Bug #15385: Ruby process hang in ensure

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 2.5.3p105 (2018-10-18 revision 65156) [x86_64-darwin18]
• Backport: 2.4: WONTFIX, 2.5: DONE

[pawelpacana (Paweł Pacana)]

While playing with mutation testing of my code I've found a weird mutation that hangs Ruby process. The isolated code which triggers this freeze is:

require 'timeout'

puts Process.pid

def raise_before_returning
  raise
  []
end

def doh
  value = raise_before_returning
ensure
  # p value # <-- that unblocks it
  value if nil
end

Timeout.timeout(3) do
  doh
end

1. Timeout is unnecessary and was added to illustrate that it has no effect to interrupt hanged process.
2. Process.pid is printed for convenience to kill hanged process.
3. Printing value in ensure changes behavior, there is no hang when encountering value if nil below.

I've also checked ruby 2.6.0preview3 (2018-11-06 trunk 65578) [x86_64-darwin18] and the problem persists.

[pawelpacana (Paweł Pacana)]

• File reproduce.rb reproduce.rb added

[ujihisa (Tatsuhiro Ujihisa)]

This reproduces even without timeout library.

def doh
  raise
ensure
  123 if false
end

doh

looks like this is because the optimizer removed if false or if nil without adding nop iseq.

[pawelpacana (Paweł Pacana)]

ujihisa (Tatsuhiro Ujihisa) wrote:
looks like this is because the optimizer removed if false or if nil without adding nop iseq.

Could it be related then to https://bugs.ruby-lang.org/issues/14959#note-7 and https://bugs.ruby-lang.org/issues/15392 then?

[shyouhei (Shyouhei Urabe)]

• Related to Bug #14897: Unexpected behavior of `if` in specific code added

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Closed

Applied in changeset trunk|r66326.

---

Fix infinite loop by ensure

• compile.c (iseq_insert_nop_between_end_and_cont): insert nop so
  that the end of rescue and continuing points are not same, to
  get rid of infinite loop. [Bug #15385]

[nagachika (Tomoyuki Chikanaga)]

• Backport changed from 2.4: UNKNOWN, 2.5: UNKNOWN to 2.4: REQUIRED, 2.5: REQUIRED

I didn't confirm this can be reproduced on ruby_2_4 too.
After a quick code reading, I think it might be.

[nagachika (Tomoyuki Chikanaga)]

Memo: r61591, r61592, r66324 are related.

[usa (Usaku NAKAMURA)]

• Backport changed from 2.4: REQUIRED, 2.5: REQUIRED to 2.4: WONTFIX, 2.5: DONE