[VPlan] Pass live-ins used as exit values straight to live-out.

Live-ins that are used as exit values don't need to be extracted, they
can be passed through directly. This fixes a crash when trying to
extract from a live-in.

Fixes llvm#106257.

Author: fhahn

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

@@ -8648,6 +8648,11 @@ addUsersInExitBlock(VPlan &Plan,
 
   // Introduce VPUsers modeling the exit values.
   for (const auto &[ExitPhi, V] : ExitingValuesToFix) {
+    // Pass live-in values used by exit phis directly through to the live-out.
+    if (V->isLiveIn()) {
+      Plan.addLiveOut(ExitPhi, V);
+      continue;
+    }
     VPValue *Ext = B.createNaryOp(
         VPInstruction::ExtractFromEnd,
         {V, Plan.getOrAddLiveIn(ConstantInt::get(

llvm/test/Transforms/LoopVectorize/extract-from-end-vector-constant.ll

@@ -0,0 +1,108 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -p loop-vectorize -force-vector-width=4 -S %s | FileCheck %s
+
+target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128"
+
+define i64 @exit_value_scalar_live_in(ptr %dst, i64 %in) {
+; CHECK-LABEL: define i64 @exit_value_scalar_live_in(
+; CHECK-SAME: ptr [[DST:%.*]], i64 [[IN:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*]]:
+; CHECK-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK:       [[VECTOR_PH]]:
+; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
+; CHECK:       [[VECTOR_BODY]]:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i16, ptr [[DST]], i64 [[TMP0]]
+; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i16, ptr [[TMP1]], i32 0
+; CHECK-NEXT:    store <4 x i16> zeroinitializer, ptr [[TMP2]], align 2
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK:       [[MIDDLE_BLOCK]]:
+; CHECK-NEXT:    br i1 false, label %[[EXIT:.*]], label %[[SCALAR_PH]]
+; CHECK:       [[SCALAR_PH]]:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT:    br label %[[LOOP:.*]]
+; CHECK:       [[LOOP]]:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT_1:%.*]], %[[LOOP]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr i16, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT:    store i16 0, ptr [[GEP]], align 2
+; CHECK-NEXT:    [[IV_NEXT_1]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[EC:%.*]] = icmp ult i64 [[IV]], 1000
+; CHECK-NEXT:    br i1 [[EC]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK:       [[EXIT]]:
+; CHECK-NEXT:    [[RES:%.*]] = phi i64 [ [[IN]], %[[LOOP]] ], [ [[IN]], %[[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret i64 [[RES]]
+;
+entry:
+  br label %loop
+
+loop:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next.1, %loop ]
+  %gep = getelementptr i16, ptr %dst, i64 %iv
+  store i16 0, ptr %gep, align 2
+  %iv.next.1 = add i64 %iv, 1
+  %ec = icmp ult i64 %iv, 1000
+  br i1 %ec, label %loop, label %exit
+
+exit:
+  %res = phi i64 [ %in, %loop ]
+  ret i64 %res
+}
+
+define <2 x i64> @exit_value_vector_live_in(ptr %dst) {
+; CHECK-LABEL: define <2 x i64> @exit_value_vector_live_in(
+; CHECK-SAME: ptr [[DST:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*]]:
+; CHECK-NEXT:    br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK:       [[VECTOR_PH]]:
+; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
+; CHECK:       [[VECTOR_BODY]]:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i16, ptr [[DST]], i64 [[TMP0]]
+; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i16, ptr [[TMP1]], i32 0
+; CHECK-NEXT:    store <4 x i16> zeroinitializer, ptr [[TMP2]], align 2
+; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; CHECK-NEXT:    br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK:       [[MIDDLE_BLOCK]]:
+; CHECK-NEXT:    br i1 false, label %[[EXIT:.*]], label %[[SCALAR_PH]]
+; CHECK:       [[SCALAR_PH]]:
+; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT:    br label %[[LOOP:.*]]
+; CHECK:       [[LOOP]]:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT_1:%.*]], %[[LOOP]] ]
+; CHECK-NEXT:    [[GEP:%.*]] = getelementptr i16, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT:    store i16 0, ptr [[GEP]], align 2
+; CHECK-NEXT:    [[IV_NEXT_1]] = add i64 [[IV]], 1
+; CHECK-NEXT:    [[EC:%.*]] = icmp ult i64 [[IV]], 1000
+; CHECK-NEXT:    br i1 [[EC]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK:       [[EXIT]]:
+; CHECK-NEXT:    [[RES:%.*]] = phi <2 x i64> [ zeroinitializer, %[[LOOP]] ], [ zeroinitializer, %[[MIDDLE_BLOCK]] ]
+; CHECK-NEXT:    ret <2 x i64> [[RES]]
+;
+entry:
+  br label %loop
+
+loop:
+  %iv = phi i64 [ 0, %entry ], [ %iv.next.1, %loop ]
+  %gep = getelementptr i16, ptr %dst, i64 %iv
+  store i16 0, ptr %gep, align 2
+  %iv.next.1 = add i64 %iv, 1
+  %ec = icmp ult i64 %iv, 1000
+  br i1 %ec, label %loop, label %exit
+
+exit:
+  %res = phi <2 x i64> [ zeroinitializer, %loop ]
+  ret <2 x i64> %res
+}
+;.
+; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
+; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
+; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META1]], [[META2]]}
+; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META2]], [[META1]]}
+;.