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

[Jan Hubicka <hubicka at ucw dot cz>]

> In message <1068041513.14345.952.camel@p4>, Andrew MacLeod writes:
>  >remove_useless_stmts_and_vars_cond could keep the cfg updated if it
>  >wanted to. It didnt before becuase it was thrown away immediately
>  >aftewards, so what was the point. Now perhaps there is a point.
> Correct.  Nothing after remove_useless_stmts_and_vars cared about the CFG
> and by not doing silly work, remove_useless_stmts_and_vars actually improved
> compilation times and kept the code reasonably simple.
> 
> So one thing we need to look at before we go and make that code CFG aware
> is whether or not it's actually going to be a good thing from the various
> angles, including the ability to eliminate obviously useless statements
> created by the out-of-ssa translation, compile time concerns, the need to
> actually eliminate all the empty statements, etc etc.

We need to preserve CFG trought RTL expansion in order to be able to
read profile into it and maintain it until end of compilation.
I have patches for the CFG aware RTL expansion and sub basic block
creation ready, just waiting for other issues to settle down (mainly the
problems with CFG being corrupt before out-of-ssa) before sending htem
out.

So I would love us to preserve CFG across remove_useless_stmts_and_vars.
Zdenek already sent patch for that AFAIK.

Honza
> 
> Jeff