New bootstrap failure on ARM systems

[Richard Earnshaw]

> > > > 
> > > > > > It looks to me as though bbro has done a partial recalculation and got it 
> > > > > > wrong.
> > > > > 
> > > > > Hmm, then it must be the cfgcleanup pass just before bbro, as bbro don't
> > > > > do any recalculation.
> > > > 
> > > > It could be.  I was just going on the rtl dump names.
> > > > 
> > > > >  We can perhaps disable the cfgcleanup when ifcvt
> > > > > made CE or make it to recompute whole info fron the scratch?
> > > > 
> > > > Doesn't the CE pass already do a full recalculation?  If so, it would seem 
> > > > that a second one was unnecessary unless further changes were being made.
> > > 
> > > It is caused by the cfgcleanups and the attached patch fix it by
> > > recomputing from scratch.  How expensive it is?
> > 
> > I've started a full bootstrap with this patch.  I should know by tomorrow 
> > roughly what the time penalty is.
> 
> Thanks.  I will try to prepare more polished version of patch for this.
> Apparently I can't come with anything significantly better.
> Sorry for the delay.
> 

The cost on a full 3-stage bootstrap for c/c++/objc/g77/java+building 
libraries is about 1.2% compared with the last time I did a full bootstrap 
(2003/12/19).  Which isn't especially good news, but isn't too bad (some 
of that may be extra java code to compile).

It does at least bootstrap now though.

Native regression tests haven't been run (my machine tripped a kernel bug 
again last night), but cross tests on arm-elf have.

R.