Subject: ufsboot doesn't build
To: None <port-sun3@NetBSD.ORG>
From: der Mouse <mouse@Collatz.McRCIM.McGill.EDU>
List: port-sun3
Date: 06/03/1996 19:14:07
With a source tree supped in the AM of Friday May 31, I find that
ufsboot won't build, because libsa doesn't contain memcpy:
===> sys/arch/sun3/stand/ufsboot
cc -O -msoft-float -DSTANDALONE -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../libsa -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../../../../lib/libsa -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../../../.. -Werror -c boot.c
cc -O -msoft-float -DSTANDALONE -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../libsa -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../../../../lib/libsa -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../../../.. -Werror -c conf.c
cc -O -msoft-float -DSTANDALONE -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../libsa -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../../../../lib/libsa -I/sources/working-usr-src/sys/arch/sun3/stand/ufsboot/../../../.. -Werror -c version.c
ld -N -T 240000 -e start -o ufsboot ../libsa/SRT0.o ../libsa/SRT1.o boot.o conf.o version.o ../libsa/libsa.a
ufs.o: Undefined symbol `_memcpy' referenced from text segment
*** Error code 1
Stop.
As I recall from last time I went looking for this, the memcpy call is
actually generated by the compiler for a large structure assignment,
not written directly in ufs.c. Certainly /sys/lib/libsa/ufs.c (which
is my best guess at the source file for the ufs.o it's complaining
about) doesn't contain "memcpy".
And yes, I rebuilt and reinstalled the compiler first.
This is a source tree that I did a "make build" in an earlier version
of, then upgraded the sources without deleting the objects and rebuilt.
I'm hoping it's some nasty skew thus introduced, and have started a
full "make build" in the hope that that will cure it. However, I don't
really think that's what's the matter, and it will take a couple of
days in any case...and we're awfully close to 1.2, and I'd hate for it
to ship in a state that couldn't rebuild itself. So I thought I'd fire
this off now, rather than waiting for a rebuild which will probably go
into an "unexplained D wait" or "everything dumps core" failure mode at
least twice, losing probably two more days because I'm too busy to
check on it every hour :-( :-)
der Mouse
mouse@collatz.mcrcim.mcgill.edu