[CodeGen][NFC] Run SROA on complex range tests #131925
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was referenced Mar 18, 2025
This stack of pull requests is managed by Graphite. Learn more about stacking. |
|
@llvm/pr-subscribers-clang @llvm/pr-subscribers-clang-codegen Author: Mészáros Gergely (Maetveis) ChangesRun mem2reg and sroa passes on the complex range tests, to make them Patch is 888.51 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/131925.diff 2 Files Affected:
diff --git a/clang/test/CodeGen/cx-complex-range-real.c b/clang/test/CodeGen/cx-complex-range-real.c
index 90c37e8669b08..a998869ac39e2 100644
--- a/clang/test/CodeGen/cx-complex-range-real.c
+++ b/clang/test/CodeGen/cx-complex-range-real.c
@@ -1,104 +1,60 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
-// RUN: %clang_cc1 %s -O0 -emit-llvm -triple x86_64-unknown-unknown \
-// RUN: -o - | FileCheck %s --check-prefix=FULL
+// RUN: %clang_cc1 %s -O0 -disable-O0-optnone -emit-llvm -triple x86_64-unknown-unknown \
+// RUN: -o - | opt -S -passes=mem2reg,sroa | FileCheck %s --check-prefix=FULL
-// RUN: %clang_cc1 %s -O0 -emit-llvm -triple x86_64-unknown-unknown \
-// RUN: -complex-range=improved -o - | FileCheck %s --check-prefix=IMPRVD
+// RUN: %clang_cc1 %s -O0 -disable-O0-optnone -emit-llvm -triple x86_64-unknown-unknown \
+// RUN: -complex-range=improved -o - | opt -S -passes=mem2reg,sroa | FileCheck %s --check-prefix=IMPRVD
-// RUN: %clang_cc1 %s -O0 -emit-llvm -triple x86_64-unknown-unknown \
-// RUN: -complex-range=promoted -o - | FileCheck %s --check-prefix=PRMTD
+// RUN: %clang_cc1 %s -O0 -disable-O0-optnone -emit-llvm -triple x86_64-unknown-unknown \
+// RUN: -complex-range=promoted -o - | opt -S -passes=mem2reg,sroa | FileCheck %s --check-prefix=PRMTD
// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -complex-range=promoted \
-// RUN: -ffp-contract=off -frounding-math -ffp-exception-behavior=strict \
-// RUN: -emit-llvm -o - %s | FileCheck %s --check-prefix=PRMTD_STRICT
+// RUN: -ffp-contract=off -frounding-math -ffp-exception-behavior=strict -disable-O0-optnone \
+// RUN: -emit-llvm -o - %s | opt -S -passes=mem2reg,sroa | FileCheck %s --check-prefix=PRMTD_STRICT
// FULL-LABEL: define dso_local <2 x float> @mulaf(
// FULL-SAME: <2 x float> noundef [[A_COERCE:%.*]], float noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {
// FULL-NEXT: [[ENTRY:.*:]]
-// FULL-NEXT: [[RETVAL:%.*]] = alloca { float, float }, align 4
-// FULL-NEXT: [[A:%.*]] = alloca { float, float }, align 4
-// FULL-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// FULL-NEXT: store <2 x float> [[A_COERCE]], ptr [[A]], align 4
-// FULL-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// FULL-NEXT: [[A_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
-// FULL-NEXT: [[A_REAL:%.*]] = load float, ptr [[A_REALP]], align 4
-// FULL-NEXT: [[A_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
-// FULL-NEXT: [[A_IMAG:%.*]] = load float, ptr [[A_IMAGP]], align 4
-// FULL-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// FULL-NEXT: [[MUL_RL:%.*]] = fmul float [[A_REAL]], [[TMP0]]
-// FULL-NEXT: [[MUL_IL:%.*]] = fmul float [[A_IMAG]], [[TMP0]]
-// FULL-NEXT: [[RETVAL_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 0
-// FULL-NEXT: [[RETVAL_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 1
-// FULL-NEXT: store float [[MUL_RL]], ptr [[RETVAL_REALP]], align 4
