Bug #17196: Segmentation Fault with Socket#close in Ractors

• Status: Closed
• Priority: Normal
• Assignee: ko1 (Koichi Sasada)
• Target version: -
• ruby -v: ruby 3.0.0preview1 (2020-09-25 master 0096d2b895) [x86_64-linux]
• Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN

[asuper (Anthony Super)]

While messing around with 3.0.0-preview1, I thought it would be fun to write a really, really simple webserver that used Ractor creation instead of forking or spawning threads. I was right, it was fun! Unfortunately it also caused a segfault. The segfault appears to be on socket closure when a lot of ractors are running at once.

The attached tarball includes the code you need to make it happen. Run bin/server in one terminal/screen session/tmux window/whatever, and ./bug_bench.sh with the other. You'll need to have the apache benchmark utility installed.

[ko1 (Koichi Sasada)]

• Assignee set to ko1 (Koichi Sasada)

[hsbt (Hiroshi SHIBATA)]

• Status changed from Open to Assigned

[wanabe (_ wanabe)]

I tried to reproduce the problem at hand, but it seems to have already been fixed in the current master.
According to git bisect, it seemed to be fixed by #18117 https://bugs.ruby-lang.org/projects/ruby-master/repository/git/revisions/bbedd29b6e98ef6e3fc2ce2b358d2b509b7cd1bb.
How about using 3.0.3 or 3.1.x or master?

[jeremyevans0 (Jeremy Evans)]

• Status changed from Assigned to Closed