SLP/RISCV: add negative test for lrint (#55208) (#65611)

The issue #55208 describes a current deficiency of the SLPVectorizer,
namely that it doesn't vectorize code written with lrint, while similar
code written with rint is vectorized. Add a test corresponding to this
issue for the RISC-V target.

Author: artagnon

llvm/test/Transforms/SLPVectorizer/RISCV/fround.ll

@@ -0,0 +1,69 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 3
+; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=riscv32 -mattr=+m,+v | FileCheck %s
+; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=riscv64 -mattr=+m,+v | FileCheck %s
+; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=riscv32 -mattr=+v,+experimental-zvbb | FileCheck %s
+; RUN: opt < %s -passes=slp-vectorizer -S -mtriple=riscv64 -mattr=+v,+experimental-zvbb | FileCheck %s
+
+define <4 x float> @rint_v4f32(ptr %a) {
+; CHECK-LABEL: define <4 x float> @rint_v4f32(
+; CHECK-SAME: ptr [[A:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load <4 x float>, ptr [[A]], align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = call <4 x float> @llvm.rint.v4f32(<4 x float> [[TMP0]])
+; CHECK-NEXT:    ret <4 x float> [[TMP1]]
+;
+entry:
+  %0 = load <4 x float>, ptr %a
+  %vecext = extractelement <4 x float> %0, i64 0
+  %1 = call float @llvm.rint.f32(float %vecext)
+  %vecins = insertelement <4 x float> undef, float %1, i64 0
+  %vecext.1 = extractelement <4 x float> %0, i64 1
+  %2 = call float @llvm.rint.f32(float %vecext.1)
+  %vecins.1 = insertelement <4 x float> %vecins, float %2, i64 1
+  %vecext.2 = extractelement <4 x float> %0, i64 2
+  %3 = call float @llvm.rint.f32(float %vecext.2)
+  %vecins.2 = insertelement <4 x float> %vecins.1, float %3, i64 2
+  %vecext.3 = extractelement <4 x float> %0, i64 3
+  %4 = call float @llvm.rint.f32(float %vecext.3)
+  %vecins.3 = insertelement <4 x float> %vecins.2, float %4, i64 3
+  ret <4 x float> %vecins.3
+}
+
+define <4 x i64> @lrint_v4i64f32(ptr %a) {
+; CHECK-LABEL: define <4 x i64> @lrint_v4i64f32(
+; CHECK-SAME: ptr [[A:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load <4 x float>, ptr [[A]], align 16
+; CHECK-NEXT:    [[VECEXT:%.*]] = extractelement <4 x float> [[TMP0]], i64 0
+; CHECK-NEXT:    [[TMP1:%.*]] = call i64 @llvm.lrint.i64.f32(float [[VECEXT]])
+; CHECK-NEXT:    [[VECINS:%.*]] = insertelement <4 x i64> undef, i64 [[TMP1]], i64 0
+; CHECK-NEXT:    [[VECEXT_1:%.*]] = extractelement <4 x float> [[TMP0]], i64 1
+; CHECK-NEXT:    [[TMP2:%.*]] = call i64 @llvm.lrint.i64.f32(float [[VECEXT_1]])
+; CHECK-NEXT:    [[VECINS_1:%.*]] = insertelement <4 x i64> [[VECINS]], i64 [[TMP2]], i64 1
+; CHECK-NEXT:    [[VECEXT_2:%.*]] = extractelement <4 x float> [[TMP0]], i64 2
+; CHECK-NEXT:    [[TMP3:%.*]] = call i64 @llvm.lrint.i64.f32(float [[VECEXT_2]])
+; CHECK-NEXT:    [[VECINS_2:%.*]] = insertelement <4 x i64> [[VECINS_1]], i64 [[TMP3]], i64 2
+; CHECK-NEXT:    [[VECEXT_3:%.*]] = extractelement <4 x float> [[TMP0]], i64 3
+; CHECK-NEXT:    [[TMP4:%.*]] = call i64 @llvm.lrint.i64.f32(float [[VECEXT_3]])
+; CHECK-NEXT:    [[VECINS_3:%.*]] = insertelement <4 x i64> [[VECINS_2]], i64 [[TMP4]], i64 3
+; CHECK-NEXT:    ret <4 x i64> [[VECINS_3]]
+;
+entry:
+  %0 = load <4 x float>, ptr %a
+  %vecext = extractelement <4 x float> %0, i64 0
+  %1 = call i64 @llvm.lrint.i64.f32(float %vecext)
+  %vecins = insertelement <4 x i64> undef, i64 %1, i64 0
+  %vecext.1 = extractelement <4 x float> %0, i64 1
+  %2 = call i64 @llvm.lrint.i64.f32(float %vecext.1)
+  %vecins.1 = insertelement <4 x i64> %vecins, i64 %2, i64 1
+  %vecext.2 = extractelement <4 x float> %0, i64 2
+  %3 = call i64 @llvm.lrint.i64.f32(float %vecext.2)
+  %vecins.2 = insertelement <4 x i64> %vecins.1, i64 %3, i64 2
+  %vecext.3 = extractelement <4 x float> %0, i64 3
+  %4 = call i64 @llvm.lrint.i64.f32(float %vecext.3)
+  %vecins.3 = insertelement <4 x i64> %vecins.2, i64 %4, i64 3
+  ret <4 x i64> %vecins.3
+}
+
+declare float @llvm.rint.f32(float)
+declare i64 @llvm.lrint.i64.f32(float)