-// FULL-NEXT: store float [[MUL_IL]], ptr [[RETVAL_IMAGP]], align 4
-// FULL-NEXT: [[TMP1:%.*]] = load <2 x float>, ptr [[RETVAL]], align 4
-// FULL-NEXT: ret <2 x float> [[TMP1]]
+// FULL-NEXT: [[A_SROA_0_0_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 0
+// FULL-NEXT: [[A_SROA_0_4_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 1
+// FULL-NEXT: [[MUL_RL:%.*]] = fmul float [[A_SROA_0_0_VEC_EXTRACT]], [[B]]
+// FULL-NEXT: [[MUL_IL:%.*]] = fmul float [[A_SROA_0_4_VEC_EXTRACT]], [[B]]
+// FULL-NEXT: [[RETVAL_SROA_0_0_VEC_INSERT:%.*]] = insertelement <2 x float> undef, float [[MUL_RL]], i32 0
+// FULL-NEXT: [[RETVAL_SROA_0_4_VEC_INSERT:%.*]] = insertelement <2 x float> [[RETVAL_SROA_0_0_VEC_INSERT]], float [[MUL_IL]], i32 1
+// FULL-NEXT: ret <2 x float> [[RETVAL_SROA_0_4_VEC_INSERT]]
//
// IMPRVD-LABEL: define dso_local <2 x float> @mulaf(
// IMPRVD-SAME: <2 x float> noundef [[A_COERCE:%.*]], float noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {
// IMPRVD-NEXT: [[ENTRY:.*:]]
-// IMPRVD-NEXT: [[RETVAL:%.*]] = alloca { float, float }, align 4
-// IMPRVD-NEXT: [[A:%.*]] = alloca { float, float }, align 4
-// IMPRVD-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// IMPRVD-NEXT: store <2 x float> [[A_COERCE]], ptr [[A]], align 4
-// IMPRVD-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// IMPRVD-NEXT: [[A_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
-// IMPRVD-NEXT: [[A_REAL:%.*]] = load float, ptr [[A_REALP]], align 4
-// IMPRVD-NEXT: [[A_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
-// IMPRVD-NEXT: [[A_IMAG:%.*]] = load float, ptr [[A_IMAGP]], align 4
-// IMPRVD-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// IMPRVD-NEXT: [[MUL_RL:%.*]] = fmul float [[A_REAL]], [[TMP0]]
-// IMPRVD-NEXT: [[MUL_IL:%.*]] = fmul float [[A_IMAG]], [[TMP0]]
-// IMPRVD-NEXT: [[RETVAL_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 0
-// IMPRVD-NEXT: [[RETVAL_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 1
-// IMPRVD-NEXT: store float [[MUL_RL]], ptr [[RETVAL_REALP]], align 4
-// IMPRVD-NEXT: store float [[MUL_IL]], ptr [[RETVAL_IMAGP]], align 4
-// IMPRVD-NEXT: [[TMP1:%.*]] = load <2 x float>, ptr [[RETVAL]], align 4
-// IMPRVD-NEXT: ret <2 x float> [[TMP1]]
+// IMPRVD-NEXT: [[A_SROA_0_0_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 0
+// IMPRVD-NEXT: [[A_SROA_0_4_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 1
+// IMPRVD-NEXT: [[MUL_RL:%.*]] = fmul float [[A_SROA_0_0_VEC_EXTRACT]], [[B]]
+// IMPRVD-NEXT: [[MUL_IL:%.*]] = fmul float [[A_SROA_0_4_VEC_EXTRACT]], [[B]]
+// IMPRVD-NEXT: [[RETVAL_SROA_0_0_VEC_INSERT:%.*]] = insertelement <2 x float> undef, float [[MUL_RL]], i32 0
+// IMPRVD-NEXT: [[RETVAL_SROA_0_4_VEC_INSERT:%.*]] = insertelement <2 x float> [[RETVAL_SROA_0_0_VEC_INSERT]], float [[MUL_IL]], i32 1
+// IMPRVD-NEXT: ret <2 x float> [[RETVAL_SROA_0_4_VEC_INSERT]]
//
// PRMTD-LABEL: define dso_local <2 x float> @mulaf(
// PRMTD-SAME: <2 x float> noundef [[A_COERCE:%.*]], float noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {
// PRMTD-NEXT: [[ENTRY:.*:]]
-// PRMTD-NEXT: [[RETVAL:%.*]] = alloca { float, float }, align 4
-// PRMTD-NEXT: [[A:%.*]] = alloca { float, float }, align 4
-// PRMTD-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// PRMTD-NEXT: store <2 x float> [[A_COERCE]], ptr [[A]], align 4
-// PRMTD-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// PRMTD-NEXT: [[A_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
