> the bootstrap behaves the same way...
> 
>  >> NetBSD/i386 BIOS, Revision 3.1
>  >> (autobuild@tgm.netbsd.org, Mon Nov  3 07:44:09 UTC 2003)
>  >> Memory: 639/65472 k
> 
> on the same machine, the 1.6.1 branch install boot disk does recognize all 
> of the memory and looks like this:
> 
>  >> NetBSD/i386 BIOS, Revision 2.13
>  >> (autobuild@tgm.daemon.org, Tue Apr  8 11:27:33 UTC 2003)
>  >> Memory: 639/326592 k
> 
> i noticed PR 21037 from back in April, but I was surprised that no one else 
> saw this behavior.  anyone else see this since then?  what source files 
> differentiate the bootstrap rev 3.1 from 2.13?

The number generated by the bootstrap (and passed to the kernel) is
generated by the code in src/sys/arch/i386/stand/lib/getextmemx.c

Setting DEBUG_MEMSIZE and rebuilding src/sys/arch/i386/stand/boot/biosboot
wll give some diagnostic output (twice) that might be informative.

	David

-- 
David Laight: david@l8s.co.uk