tyb0807 updated this revision to Diff 410225.
tyb0807 added a comment.
Cache architecture feature to avoid scanning the feature list over and over
again.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D120111/new/
https://reviews.llvm.org/D120111
Files:
clang/lib/Basic/Targets/AArch64.cpp
clang/lib/Driver/ToolChains/Arch/AArch64.cpp
clang/test/Driver/aarch64-hbc.c
clang/test/Driver/aarch64-mops.c
clang/test/Preprocessor/aarch64-target-features.c
Index: clang/test/Preprocessor/aarch64-target-features.c
===================================================================
--- clang/test/Preprocessor/aarch64-target-features.c
+++ clang/test/Preprocessor/aarch64-target-features.c
@@ -521,13 +521,17 @@
// CHECK-LSE: __ARM_FEATURE_ATOMICS 1
// ================== Check Armv8.8-A/Armv9.3-A memcpy and memset acceleration instructions (MOPS)
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.7-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.7-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.2-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.2-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.3-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
-// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.3-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.7-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.7-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a+nomops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a+nomops+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv8.8-a+mops+nomops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.2-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.2-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.3-a -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.3-a+nomops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-NOMOPS %s
+// RUN: %clang -target aarch64-arm-none-eabi -march=armv9.3-a+mops -x c -E -dM %s -o - | FileCheck --check-prefix=CHECK-MOPS %s
// CHECK-MOPS: __ARM_FEATURE_MOPS 1
// CHECK-NOMOPS-NOT: __ARM_FEATURE_MOPS 1
Index: clang/test/Driver/aarch64-mops.c
===================================================================
--- clang/test/Driver/aarch64-mops.c
+++ clang/test/Driver/aarch64-mops.c
@@ -1,6 +1,12 @@
// Test that target feature mops is implemented and available correctly
-// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a+mops %s 2>&1 | FileCheck %s
-// CHECK: "-target-feature" "+mops"
-
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.7-a+mops %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a+mops %s 2>&1 | FileCheck %s
// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a+nomops %s 2>&1 | FileCheck %s --check-prefix=NO_MOPS
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.2-a+mops %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.3-a %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.3-a+mops %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.3-a+nomops %s 2>&1 | FileCheck %s --check-prefix=NO_MOPS
+
+// CHECK: "-target-feature" "+mops"
// NO_MOPS: "-target-feature" "-mops"
Index: clang/test/Driver/aarch64-hbc.c
===================================================================
--- clang/test/Driver/aarch64-hbc.c
+++ clang/test/Driver/aarch64-hbc.c
@@ -1,6 +1,12 @@
// Test that target feature hbc is implemented and available correctly
-// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a+hbc %s 2>&1 | FileCheck %s
-// CHECK: "-target-feature" "+hbc"
-
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.7-a+hbc %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a+hbc %s 2>&1 | FileCheck %s
// RUN: %clang -### -target aarch64-none-none-eabi -march=armv8.8-a+nohbc %s 2>&1 | FileCheck %s --check-prefix=NO_HBC
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.2-a+hbc %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.3-a %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.3-a+hbc %s 2>&1 | FileCheck %s
+// RUN: %clang -### -target aarch64-none-none-eabi -march=armv9.3-a+nohbc %s 2>&1 | FileCheck %s --check-prefix=NO_HBC
+
+// CHECK: "-target-feature" "+hbc"
// NO_HBC: "-target-feature" "-hbc"
Index: clang/lib/Driver/ToolChains/Arch/AArch64.cpp
===================================================================
--- clang/lib/Driver/ToolChains/Arch/AArch64.cpp
+++ clang/lib/Driver/ToolChains/Arch/AArch64.cpp
@@ -116,7 +116,8 @@
// Check if the CPU name and feature modifiers in -mcpu are legal. If yes,
// decode CPU and feature.
static bool DecodeAArch64Mcpu(const Driver &D, StringRef Mcpu, StringRef &CPU,
- std::vector<StringRef> &Features) {
+ std::vector<StringRef> &Features,
+ StringRef &ArchFeature) {
std::pair<StringRef, StringRef> Split = Mcpu.split("+");
CPU = Split.first;
llvm::AArch64::ArchKind ArchKind = llvm::AArch64::ArchKind::ARMV8A;
@@ -131,6 +132,7 @@
if (!llvm::AArch64::getArchFeatures(ArchKind, Features))
return false;
+ ArchFeature = Features.back();
uint64_t Extension = llvm::AArch64::getDefaultExtensions(CPU, ArchKind);
if (!llvm::AArch64::getExtensionFeatures(Extension, Features))
return false;
@@ -146,15 +148,18 @@
static bool
getAArch64ArchFeaturesFromMarch(const Driver &D, StringRef March,
const ArgList &Args,
- std::vector<StringRef> &Features) {
+ std::vector<StringRef> &Features,
+ StringRef &ArchFeature) {
std::string MarchLowerCase = March.lower();
std::pair<StringRef, StringRef> Split = StringRef(MarchLowerCase).split("+");
llvm::AArch64::ArchKind ArchKind = llvm::AArch64::parseArch(Split.first);
- if (ArchKind == llvm::AArch64::ArchKind::INVALID ||
- !llvm::AArch64::getArchFeatures(ArchKind, Features))
+ // getArchFeatures() returns false only if
+ // ArchKind == llvm::AArch64::ArchKind::INVALID
+ if (!llvm::AArch64::getArchFeatures(ArchKind, Features))
return false;
+ ArchFeature = Features.back();
// Enable SVE2 by default on Armv9-A.
