[rfc] Fix PowerPC displaced stepping regression

[Ulrich Weigand]

Pedro Alves wrote:
> On Sunday 27 September 2009 22:47:13, Ulrich Weigand wrote:
> > + # the the displaced instruction identified by CLOSURE.  If false,
> 
> Double "the".

Fixed, thanks!

> Hmmm, does this mean that a breakpoint at the start of an
> atomic sequence instruction wouldn't be displaced stepped properly,
> as in, you'd trip on the same issue that happens when stepping over
> an atomic sequence without displaced stepping?

Yes, that's true.

> (If broken, this was already broken before your patch and even
> before the regression your patch fixes)

Indeed.  For now, I'm OK with restoring the state before the
regression.  The new mechanism should allow fixing this particular
corner case as well, I hope, but this will be more involved ...

> ( A nice stress test of the displaced stepping support is to run the
> whole testsuite with "set displaced-stepping on". )
> 
> I've now read through the patch carefully, and didn't spot
> anything wrong.  I think this would be safe for 7.0 as well.

OK, thanks for the review!

I've now checked in the patch to mainline.  I'll wait with
checking into the branch until Joel has agreed how we should
handle it ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com