gcc33 and amd64

[Gary Jennejohn]

Gerald Pfeifer writes:
> The gcc33 port keeps failing on our amd64 target, yet I have not seen
> any report on the gcc mailing lists.
> 
> I don't have any amd64 box with FreeBSD available -- could someone please
> try to build this port to rule out hardware issues in our build cluster?
> 
> Another possibility might be that our system compiler is broken and
> miscompiles gcc33 on this platform.  You should be able to test this
> by installing the gcc34 port and then setting CC to gcc34.
> 
> Any volunteers to help me with that?  Any further ideas?
> 
> stage1/xgcc -Bstage1/ -B/usr/local/x86_64-portbld-freebsd5.2/bin/   -g -O2 -D
> IN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -W
> traditional -pedantic -Wno-long-long   -DHAVE_CONFIG_H -DGENERATOR_FILE  -o g
> engtype \
>  gengtype.o gengtype-lex.o gengtype-yacc.o ../libiberty/libiberty.a
> ./gengtype
> gmake[2]: *** [s-gtype] Bus error (core dumped)
> gmake[2]: Leaving directory `/tmp/a/ports/lang/gcc33/work/build/gcc'
> gmake[1]: *** [stage2_build] Error 2
> gmake[1]: Leaving directory `/tmp/a/ports/lang/gcc33/work/build/gcc'
> gmake: *** [bootstrap-lean] Error 2
> *** Error code 2
> 

On my amd64 box running a freshly cvsup'ed and installed system (world
and kernel in 32 and 64 bit mode) I see:

32bit mode - ``make all'' succeeds
64bit mode - gentype fails with signal 10, just like you've observed

In 64bit mode I also see scads of ``ASM_OUTPUT_MAX_SKIP_ALIGN redefined''
warnings. I didn't watch the 32bit build closely enough to notice
whether it also emits these warnings.

---
Gary Jennejohn / garyj[at]jennejohn.org gj[at]freebsd.org gj[at]denx.de