Bug #10008: conftest crashed: installing mri 2.1.2: EXC_BAD_ACCESS (SIGABRT)

• Status: Closed
• Assignee: nobu (Nobuyoshi Nakada)
• Target version: -
• ruby -v: ruby 2.1.2p95 (2014-05-08 revision 45877) [x86_64-darwin13.0]
• Backport: 2.0.0: DONTNEED, 2.1: DONE

[jaredbeck (Jared Beck)]

crash dump: https://gist.github.com/jaredbeck/be73c376b860d8d6eed9

cd ~/.rbenv
git log --oneline | head -n1
# 13a474c Get rid of explicit exit in fish branch of `rbenv-init`
cd plugins/ruby-build/
git log --oneline | head -n1
# c819790 ruby-build 20140702

but .. the install finished!

rbenv global 2.1.2
ruby --version
# ruby 2.1.2p95 (2014-05-08 revision 45877) [x86_64-darwin13.0]
echo "puts '<3 ruby'" | ruby
# <3 ruby

Using xcode 5.1.1

[nobu (Nobuyoshi Nakada)]

Which conftest?
Some conftests are expected to segfault.

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Feedback
• Priority changed from Normal to 3

[hsbt (Hiroshi SHIBATA)]

Is conftest part of debugger gem?

[jaredbeck (Jared Beck)]

Steps to reproduce:

1. rbenv uninstall 2.1.2
2. rbenv install 2.1.2

Crash dump:
https://gist.github.com/jaredbeck/127e61aa73a16c614b6d

ruby_build log:
https://gist.github.com/jaredbeck/3f6de8e3e432a51ae707

versions:

gcc version 4.2.1 (Apple Inc. build 5666) (dot 3)
rbenv: 13a474c
ruby-build: c819790 ruby-build 20140702

[jaredbeck (Jared Beck)]

Simpler steps to reproduce:

1. Download ruby 2.1.2 source
2. ./configure

So, I don't think this has anything to do with ruby-build :)

[jaredbeck (Jared Beck)]

Which conftest?

I think conftest is defined in the configure script. For me, it's around line 20376, which looks like:

cat confdefs.h - <<_ACEOF >conftest.$ac_ext

Furthermore, line 20389 defines a sigsegv function.

void sigsegv(int signum, siginfo_t *info, void *ctx){

This sigsegv function can be seen on line 3 of the "Dispatch queue" from the crash dump:

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
0 libsystem_kernel.dylib 0x00007fff8b3f7866 __pthread_kill + 10
1 libsystem_pthread.dylib 0x00007fff8f8df35c pthread_kill + 92
2 libsystem_c.dylib 0x00007fff9000ab1a abort + 125
3 conftest 0x000000010c2dee55 sigsegv + 37
4 libsystem_platform.dylib 0x00007fff8cfdd5aa _sigtramp + 26
5 ??? 000000000000000000 0 + 0
6 libdyld.dylib 0x00007fff8efa35fd start + 1

Finally, it may be relevant that my ./configure output reports "broken backtrace":

checking for broken backtrace... yes

I hope that helps!

[jaredbeck (Jared Beck)]

• File 0001-proof-of-concept-for-bug-10008.patch 0001-proof-of-concept-for-bug-10008.patch added

If I replace the whole test for "broken backtrace" with a hardcoded value, then I can run ./configure without a crash.

See my proof of concept patch (attached) which can be compared to tag v2_1_2 in git.

[nobu (Nobuyoshi Nakada)]

• Status changed from Feedback to Rejected

It makes no sense.

[nobu (Nobuyoshi Nakada)]

• Status changed from Rejected to Open
• Backport changed from 2.0.0: UNKNOWN, 2.1: UNKNOWN to 2.0.0: DONTNEED, 2.1: REQUIRED

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Closed
• % Done changed from 0 to 100

Applied in changeset r46796.

---

configure.in: exit with failure

• configure.in (rb_cv_broken_backtrace): exit with failure
  normally, no needs to abort. [ruby-core:63678] [Bug #10008]

[nagachika (Tomoyuki Chikanaga)]

• Backport changed from 2.0.0: DONTNEED, 2.1: REQUIRED to 2.0.0: DONTNEED, 2.1: DONE

Backported into ruby_2_1 branch at r47418.