-// PRMTD-NEXT: [[A_REAL:%.*]] = load float, ptr [[A_REALP]], align 4
-// PRMTD-NEXT: [[A_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
-// PRMTD-NEXT: [[A_IMAG:%.*]] = load float, ptr [[A_IMAGP]], align 4
-// PRMTD-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// PRMTD-NEXT: [[MUL_RL:%.*]] = fmul float [[A_REAL]], [[TMP0]]
-// PRMTD-NEXT: [[MUL_IL:%.*]] = fmul float [[A_IMAG]], [[TMP0]]
-// PRMTD-NEXT: [[RETVAL_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 0
-// PRMTD-NEXT: [[RETVAL_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 1
-// PRMTD-NEXT: store float [[MUL_RL]], ptr [[RETVAL_REALP]], align 4
-// PRMTD-NEXT: store float [[MUL_IL]], ptr [[RETVAL_IMAGP]], align 4
-// PRMTD-NEXT: [[TMP1:%.*]] = load <2 x float>, ptr [[RETVAL]], align 4
-// PRMTD-NEXT: ret <2 x float> [[TMP1]]
+// PRMTD-NEXT: [[A_SROA_0_0_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 0
+// PRMTD-NEXT: [[A_SROA_0_4_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 1
+// PRMTD-NEXT: [[MUL_RL:%.*]] = fmul float [[A_SROA_0_0_VEC_EXTRACT]], [[B]]
+// PRMTD-NEXT: [[MUL_IL:%.*]] = fmul float [[A_SROA_0_4_VEC_EXTRACT]], [[B]]
+// PRMTD-NEXT: [[RETVAL_SROA_0_0_VEC_INSERT:%.*]] = insertelement <2 x float> undef, float [[MUL_RL]], i32 0
+// PRMTD-NEXT: [[RETVAL_SROA_0_4_VEC_INSERT:%.*]] = insertelement <2 x float> [[RETVAL_SROA_0_0_VEC_INSERT]], float [[MUL_IL]], i32 1
+// PRMTD-NEXT: ret <2 x float> [[RETVAL_SROA_0_4_VEC_INSERT]]
//
// PRMTD_STRICT-LABEL: define dso_local <2 x float> @mulaf(
// PRMTD_STRICT-SAME: <2 x float> noundef [[A_COERCE:%.*]], float noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {
// PRMTD_STRICT-NEXT: [[ENTRY:.*:]]
-// PRMTD_STRICT-NEXT: [[RETVAL:%.*]] = alloca { float, float }, align 4
-// PRMTD_STRICT-NEXT: [[A:%.*]] = alloca { float, float }, align 4
-// PRMTD_STRICT-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// PRMTD_STRICT-NEXT: store <2 x float> [[A_COERCE]], ptr [[A]], align 4
-// PRMTD_STRICT-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// PRMTD_STRICT-NEXT: [[A_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
-// PRMTD_STRICT-NEXT: [[A_REAL:%.*]] = load float, ptr [[A_REALP]], align 4
-// PRMTD_STRICT-NEXT: [[A_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
-// PRMTD_STRICT-NEXT: [[A_IMAG:%.*]] = load float, ptr [[A_IMAGP]], align 4
-// PRMTD_STRICT-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// PRMTD_STRICT-NEXT: [[MUL_RL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[A_REAL]], float [[TMP0]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3:[0-9]+]]
-// PRMTD_STRICT-NEXT: [[MUL_IL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[A_IMAG]], float [[TMP0]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3]]
-// PRMTD_STRICT-NEXT: [[RETVAL_REALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 0
-// PRMTD_STRICT-NEXT: [[RETVAL_IMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[RETVAL]], i32 0, i32 1
-// PRMTD_STRICT-NEXT: store float [[MUL_RL]], ptr [[RETVAL_REALP]], align 4
-// PRMTD_STRICT-NEXT: store float [[MUL_IL]], ptr [[RETVAL_IMAGP]], align 4
-// PRMTD_STRICT-NEXT: [[TMP1:%.*]] = load <2 x float>, ptr [[RETVAL]], align 4
-// PRMTD_STRICT-NEXT: ret <2 x float> [[TMP1]]
+// PRMTD_STRICT-NEXT: [[A_SROA_0_0_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 0