// It can still be disabled if +nosve2 is present.
// We must do this early so that DecodeAArch64Features has the correct state
@@ -175,24 +180,23 @@
static bool
getAArch64ArchFeaturesFromMcpu(const Driver &D, StringRef Mcpu,
const ArgList &Args,
- std::vector<StringRef> &Features) {
+ std::vector<StringRef> &Features,
+ StringRef &ArchFeature) {
StringRef CPU;
std::string McpuLowerCase = Mcpu.lower();
- if (!DecodeAArch64Mcpu(D, McpuLowerCase, CPU, Features))
- return false;
-
- return true;
+ return DecodeAArch64Mcpu(D, McpuLowerCase, CPU, Features, ArchFeature);
}
static bool
getAArch64MicroArchFeaturesFromMtune(const Driver &D, StringRef Mtune,
const ArgList &Args,
- std::vector<StringRef> &Features) {
+ std::vector<StringRef> &Features,
+ StringRef &ArchFeature) {
std::string MtuneLowerCase = Mtune.lower();
// Check CPU name is valid
std::vector<StringRef> MtuneFeatures;
StringRef Tune;
- if (!DecodeAArch64Mcpu(D, MtuneLowerCase, Tune, MtuneFeatures))
+ if (!DecodeAArch64Mcpu(D, MtuneLowerCase, Tune, MtuneFeatures, ArchFeature))
return false;
// Handle CPU name is 'native'.
@@ -209,14 +213,15 @@
static bool
getAArch64MicroArchFeaturesFromMcpu(const Driver &D, StringRef Mcpu,
const ArgList &Args,
- std::vector<StringRef> &Features) {
+ std::vector<StringRef> &Features,
+ StringRef &ArchFeature) {
StringRef CPU;
std::vector<StringRef> DecodedFeature;
std::string McpuLowerCase = Mcpu.lower();
- if (!DecodeAArch64Mcpu(D, McpuLowerCase, CPU, DecodedFeature))
+ if (!DecodeAArch64Mcpu(D, McpuLowerCase, CPU, DecodedFeature, ArchFeature))
return false;
- return getAArch64MicroArchFeaturesFromMtune(D, CPU, Args, Features);
+ return getAArch64MicroArchFeaturesFromMtune(D, CPU, Args, Features, ArchFeature);
}
void aarch64::getAArch64TargetFeatures(const Driver &D,
@@ -226,9 +231,10 @@
bool ForAS) {
Arg *A;
bool success = true;
+ StringRef AF;
// Enable NEON by default.
Features.push_back("+neon");
- llvm::StringRef WaMArch;
+ StringRef WaMArch;
if (ForAS)
for (const auto *A :
Args.filtered(options::OPT_Wa_COMMA, options::OPT_Xassembler))
@@ -239,28 +245,28 @@
// "-Xassembler -march" is detected. Otherwise it may return false
// and causes Clang to error out.
if (!WaMArch.empty())
- success = getAArch64ArchFeaturesFromMarch(D, WaMArch, Args, Features);
+ success = getAArch64ArchFeaturesFromMarch(D, WaMArch, Args, Features, AF);
else if ((A = Args.getLastArg(options::OPT_march_EQ)))
- success = getAArch64ArchFeaturesFromMarch(D, A->getValue(), Args, Features);
+ success = getAArch64ArchFeaturesFromMarch(D, A->getValue(), Args, Features, AF);
else if ((A = Args.getLastArg(options::OPT_mcpu_EQ)))
- success = getAArch64ArchFeaturesFromMcpu(D, A->getValue(), Args, Features);
+ success = getAArch64ArchFeaturesFromMcpu(D, A->getValue(), Args, Features, AF);
else if (Args.hasArg(options::OPT_arch) || isCPUDeterminedByTriple(Triple))
success = getAArch64ArchFeaturesFromMcpu(
- D, getAArch64TargetCPU(Args, Triple, A), Args, Features);
+ D, getAArch64TargetCPU(Args, Triple, A), Args, Features, AF);
else
// Default to 'A' profile if the architecture is not specified.
