dev (2012-02-21 trunk 34723)
> [x86_64-linux] and it does not segfaulted in 1000 runs. Now I'm
> wondering what was the fix, since I'd love to see it backported to
> Ruby 1.9.3.

Perhaps it was the gc.c change in r34719?
Can you revert that and see if you can reproduce the segfault?
Perhaps you can confirm my suspicion from [ruby-core:42794]