Merge remote-tracking branch 'origin/main' into vector-combine/scalarizeBinOpOfSplats-insertelement-cost

Author: lukel97

.ci/metrics/metrics.py

@@ -21,16 +21,16 @@
 # Lists the Github workflows we want to track. Maps the Github job name to
 # the metric name prefix in grafana.
 # This metric name is also used as a key in the job->name map.
-GITHUB_WORKFLOW_TO_TRACK = {"LLVM Premerge Checks": "github_llvm_premerge_checks"}
+GITHUB_WORKFLOW_TO_TRACK = {"CI Checks": "github_llvm_premerge_checks"}
 
 # Lists the Github jobs to track for a given workflow. The key is the stable
 # name (metric name) of the workflow (see GITHUB_WORKFLOW_TO_TRACK).
 # Each value is a map to link the github job name to the corresponding metric
 # name.
 GITHUB_JOB_TO_TRACK = {
     "github_llvm_premerge_checks": {
-        "Linux Premerge Checks (Test Only - Please Ignore Results)": "premerge_linux",
-        "Windows Premerge Checks (Test Only - Please Ignore Results)": "premerge_windows",
+        "Build and Test Linux (Test Only - Please Ignore Results)": "premerge_linux",
+        "Build and Test Windows (Test Only - Please Ignore Results)": "premerge_windows",
     }
 }
 

.github/workflows/libc-fullbuild-tests.yml

@@ -19,13 +19,23 @@ jobs:
         include:
           - os: ubuntu-24.04
             ccache-variant: sccache
-            c_compiler: clang-20
-            cpp_compiler: clang++-20
+            c_compiler: clang-21
+            cpp_compiler: clang++-21
+            target: x86_64-unknown-linux-llvm
+            include_scudo: ON
           # TODO: remove ccache logic when https://github.com/hendrikmuhs/ccache-action/issues/279 is resolved.
           - os: ubuntu-24.04-arm
             ccache-variant: ccache
-            c_compiler: clang-20
-            cpp_compiler: clang++-20
+            c_compiler: clang-21
+            cpp_compiler: clang++-21
+            target: aarch64-unknown-linux-llvm
+            include_scudo: ON
+          - os: ubuntu-24.04
+            ccache-variant: ccache
+            c_compiler: clang-21
+            cpp_compiler: clang++-21
+            target: x86_64-unknown-uefi-llvm
+            include_scudo: OFF
           # TODO: add back gcc build when it is fixed
           # - c_compiler: gcc
           #   cpp_compiler: g++
@@ -53,7 +63,7 @@ jobs:
       run: |
         wget https://apt.llvm.org/llvm.sh
         chmod +x llvm.sh
-        sudo ./llvm.sh 20
+        sudo ./llvm.sh 21
         sudo apt-get update
         sudo apt-get install -y libmpfr-dev libgmp-dev libmpc-dev ninja-build linux-libc-dev
         sudo ln -sf /usr/include/$(uname -p)-linux-gnu/asm /usr/include/asm
@@ -68,22 +78,31 @@ jobs:
     # Configure libc fullbuild with scudo.
     # Use MinSizeRel to reduce the size of the build.
     - name: Configure CMake
-      run: >
-        cmake -B ${{ steps.strings.outputs.build-output-dir }}
-        -DCMAKE_CXX_COMPILER=${{ matrix.cpp_compiler }}
-        -DCMAKE_C_COMPILER=${{ matrix.c_compiler }}
-        -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
-        -DCMAKE_C_COMPILER_LAUNCHER=${{ matrix.ccache-variant }}
-        -DCMAKE_CXX_COMPILER_LAUNCHER=${{ matrix.ccache-variant }}
-        -DCMAKE_INSTALL_PREFIX=${{ steps.strings.outputs.build-install-dir }}
-        -DLLVM_ENABLE_RUNTIMES="libc;compiler-rt"
-        -DLLVM_LIBC_FULL_BUILD=ON
-        -DLLVM_LIBC_INCLUDE_SCUDO=ON
-        -DCOMPILER_RT_BUILD_SCUDO_STANDALONE_WITH_LLVM_LIBC=ON
-        -DCOMPILER_RT_BUILD_GWP_ASAN=OFF
-        -DCOMPILER_RT_SCUDO_STANDALONE_BUILD_SHARED=OFF
-        -G Ninja
-        -S ${{ github.workspace }}/runtimes
+      run: |
+        export RUNTIMES="libc"
+
+        if [[ ${{ matrix.include_scudo}} == "ON" ]]; then
+          export RUNTIMES="$RUNTIMES;compiler-rt"
+          export CMAKE_FLAGS="
+            -DLLVM_LIBC_INCLUDE_SCUDO=ON
+            -DCOMPILER_RT_BUILD_SCUDO_STANDALONE_WITH_LLVM_LIBC=ON
+            -DCOMPILER_RT_BUILD_GWP_ASAN=OFF
+            -DCOMPILER_RT_SCUDO_STANDALONE_BUILD_SHARED=OFF"
+        fi
+
+        cmake -B ${{ steps.strings.outputs.build-output-dir }} \
+        -DCMAKE_CXX_COMPILER=${{ matrix.cpp_compiler }} \
+        -DCMAKE_C_COMPILER=${{ matrix.c_compiler }} \
+        -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} \
+        -DCMAKE_C_COMPILER_LAUNCHER=${{ matrix.ccache-variant }} \
+        -DCMAKE_CXX_COMPILER_LAUNCHER=${{ matrix.ccache-variant }} \
+        -DCMAKE_INSTALL_PREFIX=${{ steps.strings.outputs.build-install-dir }} \
+        -DLLVM_RUNTIMES_TARGET=${{ matrix.target }} \
+        -DLLVM_ENABLE_RUNTIMES="$RUNTIMES" \
+        -DLLVM_LIBC_FULL_BUILD=ON \
+        -G Ninja \
+        -S ${{ github.workspace }}/runtimes \
+        $CMAKE_FLAGS
 
