[clang] [C++20][Modules] Fix crash when function and lambda inside loaded from different modules (PR #109167)
aeubanks wrote: ah there's a stack trace in an asserts build of clang ``` F 00:00:1728077756.9778713296 logging.cc:62] assert.h assertion failed at third_party/llvm/llvm-project/clang/lib/Sema/SemaTemplateInstantiate.cpp:4602 in llvm::PointerUnion *clang::LocalInstantiationScope::findInstantiationOf(const Decl *): isa< LabelDecl>(D) && "declaration not instantiated in this scope" @ 0x563ef889e824 __assert_fail @ 0x563ef47a3bf4 clang::LocalInstantiationScope::findInstantiationOf() @ 0x563ef4836942 clang::Sema::FindInstantiatedDecl() @ 0x563ef4810491 clang::TreeTransform<>::TransformLambdaExpr() @ 0x563ef47ffb91 (anonymous namespace)::TemplateInstantiator::TransformLambdaExpr() @ 0x563ef47a1dc2 clang::TreeTransform<>::TransformExprs() @ 0x563ef480293d clang::TreeTransform<>::TransformCallExpr() @ 0x563ef479f7fa clang::TreeTransform<>::TransformStmt() @ 0x563ef4809f34 clang::TreeTransform<>::TransformCompoundStmt() @ 0x563ef479f788 clang::Sema::SubstStmt() @ 0x563ef484a01d clang::Sema::InstantiateFunctionDefinition() @ 0x563ef484d0bf clang::Sema::PerformPendingInstantiations() @ 0x563ef484a108 clang::Sema::InstantiateFunctionDefinition() @ 0x563ef484d0bf clang::Sema::PerformPendingInstantiations() @ 0x563ef484a10
[clang] [C++20][Modules] Fix crash when function and lambda inside loaded from different modules (PR #109167)
aeubanks wrote: this has also caused the following issue to pop up in our modules build ``` error: variable 'param' cannot be implicitly captured in a lambda with no capture-default specified ``` hopefully the error provides some guidance for where to look, but I can try to get a reduced repro https://github.com/llvm/llvm-project/pull/109167 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make Ownership of MachineModuleInfo in Its Wrapper Pass External (PR #110443)
https://github.com/aeubanks commented: I see that MMI really is a Codegen concept and not a Target concept, so that's why it takes an LLVMTargetMachine instead of TargetMachine. However, the `static_cast`s everywhere are extremely unfortunate. And it doesn't make sense to make the return type of `Target::createTargetMachine()` `LLVMTargetMachine` instead of `TargetMachine`. Perhaps alternatively we make the MMI constructor take `TargetMachine` and cast it inside the constructor to `LLVMTargetMachine`, so we only pay the cost of this weird distinction in one place rather than everywhere? wdyt? https://github.com/llvm/llvm-project/pull/110443 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make Ownership of MachineModuleInfo in Its Wrapper Pass External (PR #110443)
https://github.com/aeubanks commented: ah there's more context in https://reviews.llvm.org/D123964 https://github.com/llvm/llvm-project/pull/110443 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make Ownership of MachineModuleInfo in Its Wrapper Pass External (PR #110443)
@@ -1162,6 +1165,7 @@ void EmitAssemblyHelper::RunCodegenPipeline( // does not work with the codegen pipeline. // FIXME: make the new PM work with the codegen pipeline. legacy::PassManager CodeGenPasses; + std::unique_ptr MMI; aeubanks wrote: can this just be `MachineModuleInfo MMI;` below? https://github.com/llvm/llvm-project/pull/110443 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make Ownership of MachineModuleInfo in Its Wrapper Pass External (PR #110443)
@@ -106,16 +106,18 @@ class MachineModuleInfo {
const Function *LastRequest = nullptr; ///< Used for shortcut/cache.
MachineFunction *LastResult = nullptr; ///< Used for shortcut/cache.
- MachineModuleInfo &operator=(MachineModuleInfo &&MMII) = delete;
+ /// Deleted copy constructor
aeubanks wrote:
comment is not useful, I'd either say why it's deleted or just remove the
comment
https://github.com/llvm/llvm-project/pull/110443
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make Ownership of MachineModuleInfo in Its Wrapper Pass External (PR #110443)
https://github.com/aeubanks edited https://github.com/llvm/llvm-project/pull/110443 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make Ownership of MachineModuleInfo in Its Wrapper Pass External (PR #110443)
https://github.com/aeubanks commented: I'm still trying to wrap my head around LLVMTargetMachine vs TargetMachine, let me do some reading https://github.com/llvm/llvm-project/pull/110443 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make MMIWP not have ownership over MMI + Make MMI Only Use an External MCContext (PR #105541)
aeubanks wrote: hmm yeah it does seem that `MCContext` is hard to unentangle from `MMI`/`MachineFunction` is it possible to split out the "MMIWP doesn't own MMI" change? that seems less controversial and easier to review/merge and makes the `MCContext` changes easier to look at https://github.com/llvm/llvm-project/pull/105541 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [flang] [llvm] [mlir] Make MMIWP not have ownership over MMI + Make MMI Only Use an External MCContext (PR #105541)
aeubanks wrote: sorry for the delay after looking at MMI/MCContext, I agree that MMI shouldn't own MCContext, but do we even need a reference from MMI to MCContext? they are different layers of codegen IIUC. if it's possible to completely separate them we should do that (please correct me if this doesn't make sense since I haven't spent too much time in codegen) https://github.com/llvm/llvm-project/pull/105541 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [RFC][C++20][Modules] Fix crash when function and lambda inside loaded from different modules (PR #104512)
Valentin Clement =?utf-8?b?KOODkOODrOODsw=?=,Kyungwoo Lee ,Dmitry Polukhin ,Dmitry Polukhin Message-ID: In-Reply-To: aeubanks wrote: here's something reduced, let me know if you need the command to create the pch file as well ``` https://github.com/llvm/llvm-project/pull/104512 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang][test] Fix some windows driver tests with legacy runtime dir layout (PR #108452)
https://github.com/aeubanks created
https://github.com/llvm/llvm-project/pull/108452
With the legacy runtime directory layout, the runtime libraries are under
`lib/clang/20/lib/windows` and have an arch suffix.
>From c45f5df3c0ba7afd6c5e14e8e49ffa853abc7147 Mon Sep 17 00:00:00 2001
From: Arthur Eubanks
Date: Thu, 12 Sep 2024 20:52:20 +
Subject: [PATCH] [clang][test] Fix some windows driver tests with legacy
runtime dir layout
With the legacy runtime directory layout, the runtime libraries are under
`lib/clang/20/lib/windows` and have an arch suffix.
---
clang/test/Driver/cl-link.c | 12 ++--
clang/test/Driver/windows-cross.c | 2 +-
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/clang/test/Driver/cl-link.c b/clang/test/Driver/cl-link.c
index f5260442760452..9bf8a8137926de 100644
--- a/clang/test/Driver/cl-link.c
+++ b/clang/test/Driver/cl-link.c
@@ -13,17 +13,17 @@
// ASAN: link.exe
// ASAN: "-debug"
// ASAN: "-incremental:no"
-// ASAN: "{{[^"]*}}clang_rt.asan_dynamic.lib"
-// ASAN: "-wholearchive:{{.*}}clang_rt.asan_static_runtime_thunk.lib"
+// ASAN: "{{[^"]*}}clang_rt.asan_dynamic{{(-i386)?}}.lib"
+// ASAN:
"-wholearchive:{{.*}}clang_rt.asan_static_runtime_thunk{{(-i386)?}}.lib"
// ASAN: "{{.*}}cl-link{{.*}}.obj"
// RUN: %clang_cl -m32 -arch:IA32 --target=i386-pc-win32 /MD /Tc%s
-fuse-ld=link -### -fsanitize=address 2>&1 | FileCheck --check-prefix=ASAN-MD %s
// ASAN-MD: link.exe
// ASAN-MD: "-debug"
// ASAN-MD: "-incremental:no"
-// ASAN-MD: "{{.*}}clang_rt.asan_dynamic.lib"
+// ASAN-MD: "{{.*}}clang_rt.asan_dynamic{{(-i386)?}}.lib"
// ASAN-MD: "-include:___asan_seh_interceptor"
-// ASAN-MD: "-wholearchive:{{.*}}clang_rt.asan_dynamic_runtime_thunk.lib"
+// ASAN-MD:
"-wholearchive:{{.*}}clang_rt.asan_dynamic_runtime_thunk{{(-i386)?}}.lib"
// ASAN-MD: "{{.*}}cl-link{{.*}}.obj"
// RUN: %clang_cl /LD -fuse-ld=link -### /Tc%s 2>&1 | FileCheck
--check-prefix=DLL %s
@@ -37,8 +37,8 @@
// ASAN-DLL: "-dll"
// ASAN-DLL: "-debug"
// ASAN-DLL: "-incremental:no"
-// ASAN-DLL: "{{.*}}clang_rt.asan_dynamic.lib"
-// ASAN-DLL: "-wholearchive:{{.*}}clang_rt.asan_static_runtime_thunk.lib"
+// ASAN-DLL: "{{.*}}clang_rt.asan_dynamic{{(-i386)?}}.lib"
+// ASAN-DLL:
"-wholearchive:{{.*}}clang_rt.asan_static_runtime_thunk{{(-i386)?}}.lib"
// ASAN-DLL: "{{.*}}cl-link{{.*}}.obj"
// RUN: %clang_cl /Zi /Tc%s -fuse-ld=link -### 2>&1 | FileCheck
--check-prefix=DEBUG %s
diff --git a/clang/test/Driver/windows-cross.c
b/clang/test/Driver/windows-cross.c
index f6e831f00e13a6..096358d306323a 100644
--- a/clang/test/Driver/windows-cross.c
+++ b/clang/test/Driver/windows-cross.c
@@ -64,7 +64,7 @@
// RUN:| FileCheck %s --check-prefix CHECK-SANITIZE-ADDRESS-EXE-X86
// CHECK-SANITIZE-ADDRESS-EXE-X86: "-fsanitize=address"
-// CHECK-SANITIZE-ADDRESS-EXE-X86: "{{.*}}clang_rt.asan_dynamic.lib"
"{{.*}}clang_rt.asan_dynamic_runtime_thunk.lib" "--undefined"
"___asan_seh_interceptor"
+// CHECK-SANITIZE-ADDRESS-EXE-X86:
"{{.*}}clang_rt.asan_dynamic{{(-i386)?}}.lib"
"{{.*}}clang_rt.asan_dynamic_runtime_thunk{{(-i386)?}}.lib" "--undefined"
"___asan_seh_interceptor"
// RUN: not %clang -### --target=armv7-windows-itanium --sysroot
%S/Inputs/Windows/ARM/8.1 -B %S/Inputs/Windows/ARM/8.1/usr/bin
-fuse-ld=lld-link2 -shared -o shared.dll -fsanitize=tsan -x c++ %s 2>&1 \
// RUN:| FileCheck %s --check-prefix CHECK-SANITIZE-TSAN
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [RFC][C++20][Modules] Fix crash when function and lambda inside loaded from different modules (PR #104512)
Valentin Clement =?utf-8?b?KOODkOODrOODsw==?=,Kyungwoo Lee ,Dmitry Polukhin ,Dmitry Polukhin Message-ID: In-Reply-To: aeubanks wrote: let me try to come up with a reduced repro https://github.com/llvm/llvm-project/pull/104512 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang] Diagnose dangling issues for the "Container" case. (PR #107213)
aeubanks wrote: this looks like it has false positives: https://crbug.com/366074092 can we revert in the meantime? https://github.com/llvm/llvm-project/pull/107213 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] Reland [asan][windows] Eliminate the static asan runtime on windows (PR #107899)
