Re: [tree-ssa] New regressions as of 2003-11-04

[Daniel Berlin <dberlin at dberlin dot org>]

On Nov 4, 2003, at 3:51 PM, law@redhat.com wrote:

> In message <1067956749.23315.3.camel@frodo.toronto.redhat.com>, Diego 
> Novillo w
> rites:
> > Lots of changes yesterday produced new regressions in C, C++, Fortran
> > and mudflap.  I think I know what the mudflap problem is, so I'll take
> > care of that.  Could you folks take a look at the new regressions and
> > see if they're related to your changes?
> > 	 FAIL: gcc.dg/tree-ssa/20030814-4.c scan-tree-dump-times set = -1 0
> > 	 FAIL: gcc.dg/tree-ssa/20030814-5.c scan-tree-dump-times set = -1 0
> OK.  After looking at these some more.  These are failures that are a
> combination of the COND_EXPR lowering and some sillyness in PRE.
>
> Daniel -- did you actually bootstrap and regression test your PRE
> changes before installing them?

Of course. I bootstrap every change i commit. If I hadn't tried to 
bootstrap, i wouldn't have used the silly workaround to insert fake 
assignments instead of empty statements (which somehow screw up the 
statement linking and ends up with a cc1plus miscompilation, i'm still 
working on tracking it down)
I compared against the test results posted by diego for 
i686-pc-linux-gnu, i must have missed these two in the noise.
> Another option would be for the out-of-ssa pass to handle it.  I can 
> see
> how it's a partitioning problem, but I don't immediately see how to
> detect and optimizing this case in the out-of-ssa pass.
>
> Anyway, given that this is regression caused by the interaction between
> Zdenek's lowering code and Daniel's edge splitting
>  code, one of them
> really ought to be responsible for fixing it.

I'm not really sure how to fix it, and i'm trying to track down the 
whole "inserting empty statements breaks cc1plus" problem, so i'd 
appreciate if Zdenek could look at it.
He understands this stuff better anyway.

> jeff