On Wed, Oct 10, 2007 at 01:26:22PM +0100, Andrew Doran wrote:
> >    It was fine with a new kernel, but the new userland seems to
> > have broken it. Fallout from the new malloc, perhaps?
> 
> Thanks for all the info. I think it's likely this is due to the allocator
> changing. I'll see if I can reproduce it.

XFree86 uses a nasty hack on NetBSD-amd64. Because the XFree86 module
can't deal with addresses larger than 4GB it manually mmaps memory
starting at 1GB(*) and loads modules in that area.

Considering that the module loader is on the backtrace it is likely
that this is the cause.

	Kind regards

(*) The Linux version uses mmap() with the MAP_32BIT option to achieve that.
    But NetBSD doesn't support that.

-- 
Matthias Scheler                                  http://zhadum.org.uk/