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

[law at redhat dot com]

In message <6FFDC954-0F33-11D8-AC8B-000A95AF1FAE@dberlin.org>, Daniel Berlin wr
ites:
 >
 >On Nov 4, 2003, at 8:52 PM, Diego Novillo wrote:
 >
 >> On Tue, 2003-11-04 at 16:51, law@redhat.com wrote:
 >>
 >>> if (set != -1)
 >>>   goto lab1;
 >>> else
 >>>   goto lab2;
 >>>
 >>> lab1:
 >>>   set = 0;
 >>>   goto lab3;
 >>> lab2:
 >>>   set = -1;
 >>> lab3:
 >>>   [ ... ]
 >>>
 >>> Which looks precisely like what we've got now.
 >>>
 >> Yes, but what I had in mind is to go out of SSA in DOM order.  Andrew
 >> was thinking of doing some post-processing in the out-of-ssa pass.
 >> Maybe we could add this as well.
 >>
 >> Andrew, does that sound too gross for the out-of-ssa pass?
 >>
 >>
 >
 >Please be aware, that if you guys want, I can also temporarily fix the 
 >regression at the cost of possibly redoing dominators + dominance 
 >frontiers for every lexical expression we PRE (This is absolute worst 
 >case, where every expression has a different set of critical edges we 
 >might need to insert on.  I don't know how often this happens in 
 >practice).
 >
 >Of course, this is just a workaround, and a possibly expensive one at 
 >that, but it would make the regression go away for now.
 >
 >Just offering it as an option if none of us have time to try to fix it 
 >the right way, since it is a regression my patch exposed.
Na, let's actually get this fixed correctly.

I'm actually experimenting with some code to try and clean this up after
we've converted back to normal form, but before we run the useless
statement remover.

jeff