@@ -136,7 +160,7 @@ append_list_if(MINGW "${MINGW_LIBRARIES}" ASAN_DYNAMIC_LIBS)
add_compiler_rt_object_libraries(RTAsan_dynamic
OS ${SANITIZER_COMMON_SUPPORTED_OS}
ARCHS ${ASAN_SUPPORTED_ARCH}
- SOURCES ${ASAN_SOURCES} ${ASAN_CXX_SOURCES}
+ SOURCES ${ASAN_SOURCES}
aeubanks wrote:
this line seems to be the problem, reverting it fixes the tests
https://github.com/llvm/llvm-project/pull/107899
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] Reland [asan][windows] Eliminate the static asan runtime on windows (PR #107899)
aeubanks wrote: previous passing output ``` 2: = 3: ==7382==ERROR: AddressSanitizer: requested allocation size 0x101 (0x1001008 after adjustments for alignment, red zones etc.) exceeds maximum supported size of 0x100 (thread T0) 4: #0 0x0001032a9268 in _Znwm+0x6c (libclang_rt.asan_osx_dynamic.dylib:arm64+0x61268) 5: #1 0x000102cd775c in main allocator_returns_null.cpp:82 6: #2 0x00018cc07150 () 7: #3 0x3a08fffc () 8: 9: ==7382==HINT: if you don't care about these errors you may set allocator_may_return_null=1 10: SUMMARY: AddressSanitizer: allocation-size-too-big allocator_returns_null.cpp:82 in main 11: ==7382==ABORTING ``` failing output with this PR ``` 2: = 3: ==6635==ERROR: AddressSanitizer: requested allocation size 0x101 (0x1001008 after adjustments for alignment, red zones etc.) exceeds maximum supported size of 0x100 (thread T0) 4: #0 0x0001029b60ec in malloc+0x70 (libclang_rt.asan_osx_dynamic.dylib:arm64+0x520ec) 5: #1 0x00018cf4abd0 in operator new(unsigned long)+0x1c (libc++abi.dylib:arm64+0x16bd0) 6: #2 0xde678001023f375c () 7: #3 0x00018cc07150 () 8: #4 0x9c0d7ffc () 9: 10: ==6635==HINT: if you don't care about these errors you may set allocator_may_return_null=1 11: SUMMARY: AddressSanitizer: allocation-size-too-big (libc++abi.dylib:arm64+0x16bd0) in operator new(unsigned long)+0x1c 12: ==6635==ABORTING ``` https://github.com/llvm/llvm-project/pull/107899 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] Reland [asan][windows] Eliminate the static asan runtime on windows (PR #107899)
aeubanks wrote: I think this is also somehow affecting asan tests on macos: https://green.lab.llvm.org/job/llvm.org/job/clang-stage1-RA/2034/ was able to reproduce locally `build/cmake/runtimes/runtimes-bins/compiler-rt/test/sanitizer_common/asan-arm64-Darwin/allocator_returns_null.cpp` failing at this commit and succeeding at the previous https://github.com/llvm/llvm-project/pull/107899 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [RFC][C++20][Modules] Fix crash when function and lambda inside loaded from different modules (PR #104512)
Valentin Clement =?utf-8?b?KOODkOODrOODsw=?=,Kyungwoo Lee ,Dmitry Polukhin ,Dmitry Polukhin Message-ID: In-Reply-To: aeubanks wrote: this is causing a crash with precompiled headers. I can try to reduce, but hopefully the problem is obvious from this assert/stack trace? ``` clang++: ../../llvm/include/llvm/ADT/SmallVector.h:295: const_reference llvm::SmallVectorTemplateCommon>::operator[](size_type) const [T = std::unique_ptr]: Assertion `idx < size()' failed. PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace, preprocessed source, and associated run script. Stack dump: 0. Program arguments: ../../../../llvm-project/build/rel/bin/clang++ -MMD -MF obj/third_party/blink/renderer/core/probe/instrumentation_probes/core_probes_impl.o.d -DDCHECK_ALWAYS_ON=1 -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D_FORTIFY_SOURCE=2 -DCR_XCODE_VERSION=1500 -D_LIBCPP_HARDENING_MODE=_LIBCPP_HARDENING_MODE_EXTENSIVE -DCOMPONENT_BUILD -DCR_LIBCXX_REVISION=6ae6f38d10eda881c16d91932348fc6d4ee98332 -DTEMP_REBUILD_HACK -DNDEBUG -DNVALGRIND -DDYNAMIC_ANNOTATIONS_ENABLED=0 -DBLINK_CORE_IMPLEMENTATION=1 -DSK_ENABLE_SKSL -DSK_UNTIL_CRBUG_1187654_IS_FIXED -DSK_USER_CONFIG_HEADER=\"../../skia/config/SkUserConfig.h\" -DSK_WIN_FONTMGR_NO_SIMULATIONS -DSK_DISABLE_LEGACY_INIT_DECODERS -DSK_DISABLE_LEGACY_BACKEND_TEXTURE_FUNCS -DSK_DISABLE_LEGACY_TEXTURE_INFO_FUNCS -DSK_DISABLE_LEGACY_BACKEND_SEMAPHORE_FUNCS -DSK_DISABLE_LEGACY_GRAPHITE_IMAGES -DSK_DISABLE_LEGACY_DAWN_TEXTURE_INFO_FUNCS -DSK_DISABLE_LEGACY_DAWN_BACKEND_TEXTURE_FUNCS -DSK_CODEC_DECODES_JPEG -DSK_ENCODE_JPEG -DSK_ENCODE_PNG -DSK_ENCODE_WEBP -DSKIA_DLL -DSKCMS_API=__attribute__((visibility(\"default\"))) -DSK_BUILD_FOR_MAC -DSK_GANESH -DSK_GPU_WORKAROUNDS_HEADER=\"gpu/config/gpu_driver_bug_workaround_autogen.h\" -DSK_GL -DSK_GRAPHITE -DSK_DAWN -DSK_METAL -DCHROMIUM -DLIBYUV_DISABLE_NEON -DLIBYUV_DISABLE_SVE -DLIBYUV_DISABLE_SME -DLIBYUV_DISABLE_LSX -DLIBYUV_DISABLE_LASX -DBLINK_IMPLEMENTATION=1 -DINSIDE_BLINK -DABSL_CONSUME_DLL -DABSL_FLAGS_STRIP_NAMES=0 -DCR_CXX_INCLUDE=\"third_party/rust/chromium_crates_io/vendor/cxx-1.0.128/include/cxx.h\" -DBORINGSSL_SHARED_LIBRARY -DU_USING_ICU_NAMESPACE=0 -DU_ENABLE_DYLOAD=0 -DUSE_CHROMIUM_ICU=1 -DU_ENABLE_TRACING=1 -DU_ENABLE_RESOURCE_TRACING=0 -DICU_UTIL_DATA_IMPL=ICU_UTIL_DATA_FILE -DGOOGLE_PROTOBUF_NO_RTTI -DGOOGLE_PROTOBUF_NO_STATIC_INITIALIZER -DGOOGLE_PROTOBUF_INTERNAL_DONATE_STEAL_INLINE=0 -DHAVE_PTHREAD -DPROTOBUF_USE_DLLS -DWEBRTC_ENABLE_SYMBOL_EXPORT -DWEBRTC_ENABLE_AVX2 -DWEBRTC_CHROMIUM_BUILD -DWEBRTC_POSIX -DWEBRTC_MAC -DABSL_ALLOCATOR_NOTHROW=1 -DLOGGING_INSIDE_WEBRTC -DWGPU_SHARED_LIBRARY -DUSING_V8_SHARED -DUSING_V8_SHARED_PRIVATE -DV8_ARRAY_BUFFER_INTERNAL_FIELD_COUNT=0 -DV8_ARRAY_BUFFER_VIEW_INTERNAL_FIELD_COUNT=0 -DV8_PROMISE_INTERNAL_FIELD_COUNT=0 -DV8_COMPRESS_POINTERS -DV8_COMPRESS_POINTERS_IN_SHARED_CAGE -DV8_31BIT_SMIS_ON_64BIT_ARCH -DV8_ENABLE_SANDBOX -DV8_DEPRECATION_WARNINGS -DV8_USE_PERFETTO -DV8_HAVE_TARGET_OS -DV8_TARGET_OS_MACOS -DCPPGC_CAGED_HEAP -DCPPGC_YOUNG_GENERATION -DCPPGC_POINTER_COMPRESSION -DCPPGC_ENABLE_LARGER_CAGE -DCPPGC_SLIM_WRITE_BARRIER -DLEVELDB_PLATFORM_CHROMIUM=1 -DLEVELDB_SHARED_LIBRARY -DCRASHPAD_ZLIB_SOURCE_EXTERNAL -DUSE_LIBJPEG_TURBO=1 -DMANGLE_JPEG_NAMES -DWEBP_EXTERN=extern -DUSING_V8_BASE_SHARED -DUSING_V8_PLATFORM_SHARED -DLIBXSLT_STATIC -I../.. -Igen -I../../buildtools/third_party/libc++ -I../../third_party/perfetto/include -Igen/third_party/perfetto/build_config -Igen/third_party/perfetto -I../../third_party/skia -Igen/third_party/skia -I../../third_party/wuffs/src/release/c -Igen/third_party/dawn/include -I../../third_party/dawn/include -I../../third_party/khronos -I../../gpu -I../../net/third_party/quiche/overrides -I../../net/third_party/quiche/src/quiche/common/platform/default -I../../net/third_party/quiche/src -I../../third_party/libyuv/include -I../../base/allocator/partition_allocator/src -Igen/base/allocator/partition_allocator/src -I../../third_party/abseil-cpp -I../../third_party/boringssl/src/include -I../../third_party/protobuf/src -Igen/protoc_out -I../../third_party/ipcz/include -I../../third_party/ced/src -I../../third_party/icu/source/common -I../../third_party/icu/source/i18n -Igen/net/third_party/quiche/src -I../../third_party/webrtc_overrides -I../../third_party/webrtc -Igen/third_party/webrtc -I../../v8/include -I../../third_party/libwebm/source -I../../third_party/angle/include -I../../third_party/mesa_headers -I../../third_party/leveldatabase -I../../third_party/leveldatabase/src -I../../third_party/leveldatabase/src/include -I../../third_party/crashpad/crashpad -I../../third_party/crashpad/crashpad/compat/mac -I../../third_party/crashpad/crashpad/compat/non_win -I../../third_party/zlib -I../../third_party/libaom/source/libaom -I../../third_party/libaom/source/config/linux/x64 -I../../third_party/libjpeg_turbo -
[clang] [clang] Respect the lifetimebound in assignment operator. (PR #106997)
@@ -326,24 +326,11 @@ static bool shouldTrackFirstArgument(const FunctionDecl
*FD) {
return false;
}
-static bool implicitObjectParamIsLifetimeBound(const FunctionDecl *FD) {
- const TypeSourceInfo *TSI = FD->getTypeSourceInfo();
- if (!TSI)
-return false;
- // Don't declare this variable in the second operand of the for-statement;
- // GCC miscompiles that by ending its lifetime before evaluating the
- // third operand. See gcc.gnu.org/PR86769.
- AttributedTypeLoc ATL;
- for (TypeLoc TL = TSI->getTypeLoc();
- (ATL = TL.getAsAdjusted());
- TL = ATL.getModifiedLoc()) {
-if (ATL.getAttrAs())
- return true;
- }
-
- // Assume that all assignment operators with a "normal" return type return
- // *this, that is, an lvalue reference that is the same type as the implicit
- // object parameter (or the LHS for a non-member operator$=).
+// Return true if this is an "normal" assignment operator.
+// We assuments that a normal assingment operator always returns *this, that
is,
+// an lvalue reference that is the same type as the implicit object parameter
+// (or the LHS for a non-member operator$=).
+static bool isNormalAsisgnmentOperator(const FunctionDecl *FD) {
aeubanks wrote:
nit: "Assignment" is spelled incorrectly here and also below
https://github.com/llvm/llvm-project/pull/106997
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [-Wunsafe-buffer-usage] Warning Libc functions (PR #101583)
aeubanks wrote: There needs to be a flag to opt out of this to not break everybody who is currently using unsafe-buffer-usage. #105383 seems to do that, but it really should be in this same PR. Can this be reverted and relanded with the flag? https://github.com/llvm/llvm-project/pull/101583 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [PGO][Pipeline] Enable PGOForceFunctionAttrs in PGO optimization pipelines (PR #106790)
https://github.com/aeubanks closed https://github.com/llvm/llvm-project/pull/106790 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [PGO][Pipeline] Enable PGOForceFunctionAttrs in PGO optimization pipelines (PR #106790)
https://github.com/aeubanks updated
https://github.com/llvm/llvm-project/pull/106790
>From a469c523921c3bcb3a8e2a6ad5eaa856a6035449 Mon Sep 17 00:00:00 2001
From: Arthur Eubanks
Date: Fri, 30 Aug 2024 20:14:18 +
Subject: [PATCH 1/2] [PGO][Pipeline] Enable PGOForceFunctionAttrs in PGO
optimization pipelines
Remove flag that turns on the PGOForceFunctionAttrs pass and always add it to
default pipelines when using PGO.
This is NFC by default since PGOOpt->ColdOptType is by default
ColdFuncOpt::Default.
Remove -O2 RUN line in basic.ll since we now have the pipeline tests.