- success = getAArch64ArchFeaturesFromMarch(D, "armv8-a", Args, Features);
+ success = getAArch64ArchFeaturesFromMarch(D, "armv8-a", Args, Features, AF);
if (success && (A = Args.getLastArg(clang::driver::options::OPT_mtune_EQ)))
success =
- getAArch64MicroArchFeaturesFromMtune(D, A->getValue(), Args, Features);
+ getAArch64MicroArchFeaturesFromMtune(D, A->getValue(), Args, Features, AF);
else if (success && (A = Args.getLastArg(options::OPT_mcpu_EQ)))
success =
- getAArch64MicroArchFeaturesFromMcpu(D, A->getValue(), Args, Features);
+ getAArch64MicroArchFeaturesFromMcpu(D, A->getValue(), Args, Features, AF);
else if (success &&
(Args.hasArg(options::OPT_arch) || isCPUDeterminedByTriple(Triple)))
success = getAArch64MicroArchFeaturesFromMcpu(
- D, getAArch64TargetCPU(Args, Triple, A), Args, Features);
+ D, getAArch64TargetCPU(Args, Triple, A), Args, Features, AF);
if (!success) {
auto Diag = D.Diag(diag::err_drv_clang_unsupported);
@@ -350,7 +356,7 @@
// TargetParser rewrite.
const auto ItRNoFullFP16 = std::find(Features.rbegin(), Features.rend(), "-fullfp16");
const auto ItRFP16FML = std::find(Features.rbegin(), Features.rend(), "+fp16fml");
- if (llvm::is_contained(Features, "+v8.4a")) {
+ if (AF == "+v8.4a") {
const auto ItRFullFP16 = std::find(Features.rbegin(), Features.rend(), "+fullfp16");
if (ItRFullFP16 < ItRNoFullFP16 && ItRFullFP16 < ItRFP16FML) {
// Only entangled feature that can be to the right of this +fullfp16 is -fp16fml.
@@ -392,15 +398,9 @@
NoCrypto = true;
}
- if (std::find(ItBegin, ItEnd, "+v8.4a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v8.5a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v8.6a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v8.7a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v8.8a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v9a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v9.1a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v9.2a") != ItEnd ||
- std::find(ItBegin, ItEnd, "+v9.3a") != ItEnd) {
+ if (AF == "+v8.4a" || AF == "+v8.5a" || AF == "+v8.6a" || AF == "+v8.7a" ||
+ AF == "+v8.8a" || AF == "+v9a" || AF == "+v9.1a" || AF == "+v9.2a" ||
+ AF == "+v9.3a") {
if (HasCrypto && !NoCrypto) {
// Check if we have NOT disabled an algorithm with something like:
// +crypto, -algorithm
@@ -445,26 +445,31 @@
} else if (HasNoCrypto) {
const bool HasSHA2 = (std::find(ItBegin, ItEnd, "+sha2") != ItEnd);
const bool HasAES = (std::find(ItBegin, ItEnd, "+aes") != ItEnd);
- const bool HasV82a = (std::find(ItBegin, ItEnd, "+v8.2a") != ItEnd);
- const bool HasV83a = (std::find(ItBegin, ItEnd, "+v8.3a") != ItEnd);
- const bool HasV84a = (std::find(ItBegin, ItEnd, "+v8.4a") != ItEnd);
if (!HasSHA2)
Features.push_back("-sha2");
if (!HasAES)
Features.push_back("-aes");
- if (HasV82a || HasV83a || HasV84a) {
+ if (AF == "+v8.2a" || AF == "+v8.3a" || AF == "+v8.4a") {
Features.push_back("-sm4");
Features.push_back("-sha3");
}
}
}
- const char *Archs[] = {"+v8.6a", "+v8.7a", "+v8.8a",
- "+v9.1a", "+v9.2a", "+v9.3a"};
- auto Pos = std::find_first_of(Features.begin(), Features.end(),
- std::begin(Archs), std::end(Archs));
- if (Pos != std::end(Features))
+ // FIXME: these insertions should ideally be automated using default
+ // extensions support from the target parser.
+ if (AF == "+v8.6a" || AF == "+v8.7a" || AF == "+v8.8a" || AF == "+v9.1a" ||
+ AF == "+v9.2a" || AF == "+v9.3a") {
+ auto Pos = std::find(Features.begin(), Features.end(), AF);
Pos = Features.insert(std::next(Pos), {"+i8mm", "+bf16"});
+ }
+
+ // For Armv8.8-a/Armv9.3-a or later, FEAT_HBC and FEAT_MOPS are enabled by
+ // default.
+ if (AF == "+v8.8a" || AF == "+v9.3a") {
+ auto Pos = std::find(Features.begin(), Features.end(), AF);
+ Pos = Features.insert(std::next(Pos), {"+hbc", "+mops"});
+ }
if (Arg *A = Args.getLastArg(options::OPT_mno_unaligned_access,
options::OPT_munaligned_access)) {
Index: clang/lib/Basic/Targets/AArch64.cpp
===================================================================
--- clang/lib/Basic/Targets/AArch64.cpp
+++ clang/lib/Basic/Targets/AArch64.cpp
@@ -226,8 +226,6 @@
void AArch64TargetInfo::getTargetDefinesARMV88A(const LangOptions &Opts,
MacroBuilder &Builder) const {
- // FIXME: this does not handle the case where MOPS is disabled using +nomops
- Builder.defineMacro("__ARM_FEATURE_MOPS", "1");
// Also include the Armv8.7 defines
getTargetDefinesARMV87A(Opts, Builder);
}
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
