Bug #3528
closed"make" aborts - ruby-1.9.1-p429
Description
=begin
make aborts. Output log attached. This is highly reproducible - successive attempts to run make produce identical output (apart from a few addresses), except for one time that produced a StackOverflow.
=end
Files
Updated by naruse (Yui NARUSE) over 14 years ago
- ruby -v changed from N/A - cannot build ruby, so cannot run "ruby -v" to miniruby 1.9.1p429 (2010-07-02 revision 28523) [i686-linux]
=begin
=end
Updated by jeremy_henty (Jeremy Henty) over 14 years ago
=begin
OS is LFS (Linux From Scratch) 6.5 http://www.linuxfromscratch.org/lfs/view/6.5/
$ uname -srmp
Linux 2.6.34-0 i686 athlon-4
$ gcc --version
gcc (GCC) 4.4.1
Glibc version 2.10.1
Please suggest any other information that I may give to help.
=end
Updated by jeremy_henty (Jeremy Henty) over 14 years ago
=begin
I would like to attempt a "git bisect" to find the problem using the git mirror. Is this reasonable? I see a tag for v_191_376 in the repository, but none for 378 patch (which is my current working ruby) or the broken 429 patch. Can you suggest commits that could use to begin a bisect? Or would it be better to use the subversion repository. Please advise.
=end
Updated by naruse (Yui NARUSE) over 14 years ago
=begin
Yes, it is reasonable.
You can use http://github.com/ruby/ruby/tree/ruby_1_9_1
previous patch release is http://github.com/ruby/ruby/commits/v1_9_1_378
=end
Updated by jue (Juergen Daubert) over 14 years ago
Updated by jeremy_henty (Jeremy Henty) over 14 years ago
- File bisect.txt bisect.txt added
- File culprit.txt culprit.txt added
=begin
I have attached the results of the git bisect.
=end
Updated by jeremy_henty (Jeremy Henty) over 14 years ago
=begin
Thanks for those pointers Juergen! I find that "optflags='-O1'" does not work on p249, but
"optflags='-O0'" and "optflags='-O2 -fno-inline'" do work.
Also, 1.9.2-rc1 does not have this problem.
=end
Updated by jeremyevans0 (Jeremy Evans) about 5 years ago
- Description updated (diff)
- Status changed from Open to Closed