[llvm] [VPlan] Introduce scalar loop header in plan, remove VPLiveOut. (PR #109975)

[via llvm-commits]

================
@@ -45,10 +45,14 @@ define void @test_chained_first_order_recurrences_1(ptr %ptr) {
 ; CHECK-NEXT:  scalar.ph
 ; CHECK-NEXT:    EMIT vp<[[RESUME_1_P:%.*]]> = resume-phi vp<[[RESUME_1]]>, ir<22>
 ; CHECK-NEXT:    EMIT vp<[[RESUME_2_P:%.*]]>.1 = resume-phi vp<[[RESUME_2]]>.1, ir<33>
-; CHECK-NEXT:  No successors
+; CHECK-NEXT:  Successor(s): ir-bb<loop>
 ; CHECK-EMPTY:
-; CHECK-NEXT: Live-out i16 %for.1 = vp<[[RESUME_1_P]]>
-; CHECK-NEXT: Live-out i16 %for.2 = vp<[[RESUME_2_P]]>.1
+; CHECK-NEXT:  ir-bb<loop>:
+; CHECK-NEXT:    IR   %for.1 = phi i16 [ 22, %entry ], [ %for.1.next, %loop ] (extra operand: vp<[[RESUME_1_P]]>)
+; CHECK-NEXT:    IR   %for.2 = phi i16 [ 33, %entry ], [ %for.1, %loop ] (extra operand: vp<[[RESUME_2_P]]>.1)
----------------
ayalz wrote:

Nice!

https://github.com/llvm/llvm-project/pull/109975