llvm/test/Transforms/LoopVectorize/extract-value-widen.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
; RUN: opt < %s -passes=loop-vectorize -force-vector-width=2 -force-vector-interleave=1 -S 2>%t | FileCheck %s

; Regression test for https://github.com/llvm/llvm-project/issues/193275

define void @func(ptr %p) {
; CHECK-LABEL: define void @func(
; CHECK-SAME: ptr [[P:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    br 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:%.*]] = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 [[INDEX]], i64 1)
; CHECK-NEXT:    [[TMP2:%.*]] = extractvalue { i64, i1 } [[TMP0]], 0
; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr [8 x i8], ptr [[P]], i64 [[TMP2]]
; CHECK-NEXT:    store <2 x i64> zeroinitializer, ptr [[TMP3]], align 8
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], -9223372036854775808
; CHECK-NEXT:    br i1 [[TMP4]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %loop.header

loop.header:
  %iv.1 = phi i64 [ 0, %entry ], [ %iv.next1, %loop.header ]
  %iv.2 = phi i64 [ 0, %entry ], [ %iv.next2, %loop.header ]
  %res1 = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 %iv.1, i64 1)
  %iv.next1 = extractvalue { i64, i1 } %res1, 0
  %arrayidx37 = getelementptr [8 x i8], ptr %p, i64 %iv.next1
  store i64 0, ptr %arrayidx37, align 8
  %res2 = tail call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 %iv.2, i64 1)
  %iv.next2 = extractvalue { i64, i1 } %res2, 0
  %overflow2 = extractvalue { i64, i1 } %res2, 1
  br i1 %overflow2, label %exit, label %loop.header

exit:
  ret void
}

; The overflow bit needs all lanes, so the call is widened to a struct of
; vectors, while only the first lane of the sum is used. The sum must be
; extracted from the struct with a wide extractvalue.
define void @extract_first_lane_from_widened_struct(ptr %p, ptr noalias %q) {
; CHECK-LABEL: define void @extract_first_lane_from_widened_struct(
; CHECK-SAME: ptr [[P:%.*]], ptr noalias [[Q:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    br 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:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP0:%.*]] = call { <2 x i64>, <2 x i1> } @llvm.sadd.with.overflow.v2i64(<2 x i64> [[VEC_IND]], <2 x i64> splat (i64 1))
; CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <2 x i64>, <2 x i1> } [[TMP0]], 0
; CHECK-NEXT:    [[TMP2:%.*]] = extractvalue { <2 x i64>, <2 x i1> } [[TMP0]], 1
; CHECK-NEXT:    [[TMP3:%.*]] = zext <2 x i1> [[TMP2]] to <2 x i64>
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr i64, ptr [[Q]], i64 [[INDEX]]
; CHECK-NEXT:    store <2 x i64> [[TMP3]], ptr [[TMP4]], align 8
; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP1]], i64 0
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr i64, ptr [[P]], i64 [[TMP5]]
; CHECK-NEXT:    store <2 x i64> zeroinitializer, ptr [[TMP6]], align 8
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], splat (i64 2)
; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; CHECK-NEXT:    br i1 [[TMP7]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[EXIT:.*]]
; CHECK:       [[EXIT]]:
; CHECK-NEXT:    ret void
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %res = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 %iv, i64 1)
  %iv.next = extractvalue { i64, i1 } %res, 0
  %ov = extractvalue { i64, i1 } %res, 1
  %ov.ext = zext i1 %ov to i64
  %gep.q = getelementptr i64, ptr %q, i64 %iv
  store i64 %ov.ext, ptr %gep.q, align 8
  %gep.p = getelementptr i64, ptr %p, i64 %iv.next
  store i64 0, ptr %gep.p, align 8
  %ec = icmp eq i64 %iv.next, 1000
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}