Thread issues with ruby 1.8.0 on OpenBSD 3.3

[Rick Nooner]

I'm seeing problems with threading using ruby 1.8.0 on OpenBSD 3.3.

When I use threads, I get

longjump botch.
Illegal instruction (core dumped)

Following is a short script that reproduces the problem.
This script runs fine using ruby 1.8.0 on Solaris 8.

Are there known issues with ruby threading on OpenBSD?

Thanks,
Rick

--

require 'thread'
threads = []
q = Queue.new
threads << Thread.new { puts q.pop }
q.push("gronk")
threads.each { |t| t.join }

--
Rick Nooner
[email]rick@nooner.net[/email]
[url]http://www.nooner.net[/url]

[ts]

>>>>> "R" == Rick Nooner <rick@nooner.net> writes:

R> I'm seeing problems with threading using ruby 1.8.0 on OpenBSD 3.3.
R> When I use threads, I get

Can you give a stack trace ?

Run your script under gdb and when it crash use the command `bt'


Guy Decoux

[Rick Nooner]

Guy,

Here is the gdb session after the crash:

(gdb) bt
#0 0x8f29ae8 in _longjmp ()
#1 0x8f29aec in _longjmp ()
#2 0x7 in ?? ()
(gdb) quit

This is on a Sparc 5 which I didn't mention earlier.

Rick

On Sat, Nov 22, 2003 at 10:14:32PM +0900, ts wrote:

> >>>>> "R" == Rick Nooner <rick@nooner.net> writes:

>
> R> I'm seeing problems with threading using ruby 1.8.0 on OpenBSD 3.3.
> R> When I use threads, I get
>
> Can you give a stack trace ?
>
> Run your script under gdb and when it crash use the command `bt'
>
>
> Guy Decoux

--
Rick Nooner
[email]rick@nooner.net[/email]
[url]http://www.nooner.net[/url]

[ts]

>>>>> "R" == Rick Nooner <rick@nooner.net> writes:

R> This is on a Sparc 5 which I didn't mention earlier.

Perhaps best to see on a OpenBSD mailing list : there were well know
problems with Alpha for longjmp

[url]http://archives.neohapsis.com/archives/openbsd/2003-01/2142.html[/url]

I don't know for sparc


Guy Decoux

[Rick Nooner]

I don't think that it's a sparc issue since the same code runs fine
under Solaris. I have several other architecture/OS combinations
available that work fine as well. It is probably related to OpenBSD.

I'll dig into it further this afternoon.

Thanks for your thoughts.

Rick

On Sun, Nov 23, 2003 at 02:18:37AM +0900, ts wrote:

> >>>>> "R" == Rick Nooner <rick@nooner.net> writes:

>
> R> This is on a Sparc 5 which I didn't mention earlier.
>
> Perhaps best to see on a OpenBSD mailing list : there were well know
> problems with Alpha for longjmp
>
> [url]http://archives.neohapsis.com/archives/openbsd/2003-01/2142.html[/url]
>
> I don't know for sparc
>
>
> Guy Decoux
>

--
Rick Nooner
[email]rick@nooner.net[/email]
[url]http://www.nooner.net[/url]

[ts]

>>>>> "R" == Rick Nooner <rick@nooner.net> writes:

R> I don't think that it's a sparc issue since the same code runs fine
R> under Solaris. I have several other architecture/OS combinations
R> available that work fine as well. It is probably related to OpenBSD.

I'm agree with you

moulon% ruby -v
ruby 1.8.0 (2003-08-04) [sparc-solaris2.7]
moulon%

This is why I've said

>> Perhaps best to see on a OpenBSD mailing list

^^^^^^^^^^^^^^^^^^^^


Guy Decoux