RE: Thread/GC failure of gcj-3.3.1 on netbsd-current

["Boehm, Hans" <hans_boehm at hp dot com>]

I've attached Marc Recht's patches.  But I don't think that's the issue.

AFAICT, the collector currently doesn't have support for NetBSD threads.
(If you look at the -D flags being used to build the collector, one of them
should be of the form -DGC_<XYZ>_THREADS, where <XYZ> should probably be
NETBSD.  But I don't see support for such a flag in the collector.

The support does exist for FreeBSD, Linux, etc.  Assuming NetBSD isn't
too different from these, it probably won't require many additional lines of
code to add the support.  It probably doesn't make sense to add it to
anything before 3.4 however.

I think the symptoms are consistent with this theory; the collector is
probably running somewhere other than in the initial thread, and it's
getting confused about stack addresses, because it thinks there's only
one.

Hans 


> -----Original Message-----
> From: Tom Tromey [mailto:tromey@redhat.com]
> Sent: Friday, September 19, 2003 8:40 AM
> To: Bart Locanthi
> Cc: java@gcc.gnu.org
> Subject: Re: Thread/GC failure of gcj-3.3.1 on netbsd-current
> 
> 
> >>>>> "Bart" == Bart Locanthi <bart@sabl.com> writes:
> 
> Bart> This isn't quite a cross-post, and I suspect the 
> problem described
> Bart> here has more to do with NetBSD than GCJ, but I'm 
> hoping the symptom
> Bart> rings some bells here that would lead to understanding 
> in NetBSD-land.
> 
> I don't really have a definitive answer for you, but the problem
> might be fixed on the cvs trunk.
> 
> From boehm-gc/doc/README.changes:
> 
> Since 6.2:
>  - Integrated some NetBSD patches forwarded to me by Marc 
> Recht.  These
>    were already in the NetBSD package.    
> 
> I'm not sure what these changes are, though, so I don't know if they
> affect this area or not.
> 
> Tom
>

Attachment: patch-ab
Description: Binary data

Attachment: patch-ad
Description: Binary data