GCC build failed for native with your patch on 2003-04-17T21:46:51Z.

[Kazu Hirata]

Hi Geoff,

> I really don't understand what this patch is trying to do, it all
> looks wrong to me.  Why are you trying to make a NE look like a SET?

Consider

  if (reg0 != 100) goto L2;
  reg0 = 101;

This is equivalent to

  if (reg0 != 100) goto L2;
  reg0 += 1; /* cheaper */

because reg0 must be 100 if we don't jump to L2.  We can do this
conversion by thinking of the jump condition (reg0 != 100) as an
implicit set, much like how gcse.c detects implicit sets.

Did my follow-up patch seem to help?

> The problem appears to be bad codegen of the stage2 compiler.

Is there any way you could send me a preprocessed source of
unwind-dw2.c that reportedly failed to compile?  I'd like to try that
on powerpc-eabisim if it's not too different from
powerpc-apple-darwin6.3 at preprocessed source level.  By the way, the
last command from the failed build seems to be in stage1.  Is that
correct?  I don't see any stage1 in the command...

/Users/regress/tbox/native/build/gcc/xgcc
  -B/Users/regress/tbox/native/build/gcc/
  -B/Users/regress/tbox/objs/powerpc-apple-darwin6.3/bin/
  -B/Users/regress/tbox/objs/powerpc-apple-darwin6.3/lib/
  -isystem /Users/regress/tbox/objs/powerpc-apple-darwin6.3/include
  -isystem /Users/regress/tbox/objs/powerpc-apple-darwin6.3/sys-include
  -O2 -DIN_GCC
  -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
  -isystem ./include   -g  -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED
  -I. -I.
  -I/Users/regress/tbox/cvs-gcc/gcc/gcc
  -I/Users/regress/tbox/cvs-gcc/gcc/gcc/.
  -I/Users/regress/tbox/cvs-gcc/gcc/gcc/config
  -I/Users/regress/tbox/cvs-gcc/gcc/gcc/../include
  -fexceptions
  -c /Users/regress/tbox/cvs-gcc/gcc/gcc/unwind-dw2.c
  -o libgcc/./unwind-dw2.o

Thanks,

Kazu Hirata