---
llvm/lib/Passes/PassBuilderPipelines.cpp | 9 ++---
llvm/test/Instrumentation/PGOForceFunctionAttrs/basic.ll | 1 -
.../Other/new-pm-thinlto-postlink-samplepgo-defaults.ll | 1 +
llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll | 1 +
.../Other/new-pm-thinlto-prelink-samplepgo-defaults.ll | 1 +
5 files changed, 5 insertions(+), 8 deletions(-)
diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp
b/llvm/lib/Passes/PassBuilderPipelines.cpp
index 9c3d49cabbd38c..1fd7ef929c87d5 100644
--- a/llvm/lib/Passes/PassBuilderPipelines.cpp
+++ b/llvm/lib/Passes/PassBuilderPipelines.cpp
@@ -227,12 +227,6 @@ static cl::opt
cl::desc("Enable DFA jump threading"),
cl::init(false), cl::Hidden);
-// TODO: turn on and remove flag
-static cl::opt EnablePGOForceFunctionAttrs(
-"enable-pgo-force-function-attrs",
-cl::desc("Enable pass to set function attributes based on PGO profiles"),
-cl::init(false));
-
static cl::opt
EnableHotColdSplit("hot-cold-split",
cl::desc("Enable hot-cold splitting pass"));
@@ -1222,7 +1216,8 @@
PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
if (EnableSyntheticCounts && !PGOOpt)
MPM.addPass(SyntheticCountsPropagation());
- if (EnablePGOForceFunctionAttrs && PGOOpt)
+ if (PGOOpt && (PGOOpt->Action == PGOOptions::IRUse ||
+ PGOOpt->Action == PGOOptions::SampleUse))
MPM.addPass(PGOForceFunctionAttrsPass(PGOOpt->ColdOptType));
MPM.addPass(AlwaysInlinerPass(/*InsertLifetimeIntrinsics=*/true));
diff --git a/llvm/test/Instrumentation/PGOForceFunctionAttrs/basic.ll
b/llvm/test/Instrumentation/PGOForceFunctionAttrs/basic.ll
index 55bdb2f69ad16a..bc966f39fda148 100644
--- a/llvm/test/Instrumentation/PGOForceFunctionAttrs/basic.ll
+++ b/llvm/test/Instrumentation/PGOForceFunctionAttrs/basic.ll
@@ -5,7 +5,6 @@
; Should be no changes without profile data
; RUN: opt < %s -passes=pgo-force-function-attrs
-S -pgo-cold-func-opt=minsize | FileCheck %s --check-prefixes=NONE,CHECK
-; RUN: opt < %s -passes='default' -enable-pgo-force-function-attrs
-S -pgo-cold-func-opt=minsize | FileCheck %s --check-prefixes=O2
; NONE-NOT: Function Attrs:
; OPTSIZE: Function Attrs: optsize{{$}}
diff --git a/llvm/test/Other/new-pm-thinlto-postlink-samplepgo-defaults.ll
b/llvm/test/Other/new-pm-thinlto-postlink-samplepgo-defaults.ll
index e5aebc4850e6db..9c2025f7d1ec39 100644
--- a/llvm/test/Other/new-pm-thinlto-postlink-samplepgo-defaults.ll
+++ b/llvm/test/Other/new-pm-thinlto-postlink-samplepgo-defaults.ll
@@ -59,6 +59,7 @@
; CHECK-O-NEXT: Running analysis: PostDominatorTreeAnalysis on foo
; CHECK-O-NEXT: Running pass: SimplifyCFGPass on foo
+; CHECK-O-NEXT: Running pass: PGOForceFunctionAttrsPass
; CHECK-O-NEXT: Running pass: AlwaysInlinerPass
; CHECK-O-NEXT: Running pass: ModuleInlinerWrapperPass
; CHECK-O-NEXT: Running analysis: InlineAdvisorAnalysis
diff --git a/llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll
b/llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll
index cb49cbd22d60c0..c5fc4d57539c5f 100644
--- a/llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll
+++ b/llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll
@@ -84,6 +84,7 @@
; CHECK-O-NEXT: Running pass: PGOIndirectCallPromotion on
; CHECK-O-NEXT: Running analysis: InnerAnalysisManagerProxy
; CHECK-O-NEXT: Running analysis: OptimizationRemarkEmitterAnalysis on foo
+; CHECK-O-NEXT: Running pass: PGOForceFunctionAttrsPass
; CHECK-O-NEXT: Running pass: AlwaysInlinerPass
; CHECK-O-NEXT: Running pass: ModuleInlinerWrapperPass
; CHECK-O-NEXT: Running analysis: InlineAdvisorAnalysis
diff --git a/llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
b/llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
index 96e83493504426..096110f775b04f 100644
--- a/llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
+++ b/llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
@@ -64,6 +64,7 @@
; CHECK-O-NEXT: Running analysis: LoopAnalysis on foo
; CHECK-O-NEXT: Running analysis: PostDominatorTreeAnalysis on foo
; CHECK-O-NEXT: Running pass: SimplifyCFGPass on foo
+; CHECK-O-NEXT: Running pass: PGOForceFunctionAttrsPass
; CHECK-O-NEXT: Running pass: AlwaysInlinerPass
; CHECK-O-NEXT: Running pass:
[clang] [flang] [llvm] [mlir] Make MMIWP not have ownership over MMI + Remove Move Constructor of MMI + Make MMI Only Use and Externally-Created MCContext (PR #105541)
aeubanks wrote: I'm just back from vacation, taking a look looks like this has merge conflicts could you simplify the commit title (maybe move some details into the description) https://github.com/llvm/llvm-project/pull/105541 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [inline] Clone return range attribute on the callsite into inlined call (PR #92666)
aeubanks wrote: for future reference, https://github.com/llvm/llvm-project/commit/3cd67eeca28ab1084d02b7976de1af4c4c8d37d5 fixes a miscompile that this patch uncovered https://github.com/llvm/llvm-project/pull/92666 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [analyzer] Refine invalidation caused by `fread` (PR #93408)
@@ -907,6 +945,76 @@ void StreamChecker::preWrite(const FnDescription *Desc,
const CallEvent &Call,
C.addTransition(State);
}
+static std::optional getPointeeType(const MemRegion *R) {
+ if (!R)
+return std::nullopt;
+ if (const auto *ER = dyn_cast(R))
+return ER->getElementType();
+ if (const auto *TR = dyn_cast(R))
+return TR->getValueType();
+ if (const auto *SR = dyn_cast(R))
+return SR->getPointeeStaticType();
+ return std::nullopt;
+}
+
+static std::optional getStartIndex(SValBuilder &SVB,
+ const MemRegion *R) {
+ if (!R)
+return std::nullopt;
+
+ auto Zero = [&SVB] {
+BasicValueFactory &BVF = SVB.getBasicValueFactory();
+return nonloc::ConcreteInt(BVF.getIntValue(0, /*isUnsigned=*/false));
+ };
+
+ if (const auto *ER = dyn_cast(R))
+return ER->getIndex();
+ if (const auto *TR = dyn_cast(R))
aeubanks wrote:
unused variable (and ditto below), please use `isa<>`
https://github.com/llvm/llvm-project/pull/93408
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang] Reland Add tanf16 builtin and support for tan constrained intrinsic (PR #94559)
aeubanks wrote: @alexey-bataev https://github.com/llvm/llvm-project/pull/94559 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang] Reland Add tanf16 builtin and support for tan constrained intrinsic (PR #94559)
aeubanks wrote:
SLPVectorizer can introduce `llvm.tan.v2f32`. For example, running `opt -O3` on
the following introduces `llvm.tan.v2f32`
```
target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
target triple = "thumbv7-unknown-linux-android26"
; Function Attrs: null_pointer_is_valid
define void @foo() #0 {
bb:
%call = call float @pluto(float 1.00e+00)
%call1 = call i1 @ham(float %call)
%call2 = call float @pluto(float 0.00e+00)
%call3 = call i1 @ham(float %call2)
%select = select i1 %call3, float 0.00e+00, float 1.00e+00
%select4 = select i1 %call1, float 0.00e+00, float 1.00e+00
store float %select, ptr null, align 4
%call5 = call ptr null(ptr null, float %select4, float 0.00e+00, float
0.00e+00, float 0.00e+00)
ret void
}
define float @pluto(float %arg) {
bb:
%call = call float @llvm.tan.f32(float %arg)
ret float %call
}
define i1 @ham(float %arg) {
bb:
%fcmp = fcmp ole float %arg, 0.00e+00
ret i1 %fcmp
}
; Function Attrs: nocallback nofree nosync nounwind speculatable willreturn
memory(none)
declare float @llvm.tan.f32(float) #1
attributes #0 = { null_pointer_is_valid }
attributes #1 = { nocallback nofree nosync nounwind speculatable willreturn
memory(none) }
```
https://github.com/llvm/llvm-project/pull/94559
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] Support [[guarded_by(mutex)]] attribute inside C struct (PR #94216)
aeubanks wrote: > @aeubanks I'll revert. Is this example C or C++? C++ https://github.com/llvm/llvm-project/pull/94216 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] Support [[guarded_by(mutex)]] attribute inside C struct (PR #94216)
aeubanks wrote:
reduced:
```
struct Lock {};
struct A {
Lock lock;
union {
bool b __attribute__((guarded_by(lock)));
};
};
```
seems like a regression, can we revert for now?
https://github.com/llvm/llvm-project/pull/94216
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] Support [[guarded_by(mutex)]] attribute inside C struct (PR #94216)
aeubanks wrote:
this seems to have broken code like
[this](https://crsrc.org/c/base/allocator/partition_allocator/src/partition_alloc/random.cc;drc=36293863bf9bbc8131797eb8f4d2bafe8af3de79;l=33)
with
```
../../base/allocator/partition_allocator/src/partition_alloc/random.cc:33:69:
error: invalid use of non-static data member 'lock_'
33 | internal::base::InsecureRandomGenerator instance_
PA_GUARDED_BY(lock_);
|
^
../../base/allocator/partition_allocator/src/partition_alloc/partition_alloc_base/thread_annotations.h:59:70:
note: expanded from macro 'PA_GUARDED_BY'
59 | #define PA_GUARDED_BY(x) PA_THREAD_ANNOTATION_ATTRIBUTE__(guarded_by(x))
| ^
../../base/allocator/partition_allocator/src/partition_alloc/partition_alloc_base/thread_annotations.h:44:60:
note: expanded from macro 'PA_THREAD_ANNOTATION_ATTRIBUTE__'
44 | #define PA_THREAD_ANNOTATION_ATTRIBUTE__(x) __attribute__((x))
|^
../../base/allocator/partition_allocator/src/partition_alloc/random.cc:35:65:
error: invalid use of non-static data member 'lock_'
35 | internal::base::InsecureRandomGenerator)] PA_GUARDED_BY(lock_)
= {};
| ^
../../base/allocator/partition_allocator/src/partition_alloc/partition_alloc_base/thread_annotations.h:59:70:
note: expanded from macro 'PA_GUARDED_BY'
59 | #define PA_GUARDED_BY(x) PA_THREAD_ANNOTATION_ATTRIBUTE__(guarded_by(x))
| ^
../../base/allocator/partition_allocator/src/partition_alloc/partition_alloc_base/thread_annotations.h:44:60:
note: expanded from macro 'PA_THREAD_ANNOTATION_ATTRIBUTE__'
44 | #define PA_THREAD_ANNOTATION_ATTRIBUTE__(x) __attribute__((x))
|^
```
is that intentional?
https://github.com/llvm/llvm-project/pull/94216
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-repl] Lay the foundation of pretty printing for C. (PR #89811)
aeubanks wrote: This was with `-DLLVM_ENABLE_LLD=ON`, so probably not? I'm guessing there's some CMake logic that doesn't fit with clang-repl's expectation of symbol availability, but that may be wrong. https://github.com/llvm/llvm-project/pull/89811 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang] Reland Add tanf16 builtin and support for tan constrained intrinsic (PR #94559)
aeubanks wrote:
I believe this has exposed a preexisting issue in isel with tan:
```
$ cat /tmp/b.ll
target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
target triple = "thumbv7-unknown-linux-android26"
define <2 x float> @g() {
entry:
%0 = call <2 x float> @llvm.tan.v2f32(<2 x float> zeroinitializer)
ret <2 x float> %0
}
$ llc /tmp/b.ll -o /dev/null
LLVM ERROR: Cannot select: t3: v2f32 = ftan t15
t15: v2f32 = bitcast t14
t14: v2i32 = ARMISD::VMOVIMM TargetConstant:i32<0>
t13: i32 = TargetConstant<0>
```
https://github.com/llvm/llvm-project/pull/94559
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang-repl] Lay the foundation of pretty printing for C. (PR #89811)
aeubanks wrote: So actually even the `export_executable_symbols_for_plugins` doesn't fix our bots. I've narrowed it down to `-DLLVM_ENABLE_PIC=ON/OFF`. Perhaps we're not exporting symbols when `-DLLVM_ENABLE_PIC=OFF`. https://github.com/llvm/llvm-project/pull/89811 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [ConstantFold] Drop gep of gep fold entirely (PR #95126)
https://github.com/aeubanks approved this pull request. https://github.com/llvm/llvm-project/pull/95126 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
https://github.com/aeubanks closed https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
aeubanks wrote: thanks for doing this! https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
https://github.com/aeubanks approved this pull request. https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [ConstantFold] Remove non-trivial gep-of-gep fold (PR #93823)
https://github.com/aeubanks approved this pull request. this PR looks good, and further dropping the fold is also good https://github.com/llvm/llvm-project/pull/93823 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
aeubanks wrote: looks like `CodeGen/AMDGPU/llc-pipeline.ll` is failing https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
https://github.com/aeubanks approved this pull request. https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] [llvm] [mlir] [ConstantFold] Remove notional over-indexing fold (PR #93697)
https://github.com/aeubanks approved this pull request. https://github.com/llvm/llvm-project/pull/93697 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
@@ -1030,6 +1036,12 @@
PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
Phase != ThinOrFullLTOPhase::ThinLTOPostLink)
MPM.addPass(SampleProfileProbePass(TM));
+ // Instrument function entry and exit before all inlining.
+ if (!isLTOPostLink(Phase)) {
+MPM.addPass(createModuleToFunctionPassAdaptor(
aeubanks wrote:
if you have a small local repro, you can pass `-Wl,-mllvm,-print-after-all` to
the link command to have it print IR after every pass and see where the calls
are getting added. or to see changes in the pre-link optimization pipeline,
`-mllvm=-print-after-all` (or `-mllvm=-print-changed=quiet`, which doesn't work
with the IR parts of the codegen pipeline for *reasons*)
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines (PR #92171)
@@ -1030,6 +1036,12 @@
PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
Phase != ThinOrFullLTOPhase::ThinLTOPostLink)
MPM.addPass(SampleProfileProbePass(TM));
+ // Instrument function entry and exit before all inlining.
