Subject: unpredictable gcc behavior (not signal 11) with recent slink
system
Date: Wed, Aug 04, 1999 at 03:29:13PM -0400
In reply to:Timothy Burt
Quoting Timothy Burt([EMAIL PROTECTED]):
> Hello All,
>
> I just last week installed slink 2.1_r2 on a new machine and am now
> having trouble with gcc during large compiles (the kernel, for
> example). So far, gcc has never reported signal 11, only signal 6.
> The problem is unpredictable in the sense that the compile always
> fails but rarely in the same place. Here are four snippets of the
> actual output during a make zImage (after running make menuconfig;
> make dep):
[ snip results ]
Tim
I just had to replace a motherboard thæt died, with a FIC VA-503+
and an AMD K6-2/450. I hd applied the heat sink compound and the
heat-sink/fan combo before fireing up the beast. Have been tweeking
the bios settings and decided to recompile the kernel as it had been
done last with a Cyrix6x86MX166. I then started getting results
similar to yours
[1]
-m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2 -DCPU=586
-c -o af_inet.o af_inet.c
Next token is 93 (']')
af_inet.c: In function `inet_ioctl':
af_inet.c:954: internal error--unrecognizable insn:
(jump_insn 56 55 57 (return) -1 (nil)
(nil))
gcc: Internal compiler error: program cc1 got fatal signal 6
make[3]: *** [af_inet.o] Error 1
[2]
vm86.c: In function `save_v86_state':
vm86.c:90: internal error--unrecognizable insn:
(jump_insn 514 510 515 (return) -1 (nil)
(nil))
gcc: Internal compiler error: program cc1 got fatal signal 6
make[1]: *** [vm86.o] Error 1
[3]
gcc -D__KERNEL__ -I/usr/src/linux-2.2.11/include -Wall
-Wstrict-prototypes -O2 -fomit-frame-pointer -pipe
-fno-strength-reduce
-m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2 -DCPU=586
-DMODULE -DMODVERSIONS -include
/usr/src/linux-2.2.11/include/linux/modversions.h -c -o ip6_fib.o
ip6_fib.c
ip6_fib.c: In function `fib6_del_route':
ip6_fib.c:921: internal error--unrecognizable insn:
(jump_insn 186 177 187 (return) -1 (nil)
(nil))
gcc: Internal compiler error: program cc1 got fatal signal 6
make[2]: *** [ip6_fib.o] Error 1
and on and on.
The only way to compile the kernel is to keep repeating the make
bzImage. After doing that 10-15 tikes the compile completes and then
you have to do the same for the modules compile!
I am running at 100Mhz, not overclocked, and have tried setting the
Gate A20 Option to normal and Fast with no change.
I will let you know what else does/dosen't work when I finish. My
last try will be loading slackware 3.5 back up on the new HD and see
if it just might be a glib problem.
Wayne
--
"... one of the main causes of the fall of the Roman Empire was that,
lacking zero, they had no way to indicate successful termination of
their C programs." -- Robert Firth
_______________________________________________________
Wayne T. Topa <[EMAIL PROTECTED]>
