Subject: Cross-building from amd64-freebsd-6.1 ?
To: None <current-users@netbsd.org>
From: Andrew Reilly <andrew-netbsd@areilly.bpc-users.org>
List: current-users
Date: 06/22/2006 06:29:27
Hi all,

Is there any particular reason that this wouldn't work?  I've
got a freshly checked-out -current source tree, and I've tried
building thusly:

./build.sh -U -O ../obj -m i386 -N 1 -w ../nbmake -V HOST_CC=gcc42 -V  
HOST_CXX=g++42 tools

(and also without the HOST_CC and HOST_CXX flags, and also with
-m evbarm)

but after a little while the build falls over thusly:


*** Configuration amd64-unknown-freebsd6.1 not supported
Configure in /usr/home/andrew/rtos-comp/netbsd/obj/tools/gcc/build/ 
gcc failed, exiting.

*** Failed target:  .configure_done
*** Failed command: (cd build && CC=gcc42 CFLAGS=-O CPPFLAGS= CXX=g+ 
+42 CXXFLAGS=-O INSTALL=/usr/home/andrew/rtos-comp/netbsd/src/../obj/ 
tooldir.FreeBSD-6.1-STABLE-amd64/bin/nbinstall\ -c\ \ -r LDFLAGS=  
AR=ar RANLIB=ranlib LEX=/usr/home/andrew/rtos-comp/netbsd/src/../obj/ 
tooldir.FreeBSD-6.1-STABLE-amd64/bin/nblex MAKE=/usr/home/andrew/rtos- 
comp/netbsd/src/../obj/tooldir.FreeBSD-6.1-STABLE-amd64/bin/nbmake  
PATH="/usr/home/andrew/rtos-comp/netbsd/src/../obj/ 
tooldir.FreeBSD-6.1-STABLE-amd64/bin:$PATH" YACC=/usr/home/andrew/ 
rtos-comp/netbsd/src/../obj/tooldir.FreeBSD-6.1-STABLE-amd64/bin/ 
nbyacc sh /usr/home/andrew/rtos-comp/netbsd/src/tools/gcc/../../gnu/ 
dist/gcc/configure --target=i386--netbsdelf --disable-nls --enable- 
long-long --disable-multilib --enable-threads --program-transform- 
name="s,^,i386--netbsdelf-," --enable-languages="c c++ objc f77" -- 
prefix=/usr/home/andrew/rtos-comp/netbsd/src/../obj/ 
tooldir.FreeBSD-6.1-STABLE-amd64)
*** Error code 1

Stop.
nbmake: stopped in /usr/home/andrew/rtos-comp/netbsd/src/tools/gcc

Is this an indication that there is something fundamentally
wrong with FreeBSD/amd64, or is this just a config file case
statement that hasn't been taught that version 6.1 is OK?

I this seems to be working fine on my older pentium3 box, but that
is destined to become the crash-box for a bunch of experimental
configurations, and so is not really going to be up to repeating
the exercise.  It's also vastly slower, of course.

Best regards,

-- 
Andrew