+ if (!isLTOPostLink(Phase)) {
+MPM.addPass(createModuleToFunctionPassAdaptor(
aeubanks wrote:
the check for `Phase != ThinOrFullLTOPhase::FullLTOPostLink` is unnecessary,
`buildModuleSimplificationPipeline` isn't called for FullLTO post link
I'm not sure why you're seeing that behavior if we're only running the
post-inline instrumenter once in the codegen pipeline
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [lld] [llvm] Run ObjCContractPass in Default Codegen Pipeline (PR #92331)
@@ -31,6 +31,10 @@ ; CHECK-NEXT: AArch64 Stack Tagging ; CHECK-NEXT: SME ABI Pass ; CHECK-NEXT: Exception handling preparation +; CHECK-NEXT: Dominator Tree Construction +; CHECK-NEXT: Basic Alias Analysis (stateless AA impl) +; CHECK-NEXT: Function Alias Analysis Results aeubanks wrote: this is where having the new pass manager for codegen would be nice https://github.com/llvm/llvm-project/pull/92331 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] Move instrumentation passes (PR #92171)
@@ -1030,6 +1036,12 @@
PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
Phase != ThinOrFullLTOPhase::ThinLTOPostLink)
MPM.addPass(SampleProfileProbePass(TM));
+ // Instrument function entry and exit before all inlining.
+ if (!isLTOPostLink(Phase)) {
+MPM.addPass(createModuleToFunctionPassAdaptor(
aeubanks wrote:
actually is it possible to add this into `EarlyFPM` below? it's nice to have
fewer module->function adaptors. (e.g. in `EarlyFPM` we run all the function
passes on a single function before moving to the next one which is nice for
memory locality)
sorry for suggesting this so late, since it'll require updating all the
pipeline tests... but first check that none of the other tests fail
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] Move instrumentation passes (PR #92171)
@@ -1,5 +1,5 @@ // RUN: %clang_cc1 -triple x86_64-apple-darwin10 -debug-info-kind=standalone -emit-llvm -o - %s -finstrument-functions | FileCheck %s aeubanks wrote: this should also not be testing the pass and should have -disable-llvm-passes https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] Move instrumentation passes (PR #92171)
https://github.com/aeubanks commented: looks pretty good, can you update the commit title to something like `[EntryExitInstrumenter] Move passes out of clang into LLVM default pipelines` I've added test coverage for mcount-aix in 3ec57a7ed60a19c5e3e18655e19c997a469d10ec, can you merge that in? https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] Move instrumentation passes (PR #92171)
https://github.com/aeubanks edited https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] Move instrumentation passes (PR #92171)
@@ -1,25 +1,13 @@ // RUN: %clang_cc1 -pg -triple powerpc-ibm-aix7.2.0.0 -emit-llvm %s -o - | FileCheck %s aeubanks wrote: also -disable-llvm-passes here https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -1,37 +1,27 @@
// REQUIRES: x86-registered-target
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown -finstrument-functions -O0
-o - -emit-llvm %s | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown -finstrument-functions -O2
-o - -emit-llvm %s | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown
-finstrument-functions-after-inlining -O2 -o - -emit-llvm %s | FileCheck
-check-prefix=NOINLINE %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions -O0 -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions -O2 -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions-after-inlining -O2 -o - -emit-llvm %s | FileCheck
-check-prefix=NOINLINE %s
__attribute__((always_inline)) int leaf(int x) {
return x;
// CHECK-LABEL: define {{.*}} @leaf
-// CHECK: call void @__cyg_profile_func_enter
-// CHECK-NOT: cyg_profile
-// CHECK: call void @__cyg_profile_func_exit
// CHECK-NOT: cyg_profile
// CHECK: ret
}
int root(int x) {
return leaf(x);
// CHECK-LABEL: define {{.*}} @root
-// CHECK: call void @__cyg_profile_func_enter
-// CHECK-NOT: cyg_profile
-
-// Inlined from leaf():
-// CHECK: call void @__cyg_profile_func_enter
-// CHECK-NOT: cyg_profile
-// CHECK: call void @__cyg_profile_func_exit
// CHECK-NOT: cyg_profile
-
-// CHECK: call void @__cyg_profile_func_exit
// CHECK: ret
// NOINLINE-LABEL: define {{.*}} @root
-// NOINLINE: call void @__cyg_profile_func_enter
-// NOINLINE-NOT: cyg_profile
-// NOINLINE: call void @__cyg_profile_func_exit
// NOINLINE-NOT: cyg_profile
// NOINLINE: ret
}
+
+// CHECK: attributes #0 = { alwaysinline {{.*}}
"instrument-function-entry"="__cyg_profile_func_enter"
"instrument-function-exit"="__cyg_profile_func_exit"
+// CHECK: attributes #1 = { {{.*}}
"instrument-function-entry"="__cyg_profile_func_enter"
"instrument-function-exit"="__cyg_profile_func_exit"
+// NOINLINE: attributes #0 = { alwaysinline {{.*}}
"instrument-function-entry-inlined"="__cyg_profile_func_enter"
"instrument-function-exit-inlined"="__cyg_profile_func_exit"
aeubanks wrote:
drop the `alwaysinline` check since that's not relevant (we can also remove it
from the source), and check ` { {{.*}}"instrument-function-entry"=...` in case
`"instrument-function-entry"` is the first attribute in the group and there's
only one space before it
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -1,37 +1,27 @@
// REQUIRES: x86-registered-target
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown -finstrument-functions -O0
-o - -emit-llvm %s | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown -finstrument-functions -O2
-o - -emit-llvm %s | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown
-finstrument-functions-after-inlining -O2 -o - -emit-llvm %s | FileCheck
-check-prefix=NOINLINE %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions -O0 -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions -O2 -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions-after-inlining -O2 -o - -emit-llvm %s | FileCheck
-check-prefix=NOINLINE %s
__attribute__((always_inline)) int leaf(int x) {
return x;
// CHECK-LABEL: define {{.*}} @leaf
-// CHECK: call void @__cyg_profile_func_enter
-// CHECK-NOT: cyg_profile
-// CHECK: call void @__cyg_profile_func_exit
// CHECK-NOT: cyg_profile
aeubanks wrote:
no need to CHECK the IR here
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
https://github.com/aeubanks edited https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
https://github.com/aeubanks commented: forgot to hit "submit review"... https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -1,37 +1,27 @@
// REQUIRES: x86-registered-target
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown -finstrument-functions -O0
-o - -emit-llvm %s | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown -finstrument-functions -O2
-o - -emit-llvm %s | FileCheck %s
-// RUN: %clang_cc1 -triple x86_64-unknown-unknown
-finstrument-functions-after-inlining -O2 -o - -emit-llvm %s | FileCheck
-check-prefix=NOINLINE %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions -O0 -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions -O2 -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -disable-llvm-passes -triple x86_64-unknown-unknown
-finstrument-functions-after-inlining -O2 -o - -emit-llvm %s | FileCheck
-check-prefix=NOINLINE %s
__attribute__((always_inline)) int leaf(int x) {
return x;
// CHECK-LABEL: define {{.*}} @leaf
aeubanks wrote:
check `#0` here
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,29 @@
+; RUN: llc -mtriple=x86_64-- -O0 < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-- -O1 < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-- -O2 < %s | FileCheck %s
+
+; The codegen should insert post-inlining instrumentation calls and should not
+; insert pre-inlining instrumentation.
+
+; CHECK-NOT: callq __cyg_profile_func_enter
+
+define void @leaf_function() #0 {
+; CHECK-LABEL: leaf_function:
+; CHECK: callq __cyg_profile_func_enter_bare
+; CHECK: movqleaf_function@GOTPCREL(%rip), %rdi
aeubanks wrote:
checking the specifics of how the function is called shouldn't be in this test
that tests the pipeline, the point of this test is just a quick end-to-end test
testing we inserted instrumentation. if we want to test this we should have a
separate test focused on the exact instruction sequence. but I don't think we
need to add that test coverage in this PR
so I'd remove the `%rdi` CHECK line
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,44 @@
+; RUN: opt -passes="default" -S < %s | FileCheck -check-prefix=PRELTO %s
+; RUN: opt -passes="default" -S < %s | FileCheck -check-prefix=PRELTO %s
+; RUN: opt -passes="thinlto-pre-link,thinlto" -S < %s | FileCheck
-check-prefix=PRELTO %s
+; RUN: opt -passes="thinlto-pre-link" -S < %s | FileCheck
-check-prefix=PRELTO %s
+; RUN: opt -passes="thinlto" -S < %s | FileCheck -check-prefix=LTO %s
+; RUN: opt -passes="lto" -S < %s | FileCheck -check-prefix=LTO %s
+
+; Pre-inline instrumentation should be inserted, but not by LTO/ThinLTO passes.
+
+target triple = "x86_64-unknown-linux"
+
+define void @leaf_function() #0 {
+entry:
+ ret void
+; PRELTO-LABEL: entry:
+; PRELTO-NEXT: %0 ={{( tail)?}} call ptr @llvm.returnaddress(i32 0)
aeubanks wrote:
no need to add `{{( tail)?}}` if it's at the very beginning. for this specific
CHECK line I'd simplify to `{{.*}}`
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -1,41 +0,0 @@
-// REQUIRES: arm-registered-target,aarch64-registered-target
-
-// RUN: %clang -target armv7-unknown-none-eabi -pg -S -emit-llvm -o - %s |
FileCheck %s -check-prefixes=CHECK,UNSUPPORTED
-// RUN: %clang -target armv7-unknown-none-eabi -pg -meabi gnu -S -emit-llvm -o
- %s | FileCheck %s --check-prefixes=CHECK,UNSUPPORTED
-// RUN: %clang --target=aarch64-unknown-none-gnu -pg -S -emit-llvm -o - %s |
FileCheck %s --check-prefixes=CHECK,MCOUNT
-// RUN: %clang -target armv7-unknown-linux-gnueabi -pg -S -emit-llvm -o - %s |
FileCheck %s -check-prefix CHECK -check-prefix CHECK-ARM-EABI
-// RUN: %clang -target armv7-unknown-linux-gnueabi -meabi gnu -pg -S
-emit-llvm -o - %s | FileCheck %s -check-prefix CHECK -check-prefix
CHECK-ARM-EABI-MEABI-GNU
-// RUN: %clang --target=aarch64-unknown-linux -pg -S -emit-llvm -o - %s |
FileCheck %s --check-prefixes=CHECK,UNDER
-// RUN: %clang -target armv7-unknown-linux-gnueabihf -pg -S -emit-llvm -o - %s
| FileCheck %s -check-prefix CHECK -check-prefix CHECK-ARM-EABI
-// RUN: %clang -target armv7-unknown-linux-gnueabihf -meabi gnu -pg -S
-emit-llvm -o - %s | FileCheck %s -check-prefix CHECK -check-prefix
CHECK-ARM-EABI-MEABI-GNU
-// RUN: %clang -target armv7-unknown-freebsd-gnueabihf -pg -S -emit-llvm -o -
%s | FileCheck %s --check-prefixes=CHECK,UNDER_UNDER
-// RUN: %clang -target armv7-unknown-freebsd-gnueabihf -meabi gnu -pg -S
-emit-llvm -o - %s | FileCheck %s --check-prefixes=CHECK,UNDER_UNDER
-// RUN: %clang --target=aarch64-unknown-freebsd -pg -S -emit-llvm -o - %s |
FileCheck %s -check-prefix CHECK -check-prefix CHECK-ARM64-EABI-FREEBSD
-// RUN: %clang -target armv7-unknown-openbsd-gnueabihf -pg -S -emit-llvm -o -
%s | FileCheck %s -check-prefix CHECK -check-prefix UNDER_UNDER
-// RUN: %clang -target armv7-unknown-openbsd-gnueabihf -meabi gnu -pg -S
-emit-llvm -o - %s | FileCheck %s -check-prefix CHECK -check-prefix UNDER_UNDER
-// RUN: %clang --target=aarch64-unknown-openbsd -pg -S -emit-llvm -o - %s |
FileCheck %s -check-prefix CHECK -check-prefix UNDER_UNDER
-// RUN: %clang -target armv7-unknown-netbsd-gnueabihf -pg -S -emit-llvm -o -
%s | FileCheck %s --check-prefixes=CHECK,UNDER_UNDER
-// RUN: %clang -target armv7-unknown-netbsd-gnueabihf -meabi gnu -pg -S
-emit-llvm -o - %s | FileCheck %s --check-prefixes=CHECK,UNDER_UNDER
-// RUN: %clang --target=aarch64-unknown-netbsd -pg -S -emit-llvm -o - %s |
FileCheck %s --check-prefixes=CHECK,UNDER_UNDER
-// RUN: %clang -target armv7-apple-ios -pg -S -emit-llvm -o - %s | FileCheck
%s --check-prefixes=CHECK,UNSUPPORTED
-// RUN: %clang -target armv7-apple-ios -pg -meabi gnu -S -emit-llvm -o - %s |
FileCheck %s --check-prefixes=CHECK,UNSUPPORTED
-// RUN: %clang -target arm64-apple-ios -pg -S -emit-llvm -o - %s | FileCheck
%s --check-prefixes=CHECK,UNSUPPORTED
-// RUN: %clang -target arm64-apple-ios -pg -meabi gnu -S -emit-llvm -o - %s |
FileCheck %s --check-prefixes=CHECK,UNSUPPORTED
-// RUN: %clang -target armv7-unknown-rtems-gnueabihf -pg -S -emit-llvm -o - %s
| FileCheck %s --check-prefixes=CHECK,MCOUNT
-// RUN: %clang -target armv7-unknown-rtems-gnueabihf -meabi gnu -pg -S
-emit-llvm -o - %s | FileCheck %s --check-prefixes=CHECK,MCOUNT
-// RUN: %clang --target=aarch64-unknown-rtems -pg -S -emit-llvm -o - %s |
FileCheck %s -check-prefixes=CHECK,MCOUNT
-// RUN: %clang --target=aarch64-unknown-rtems -pg -S -emit-llvm -o - %s |
FileCheck %s -check-prefixes=CHECK,MCOUNT
-
-int f() {
- return 0;
-}
-
-// CHECK-LABEL: f
-// TODO: add profiling support for arm-baremetal
aeubanks wrote:
we should keep both these tests you're deleting to ensure that we're emitting
the correct function attributes. add `-Xclang -disable-llvm-passes` to the
clang invocations to not run any passes and check the function attributes.
