[llvm] [SCEV] Use power of two facts involving vscale when inferring wrap flags (PR #101380)

[Nikita Popov via llvm-commits]

================
@@ -374,15 +374,16 @@ define void @vscale_slt_noflags(ptr nocapture %A, i32 %n) mustprogress vscale_ra
 ; CHECK-NEXT:    %vscale = call i32 @llvm.vscale.i32()
 ; CHECK-NEXT:    --> vscale U: [2,1025) S: [2,1025)
 ; CHECK-NEXT:    %i.05 = phi i32 [ %add, %for.body ], [ 0, %entry ]
-; CHECK-NEXT:    --> {0,+,vscale}<%for.body> U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %for.body: Computable }
+; CHECK-NEXT:    --> {0,+,vscale}<%for.body> U: full-set S: full-set Exits: (vscale * ((-1 + %n) /u vscale))<nuw> LoopDispositions: { %for.body: Computable }
----------------
nikic wrote:

Drop TODO above.

https://github.com/llvm/llvm-project/pull/101380