     - name: Build
       run: >
@@ -93,6 +112,8 @@ jobs:
         --target install
 
     - name: Test
+      # Skip UEFI tests until we have testing set up.
+      if: ${{ ! endsWith(matrix.target, '-uefi-llvm') }}
       run: >
         cmake 
         --build ${{ steps.strings.outputs.build-output-dir }} 

.github/workflows/libcxx-build-and-test.yaml

@@ -37,7 +37,7 @@ jobs:
   stage1:
     if: github.repository_owner == 'llvm'
     runs-on: libcxx-self-hosted-linux
-    container: ghcr.io/llvm/libcxx-linux-builder:b6bb9dc5abd7c6452c13a53fa8949cb259db459b
+    container: ghcr.io/llvm/libcxx-linux-builder:2b57ebb50b6d418e70382e655feaa619b558e254
     continue-on-error: false
     strategy:
       fail-fast: false
@@ -75,7 +75,7 @@ jobs:
   stage2:
     if: github.repository_owner == 'llvm'
     runs-on: libcxx-self-hosted-linux
-    container: ghcr.io/llvm/libcxx-linux-builder:b6bb9dc5abd7c6452c13a53fa8949cb259db459b
+    container: ghcr.io/llvm/libcxx-linux-builder:2b57ebb50b6d418e70382e655feaa619b558e254
     needs: [ stage1 ]
     continue-on-error: false
     strategy:
@@ -163,7 +163,7 @@ jobs:
         - config: 'generic-msan'
           machine: libcxx-self-hosted-linux
     runs-on: ${{ matrix.machine }}
-    container: ghcr.io/llvm/libcxx-linux-builder:b6bb9dc5abd7c6452c13a53fa8949cb259db459b
+    container: ghcr.io/llvm/libcxx-linux-builder:2b57ebb50b6d418e70382e655feaa619b558e254
     steps:
       - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
       - name: ${{ matrix.config }}