this test is important to check that we're calling the correct `mcount`, and
the other test is important to test that
`-finstrument-functions-after-inlining` is emitting the correct function
attribute (that test is the only instance of
`-finstrument-functions-after-inlining`)
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,44 @@ +; RUN: opt -passes="default" -S < %s | FileCheck -check-prefix=PRELTO %s aeubanks wrote: `PRELTO` isn't quite accurate since we also instrument in the default pipelines, I'd use `INSTRUMENT`/`NOINSTRUMENT` or even just `YES`/`NO` :) https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,44 @@ +; RUN: opt -passes="default" -S < %s | FileCheck -check-prefix=PRELTO %s +; RUN: opt -passes="default" -S < %s | FileCheck -check-prefix=PRELTO %s +; RUN: opt -passes="thinlto-pre-link,thinlto" -S < %s | FileCheck -check-prefix=PRELTO %s aeubanks wrote: don't run two pipelines in a single `opt` invocation, and we need tests for `thinlto`/`lto` https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [lldb] [llvm] Remove some `try_compile` CMake checks for compiler flags (PR #92953)
https://github.com/aeubanks approved this pull request. awesome, thanks! https://github.com/llvm/llvm-project/pull/92953 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,29 @@
+; RUN: llc -mtriple=x86_64-- -O0 < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-- -O1 < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-- -O2 < %s | FileCheck %s
+
+; The codegen should insert post-inlining instrumentation calls and should not
+; insert pre-inlining instrumentation.
+
+; CHECK-NOT: callq __cyg_profile_func_enter
+
+define void @leaf_function() #0 {
+; CHECK-LABEL: leaf_function:
+; CHECK: callq __cyg_profile_func_enter_bare
+; CHECK: movqleaf_function@GOTPCREL(%rip), %rdi
aeubanks wrote:
probably don't need this CHECK line, it just makes the test more brittle
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,31 @@
+; RUN: opt -passes="default" -S < %s | FileCheck %s
+; RUN: opt -passes="thinlto-pre-link" -S < %s | FileCheck %s
+; RUN: opt -passes="thinlto-pre-link,thinlto" -S < %s | FileCheck %s
+
+target triple = "x86_64-unknown-linux"
+
+define void @leaf_function() #0 {
aeubanks wrote:
mark this function as `alwaysinline` so testing `-O0` works
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,31 @@ +; RUN: opt -passes="default" -S < %s | FileCheck %s aeubanks wrote: should also test `O0` versions of these, plus `lto-pre-link` https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -0,0 +1,31 @@ +; RUN: opt -passes="default" -S < %s | FileCheck %s +; RUN: opt -passes="thinlto-pre-link" -S < %s | FileCheck %s +; RUN: opt -passes="thinlto-pre-link,thinlto" -S < %s | FileCheck %s aeubanks wrote: no need to run `thinlto` here, but we should test that `thinlto`/`lto` don't run this pass by adding different RUN lines with a different `FileCheck %s --checkprefix=NO` https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [lld] [llvm] Run ObjCContractPass in Distributed Thin-LTO Pipeline (PR #92331)
aeubanks wrote: it seems like this should just be in the default codegen pipeline? you'd need to change the pass to bail out early if there are no relevant intrinsics (by checking if the module contains the intrinsic declaration) to not affect compile times https://github.com/llvm/llvm-project/pull/92331 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] [llvm] [ConstantFolding] Canonicalize constexpr GEPs to i8 (PR #89872)
https://github.com/aeubanks approved this pull request. we've resolved the performance regression from the previous patch internally, thanks for waiting! https://github.com/llvm/llvm-project/pull/89872 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [IR] Add getelementptr nusw and nuw flags (PR #90824)
@@ -0,0 +1,86 @@
+//===-- llvm/GEPNoWrapFlags.h - NoWrap flags for GEPs ---*- C++
-*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM
Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===--===//
+//
+// This file defines the nowrap flags for getelementptr operators.
+//
+//===--===//
+
+#ifndef LLVM_IR_GEPNOWRAPFLAGS_H
+#define LLVM_IR_GEPNOWRAPFLAGS_H
+
+namespace llvm {
+
+class GEPNoWrapFlags {
+ enum : unsigned{
+InBoundsFlag = (1 << 0),
+NUSWFlag = (1 << 1),
+NUWFlag = (1 << 2),
+ };
+
+ unsigned Flags;
+ GEPNoWrapFlags(unsigned Flags) : Flags(Flags) {
+assert((!isInBounds() || hasNoUnsignedSignedWrap()) &&
+ "inbounds implies nusw");
+ }
+
+public:
+ GEPNoWrapFlags() : Flags(0) {}
+ // For historical reasons, interpret plain boolean as InBounds.
aeubanks wrote:
TODO: remove?
https://github.com/llvm/llvm-project/pull/90824
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [IR] Add getelementptr nusw and nuw flags (PR #90824)
https://github.com/aeubanks approved this pull request. thanks, I think abstracting out GEPNoWrapFlags is good https://github.com/llvm/llvm-project/pull/90824 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [IR] Add getelementptr nusw and nuw flags (PR #90824)
@@ -0,0 +1,86 @@
+//===-- llvm/GEPNoWrapFlags.h - NoWrap flags for GEPs ---*- C++
-*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM
Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===--===//
+//
+// This file defines the nowrap flags for getelementptr operators.
+//
+//===--===//
+
+#ifndef LLVM_IR_GEPNOWRAPFLAGS_H
+#define LLVM_IR_GEPNOWRAPFLAGS_H
+
+namespace llvm {
+
+class GEPNoWrapFlags {
+ enum : unsigned{
aeubanks wrote:
clang-format
https://github.com/llvm/llvm-project/pull/90824
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [IR] Add getelementptr nusw and nuw flags (PR #90824)
https://github.com/aeubanks edited https://github.com/llvm/llvm-project/pull/90824 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -1028,6 +1029,14 @@
PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level,
Phase != ThinOrFullLTOPhase::ThinLTOPostLink)
MPM.addPass(SampleProfileProbePass(TM));
+ // Instrument function entry and exit before all inlining.
+ if (Phase != ThinOrFullLTOPhase::ThinLTOPostLink &&
+ Phase != ThinOrFullLTOPhase::FullLTOPostLink &&
+ Phase != ThinOrFullLTOPhase::None) {
aeubanks wrote:
we need this pass when `Phase == ThinOrFullLTOPhase::None` right?
I'd extract out `isLTOPostLink` to mirror the existing `isLTOPreLink`
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
https://github.com/aeubanks commented: looks pretty good to me for testing the pre-inliner one, we should add some tests in `llvm/test/Transforms/EntryExitInstrumenter/` that invoke things like `opt -passes='default'`, `opt -passes='thinlto-pre-link'`, `opt -passes='thinlto'` to make sure that the pass did/didn't insert the call given some IR with the appropriate function attribute for testing the post-inliner one, an `llc` x86-64 test that checks that a call to the function was generated in the assembly given some IR with the appropriate function attribute is enough https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
https://github.com/aeubanks edited https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -101,6 +101,7 @@ void initializeEarlyMachineLICMPass(PassRegistry&); void initializeEarlyTailDuplicatePass(PassRegistry&); void initializeEdgeBundlesPass(PassRegistry&); void initializeEHContGuardCatchretPass(PassRegistry &); +void initializeEntryExitInstrumenterPass(PassRegistry&); aeubanks wrote: not necessary https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
aeubanks wrote: can you add links to https://reviews.llvm.org/D97608, https://github.com/rust-lang/rust/issues/92109, https://github.com/llvm/llvm-project/issues/52853 https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -135,6 +138,65 @@ static bool runOnFunction(Function &F, bool PostInlining) {
return Changed;
}
+namespace {
+struct EntryExitInstrumenter : public FunctionPass {
+ static char ID;
+ EntryExitInstrumenter() : FunctionPass(ID) {
+initializeEntryExitInstrumenterPass(*PassRegistry::getPassRegistry());
+ }
+ void getAnalysisUsage(AnalysisUsage &AU) const override {
+AU.addPreserved();
+AU.addPreserved();
+ }
+ bool runOnFunction(Function &F) override { return ::runOnFunction(F, false);
}
+};
+char EntryExitInstrumenter::ID = 0;
+
+struct PostInlineEntryExitInstrumenter : public FunctionPass {
+ static char ID;
+ PostInlineEntryExitInstrumenter() : FunctionPass(ID) {
+initializePostInlineEntryExitInstrumenterPass(
+*PassRegistry::getPassRegistry());
+ }
+ void getAnalysisUsage(AnalysisUsage &AU) const override {
+AU.addPreserved();
+AU.addPreserved();
+ }
+ bool runOnFunction(Function &F) override { return ::runOnFunction(F, true); }
+};
+char PostInlineEntryExitInstrumenter::ID = 0;
+}
+
+INITIALIZE_PASS_BEGIN(
+EntryExitInstrumenter, "ee-instrument",
+"Instrument function entry/exit with calls to e.g. mcount() (pre
inlining)",
+false, false)
+INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
+INITIALIZE_PASS_END(
+EntryExitInstrumenter, "ee-instrument",
+"Instrument function entry/exit with calls to e.g. mcount() (pre
inlining)",
+false, false)
+
+INITIALIZE_PASS_BEGIN(
+PostInlineEntryExitInstrumenter, "post-inline-ee-instrument",
+"Instrument function entry/exit with calls to e.g. mcount() "
+"(post inlining)",
+false, false)
+INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
+INITIALIZE_PASS_END(
+PostInlineEntryExitInstrumenter, "post-inline-ee-instrument",
+"Instrument function entry/exit with calls to e.g. mcount() "
+"(post inlining)",
+false, false)
+
+FunctionPass *llvm::createEntryExitInstrumenterPass() {
aeubanks wrote:
we don't need this one since we're adding it in the optimization pipeline which
only works with the new pass manager
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -670,9 +670,6 @@ void CodeGenPassBuilder::addIRPasses( !Opt.DisablePartialLibcallInlining) addPass(PartiallyInlineLibCallsPass()); - // Instrument function entry and exit, e.g. with calls to mcount(). - addPass(EntryExitInstrumenterPass(/*PostInlining=*/true)); aeubanks wrote: don't touch this one, this is the WIP port of the codegen pipeline to the new pass manager (and this is actually what we want, that the codegen pipeline does this) https://github.com/llvm/llvm-project/pull/92171 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] wip: Move instrumentation passes (PR #92171)
@@ -1016,6 +1000,11 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
if (!IsThinLTOPostLink) {
addSanitizers(TargetTriple, CodeGenOpts, LangOpts, PB);
addKCFIPass(TargetTriple, LangOpts, PB);
+ PB.registerPipelineStartEPCallback(
aeubanks wrote:
we should be consistent and make both entry exit instrumenters part of the
default pipelines, not a clang-specific add-on
this would probably be somewhere in `buildModuleSimplificationPipeline`,
checking `Phase` (and make sure -O0 works by modifying `buildO0DefaultPipeline`)
https://github.com/llvm/llvm-project/pull/92171
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [CodeGen] Revert "Generate assume loads only with -fstrict-vtable-pointers" (PR #91900)
aeubanks wrote: > -fstrict-vtable-pointers IS experimental, but if you recall, this particular > optimization was added to -fstrict-vtable-pointers because of the effects it > had on compile-time, not because of correctness issues. can you clarify what you mean by "this particular optimization"? you mean adding or not adding assume loads? when I said "regress", I meant runtime performance, not compile times, I should have been clearer https://github.com/llvm/llvm-project/pull/91900 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [CodeGen] Revert "Generate assume loads only with -fstrict-vtable-pointers" (PR #91900)
aeubanks wrote: adding assumes in general has issues: https://discourse.llvm.org/t/llvm-assume-blocks-optimization/71609 do you have proof that this change helps binaries and doesn't regress things? I have a feeling this will regress many things. `-fstrict-vtable-pointers` is still somewhat experimental https://github.com/llvm/llvm-project/pull/91900 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] Revert "[Clang][Sema] Fix lookup of dependent operator= outside of complete-class contexts (#91498)" (PR #91620)
aeubanks wrote: should be fine to revert as much as you want in a single PR, just make sure to mention what you're reverting in the description https://github.com/llvm/llvm-project/pull/91620 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [Clang][Sema] Fix lookup of dependent operator= outside of complete-class contexts (PR #91498)
aeubanks wrote: Chromium is also seeing similar breakages. @sdkrystian is this breaking valid code? I can't tell from your latest comment. (if it is breaking valid code we should revert) https://github.com/llvm/llvm-project/pull/91498 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] [llvm] [ConstantFolding] Canonicalize constexpr GEPs to i8 (PR #89872)
aeubanks wrote: btw we're still looking into a performance regression caused by #68882 that still repros with LLVM head, even after the SROA enhancements. this patch looks good, but can we hold off a bit on submitting this? https://github.com/llvm/llvm-project/pull/89872 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [compiler-rt] [llvm] [ConstantFolding] Canonicalize constexpr GEPs to i8 (PR #89872)
@@ -944,43 +943,18 @@ Constant *SymbolicallyEvaluateGEP(const GEPOperator *GEP,
return ConstantExpr::getIntToPtr(C, ResTy);
}
- // Otherwise form a regular getelementptr. Recompute the indices so that
- // we eliminate over-indexing of the notional static type array bounds.
