[parisc-linux] abort in eliminate_regs compiling glob.c from glibc

[John David Anglin]

The following abort occurs with a relatively current version of gcc from
the cvs under hpux and the parisc-linux version of the compiler:

hppa-linux-gcc ../sysdeps/generic/glob.c -c -O3 -Wall -Winline -Wstrict-prototypes -Wwrite-strings      -I../include -I. -I/home/dave/puffin/glibc/objdir/posix -I.. -I../libio  -I/home/dave/puffin/glibc/objdir -I../sysdeps/hppa/elf -I../linuxthreads/sysdeps/unix/sysv/linux/hppa -I../linuxthreads/sysdeps/unix/sysv/linux -I../linuxthreads/sysdeps/pthread -I../linuxthreads/sysdeps/unix/sysv -I../linuxthreads/sysdeps/unix -I../linuxthreads/sysdeps/hppa -I../sysdeps/unix/sysv/linux/hppa -I../sysdeps/unix/sysv/linux -I../sysdeps/gnu -I../sysdeps/unix/common -I../sysdeps/unix/mman -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/hppa/hppa1.1 -I../sysdeps/wordsize-32 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/ldbl-128 -I../sysdeps/hppa/fpu -I../sysdeps/hppa -I../sysdeps/ieee754 -I../sysdeps/generic/elf -I../sysdeps/generic  -nostdinc -isystem /usr/local/puffin/lib/gcc-lib/hppa-linux/2.96/include -isystem !
/home/dave/puffin/linux/include -D_LIBC_REENTRANT -include ../include/libc-symbols.h     -o /home/dave/puffin/glibc/objdir/posix/glob.o
../sysdeps/generic/glob.c: In function `glob_in_dir':
../sysdeps/generic/glob.c:1446: Internal compiler error in , at reload1.c:2516
   Please submit a full bug report.
   See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
make[2]: *** [/home/dave/puffin/glibc/objdir/posix/glob.o] Error 1
make[2]: Leaving directory `/home/dave/puffin/glibc/posix'
make[1]: *** [posix/subdir_lib] Error 2
make[1]: Leaving directory `/home/dave/puffin/glibc'
make: *** [all] Error 2

The insn that causes the fault is the following:

Breakpoint 2, eliminate_regs_in_insn (insn=0x406a0ba0, replace=0)
    at ../../gcc/reload1.c:2826
2826      if (! insn_is_asm && icode < 0)
(gdb) p debug_rtx (insn)
(insn/s 2711 2709 2719 (set (reg:SI 6 %r6)
        (reg:SI 28 %r28)) 69 {pre_ldw-4} (insn_list 2708 (insn_list:REG_DEP_ANTI 2696 (insn_list:REG_DEP_ANTI 2702 (insn_list:REG_DEP_ANTI 2697 (nil)))))
    (expr_list:REG_DEAD (reg:SI 28 %r28)
        (insn_list:REG_RETVAL 2708 (expr_list:REG_EQUAL (expr_list (use (mem:BLK (scratch) 0))
                    (expr_list (symbol_ref/v:SI ("@strlen"))
                        (expr_list (reg/v:SI 4 %r4)
                            (nil))))
                (nil)))))

As can be seen, there is a use in the REG notes which will cause eliminate_regs
to abort if it is called to process the notes of this insn.  It is called
from this code which is near the end of eliminate_regs_in_insn:

  for (ep = reg_eliminate; ep < &reg_eliminate[NUM_ELIMINABLE_REGS]; ep++)
    {
      if (ep->previous_offset != ep->offset && ep->ref_outside_mem)
        ep->can_eliminate = 0;

      ep->ref_outside_mem = 0;

      if (ep->previous_offset != ep->offset)
	val = 1;
    }

 done:
  /* If we changed something, perform elimination in REG_NOTES.  This is
     needed even when REPLACE is zero because a REG_DEAD note might refer
     to a register that we eliminate and could cause a different number
     of spill registers to be needed in the final reload pass than in
     the pre-passes.  */
  if (val && REG_NOTES (insn) != 0)
    REG_NOTES (insn) = eliminate_regs (REG_NOTES (insn), 0, REG_NOTES (insn));

ep->previous_offset is not equal ep->offset here and thus val is 1.

The use would appear to arise from the rtl generated by expand_builtin_strlen.
Anybody got any thoughts on how to fix this.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)