egcs-ss-970821 on m68k-next-nextstep3

egcs-ss-970821 on m68k-next-nextstep3

[Toon Moene]

After Jim removed the obstacles to make a 3-stage build of gcc on  
m68k-next-nextstep3 (with make compare succeeding), another problem  
popped up:

stage2/xgcc -Bstage2/ -c  -DIN_GCC   -O -g -O3  -W -Wall    -If  
-I./f -I. -I. -I./config f/data.c -o f/data.o
f/data.c: In function `ffedata_gather_':
f/data.c:1333: warning: `unexp' might be used uninitialized in this  
function
xgcc: Internal compiler error: program cc1 got fatal signal 6

i.e. the stage3 compiler (hope I have the stageN numbering right  
now *) can't compile the data.c file in the Fortran language  
directory.

This is what GDB has to say about it:

% gdb stage2/cc1
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 4.7 (NeXT 3.1), Copyright 1992 Free Software Foundation, Inc...
Reading symbols from  
/Users/toon/Unix/compilers/egcs-ss-970821/stage2/cc1...Value with  
psymtab but no minimal symbol: 165048 in tree.c.
Value with psymtab but no minimal symbol: 421048 in varasm.c.
Value with psymtab but no minimal symbol: 421048 in varasm.c.
....
Value with psymtab but no minimal symbol: 861146 in recog.c.
done.
Reading symbols from /usr/shlib/libsys_s.B.shlib...done.

Breakpoint 1 at 0x505676a
(gdb) run data.i  -quiet -dumpbase data.c -g -O -O3 -W -Wall -o data.s
Starting program:  
/Users/toon/Unix/compilers/egcs-ss-970821/stage2/cc1 data.i  -quiet  
-dumpbase data.c -g -O -O3 -W -Wall -o data.s
f/data.c: In function `ffedata_gather_':
f/data.c:1333: warning: `unexp' might be used uninitialized in this  
function

Breakpoint 1, 0x505676a in abort ()
(gdb) bt
Reading in symbols for reload.c...done.
Reading in symbols for reload1.c...done.
Reading in symbols for global.c...done.
Reading in symbols for toplev.c...done.
Reading in symbols for c-decl.c...done.
Reading in symbols for ./c-parse.c...done.
#0  0x505676a in abort ()
#1  0xb9cce in push_reload (in=0x586b06, out=0x586ade,  
inloc=0x586b9a, outloc=0x586ba6, class=NO_REGS, inmode=DImode,  
outmode=DImode, strict_low=0, optional=0, opnum=0,  
type=RELOAD_OTHER) at reload.c:1156
#2  0xbe656 in find_reloads (ind_levels=0, insn=0x586bee,  
live_known=1, reload_reg_p=0x12c11c, replace=0) at reload.c:3613
#3  0xc2b98 in reload (first=0x58f6fc, global=1, dumpfile=0x0) at  
reload1.c:1120
#4  0xb752e in global_alloc (file=0x0) at global.c:580
#5  0x22508 in rest_of_compilation (decl=void) at toplev.c:3448
#6  0x10084 in finish_function (nested=0) at c-decl.c:7012
#7  0x31a2 in yyparse () at c-parse.y:352
#8  0x1f6d8 in compile_file (name=0x3fffc55 "data.i") at toplev.c:2470
#9  0x237dc in main (argc=12, argv=0x2, envp=0x3fffbf4) at toplev.c:4309

i.e. push_reload is called with class == NO_REGS.which it shouldn't.

Hope someone can provide some insight, or ideas how to investigate this.

TIA,
Toon.

*) There's an interesting `problem' with stage numbering.  I  
assumed, subconsciously, dealing with C programmers, that stage  
numbering would start at 0.  Unfortunately, it starts at 1, leading  
to misunderstanding between Fortran users (who try to adjust for  
0-based counting) and C programmers, somehow getting used to  
counting - in this specific case - starting at 1.

Re: egcs-ss-970821 on m68k-next-nextstep3

[Jeffrey A Law]

  In message <9708231749.AA05178@moene.indiv.nluug.nl>you write:
  > stage2/xgcc -Bstage2/ -c  -DIN_GCC   -O -g -O3  -W -Wall    -If  
  > -I./f -I. -I. -I./config f/data.c -o f/data.o
  > f/data.c: In function `ffedata_gather_':
  > f/data.c:1333: warning: `unexp' might be used uninitialized in this  
  > function
  > xgcc: Internal compiler error: program cc1 got fatal signal 6
  > 
  > i.e. the stage3 compiler (hope I have the stageN numbering right  
  > now *) can't compile the data.c file in the Fortran language  
  > directory.
Close :-) the stage2 compiler chokes on data.c during the stage3
build.

I've got a 68k bootstrap running right now; it'll take a _long_
time to finish, but hopefully it'll enable me to track down some
of your problems (assuming they're generic 68k bugs).

Jeff

Re: egcs-ss-970821 on m68k-next-nextstep3

[Jeffrey A Law]

  In message <9708231749.AA05178@moene.indiv.nluug.nl>you write:
  > After Jim removed the obstacles to make a 3-stage build of gcc on  
  > m68k-next-nextstep3 (with make compare succeeding), another problem  
  > popped up:
  > 
  > stage2/xgcc -Bstage2/ -c  -DIN_GCC   -O -g -O3  -W -Wall    -If  
  > -I./f -I. -I. -I./config f/data.c -o f/data.o
  > f/data.c: In function `ffedata_gather_':
  > f/data.c:1333: warning: `unexp' might be used uninitialized in this  
  > function
  > xgcc: Internal compiler error: program cc1 got fatal signal 6
Just a note, I've been able to reproduce this on a 68k box I've
still got access to -- I'm working on it.

jeff