How to eliminate PHI nodes on pointer types?

Project Infrastructure LLVM Dev List Archives
#1

Somewhere during optimization PHI nodes on pointer types (including alloca instructions) are being introduced, and they persist through the scalar replacement of aggregates pass and others. I can't seem to find a combination of passes or transformations to get rid of them. Has anyone had this problem before, and know a transformation to eliminate it? My optimization passes rely on everything being in pure SSA form.

One more note, this particular optimization problem only seems to occur in llvm 2.8, not 3.0 - though I can't say for sure whether it never happens in 3.0.

Here's an example from the compiled code:

define private void @"vop_bindST@FFIFFI"(double* %news, double* %ss, i32* %isSConnected, double* %newt, double* %tt, i32* %isTConnected) nounwind alwaysinline {
entry:
   %"reg2mem alloca point" = bitcast i32 0 to i32
   %isSConnected1 = load i32* %isSConnected
   %0 = icmp eq i32 %isSConnected1, 0
   br i1 %0, label %false, label %true

true: ; preds = %entry
   br label %end

false: ; preds = %entry
   br label %end

end: ; preds = %false, %true
   %phi.in.reg2mem.0 = phi double* [ @s, %false ], [ %ss, %true ]
   %phi = load double* %phi.in.reg2mem.0
   store double %phi, double* %news
   %isTConnected4 = load i32* %isTConnected
   %1 = icmp eq i32 %isTConnected4, 0
   br i1 %1, label %false6, label %true5

true5: ; preds = %end
   br label %end7

false6: ; preds = %end
   br label %end7

end7: ; preds = %false6, %true5
   %phi9.in.reg2mem.0 = phi double* [ @t, %false6 ], [ %tt, %true5 ]
   %phi9 = load double* %phi9.in.reg2mem.0
   store double %phi9, double* %newt
   ret void
}

#2

Hi Andrew,

Somewhere during optimization PHI nodes on pointer types (including
alloca instructions) are being introduced, and they persist through the
scalar replacement of aggregates pass and others. I can't seem to find
a combination of passes or transformations to get rid of them. Has
anyone had this problem before, and know a transformation to eliminate
it? My optimization passes rely on everything being in pure SSA form.

why is this a problem? Presumably you are happy with phi nodes of integer
or float type, why are phi nodes of pointer type problematic?

Ciao, Duncan.

PS: I'm not sure what you mean by "pure SSA form". Phi nodes are needed for
values of pointer type exactly because these values are in SSA form...

#3

By "pure" I mean that the IR has no alloca instructions, ie. it all has been promoted to registers. It's easy to see that this is possible in the program I posted (by changing the PHI to one on loaded values rather than pointers). The presence of allocas is problematic since the analysis and lowering passes in my optimizer don't handle them.

Andrew

#4

Hi!

I'm not sure if I understand well your problem, but you are trying to get rid of the phi node right?
If so, you can use the -reg2mem pass, wich eliminates all phinode, but introduce sometimes others stuffs (like 'critical egdes').

I think I'm trying to do the same: eliminate all phinode from the code, using pointer to get rid of the dominance problem...
I'm trying with DemotePHIToStack() (you can find it in Utils/DemoteRegToStack.cpp), but it doesn't work for what I want.
Maybe it will help you...

Cheers