+// PRMTD_STRICT-NEXT: [[A_SROA_0_4_VEC_EXTRACT:%.*]] = extractelement <2 x float> [[A_COERCE]], i32 1
+// PRMTD_STRICT-NEXT: [[MUL_RL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[A_SROA_0_0_VEC_EXTRACT]], float [[B]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3:[0-9]+]]
+// PRMTD_STRICT-NEXT: [[MUL_IL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[A_SROA_0_4_VEC_EXTRACT]], float [[B]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3]]
+// PRMTD_STRICT-NEXT: [[RETVAL_SROA_0_0_VEC_INSERT:%.*]] = insertelement <2 x float> undef, float [[MUL_RL]], i32 0
+// PRMTD_STRICT-NEXT: [[RETVAL_SROA_0_4_VEC_INSERT:%.*]] = insertelement <2 x float> [[RETVAL_SROA_0_0_VEC_INSERT]], float [[MUL_IL]], i32 1
+// PRMTD_STRICT-NEXT: ret <2 x float> [[RETVAL_SROA_0_4_VEC_INSERT]]
//
_Complex float mulaf(_Complex float a, float b) {
return a * b;
@@ -107,20 +63,14 @@ _Complex float mulaf(_Complex float a, float b) {
// FULL-LABEL: define dso_local void @mulassignf(
// FULL-SAME: ptr noundef [[A:%.*]], float noundef [[B:%.*]]) #[[ATTR1:[0-9]+]] {
// FULL-NEXT: [[ENTRY:.*:]]
-// FULL-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
-// FULL-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// FULL-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
-// FULL-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// FULL-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// FULL-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
-// FULL-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
+// FULL-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
// FULL-NEXT: [[DOTREAL:%.*]] = load float, ptr [[DOTREALP]], align 4
-// FULL-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// FULL-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// FULL-NEXT: [[DOTIMAG:%.*]] = load float, ptr [[DOTIMAGP]], align 4
-// FULL-NEXT: [[MUL_RL:%.*]] = fmul float [[DOTREAL]], [[TMP0]]
-// FULL-NEXT: [[MUL_IL:%.*]] = fmul float [[DOTIMAG]], [[TMP0]]
-// FULL-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
-// FULL-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// FULL-NEXT: [[MUL_RL:%.*]] = fmul float [[DOTREAL]], [[B]]
+// FULL-NEXT: [[MUL_IL:%.*]] = fmul float [[DOTIMAG]], [[B]]
+// FULL-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
+// FULL-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// FULL-NEXT: store float [[MUL_RL]], ptr [[DOTREALP1]], align 4
// FULL-NEXT: store float [[MUL_IL]], ptr [[DOTIMAGP2]], align 4
// FULL-NEXT: ret void
@@ -128,20 +78,14 @@ _Complex float mulaf(_Complex float a, float b) {
// IMPRVD-LABEL: define dso_local void @mulassignf(
// IMPRVD-SAME: ptr noundef [[A:%.*]], float noundef [[B:%.*]]) #[[ATTR1:[0-9]+]] {
// IMPRVD-NEXT: [[ENTRY:.*:]]
-// IMPRVD-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
-// IMPRVD-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// IMPRVD-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
-// IMPRVD-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// IMPRVD-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// IMPRVD-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
-// IMPRVD-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
+// IMPRVD-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
// IMPRVD-NEXT: [[DOTREAL:%.*]] = load float, ptr [[DOTREALP]], align 4
-// IMPRVD-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// IMPRVD-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// IMPRVD-NEXT: [[DOTIMAG:%.*]] = load float, ptr [[DOTIMAGP]], align 4