.github/workflows/premerge.yaml

@@ -1,4 +1,4 @@
-name: LLVM Premerge Checks
+name: CI Checks
 
 permissions:
   contents: read
@@ -25,7 +25,7 @@ concurrency:
 
 jobs:
   premerge-checks-linux:
-    name: Linux Premerge Checks (Test Only - Please Ignore Results)
+    name: Build and Test Linux (Test Only - Please Ignore Results)
     if: >-
         github.repository_owner == 'llvm' &&
         (github.event_name != 'pull_request' || github.event.action != 'closed')
@@ -74,7 +74,7 @@ jobs:
           include-hidden-files: 'true'
 
   premerge-checks-windows:
-    name: Windows Premerge Checks (Test Only - Please Ignore Results)
+    name: Build and Test Windows (Test Only - Please Ignore Results)
     if: >-
         github.repository_owner == 'llvm' &&
         (github.event_name != 'pull_request' || github.event.action != 'closed')

bolt/docs/BinaryAnalysis.md

@@ -180,12 +180,6 @@ The following are current known cases of false negatives:
    [prototype branch](
    https://github.com/llvm/llvm-project/compare/main...kbeyls:llvm-project:bolt-gadget-scanner-prototype).
 
-BOLT cannot currently handle functions with `cfi_negate_ra_state` correctly,
-i.e. any binaries built with `-mbranch-protection=pac-ret`. The scanner is meant
-to be used on specifically such binaries, so this is a major limitation! Work is
-going on in PR [#120064](https://github.com/llvm/llvm-project/pull/120064) to
-fix this.
-
 ## How to add your own binary analysis
 
 _TODO: this section needs to be written. Ideally, we should have a simple

bolt/lib/Core/BinaryFunction.cpp

@@ -65,6 +65,8 @@ extern cl::opt<bool> StrictMode;
 extern cl::opt<bool> UpdateDebugSections;
 extern cl::opt<unsigned> Verbosity;
 
+extern bool BinaryAnalysisMode;
+extern bool HeatmapMode;
 extern bool processAllFunctions();
 
 static cl::opt<bool> CheckEncoding(
@@ -2760,13 +2762,19 @@ struct CFISnapshot {
     }
     case MCCFIInstruction::OpAdjustCfaOffset:
     case MCCFIInstruction::OpWindowSave:
-    case MCCFIInstruction::OpNegateRAState:
     case MCCFIInstruction::OpNegateRAStateWithPC:
     case MCCFIInstruction::OpLLVMDefAspaceCfa:
     case MCCFIInstruction::OpLabel:
     case MCCFIInstruction::OpValOffset:
       llvm_unreachable("unsupported CFI opcode");
       break;
+    case MCCFIInstruction::OpNegateRAState:
+      if (!(opts::BinaryAnalysisMode || opts::HeatmapMode)) {
+        llvm_unreachable("BOLT-ERROR: binaries using pac-ret hardening (e.g. "
+                         "as produced by '-mbranch-protection=pac-ret') are "
+                         "currently not supported by BOLT.");
+      }
+      break;
     case MCCFIInstruction::OpRememberState:
     case MCCFIInstruction::OpRestoreState:
     case MCCFIInstruction::OpGnuArgsSize:
@@ -2900,13 +2908,19 @@ struct CFISnapshotDiff : public CFISnapshot {
       return CFAReg == Instr.getRegister() && CFAOffset == Instr.getOffset();
     case MCCFIInstruction::OpAdjustCfaOffset:
     case MCCFIInstruction::OpWindowSave:
-    case MCCFIInstruction::OpNegateRAState:
     case MCCFIInstruction::OpNegateRAStateWithPC:
     case MCCFIInstruction::OpLLVMDefAspaceCfa:
     case MCCFIInstruction::OpLabel:
     case MCCFIInstruction::OpValOffset:
       llvm_unreachable("unsupported CFI opcode");
       return false;
+    case MCCFIInstruction::OpNegateRAState:
+      if (!(opts::BinaryAnalysisMode || opts::HeatmapMode)) {
+        llvm_unreachable("BOLT-ERROR: binaries using pac-ret hardening (e.g. "
+                         "as produced by '-mbranch-protection=pac-ret') are "
+                         "currently not supported by BOLT.");
+      }
+      break;
     case MCCFIInstruction::OpRememberState:
     case MCCFIInstruction::OpRestoreState:
     case MCCFIInstruction::OpGnuArgsSize:
@@ -3051,13 +3065,19 @@ BinaryFunction::unwindCFIState(int32_t FromState, int32_t ToState,
       break;
     case MCCFIInstruction::OpAdjustCfaOffset:
     case MCCFIInstruction::OpWindowSave:
-    case MCCFIInstruction::OpNegateRAState:
     case MCCFIInstruction::OpNegateRAStateWithPC:
     case MCCFIInstruction::OpLLVMDefAspaceCfa:
     case MCCFIInstruction::OpLabel:
     case MCCFIInstruction::OpValOffset:
       llvm_unreachable("unsupported CFI opcode");
       break;
+    case MCCFIInstruction::OpNegateRAState:
+      if (!(opts::BinaryAnalysisMode || opts::HeatmapMode)) {
+        llvm_unreachable("BOLT-ERROR: binaries using pac-ret hardening (e.g. "
+                         "as produced by '-mbranch-protection=pac-ret') are "
+                         "currently not supported by BOLT.");
+      }
+      break;
     case MCCFIInstruction::OpGnuArgsSize:
       // do not affect CFI state
       break;

