Re: [Gc] Re: Boehm GC on NetBSD 5.1

[Ivan Maidanski]

Hello Bruno,

17 01 2012, 11:35 Bruno Haible <bruno at clisp.org>:
> Hi Ivan,
> 
> > please checkout code from Git "release" (or "master") branch.
> 
> Thanks. It is not immediate how to build from there, because the
> 'configure' file, in particular, is not under version control.

In "release" branch 'configure' still exists (so, you don't need to run "autoreconf -vif").
And we definitely supply configure in the release tarballs.

> It took me a while of trial-and-error to get it created.
> How about adding these two files to the repository? Some packages
> have an executable shell script called 'autogen.sh' that re-creates
> the files that have been omitted from version control. Some packages call
> this script 'bootstrap'. Some other packages have a file 'README-HACKING'
> which contains instructions.
> 
> ====================== autogen.sh for bdwgc =============================

bdwgc already has autogen.sh

> #!/bin/sh
> # Convenience script for regenerating all autogeneratable files that are
> # omitted from the version control repository. In particular, this script
> # also regenerates all aclocal.m4, config.h.in, Makefile.in, configure files
> # with new versions of autoconf or automake.
> 
> test -f m4/pkg.m4 || {
>   wget -q --timeout=5 -O m4/pkg.m4.tmp http://cgit.freedesktop.org/pkg-config/plain/pkg.m4 \
>     && mv m4/pkg.m4.tmp m4/pkg.m4
> }
> 
> autoreconf -i
> =========================================================================
> 
> ====================== autogen.sh for libatomic_ops =====================

Probably we could copy that from bdwgc

> #!/bin/sh
> # Convenience script for regenerating all autogeneratable files that are
> # omitted from the version control repository. In particular, this script
> # also regenerates all aclocal.m4, config.h.in, Makefile.in, configure files
> # with new versions of autoconf or automake.
> 
> autoreconf -i
> =========================================================================
> 
> > Please retry with fresh snapshot and report whether a problem with
> > threadkey_test still exists.
> 
> Compiles fine. "All 14 tests passed." twice. But on the 3rd "make check"

Something wrong in pthread_stop_world.c or pthread_support.c for NetBSD (regarding thread registration).

> threadkey_test hangs. 'ps' shows
> 
> USER        PID %CPU %MEM     VSZ    RSS TTY    STAT STARTED      TIME COMMAND
> haible    18720  0.0  0.5 2552764  16380 ttyp0  DEl+ 10:42PM   0:00.71 /home/haible/bdwgc/.libs/threadkey_test
> 
> This means, the process has allocated 2.5 GB of memory?!

How many threads are running at that time?
> 
> Trying to attach a gdb does not succeed.

Launch it under gdb for several times. (Compile without optimizations)
Try to get stack traces.

Regards.

> 
> Bruno