[Mono-devel-list] bootstrap-check on PowerPC / Mac OS X

[Patrick Beard]

On Jun 18, 2004, at 10:22 AM, Paolo Molaro wrote:

> On 06/18/04 Patrick Beard wrote:
>> env ACLOCAL_FLAGS="-I /sw/share/aclocal" ./autogen.sh
>> --prefix=/usr/local --without-sigaltstack --with-gc=boehm
>> --with-jit=yes
>
> You don't need any of --with-jit=yes --without-sigaltstack since the
> default should be the same. Using --with-gc=boehm is discouraged,
> since we have a few additional fixes in our version of libgc that Hans
> hasn't released yet.
>
>>> When I pull the latest mono runtime sources from CVS (from
>>> anoncvs.go-mono.com), and successfully bootstrap (that took a while 
>>> to
>>> get right this morning, I had to pull fresh from CVS and reapply my
>>> patches), I tried to run 'make -k bootstrap-test'. I found that mono
>>> hung when running System.Data_test.dll. The stack crawls of the hung
>>> process seem to indicate a deadlock:
>
> Dick today fixed an issue in libgc to fix a deadlock, so you may want 
> to
> try with current cvs, but please don't use --with-gc=boehm or you
> won't get the fix.
>
>>> I am trying to validate that my changes to mini-ppc.c and aot.c 
>>> aren't
>>> the cause of this. Are these known problems? I'm running on Mac OS X
>>> 10.3.4 on a PB G4 with a gig of memory.
>
> If you want to check if your patches cause this, you can just unapply
> them (cvs diff >my-patch; patch -R < my-patch, basically).
> I'll test your patches today as well.

Things are good rebuilding with libgc rather than using the one 
provided by fink.

One thing that keeps happening when I cvs update from the anoncvs 
mirrors is that all of libgc keeps coming up as "in the way" which 
means I have to rm -rf libgc, and repull. What's up with that?

- Patrick