Re: update dwarf2 asm unwind info [hppa64-*-* failures]

[Steve Ellcey <sje at cup dot hp dot com>]

On Fri, 2008-08-08 at 17:12 -0700, Richard Henderson wrote:
> Steve Ellcey wrote:
> > Any suggestions on what to look for to see if this is a problem with the
> > CFI changes?
> 
> You could compare the output of readelf -wf with and without
> -fno-dwarf2-cfi-asm.  The outputs should be the same.
> 
> 
> r~

I am seeing lots of difference on hppa64-*-*.  If I run readelf on two
executables built with and without the changes I see differences right
away.  The changes causing the problem seem to be in the unwind code
built into libgcc, not in the test program.  Using the flag to change
how I compile the test doesn't affect things.  Changing how I build
libgcc does change the behavour.  Most of the C++ tests that involve
throw/catch are failing with the CFI changes.

Steve Ellcey
sje@cup.hp.com

without the CFI changes:


The section .eh_frame contains:

00000000 00000024 00000000 CIE
  Version:               1
  Augmentation:          "zPL"
  Code alignment factor: 1
  Data alignment factor: 8
  Return address column: 2
  Augmentation data:     50 00 00 00 00 00 00 40 00 00 00 00 02 05 00 00

With the CFI changes:

The section .eh_frame contains:

00000000 00000010 00000000 CIE
  Version:               1
  Augmentation:          "zR"
  Code alignment factor: 4
  Data alignment factor: -8
  Return address column: 2
  Augmentation data:     1b