bolt/lib/Passes/PatchEntries.cpp

@@ -98,21 +98,10 @@ Error PatchEntries::runOnFunctions(BinaryContext &BC) {
     });
 
     if (!Success) {
-      // We can't change output layout for AArch64 due to LongJmp pass
-      if (BC.isAArch64()) {
-        if (opts::ForcePatch) {
-          BC.errs() << "BOLT-ERROR: unable to patch entries in " << Function
-                    << "\n";
-          return createFatalBOLTError("");
-        }
-
-        continue;
-      }
-
       // If the original function entries cannot be patched, then we cannot
       // safely emit new function body.
       BC.errs() << "BOLT-WARNING: failed to patch entries in " << Function
-                << ". The function will not be optimized.\n";
+                << ". The function will not be optimized\n";
       Function.setIgnored();
       continue;
     }

bolt/lib/Rewrite/BinaryPassManager.cpp

@@ -497,6 +497,10 @@ Error BinaryFunctionPassManager::runAllPasses(BinaryContext &BC) {
   // memory profiling data.
   Manager.registerPass(std::make_unique<ReorderData>());
 
+  // Patch original function entries
+  if (BC.HasRelocations)
+    Manager.registerPass(std::make_unique<PatchEntries>());
+
   if (BC.isAArch64()) {
     Manager.registerPass(
         std::make_unique<ADRRelaxationPass>(PrintAdrRelaxation));
@@ -524,10 +528,6 @@ Error BinaryFunctionPassManager::runAllPasses(BinaryContext &BC) {
   // Assign each function an output section.
   Manager.registerPass(std::make_unique<AssignSections>());
 
-  // Patch original function entries
-  if (BC.HasRelocations)
-    Manager.registerPass(std::make_unique<PatchEntries>());
-
   // This pass turns tail calls into jumps which makes them invisible to
   // function reordering. It's unsafe to use any CFG or instruction analysis
   // after this point.

bolt/lib/Rewrite/RewriteInstance.cpp

@@ -2697,20 +2697,19 @@ void RewriteInstance::handleRelocation(const SectionRef &RelocatedSection,
                BD->nameStartsWith("_ZTCN"))) { // construction vtable
       BinaryFunction *BF = BC->getBinaryFunctionContainingAddress(
           SymbolAddress, /*CheckPastEnd*/ false, /*UseMaxSize*/ true);
-      if (!BF || BF->getAddress() != SymbolAddress) {
-        BC->errs()
-            << "BOLT-ERROR: the virtual function table entry at offset 0x"
-            << Twine::utohexstr(Rel.getOffset());
-        if (BF)
-          BC->errs() << " points to the middle of a function @ 0x"
-                     << Twine::utohexstr(BF->getAddress()) << "\n";
-        else
-          BC->errs() << " does not point to any function\n";
-        exit(1);
+      if (BF) {
+        if (BF->getAddress() != SymbolAddress) {
+          BC->errs()
+              << "BOLT-ERROR: the virtual function table entry at offset 0x"
+              << Twine::utohexstr(Rel.getOffset())
+              << " points to the middle of a function @ 0x"
+              << Twine::utohexstr(BF->getAddress()) << "\n";
+          exit(1);
+        }
+        BC->addRelocation(Rel.getOffset(), BF->getSymbol(), RType, Addend,
+                          ExtractedValue);
+        return;
       }
-      BC->addRelocation(Rel.getOffset(), BF->getSymbol(), RType, Addend,
-                        ExtractedValue);
-      return;
     }
   }
 