- // This makes it easy to determine if the getelementptr is "inbounds".
-
- // For GEPs of GlobalValues, use the value type, otherwise use an i8 GEP.
- if (auto *GV = dyn_cast(Ptr))
-SrcElemTy = GV->getValueType();
- else
-SrcElemTy = Type::getInt8Ty(Ptr->getContext());
-
- if (!SrcElemTy->isSized())
-return nullptr;
-
- Type *ElemTy = SrcElemTy;
- SmallVector Indices = DL.getGEPIndicesForOffset(ElemTy, Offset);
- if (Offset != 0)
-return nullptr;
-
- // Try to add additional zero indices to reach the desired result element
- // type.
- // TODO: Should we avoid extra zero indices if ResElemTy can't be reached and
- // we'll have to insert a bitcast anyway?
- while (ElemTy != ResElemTy) {
-Type *NextTy = GetElementPtrInst::getTypeAtIndex(ElemTy, (uint64_t)0);
-if (!NextTy)
- break;
-
-Indices.push_back(APInt::getZero(isa(ElemTy) ? 32 : BitWidth));
-ElemTy = NextTy;
+ // Try to infer inbounds for GEPs of globals.
+ if (!InBounds && Offset.isNonNegative()) {
aeubanks wrote:
is there a test for the case where `!Offset.isNonNegative()`?
https://github.com/llvm/llvm-project/pull/89872
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [ConstantFolding] Canonicalize constexpr GEPs to i8 (PR #89872)
@@ -944,43 +943,18 @@ Constant *SymbolicallyEvaluateGEP(const GEPOperator *GEP,
return ConstantExpr::getIntToPtr(C, ResTy);
}
- // Otherwise form a regular getelementptr. Recompute the indices so that
- // we eliminate over-indexing of the notional static type array bounds.
- // This makes it easy to determine if the getelementptr is "inbounds".
-
- // For GEPs of GlobalValues, use the value type, otherwise use an i8 GEP.
- if (auto *GV = dyn_cast(Ptr))
-SrcElemTy = GV->getValueType();
- else
-SrcElemTy = Type::getInt8Ty(Ptr->getContext());
-
- if (!SrcElemTy->isSized())
-return nullptr;
-
- Type *ElemTy = SrcElemTy;
- SmallVector Indices = DL.getGEPIndicesForOffset(ElemTy, Offset);
- if (Offset != 0)
-return nullptr;
-
- // Try to add additional zero indices to reach the desired result element
- // type.
- // TODO: Should we avoid extra zero indices if ResElemTy can't be reached and
- // we'll have to insert a bitcast anyway?
- while (ElemTy != ResElemTy) {
-Type *NextTy = GetElementPtrInst::getTypeAtIndex(ElemTy, (uint64_t)0);
-if (!NextTy)
- break;
-
-Indices.push_back(APInt::getZero(isa(ElemTy) ? 32 : BitWidth));
-ElemTy = NextTy;
+ // Try to infer inbounds for GEPs of globals.
+ if (!InBounds && Offset.isNonNegative()) {
+bool CanBeNull, CanBeFreed;
+uint64_t DerefBytes =
+Ptr->getPointerDereferenceableBytes(DL, CanBeNull, CanBeFreed);
+InBounds = DerefBytes != 0 && !CanBeNull && Offset.sle(DerefBytes);
aeubanks wrote:
can we remove the other constant folding?
https://github.com/llvm/llvm-project/pull/89872
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang] Add flag to experiment with cold function attributes (PR #89298)
https://github.com/aeubanks closed https://github.com/llvm/llvm-project/pull/89298 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [clang] Add flag to experiment with cold function attributes (PR #89298)
https://github.com/aeubanks created
https://github.com/llvm/llvm-project/pull/89298
To be removed and promoted to a proper driver flag if experiments turn out
fruitful.
For now, this can be experimented with `-mllvm
-pgo-cold-func-opt=[optsize|minsize|optnone|default] -mllvm
-enable-pgo-force-function-attrs`.
Original LLVM patch for this functionality: #69030
>From 1c69510ab5a92998cc443100ecfb551776fc03a0 Mon Sep 17 00:00:00 2001
From: Arthur Eubanks
Date: Mon, 15 Apr 2024 20:40:43 +
Subject: [PATCH] [clang] Add flag to experiment with cold function attributes
To be removed and promoted to a proper driver flag if experiments turn out
fruitful.
For now, this can be experimented with `-mllvm
-pgo-cold-func-opt=[optsize|minsize|optnone|default] -mllvm
-enable-pgo-force-function-attrs`.
Original LLVM patch for this functionality: #69030
---
clang/lib/CodeGen/BackendUtil.cpp | 57 ---
.../test/CodeGen/pgo-force-function-attrs.ll | 12
2 files changed, 47 insertions(+), 22 deletions(-)
create mode 100644 clang/test/CodeGen/pgo-force-function-attrs.ll
diff --git a/clang/lib/CodeGen/BackendUtil.cpp
b/clang/lib/CodeGen/BackendUtil.cpp
index 6cc00b85664f41..22c3f8642ad8eb 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/clang/lib/CodeGen/BackendUtil.cpp
@@ -104,6 +104,21 @@ static cl::opt ClSanitizeOnOptimizerEarlyEP(
"sanitizer-early-opt-ep", cl::Optional,
cl::desc("Insert sanitizers on OptimizerEarlyEP."));
+// Experiment to mark cold functions as optsize/minsize/optnone.
+// TODO: remove once this is exposed as a proper driver flag.
+static cl::opt ClPGOColdFuncAttr(
+"pgo-cold-func-opt", cl::init(PGOOptions::ColdFuncOpt::Default),
cl::Hidden,
+cl::desc(
+"Function attribute to apply to cold functions as determined by PGO"),
+cl::values(clEnumValN(PGOOptions::ColdFuncOpt::Default, "default",
+ "Default (no attribute)"),
+ clEnumValN(PGOOptions::ColdFuncOpt::OptSize, "optsize",
+ "Mark cold functions with optsize."),
+ clEnumValN(PGOOptions::ColdFuncOpt::MinSize, "minsize",
+ "Mark cold functions with minsize."),
+ clEnumValN(PGOOptions::ColdFuncOpt::OptNone, "optnone",
+ "Mark cold functions with optnone.")));
+
extern cl::opt ProfileCorrelate;
// Re-link builtin bitcodes after optimization
@@ -768,42 +783,41 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
CodeGenOpts.InstrProfileOutput.empty() ? getDefaultProfileGenName()
:
CodeGenOpts.InstrProfileOutput,
"", "", CodeGenOpts.MemoryProfileUsePath, nullptr, PGOOptions::IRInstr,
-PGOOptions::NoCSAction, PGOOptions::ColdFuncOpt::Default,
+PGOOptions::NoCSAction, ClPGOColdFuncAttr,
CodeGenOpts.DebugInfoForProfiling,
/*PseudoProbeForProfiling=*/false, CodeGenOpts.AtomicProfileUpdate);
else if (CodeGenOpts.hasProfileIRUse()) {
// -fprofile-use.
auto CSAction = CodeGenOpts.hasProfileCSIRUse() ? PGOOptions::CSIRUse
: PGOOptions::NoCSAction;
-PGOOpt = PGOOptions(
-CodeGenOpts.ProfileInstrumentUsePath, "",
-CodeGenOpts.ProfileRemappingFile, CodeGenOpts.MemoryProfileUsePath,
VFS,
-PGOOptions::IRUse, CSAction, PGOOptions::ColdFuncOpt::Default,
-CodeGenOpts.DebugInfoForProfiling);
+PGOOpt = PGOOptions(CodeGenOpts.ProfileInstrumentUsePath, "",
+CodeGenOpts.ProfileRemappingFile,
+CodeGenOpts.MemoryProfileUsePath, VFS,
+PGOOptions::IRUse, CSAction, ClPGOColdFuncAttr,
+CodeGenOpts.DebugInfoForProfiling);
} else if (!CodeGenOpts.SampleProfileFile.empty())
// -fprofile-sample-use
PGOOpt = PGOOptions(
CodeGenOpts.SampleProfileFile, "", CodeGenOpts.ProfileRemappingFile,
CodeGenOpts.MemoryProfileUsePath, VFS, PGOOptions::SampleUse,
-PGOOptions::NoCSAction, PGOOptions::ColdFuncOpt::Default,
+PGOOptions::NoCSAction, ClPGOColdFuncAttr,
CodeGenOpts.DebugInfoForProfiling,
CodeGenOpts.PseudoProbeForProfiling);
else if (!CodeGenOpts.MemoryProfileUsePath.empty())
// -fmemory-profile-use (without any of the above options)
PGOOpt = PGOOptions("", "", "", CodeGenOpts.MemoryProfileUsePath, VFS,
PGOOptions::NoAction, PGOOptions::NoCSAction,
-PGOOptions::ColdFuncOpt::Default,
-CodeGenOpts.DebugInfoForProfiling);
+ClPGOColdFuncAttr, CodeGenOpts.DebugInfoForProfiling);
else if (CodeGenOpts.PseudoProbeForProfiling)
// -fpseudo-probe-for-profiling
-PGOOpt = PGOOptions("", "", "", /*MemoryProfile=*/"", nullptr,
-PGOOptions::NoAction, PGOOptions::NoCSAction,
-
[clang] [clang] Add flag to experiment with cold function attributes (PR #88793)
https://github.com/aeubanks created
https://github.com/llvm/llvm-project/pull/88793
To be removed and promoted to a proper driver flag if experiments turn out
fruitful.
Original LLVM patch for this functionality: #69030
>From 52cd9974be908bf693832012e56e945e9e34f389 Mon Sep 17 00:00:00 2001
From: Arthur Eubanks
Date: Mon, 15 Apr 2024 20:40:43 +
Subject: [PATCH] [clang] Add flag to experiment with cold function attributes
To be removed and promoted to a proper driver flag if experiments turn out
fruitful.
Original LLVM patch for this functionality: #69030
---
clang/lib/CodeGen/BackendUtil.cpp | 57 +++
1 file changed, 35 insertions(+), 22 deletions(-)
diff --git a/clang/lib/CodeGen/BackendUtil.cpp
b/clang/lib/CodeGen/BackendUtil.cpp
index 6cc00b85664f41..22c3f8642ad8eb 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/clang/lib/CodeGen/BackendUtil.cpp
@@ -104,6 +104,21 @@ static cl::opt ClSanitizeOnOptimizerEarlyEP(
"sanitizer-early-opt-ep", cl::Optional,
cl::desc("Insert sanitizers on OptimizerEarlyEP."));
+// Experiment to mark cold functions as optsize/minsize/optnone.
+// TODO: remove once this is exposed as a proper driver flag.
