richard zidlicky

done that already

got a new problem though, this time it is happening randomly and
pretty rare in a threaded program so could not get gdb backtrace
yet:

/usr/lib/site_ruby/1.6/yaml/emitter.rb:285: [BUG] Segmentation fault
ruby 1.6.7 (2002-03-19) [i386-linux]
Aborted

Any idea?

Richard

why the lucky stiff

Hard to say. What's the data like that you're dumping (calling
to_yaml)?

Incidentally, I'm nearing completion on a C emitter. I don't think I'll
try to merge anything new before 1.8.0, but in CVS and thereafter it
will appear.

_why

nigel gilbert

Could your segmentation fault be due to the same problem as in the
following?

From: Mauricio Fernández <batsman.geo@yahoo.com>
Date: Sat Jul 5, 2003 7:20:38 PM Europe/London
To: ruby-talk@ruby-lang.org (ruby-talk ML)
Subject: Re: Tracking down a seg fault
Reply-To: ruby-talk@ruby-lang.org

I had a similar problem, with the same code (I wrote
ObjectPoolingAndThreading :-)

I was getting random segfaults with a heavily threaded script that was
doing 200 connections to a dictd server. After some time, I found out
that the problem seemed to be associated to GCC, as the segfault would
only happen when targeting i386 and using gcc 3.2...


You could check if the problem persists w/ a different GCC (although
I don't know if you are limited to gcc 3.1 in OSX). Other than that,
I guess there's not much to do :-( In my case, it was impossible to
isolate the bug, as any attempt to remove code would make it impossible
to reproduce it.

--
_ _
| |__ __ _| |_ ___ _ __ ___ __ _ _ __
| '_ \ / _` | __/ __| '_ ` _ \ / _` | '_ \
| |_) | (_| | |_\__ \ | | | | | (_| | | | |
|_.__/ \__,_|\__|___/_| |_| |_|\__,_|_| |_|
Running Debian GNU/Linux Sid (unstable)
batsman dot geo at yahoo dot com
<Tazman> damn my office is cold.
<Tazman> need a hot secretary to warm it up.
-- Seen on #Linux


On Sunday, July 6, 2003, at 11:27 AM, Richard Zidlicky wrote:


__________________________________________________ ______________________
__
Professor Nigel Gilbert, FREng, AcSS, Pro Vice-Chancellor and Professor
of
Sociology, University of Surrey, Guildford GU2 7XH, UK. +44 (0)1483
689173

richard zidlicky

it could be the same or related, but so far all symptoms
described seem much to random to be debugged.
Does anyone have a truly repeatable testcase?

BTW does ruby have an option to dump cure instead of
catching the segfault?

Richard

mauricio fernández

I did the following:
* comment the following lines in signal.c (around line 736)
#ifdef SIGSEGV
install_sighandler(SIGSEGV, sigsegv);
#endif
* compile and rename the binary to gruby, then put it in my path

This way I can get core dumps when debugging rjni.

--
_ _
| |__ __ _| |_ ___ _ __ ___ __ _ _ __
| '_ \ / _` | __/ __| '_ ` _ \ / _` | '_ \
| |_) | (_| | |_\__ \ | | | | | (_| | | | |
|_.__/ \__,_|\__|___/_| |_| |_|\__,_|_| |_|
Running Debian GNU/Linux Sid (unstable)
batsman dot geo at yahoo dot com
<dark> Turns out that grep returns error code 1 when there are no matches.
I KNEW that. Why did it take me half an hour?
-- Seen on #Debian

why the lucky stiff

I tested thousands of operations loading/dumping the sample file you gave me.
I'm trying to get some random data in there to test as well, but so far
things seem fine.

What version of Syck are you using? If you're using 0.35, then I would
upgrade to the CVS version. You can find cases of dump segfaults on
ruby-core that have been fixed since 0.35.

Don't worry. We'll beat this thing.

_why