bolt/test/AArch64/skip-non-vfuncptr-reloc-in-relative-vtable.s

@@ -0,0 +1,35 @@
+// Test the fix that BOLT should skip special handling of any non-virtual
+// function pointer relocations in relative vtable.
+
+// RUN: llvm-mc -filetype=obj -triple aarch64-unknown-gnu %s -o %t.o
+// RUN: %clang %cxxflags -fuse-ld=lld %t.o -o %t.so -Wl,-q
+// RUN: llvm-bolt %t.so -o %t.bolted.so
+
+	.text
+	.p2align	2
+	.type	foo,@function
+foo:
+	.cfi_startproc
+	adrp	x8, _ZTV3gooE
+	add	x8, x8, :lo12:_ZTV3gooE
+	ldr	x0, [x8]
+	ret
+.Lfunc_end0:
+	.size	foo, .Lfunc_end0-foo
+	.cfi_endproc
+
+	.type	_fake_rtti_data,@object
+	.section	.rodata.cst16._fake_rtti_data,"aMG",@progbits,16,_fake_rtti_data,comdat
+	.p2align	3, 0x0
+_fake_rtti_data:
+	.ascii	"_FAKE_RTTI_DATA_"
+	.size	_fake_rtti_data, 16
+
+	.type	_ZTV3gooE,@object
+	.section	.rodata,"a",@progbits
+	.p2align	2, 0x0
+_ZTV3gooE:
+	.word	0
+	.word	_fake_rtti_data-_ZTV3gooE-8
+	.word	foo@PLT-_ZTV3gooE-8
+	.size	_ZTV3gooE, 12

bolt/test/non-empty-debug-line.test

@@ -4,7 +4,7 @@
 REQUIRES: system-linux
 
 RUN: %clang %cflags %S/Inputs/hello.c -g -o %t
-RUN: llvm-bolt %t -o %t1 --update-debug-sections --funcs=_start
+RUN: llvm-bolt %t -o %t1 --update-debug-sections --funcs=_start --keep-nops
 RUN: llvm-readobj -S %t > %t2
 RUN: llvm-readobj -S %t1 >> %t2
 RUN: FileCheck %s --input-file %t2

clang-tools-extra/clang-doc/BitcodeReader.cpp

@@ -79,7 +79,8 @@ static llvm::Error decodeRecord(const Record &R, std::optional<Location> &Field,
   if (R[0] > INT_MAX)
     return llvm::createStringError(llvm::inconvertibleErrorCode(),
                                    "integer too large to parse");
-  Field.emplace(static_cast<int>(R[0]), Blob, static_cast<bool>(R[1]));
+  Field.emplace(static_cast<int>(R[0]), static_cast<int>(R[1]), Blob,
+                static_cast<bool>(R[2]));
   return llvm::Error::success();
 }
 
@@ -130,7 +131,8 @@ static llvm::Error decodeRecord(const Record &R,
   if (R[0] > INT_MAX)
     return llvm::createStringError(llvm::inconvertibleErrorCode(),
                                    "integer too large to parse");
-  Field.emplace_back(static_cast<int>(R[0]), Blob, static_cast<bool>(R[1]));
+  Field.emplace_back(static_cast<int>(R[0]), static_cast<int>(R[1]), Blob,
+                     static_cast<bool>(R[2]));
   return llvm::Error::success();
 }