This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tcb]: Add RHS to V_MUST_DEF
On Mon, 2004-10-04 at 15:03, Jeffrey A Law wrote:
> On Mon, 2004-10-04 at 08:05, Andrew MacLeod wrote:
> > On Sun, 2004-10-03 at 21:48, Daniel Berlin wrote:
> > > On Fri, 2004-10-01 at 14:48 -0400, Andrew MacLeod wrote:
> > > > On Fri, 2004-10-01 at 14:34, Daniel Berlin wrote:
> > Have you thought of any other solutions? Im just realizing this now, so
> > I havent had time to consider it yet. I think I might like Diego hacky
> > solution of looking at the base variable for code sinking if we have to
> > do this.
> FWIW, DSE has to solve the same core problem. Luckily we have
> immediate use information in DSE and it's pretty easy to fix up
> the RHS of the V_{MAY,MUST}_DEF operands.
Yeah, we discussed it a bit, and I think exposing the RHS in the MUST
defs is the right thing.
All we need to avoid the renaming in DCE is to figure out which argument
of the PHI is to be propagated. Dan thinks he can figure that out, then
its simply a matter of propagating the value and we have no more
renaming.
SO I think where we currently stand is:
- expose the RHS of the V_MUST_DEF as a KILL.
- This kill will also be a reference which shows up in the
immeidate_uses info since it is now a reference in an operand.
- DCE will use immediate_uses to do this, as it will be free in 4.1
Dan has issues with partial renaming (wants multiple RHS arguments to a
MUST DEF), but we'll visit that later :-)
Andrew