Re: 108 compile problems for alpha

Jiann-Ming Su (js1@microwave.ph.msstate.edu)
Fri, 10 Jul 1998 12:10:41 -0500 (CDT)

> Looks like the same error that I had with previous versions. The problem
> lies in the Hiafa scheduler in egcs. If you get a copy of egcs that was
> compiled with the --disable-hiafa flag, then it should work. Otherwise,
> you'll have to rebuild egcs with that and it should work after that. For
> some reason, alot of Redhat users are installing the hiafa-enabled egcs...
>
> FYI, Richard Henderson is looking into this problem....
>

Same error with egcs-2.90.29 980515 (egcs-1.0.3 release):

gcc -D__KERNEL__ -I/usr/src/linux-2.1/include -Wall -Wstrict-prototypes
-O2 -fomit-frame-pointer -pipe -mno-fp-regs -ffixed-8 -Wa,-m21164a
-DBWX_USABLE -DBWIO_ENABLED -c -o page_alloc.o page_alloc.c
page_alloc.c: In function `__free_page':
page_alloc.c:190: internal error--unrecognizable insn:
(jump_insn 300 292 301 (return) -1 (nil)
(nil))
gcc: Internal compiler error: program cc1 got fatal signal 6{standard
input}:
make[2]: *** [page_alloc.o] Error 1
make[2]: Leaving directory `/usr/src/linux-2.1/mm'
make[1]: *** [first_rule] Error 2
make[1]: Leaving directory `/usr/src/linux-2.1/mm'
make: *** [linuxsubdirs] Error 2
Assembler messages:
{standard input}:197: Warning: Missing .end or .bend at end of file

I'm not sure why RH went with egcs in 5.1 when Linus has stated publically
that there are problems with egcs. If gcc-2.8.1 was stable, why not stick
with it? Oh well...

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu