Re: evbppc and alpha build breakage, but not in releng autobuild

[Taylor R Campbell <campbell+netbsd-current-users%mumble.net@localhost>]

   Date: Sat, 16 Mar 2013 07:43:08 -0400
   From: Greg Troxel <gdt%ir.bbn.com@localhost>

   Taylor R Campbell <campbell+netbsd-current-users%mumble.net@localhost> 
writes:

   > ./powerpc/ibm4xx/cpu.h: In function 'mtdcr':
   > ./powerpc/ibm4xx/cpu.h:94:2: error: asm operand 0 probably doesn't
   > match constraints
   > ./powerpc/ibm4xx/cpu.h: In function 'mfdcr':
   > ./powerpc/ibm4xx/cpu.h:102:2: error: asm operand 1 probably doesn't
   > match constraints

   I would guess that you have some optimization flags different, perhaps
   adding -g?  I have seen this before, and the problem was I think wrong
   constraints, or a compiler bug, but with -O2 it worked anyway, or
   something like that.

Good thinking -- I set MKDEBUG=yes in my mk.conf.  Maybe the releng
autobuilds should do this too?  I don't know how tight we are on space
for them, but at least the added weight of MKDEBUG=yes has been
relegated to a separate debug.tgz set now.