-// IMPRVD-NEXT: [[MUL_RL:%.*]] = fmul float [[DOTREAL]], [[TMP0]]
-// IMPRVD-NEXT: [[MUL_IL:%.*]] = fmul float [[DOTIMAG]], [[TMP0]]
-// IMPRVD-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
-// IMPRVD-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// IMPRVD-NEXT: [[MUL_RL:%.*]] = fmul float [[DOTREAL]], [[B]]
+// IMPRVD-NEXT: [[MUL_IL:%.*]] = fmul float [[DOTIMAG]], [[B]]
+// IMPRVD-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
+// IMPRVD-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// IMPRVD-NEXT: store float [[MUL_RL]], ptr [[DOTREALP1]], align 4
// IMPRVD-NEXT: store float [[MUL_IL]], ptr [[DOTIMAGP2]], align 4
// IMPRVD-NEXT: ret void
@@ -149,20 +93,14 @@ _Complex float mulaf(_Complex float a, float b) {
// PRMTD-LABEL: define dso_local void @mulassignf(
// PRMTD-SAME: ptr noundef [[A:%.*]], float noundef [[B:%.*]]) #[[ATTR1:[0-9]+]] {
// PRMTD-NEXT: [[ENTRY:.*:]]
-// PRMTD-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
-// PRMTD-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// PRMTD-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
-// PRMTD-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// PRMTD-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// PRMTD-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
-// PRMTD-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
+// PRMTD-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
// PRMTD-NEXT: [[DOTREAL:%.*]] = load float, ptr [[DOTREALP]], align 4
-// PRMTD-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// PRMTD-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// PRMTD-NEXT: [[DOTIMAG:%.*]] = load float, ptr [[DOTIMAGP]], align 4
-// PRMTD-NEXT: [[MUL_RL:%.*]] = fmul float [[DOTREAL]], [[TMP0]]
-// PRMTD-NEXT: [[MUL_IL:%.*]] = fmul float [[DOTIMAG]], [[TMP0]]
-// PRMTD-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
-// PRMTD-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// PRMTD-NEXT: [[MUL_RL:%.*]] = fmul float [[DOTREAL]], [[B]]
+// PRMTD-NEXT: [[MUL_IL:%.*]] = fmul float [[DOTIMAG]], [[B]]
+// PRMTD-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
+// PRMTD-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// PRMTD-NEXT: store float [[MUL_RL]], ptr [[DOTREALP1]], align 4
// PRMTD-NEXT: store float [[MUL_IL]], ptr [[DOTIMAGP2]], align 4
// PRMTD-NEXT: ret void
@@ -170,20 +108,14 @@ _Complex float mulaf(_Complex float a, float b) {
// PRMTD_STRICT-LABEL: define dso_local void @mulassignf(
// PRMTD_STRICT-SAME: ptr noundef [[A:%.*]], float noundef [[B:%.*]]) #[[ATTR2:[0-9]+]] {
// PRMTD_STRICT-NEXT: [[ENTRY:.*:]]
-// PRMTD_STRICT-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
-// PRMTD_STRICT-NEXT: [[B_ADDR:%.*]] = alloca float, align 4
-// PRMTD_STRICT-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
-// PRMTD_STRICT-NEXT: store float [[B]], ptr [[B_ADDR]], align 4
-// PRMTD_STRICT-NEXT: [[TMP0:%.*]] = load float, ptr [[B_ADDR]], align 4
-// PRMTD_STRICT-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
-// PRMTD_STRICT-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
+// PRMTD_STRICT-NEXT: [[DOTREALP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
// PRMTD_STRICT-NEXT: [[DOTREAL:%.*]] = load float, ptr [[DOTREALP]], align 4
