Subject: cross-building for sparc and i386
To: None <tech-toolchain@netbsd.org>
From: Wolfgang S. Rupprecht <wolfgang+gnus20031214T105550@dailyplanet.dontspam.wsrcc.com>
List: tech-toolchain
Date: 12/14/2003 11:05:24
I just noticed this after attempting a sparc crossbuild from the same source
tree I had just completed a native i386 build on:

    # first build -- it worked fine.
    unset MACHINE
    ./build.sh -U tools kernel=WSRCC_ATHLON distribution

    # second build, it broke with the following error 
    export MACHINE=sparc
    ./build.sh -U tools kernel=WSRCC distribution

    =======
    #      link  ldexp/ldexp
    /v/src/netbsd/src/obj/tooldir.NetBSD-1.6ZG-i386/bin/sparc--netbsdelf-gcc    -o ldexp -nostdlib  -Wl,-rpath-link,/v/src/netbsd/src/obj/destdir.sparc/lib:/v/src/netbsd/src/obj/destdir.sparc/usr/lib  -L/v/src/netbsd/src/obj/destdir.sparc/lib /v/src/netbsd/src/obj/destdir.sparc/usr/lib/crt0.o /v/src/netbsd/src/obj/destdir.sparc/usr/lib/crti.o /v/src/netbsd/src/obj/destdir.sparc/usr/lib/crtbegin.o ldexp.o  -L/v/src/netbsd/src/obj/destdir.sparc/usr/lib -L/v/src/netbsd/src/obj/destdir.sparc/usr/lib  -lgcc -lc -lgcc /v/src/netbsd/src/obj/destdir.sparc/usr/lib/crtend.o /v/src/netbsd/src/obj/destdir.sparc/usr/lib/crtn.o
    dependall ===> regress/lib/libc/siginfo
    dependall ===> regress/lib/libc/siginfo/sigalrm
    nbmake: don't know how to make /v/src/netbsd/src/obj/destdir.i386/usr/include/assert.h. Stop
=================================================================^^^^
This is a sparc build.  It is looking for something in the i386 destdir.

    nbmake: stopped in /v/src/netbsd/src/regress/lib/libc/siginfo/sigalrm

    *** Failed target:  dependall
    *** Failed command: cd /v/src/netbsd/src/regress/lib/libc/siginfo/sigalrm; /v/src/netbsd/src/obj/tooldir.NetBSD-1.6ZG-i386/bin/nbmake realall
    *** Error code 2

    Stop.
    nbmake: stopped in /v/src/netbsd/src/regress/lib/libc/siginfo/sigalrm

I'm using:

    MKOBJ=yes
    MKOBJDIRS=yes			# remake obj dirs on each make-build

Is this operator error, or do the regression tests tangle state from
one build into the other?

-wolfgang
-- 
Wolfgang S. Rupprecht 		     http://www.wsrcc.com/wolfgang/
       The above "From:" address is valid.  Don't mess with it.