Bug #9757: Segfault in 2.1.1 from RSpec expect_any_instance_of

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 2.1.1p76 (2014-02-24 revision 45161) [x86_64-darwin13.0]
• Backport: 2.0.0: UNKNOWN, 2.1: UNKNOWN

[dafaber (David Faber)]

The following spec causes a segfault for me in ruby 2.1.1 on OS X 10.9.2:

class A
  def a
    'a'
  end
end

class B
  def b
    'b'
  end
end

describe 'segfault' do
  it 'should make unused recorders' do
    expect_any_instance_of(A).to receive(:a)
    expect_any_instance_of(B).to receive(:b)
  end
  
  it 'should crash ruby' do
    A.new.a
    B.new.b
  end
end

Some more info:

$ ruby -v
ruby 2.1.1p76 (2014-02-24 revision 45161) [x86_64-darwin13.0]
$ bundle exec gem list

*** LOCAL GEMS ***

bundler (1.6.0.rc)
diff-lcs (1.2.5)
rspec (3.0.0.beta2)
rspec-core (3.0.0.beta2)
rspec-expectations (3.0.0.beta2)
rspec-mocks (3.0.0.beta2)
rspec-support (3.0.0.beta2)

I've also attached the crash report generated by OS X. Let me know if you need some more info. Sorry if this is a duplicate, I found some similar looking reports (#9309, #9461, #9315), but I don't really know enough to recognize if they are exactly the same issue.

[nobu (Nobuyoshi Nakada)]

• Category set to doc
• Status changed from Open to Feedback

Seems like stack overflow.
Could you try with the trunk?

[nobu (Nobuyoshi Nakada)]

• Category changed from doc to core

Set the category to "doc" by mistake.

[dafaber (David Faber)]

Sorry I missed the notification about your message. I've retried this with trunk (ruby 2.2.0dev (2014-04-25 trunk 45720) [x86_64-darwin13]) and it just gives a SystemStackError, with message "stack level too deep" now, so it does seem like stack overflow. Does that indicate that the segfault is fixed in trunk, and there may be a different bug in RSpec causing the stack overflow in the first place?

[jeremyevans0 (Jeremy Evans)]

• Status changed from Feedback to Closed