-// PRMTD_STRICT-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// PRMTD_STRICT-NEXT: [[DOTIMAGP:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// PRMTD_STRICT-NEXT: [[DOTIMAG:%.*]] = load float, ptr [[DOTIMAGP]], align 4
-// PRMTD_STRICT-NEXT: [[MUL_RL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[DOTREAL]], float [[TMP0]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3]]
-// PRMTD_STRICT-NEXT: [[MUL_IL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[DOTIMAG]], float [[TMP0]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3]]
-// PRMTD_STRICT-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 0
-// PRMTD_STRICT-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[TMP1]], i32 0, i32 1
+// PRMTD_STRICT-NEXT: [[MUL_RL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[DOTREAL]], float [[B]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3]]
+// PRMTD_STRICT-NEXT: [[MUL_IL:%.*]] = call float @llvm.experimental.constrained.fmul.f32(float [[DOTIMAG]], float [[B]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR3]]
+// PRMTD_STRICT-NEXT: [[DOTREALP1:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 0
+// PRMTD_STRICT-NEXT: [[DOTIMAGP2:%.*]] = getelementptr inbounds nuw { float, float }, ptr [[A]], i32 0, i32 1
// PRMTD_STRICT-NEXT: store float [[MUL_RL]], ptr [[DOTREALP1]], align 4
// PRMTD_STRICT-NEXT: store float [[MUL_IL]], ptr [[DOTIMAGP2]], align 4
// PRMTD_STRICT-NEXT: ret void
@@ -195...
[truncated]
|
You can test this locally with the following command:git diff -U0 --pickaxe-regex -S '([^a-zA-Z0-9#_-]undef[^a-zA-Z0-9_-]|UndefValue::get)' f017073cd8ac9e627db17678c6b64e9abe4fbec7 6dbcb05340584187761da8dd46193dab0db0ae5d clang/test/CodeGen/cx-complex-range-real.c clang/test/CodeGen/cx-complex-range.cThe following files introduce new uses of undef:
Undef is now deprecated and should only be used in the rare cases where no replacement is possible. For example, a load of uninitialized memory yields In tests, avoid using For example, this is considered a bad practice: define void @fn() {
...
br i1 undef, ...
}Please use the following instead: define void @fn(i1 %cond) {
...
br i1 %cond, ...
}Please refer to the Undefined Behavior Manual for more information. |
Hmm these were added by sroa I believe, for the "buildvector" idiom, curious that some instances got poison, but not all. EDIT: insertelement seems to be undef, while insertvalue is poison. |
Maetveis
force-pushed
the
users/Maetveis/simplify_complex_tests_with_opt
branch
from
ce2752e to
6da24fc
Compare
Maetveis
force-pushed
the
users/Maetveis/complex-promoted-fix-assign
branch
from
f02c8a6 to
57b488a
Compare
Maetveis
force-pushed
the
users/Maetveis/complex-promoted-fix-assign
branch
from
57b488a to
f2a2bf5
Compare
Base automatically changed from
users/Maetveis/complex-promoted-fix-assign
to
main
March 19, 2025 06:29
Run mem2reg and sroa passes on the complex range tests, to make them shorter and easier to read. This removes ~2000 lines of cruft.
Maetveis
force-pushed
the
users/Maetveis/simplify_complex_tests_with_opt
branch
from
6da24fc to
1ecc06b
Compare
|
Running both mem2reg and sroa is redunant; just run sroa. If the undefs are getting introduced by sroa, that's fine; they'll go away when we fix sroa. |
Maetveis
changed the title
Okay, thanks. I've updated the PR to remove the explicit mem2reg. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
... to make them shorter and easier to read. This removes ~2000 lines of cruft.