+static cl::opt ClPGOColdFuncAttr(
+"pgo-cold-func-opt", cl::init(PGOOptions::ColdFuncOpt::Default),
cl::Hidden,
+cl::desc(
+"Function attribute to apply to cold functions as determined by PGO"),
+cl::values(clEnumValN(PGOOptions::ColdFuncOpt::Default, "default",
+ "Default (no attribute)"),
+ clEnumValN(PGOOptions::ColdFuncOpt::OptSize, "optsize",
+ "Mark cold functions with optsize."),
+ clEnumValN(PGOOptions::ColdFuncOpt::MinSize, "minsize",
+ "Mark cold functions with minsize."),
+ clEnumValN(PGOOptions::ColdFuncOpt::OptNone, "optnone",
+ "Mark cold functions with optnone.")));
+
extern cl::opt ProfileCorrelate;
// Re-link builtin bitcodes after optimization
@@ -768,42 +783,41 @@ void EmitAssemblyHelper::RunOptimizationPipeline(
CodeGenOpts.InstrProfileOutput.empty() ? getDefaultProfileGenName()
:
CodeGenOpts.InstrProfileOutput,
"", "", CodeGenOpts.MemoryProfileUsePath, nullptr, PGOOptions::IRInstr,
-PGOOptions::NoCSAction, PGOOptions::ColdFuncOpt::Default,
+PGOOptions::NoCSAction, ClPGOColdFuncAttr,
CodeGenOpts.DebugInfoForProfiling,
/*PseudoProbeForProfiling=*/false, CodeGenOpts.AtomicProfileUpdate);
else if (CodeGenOpts.hasProfileIRUse()) {
// -fprofile-use.
auto CSAction = CodeGenOpts.hasProfileCSIRUse() ? PGOOptions::CSIRUse
: PGOOptions::NoCSAction;
-PGOOpt = PGOOptions(
-CodeGenOpts.ProfileInstrumentUsePath, "",
-CodeGenOpts.ProfileRemappingFile, CodeGenOpts.MemoryProfileUsePath,
VFS,
-PGOOptions::IRUse, CSAction, PGOOptions::ColdFuncOpt::Default,
-CodeGenOpts.DebugInfoForProfiling);
+PGOOpt = PGOOptions(CodeGenOpts.ProfileInstrumentUsePath, "",
+CodeGenOpts.ProfileRemappingFile,
+CodeGenOpts.MemoryProfileUsePath, VFS,
+PGOOptions::IRUse, CSAction, ClPGOColdFuncAttr,
+CodeGenOpts.DebugInfoForProfiling);
} else if (!CodeGenOpts.SampleProfileFile.empty())
// -fprofile-sample-use
PGOOpt = PGOOptions(
CodeGenOpts.SampleProfileFile, "", CodeGenOpts.ProfileRemappingFile,
CodeGenOpts.MemoryProfileUsePath, VFS, PGOOptions::SampleUse,
-PGOOptions::NoCSAction, PGOOptions::ColdFuncOpt::Default,
+PGOOptions::NoCSAction, ClPGOColdFuncAttr,
CodeGenOpts.DebugInfoForProfiling,
CodeGenOpts.PseudoProbeForProfiling);
else if (!CodeGenOpts.MemoryProfileUsePath.empty())
// -fmemory-profile-use (without any of the above options)
PGOOpt = PGOOptions("", "", "", CodeGenOpts.MemoryProfileUsePath, VFS,
PGOOptions::NoAction, PGOOptions::NoCSAction,
-PGOOptions::ColdFuncOpt::Default,
-CodeGenOpts.DebugInfoForProfiling);
+ClPGOColdFuncAttr, CodeGenOpts.DebugInfoForProfiling);
else if (CodeGenOpts.PseudoProbeForProfiling)
// -fpseudo-probe-for-profiling
-PGOOpt = PGOOptions("", "", "", /*MemoryProfile=*/"", nullptr,
-PGOOptions::NoAction, PGOOptions::NoCSAction,
-PGOOptions::ColdFuncOpt::Default,
-CodeGenOpts.DebugInfoForProfiling, true);
+PGOOpt =
+PGOOptions("", "", "", /*MemoryProfile=*/"", nullptr,
+ PGOOptions::NoAction, PGOOptions::NoCSAction,
+ ClPGOColdFuncAttr, CodeGenOpts.DebugInfoForProfiling, true);
else if (CodeGenOpts.DebugInfoForProfiling)
// -fdebug-info-for-
[clang] [flang] [Driver] Ensure ToolChain::LibraryPaths is not empty for non-Darwin (PR #88661)
https://github.com/aeubanks approved this pull request. lg, but update the commit message `Pull Request: https://github.com/llvm/llvm-project/pull/87866`, that's obsolete https://github.com/llvm/llvm-project/pull/88661 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang][llvm] Remove "implicit-section-name" attribute (PR #87906)
https://github.com/aeubanks closed https://github.com/llvm/llvm-project/pull/87906 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [Driver] Ensure ToolChain::LibraryPaths is not empty for non-Darwin (PR #87866)
aeubanks wrote: yeah that patch makes those test pass with this PR, lgtm (you could also test locally by touching the files I mentioned above, e.g. even just `touch lib/clang/19/lib/linux/libclang_rt.builtins-aarch64-android.a` repro'd the test failure on my machine) https://github.com/llvm/llvm-project/pull/87866 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] be10070 - Revert "[Driver] Ensure ToolChain::LibraryPaths is not empty for non-Darwin"
Author: Arthur Eubanks
Date: 2024-04-10T23:41:51Z
New Revision: be10070f91b86a6f126d2451852242bfcb2cd366
URL:
https://github.com/llvm/llvm-project/commit/be10070f91b86a6f126d2451852242bfcb2cd366
DIFF:
https://github.com/llvm/llvm-project/commit/be10070f91b86a6f126d2451852242bfcb2cd366.diff
LOG: Revert "[Driver] Ensure ToolChain::LibraryPaths is not empty for
non-Darwin"
This reverts commit ccdebbae4d77d3efc236af92c22941de5d437e01.
Causes test failures in the presence of Android runtime libraries in
resource-dir.
See comments on https://github.com/llvm/llvm-project/pull/87866.
Added:
Modified:
clang/lib/Driver/ToolChain.cpp
clang/test/Driver/arm-compiler-rt.c
clang/test/Driver/cl-link.c
clang/test/Driver/compiler-rt-unwind.c
clang/test/Driver/coverage-ld.c
clang/test/Driver/instrprof-ld.c
clang/test/Driver/linux-ld.c
clang/test/Driver/mingw-sanitizers.c
clang/test/Driver/msp430-toolchain.c
clang/test/Driver/print-libgcc-file-name-clangrt.c
clang/test/Driver/print-runtime-dir.c
clang/test/Driver/riscv32-toolchain-extra.c
clang/test/Driver/riscv32-toolchain.c
clang/test/Driver/riscv64-toolchain-extra.c
clang/test/Driver/riscv64-toolchain.c
clang/test/Driver/sanitizer-ld.c
clang/test/Driver/wasm-toolchain.c
clang/test/Driver/wasm-toolchain.cpp
clang/test/Driver/windows-cross.c
clang/test/Driver/zos-ld.c
flang/test/Driver/msvc-dependent-lib-flags.f90
Removed:
diff --git a/clang/lib/Driver/ToolChain.cpp b/clang/lib/Driver/ToolChain.cpp
index 237092ed07e5dc..03450fc0f57b93 100644
--- a/clang/lib/Driver/ToolChain.cpp
+++ b/clang/lib/Driver/ToolChain.cpp
@@ -796,13 +796,7 @@ ToolChain::getTargetSubDirPath(StringRef BaseDir) const {
std::optional ToolChain::getRuntimePath() const {
SmallString<128> P(D.ResourceDir);
llvm::sys::path::append(P, "lib");
- if (auto Ret = getTargetSubDirPath(P))
-return Ret;
- // Darwin does not use per-target runtime directory.
- if (Triple.isOSDarwin())
-return {};
- llvm::sys::path::append(P, Triple.str());
- return std::string(P);
+ return getTargetSubDirPath(P);
}
std::optional ToolChain::getStdlibPath() const {
diff --git a/clang/test/Driver/arm-compiler-rt.c
b/clang/test/Driver/arm-compiler-rt.c
index cb6c29f48a7814..5e9e528400d08e 100644
--- a/clang/test/Driver/arm-compiler-rt.c
+++ b/clang/test/Driver/arm-compiler-rt.c
@@ -10,47 +10,47 @@
// RUN: -resource-dir=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -rtlib=compiler-rt -### %s 2>&1 \
// RUN: | FileCheck %s -check-prefix ARM-GNUEABI
-// ARM-GNUEABI: "{{.*[/\\]}}libclang_rt.builtins.a"
+// ARM-GNUEABI: "{{.*[/\\]}}libclang_rt.builtins-arm.a"
// RUN: %clang -target arm-linux-gnueabi \
// RUN: --sysroot=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -resource-dir=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -rtlib=compiler-rt -mfloat-abi=hard -### %s 2>&1 \
// RUN: | FileCheck %s -check-prefix ARM-GNUEABI-ABI
-// ARM-GNUEABI-ABI: "{{.*[/\\]}}libclang_rt.builtins.a"
+// ARM-GNUEABI-ABI: "{{.*[/\\]}}libclang_rt.builtins-armhf.a"
// RUN: %clang -target arm-linux-gnueabihf \
// RUN: --sysroot=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -resource-dir=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -rtlib=compiler-rt -### %s 2>&1 \
// RUN: | FileCheck %s -check-prefix ARM-GNUEABIHF
-// ARM-GNUEABIHF: "{{.*[/\\]}}libclang_rt.builtins.a"
+// ARM-GNUEABIHF: "{{.*[/\\]}}libclang_rt.builtins-armhf.a"
// RUN: %clang -target arm-linux-gnueabihf \
// RUN: --sysroot=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -resource-dir=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -rtlib=compiler-rt -mfloat-abi=soft -### %s 2>&1 \
// RUN: | FileCheck %s -check-prefix ARM-GNUEABIHF-ABI
-// ARM-GNUEABIHF-ABI: "{{.*[/\\]}}libclang_rt.builtins.a"
+// ARM-GNUEABIHF-ABI: "{{.*[/\\]}}libclang_rt.builtins-arm.a"
// RUN: %clang -target arm-windows-itanium \
// RUN: --sysroot=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -resource-dir=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -rtlib=compiler-rt -### %s 2>&1 \
// RUN: | FileCheck %s -check-prefix ARM-WINDOWS
-// ARM-WINDOWS: "{{.*[/\\]}}clang_rt.builtins.lib"
+// ARM-WINDOWS: "{{.*[/\\]}}clang_rt.builtins-arm.lib"
// RUN: %clang -target arm-linux-androideabi \
// RUN: --sysroot=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -resource-dir=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -rtlib=compiler-rt -### %s 2>&1 \
// RUN: | FileCheck %s -check-prefix ARM-ANDROID
-// ARM-ANDROID: "{{.*[/\\]}}libclang_rt.builtins.a"
+// ARM-ANDROID: "{{.*[/\\]}}libclang_rt.builtins-arm-android.a"
// RUN: not %clang --target=arm-linux-androideabi \
// RUN: --sysroot=%S/Inputs/resource_dir_with_arch_subdir \
// RUN: -resource-dir=%S/Inputs/r
[clang] [Driver] Ensure ToolChain::LibraryPaths is not empty for non-Darwin (PR #87866)
aeubanks wrote: it seems that the test depends on if certain android runtime libraries are present or not in the resource dir (without per-target runtime directory since that's still an issue on Android). perhaps the tests need `-resource-dir` to make them more hermetic? anyway, will revert, feel free to reland once the tests are fixed. this is what's in my resource dir if it's helpful for reproducing ``` $ ls lib/clang/19/lib/linux/ aarch64 libclang_rt.asan_cxx-aarch64-android.a libclang_rt.builtins-aarch64-android.a libclang_rt.tsan_cxx-aarch64-android.a libclang_rt.ubsan_standalone-aarch64-android.so libclang_rt.asan-aarch64-android.a libclang_rt.asan-preinit-aarch64-android.a libclang_rt.profile-aarch64-android.a libclang_rt.ubsan_minimal-aarch64-android.a libclang_rt.ubsan_standalone_cxx-aarch64-android.a libclang_rt.asan-aarch64-android.so libclang_rt.asan_static-aarch64-android.a libclang_rt.tsan-aarch64-android.a libclang_rt.ubsan_standalone-aarch64-android.a ``` https://github.com/llvm/llvm-project/pull/87866 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [Driver] Ensure ToolChain::LibraryPaths is not empty for non-Darwin (PR #87866)
aeubanks wrote: this seemes to be causing some test failures for us: https://logs.chromium.org/logs/chromium/buildbucket/cr-buildbucket/8751043232043110529/+/u/package_clang/stdout?format=raw ``` TEST 'Clang :: Driver/linux-ld.c' FAILED RUN: at line 92: ... /b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/linux-ld.c:102:25: error: CHECK-LD-RT-ANDROID: expected string not found in input // CHECK-LD-RT-ANDROID: libclang_rt.builtins.a" ^ :6:331: note: scanning from here "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/bin/ld.lld" "--sysroot=/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot" "-EL" "-z" "now" "-z" "relro" "-z" "max-page-size=4096" "-X" "--hash-style=both" "--eh-frame-hdr" "-m" "armelf_linux_eabi" "-dynamic-linker" "/system/bin/linker" "-o" "a.out" "/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib/crtbegin_dynamic.o" "-L/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib" "-L/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib" "/b/s/w/ir/x/t/lit-tmp-jnv32xv9/linux-ld-60ec02.o" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.builtins-arm-android.a" "-l:libunwind.a" "-ldl" "-lc" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.builtins-arm-android.a" "-l:libunwind.a" "-ldl" "/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib/crtend_android.o" ^ :6:866: note: possible intended match here "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/bin/ld.lld" "--sysroot=/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot" "-EL" "-z" "now" "-z" "relro" "-z" "max-page-size=4096" "-X" "--hash-style=both" "--eh-frame-hdr" "-m" "armelf_linux_eabi" "-dynamic-linker" "/system/bin/linker" "-o" "a.out" "/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib/crtbegin_dynamic.o" "-L/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib" "-L/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib" "/b/s/w/ir/x/t/lit-tmp-jnv32xv9/linux-ld-60ec02.o" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.builtins-arm-android.a" "-l:libunwind.a" "-ldl" "-lc" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.builtins-arm-android.a" "-l:libunwind.a" "-ldl" "/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib/crtend_android.o" TEST 'Clang :: Driver/sanitizer-ld.c' FAILED RUN: at line 177: ... /b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/sanitizer-ld.c:187:24: error: CHECK-ASAN-ANDROID: expected string not found in input // CHECK-ASAN-ANDROID: libclang_rt.asan.so" ^ :6:320: note: scanning from here "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/bin/ld.lld" "--sysroot=/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot" "-EL" "-z" "now" "-z" "relro" "-z" "max-page-size=4096" "-X" "--hash-style=both" "--eh-frame-hdr" "-m" "armelf_linux_eabi" "-pie" "-dynamic-linker" "/system/bin/linker" "-o" "a.out" "/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib/crtbegin_dynamic.o" "-L/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib/../lib" "-L/b/s/w/ir/cache/builder/src/third_party/llvm/clang/test/Driver/Inputs/basic_android_tree/sysroot/usr/lib" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.asan-arm-android.so" "--whole-archive" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.asan_static-arm-android.a" "--no-whole-archive" "/b/s/w/ir/x/t/lit-tmp-jnv32xv9/sanitizer-ld-d1ddb3.o" "/b/s/w/ir/cache/builder/src/third_party/llvm-build/Release+Asserts/lib/clang/19/lib/linux/libclang_rt.builtins-arm-android.a"
[clang] Reland "[Win32][ELF] Make CodeView a DebugInfoFormat only for COFF format" (PR #87987)
https://github.com/aeubanks approved this pull request. https://github.com/llvm/llvm-project/pull/87987 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang][llvm] Remove "implicit-section-name" attribute (PR #87906)
https://github.com/aeubanks edited https://github.com/llvm/llvm-project/pull/87906 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang][llvm] Remove "implicit-section-name" attribute (PR #87906)
aeubanks wrote: > I'd suggest adding bitcode upgrade if it isn't too hard (I don't think it > should be?) done https://github.com/llvm/llvm-project/pull/87906 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang][llvm] Remove "implicit-section-name" attribute (PR #87906)
https://github.com/aeubanks updated
https://github.com/llvm/llvm-project/pull/87906
>From 7a9df42b4c4f4f1b02dc3158d24800f3d4b68d8f Mon Sep 17 00:00:00 2001
From: Arthur Eubanks
Date: Sun, 7 Apr 2024 05:29:36 +
Subject: [PATCH 1/2] [clang][llvm] Remove "implicit-section-name" attribute
D33412/D33413 introduced this to support a clang pragma to set section
names for a symbol depending on if it would be placed in
bss/data/rodata/text, which may not be known until the backend. However,
for text we know that only functions will go there, so just
directly set the section in clang instead of going through a
completely separate attribute.
