Does someone object to this patch?
Tue Feb 1 07:56:00 GMT 2000
On Tue, 1 Feb 2000, Jeffrey A Law wrote:
> In message <Pine.LNX.email@example.com>you
> > I have a patch that splits off some code from reload_cse_regs and makes
> > it more general so it can be used from other optimizers that need to
> > do some cse.
> > One optimization is lost in that patch; some other optimization
> > opportunities are gained. Joern Rennecke who is reviewing the patch
> > suggested asking the gcc list whether anyone objects to removing the
> > optimization below (this patch obviously isn't meant to go in in this
> > form). I haven't found a test case that makes the removed code actually
> > trigger, but maybe I'm testing the wrong code.
> > Comments, anyone?
> What's unclear to me is why this code is problematical when you split off
> the reload_cse_regs code. Knowing that might help people evaluate the
> pros and cons of your suggestion.
It's not exactly problematical - it's just that keeping it adds a bit more
code in cselib and I'd rather avoid that for now. I'd rather we get this
installed without making too many modifications (each of which risks new
bugs) and add enhancements after that.
I've done a lot of tests and haven't come across a single example where
reload_cse_regs would generate worse code after the patch, so I just think
that it's better not to try to add too much additional code before installing
it. The patch is already fairly large.
More information about the Gcc-patches