rtlanal faults during bootstrap

[Robert Lipe]

After applying Jeff's patch, (Thanx, Jeff!) the 'make bootstrap' gets
further on i586-pc-sco3.2v5.0.4.

When we start to build libgcc2 with the freshly built cc1 for ELF/PIC,
we get into trouble.   Specifically, the line that kills it is:

./xgcc -B./ -O2   -DIN_GCC    -g -I./include   -g1 -DIN_LIBGCC2 -D__GCC_FLOAT_NO
T_NEEDED  -fPIC -I. -I../../egcs-970917/gcc -I../../egcs-970917/gcc/config -c -D
L_bb ../../egcs-970917/gcc/libgcc2.c -o _bb.o


If I remove the '-fPIC' from that line, it doesn't keel.
If I put back -fPIC and change -O2 to -O0, it doesn't keel.


To make matters worse, I currently have only an ancient GDB on this 
system, but I'll try to remedy that quickly.  The absence of an exact
line of fault and a good backtrace does not help me very much.

$ gdb cc1 core
GDB is free software and you are welcome to distribute copies of it
 under certain conditions; type "show copying" to see the conditions.
There is absolutely no warranty for GDB; type "show warranty" for details.
GDB 4.14-95q4 (i486-unknown-sco3.2v5.0.0elf),
Copyright 1995 Free Software Foundation, Inc...
Core was generated by `cc1'.
Program terminated with signal 11, Segmentation fault.
#0  0x87d1d in jmp_uses_reg_or_mem (x=0x2000003)
Breakpoint 1 at 0x160d4d
(gdb) where
#0  0x87d1d in jmp_uses_reg_or_mem (x=0x2000003)
(gdb) print code
$1 = INSN_LIST
(gdb) print fmt
$2 = 0x3 <Address 0x3 out of bounds>
(gdb) print i
$3 = 5015084



I'll go get a more believable gdb installed...

RJL