---
clang/lib/CodeGen/CodeGenModule.cpp | 2 +-
clang/test/CodeGen/clang-sections-attribute.c | 3 ---
clang/test/CodeGenCXX/clang-sections.cpp | 18 ++---
llvm/lib/CodeGen/TargetInstrInfo.cpp | 3 +--
.../CodeGen/TargetLoweringObjectFileImpl.cpp | 11 +---
llvm/lib/Target/TargetLoweringObjectFile.cpp | 5
.../CodeGen/AArch64/clang-section-macho.ll| 22 ---
llvm/test/CodeGen/ARM/clang-section.ll| 24 -
.../Generic/machine-function-splitter.ll | 27 +++
.../basic-block-sections-pragma-sections.ll | 4 +--
10 files changed, 15 insertions(+), 104 deletions(-)
delete mode 100644 llvm/test/CodeGen/AArch64/clang-section-macho.ll
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index 00b3bfcaa0bc25..f4dbfe7a21f83c 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -2627,7 +2627,7 @@ void CodeGenModule::setNonAliasAttributes(GlobalDecl GD,
addUsedGlobal(F);
if (auto *SA = D->getAttr())
if (!D->getAttr())
- F->addFnAttr("implicit-section-name", SA->getName());
+ F->setSection(SA->getName());
llvm::AttrBuilder Attrs(F->getContext());
if (GetCPUAndFeaturesAttributes(GD, Attrs)) {
diff --git a/clang/test/CodeGen/clang-sections-attribute.c
b/clang/test/CodeGen/clang-sections-attribute.c
index 70ed24ed07a280..768bdd4d87649e 100644
--- a/clang/test/CodeGen/clang-sections-attribute.c
+++ b/clang/test/CodeGen/clang-sections-attribute.c
@@ -69,8 +69,5 @@ static int int_zvar;
// CHECK: define internal void @int_fun() #0 section ".int_fun_attr"
// CHECK: define internal void @int_fun2() #0 section ".int_fun2_attr"
//
-// Function attributes should not include implicit-section-name.
-// CHECK-NOT: attributes #0 = {{.*}}implicit-section-name
-//
// No other attribute group should be present in the file.
// CHECK-NOT: attributes #1
diff --git a/clang/test/CodeGenCXX/clang-sections.cpp
b/clang/test/CodeGenCXX/clang-sections.cpp
index a444f2d0cae59c..aa159e552b1b3c 100644
--- a/clang/test/CodeGenCXX/clang-sections.cpp
+++ b/clang/test/CodeGenCXX/clang-sections.cpp
@@ -81,24 +81,22 @@ int hoo(void) {
//CHECK: @p ={{.*}} constant i32 7, align 4
//CHECK: @_ZL5fptrs = internal constant [2 x ptr] [ptr @foo, ptr @goo], align
{{4|8}} #3
-//CHECK: define{{.*}} i32 @foo() #5 {
-//CHECK: define{{.*}} i32 @goo() #6 {
-//CHECK: declare i32 @zoo(ptr noundef, ptr noundef) #7
-//CHECK: define{{.*}} i32 @hoo() #8 {
+//ELF: define{{.*}} i32 @foo(){{.*}} section "my_text.1" {
+//ELF: define{{.*}} i32 @goo(){{.*}} section "my_text.2" {
+//MACHO: define{{.*}} i32 @foo(){{.*}} section "__TEXT,__mytext1" {
+//MACHO: define{{.*}} i32 @goo(){{.*}} section "__TEXT,__mytext2" {
+
+// ensure zoo/hoo don't have a section
+//CHECK: declare i32 @zoo(ptr noundef, ptr noundef) #6{{$}}
+//CHECK: define{{.*}} i32 @hoo() #5 {
//ELF: attributes #0 = { "bss-section"="my_bss.1" "data-section"="my_data.1"
"rodata-section"="my_rodata.1" }
//ELF: attributes #1 = { "data-section"="my_data.1"
"rodata-section"="my_rodata.1" }
//ELF: attributes #2 = { "bss-section"="my_bss.2"
"rodata-section"="my_rodata.1" }
//ELF: attributes #3 = { "bss-section"="my_bss.2" "data-section"="my_data.2"
"relro-section"="my_relro.2" "rodata-section"="my_rodata.2" }
//ELF: attributes #4 = { "relro-section"="my_relro.2" }
-//ELF: attributes #5 = { {{.*"implicit-section-name"="my_text.1".*}} }
-//ELF: attributes #6 = { {{.*"implicit-section-name"="my_text.2".*}} }
//MACHO: attributes #0 = { "bss-section"="__BSS,__mybss1"
"data-section"="__DATA,__mydata1" "rodata-section"="__RODATA,__myrodata1" }
//MACHO: attributes #1 = { "data-section"="__DATA,__mydata1"
"rodata-section"="__RODATA,__myrodata1" }
//MACHO: attributes #2 = { "bss-section"="__BSS,__mybss2"
"rodata-section"="__RODATA,__myrodata1" }
//MACHO: attributes #3 = { "bss-section"="__BSS,__mybss2"
"data-section"="__DATA,__mydata2" "relro-section"="__RELRO,__myrelro2"
"rodata-section"="__RODATA,__myrodata2" }
//MACHO: attributes #4 = { "relro-section"="__RELRO,__myrelro2" }
-//MACHO: attributes #5 = { {{.*"implicit-section-name"="__TEXT,__mytext1".*}} }
-//MACHO: attributes #6 = { {{.*"implicit-section-name"="__TEXT,__mytext2".
[clang] [llvm] [clang][llvm] Remove "implicit-section-name" attribute (PR #87906)
aeubanks wrote: this probably needs bitcode upgrade? https://github.com/llvm/llvm-project/pull/87906 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] [llvm] [clang][llvm] Remove "implicit-section-name" attribute (PR #87906)
https://github.com/aeubanks created
https://github.com/llvm/llvm-project/pull/87906
D33412/D33413 introduced this to support a clang pragma to set section names
for a symbol depending on if it would be placed in bss/data/rodata/text, which
may not be known until the backend. However, for text we know that only
functions will go there, so just directly set the section in clang instead of
going through a completely separate attribute.
>From 7a9df42b4c4f4f1b02dc3158d24800f3d4b68d8f Mon Sep 17 00:00:00 2001
From: Arthur Eubanks
Date: Sun, 7 Apr 2024 05:29:36 +
Subject: [PATCH] [clang][llvm] Remove "implicit-section-name" attribute
D33412/D33413 introduced this to support a clang pragma to set section
names for a symbol depending on if it would be placed in
bss/data/rodata/text, which may not be known until the backend. However,
for text we know that only functions will go there, so just
directly set the section in clang instead of going through a
completely separate attribute.
---
clang/lib/CodeGen/CodeGenModule.cpp | 2 +-
clang/test/CodeGen/clang-sections-attribute.c | 3 ---
clang/test/CodeGenCXX/clang-sections.cpp | 18 ++---
llvm/lib/CodeGen/TargetInstrInfo.cpp | 3 +--
.../CodeGen/TargetLoweringObjectFileImpl.cpp | 11 +---
llvm/lib/Target/TargetLoweringObjectFile.cpp | 5
.../CodeGen/AArch64/clang-section-macho.ll| 22 ---
llvm/test/CodeGen/ARM/clang-section.ll| 24 -
.../Generic/machine-function-splitter.ll | 27 +++
.../basic-block-sections-pragma-sections.ll | 4 +--
10 files changed, 15 insertions(+), 104 deletions(-)
delete mode 100644 llvm/test/CodeGen/AArch64/clang-section-macho.ll
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index 00b3bfcaa0bc25..f4dbfe7a21f83c 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -2627,7 +2627,7 @@ void CodeGenModule::setNonAliasAttributes(GlobalDecl GD,
addUsedGlobal(F);
if (auto *SA = D->getAttr())
if (!D->getAttr())
- F->addFnAttr("implicit-section-name", SA->getName());
+ F->setSection(SA->getName());
llvm::AttrBuilder Attrs(F->getContext());
if (GetCPUAndFeaturesAttributes(GD, Attrs)) {
diff --git a/clang/test/CodeGen/clang-sections-attribute.c
b/clang/test/CodeGen/clang-sections-attribute.c
index 70ed24ed07a280..768bdd4d87649e 100644
--- a/clang/test/CodeGen/clang-sections-attribute.c
+++ b/clang/test/CodeGen/clang-sections-attribute.c
@@ -69,8 +69,5 @@ static int int_zvar;
// CHECK: define internal void @int_fun() #0 section ".int_fun_attr"
// CHECK: define internal void @int_fun2() #0 section ".int_fun2_attr"
//
-// Function attributes should not include implicit-section-name.
-// CHECK-NOT: attributes #0 = {{.*}}implicit-section-name
-//
// No other attribute group should be present in the file.
// CHECK-NOT: attributes #1
diff --git a/clang/test/CodeGenCXX/clang-sections.cpp
b/clang/test/CodeGenCXX/clang-sections.cpp
index a444f2d0cae59c..aa159e552b1b3c 100644
--- a/clang/test/CodeGenCXX/clang-sections.cpp
+++ b/clang/test/CodeGenCXX/clang-sections.cpp
@@ -81,24 +81,22 @@ int hoo(void) {
//CHECK: @p ={{.*}} constant i32 7, align 4
//CHECK: @_ZL5fptrs = internal constant [2 x ptr] [ptr @foo, ptr @goo], align
{{4|8}} #3
-//CHECK: define{{.*}} i32 @foo() #5 {
-//CHECK: define{{.*}} i32 @goo() #6 {
-//CHECK: declare i32 @zoo(ptr noundef, ptr noundef) #7
-//CHECK: define{{.*}} i32 @hoo() #8 {
+//ELF: define{{.*}} i32 @foo(){{.*}} section "my_text.1" {
+//ELF: define{{.*}} i32 @goo(){{.*}} section "my_text.2" {
+//MACHO: define{{.*}} i32 @foo(){{.*}} section "__TEXT,__mytext1" {
+//MACHO: define{{.*}} i32 @goo(){{.*}} section "__TEXT,__mytext2" {
+
+// ensure zoo/hoo don't have a section
+//CHECK: declare i32 @zoo(ptr noundef, ptr noundef) #6{{$}}
+//CHECK: define{{.*}} i32 @hoo() #5 {
//ELF: attributes #0 = { "bss-section"="my_bss.1" "data-section"="my_data.1"
"rodata-section"="my_rodata.1" }
//ELF: attributes #1 = { "data-section"="my_data.1"
"rodata-section"="my_rodata.1" }
//ELF: attributes #2 = { "bss-section"="my_bss.2"
"rodata-section"="my_rodata.1" }
//ELF: attributes #3 = { "bss-section"="my_bss.2" "data-section"="my_data.2"
"relro-section"="my_relro.2" "rodata-section"="my_rodata.2" }
//ELF: attributes #4 = { "relro-section"="my_relro.2" }
-//ELF: attributes #5 = { {{.*"implicit-section-name"="my_text.1".*}} }
-//ELF: attributes #6 = { {{.*"implicit-section-name"="my_text.2".*}} }
//MACHO: attributes #0 = { "bss-section"="__BSS,__mybss1"
"data-section"="__DATA,__mydata1" "rodata-section"="__RODATA,__myrodata1" }
//MACHO: attributes #1 = { "data-section"="__DATA,__mydata1"
"rodata-section"="__RODATA,__myrodata1" }
//MACHO: attributes #2 = { "bss-section"="__BSS,__mybss2"
"rodata-section"="__RODATA,__myrodata1" }
//MACHO: attributes #3 = { "bss-section"="
