Merged main:7b97d5048a8f into amd-gfx:3ab181f6d3ad

Local branch amd-gfx 3ab181f Merge amd-gfx into merge-candidate branch
Remote branch main 7b97d50 Fix MSVC "not all control paths return a value" warning. NFC.

Author: SC llvm team

.github/CODEOWNERS

@@ -42,6 +42,10 @@
 /mlir/lib/Dialect/Vector/Transforms/VectorEmulateNarrowType.cpp @MaheshRavishankar
 /mlir/lib/Interfaces/TilingInterface.* @MaheshRavishankar
 
+# Transform Dialect in MLIR.
+/mlir/include/mlir/Dialect/Transform/* @ftynse
+/mlir/lib/Dialect/Transform/* @ftynse
+
 # SPIR-V in MLIR.
 /mlir/**/SPIRV/ @antiagainst @kuhar
 /mlir/**/SPIRVTo*/ @antiagainst @kuhar

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

@@ -44,7 +44,7 @@ env:
 jobs:
   stage1:
     runs-on:
-      group: libcxx-runners-16
+      group: libcxx-runners-8
     continue-on-error: false
     strategy:
       fail-fast: true
@@ -163,10 +163,10 @@ jobs:
           machine: libcxx-runners-8
           std_modules: 'ON'
         - config: 'generic-asan'
-          machine: libcxx-runners-16
+          machine: libcxx-runners-8
           std_modules: 'OFF'
         - config: 'generic-tsan'
-          machine: libcxx-runners-16
+          machine: libcxx-runners-8
           std_modules: 'OFF'
         - config: 'generic-ubsan'
           machine: libcxx-runners-8

.github/workflows/release-binaries.yml

@@ -57,17 +57,52 @@ jobs:
         fi
         bash .github/workflows/set-release-binary-outputs.sh "${{ github.actor }}" "$tag" "$upload"
 
+  # Try to get around the 6 hour timeout by first running a job to fill
+  # the build cache.
+  fill-cache:
+    name: "Fill Cache ${{ matrix.os }}"
+    needs: prepare
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os:
+          - ubuntu-22.04
+    steps:
+    - name: Checkout LLVM
+      uses: actions/checkout@v4
+      with:
+        ref: ${{ inputs.tag || github.ref_name }}
+
+    - name: Install Ninja
+      uses: llvm/actions/install-ninja@main
+
+    - name: Setup sccache
+      uses: hendrikmuhs/ccache-action@v1
+      with:
+        max-size: 250M
+        key: sccache-${{ matrix.os }}-release
+        variant: sccache
+
+    - name: Build Clang
+      run: |
+        cmake -G Ninja  -DCMAKE_C_COMPILER_LAUNCHER=sccache -DCMAKE_CXX_COMPILER_LAUNCHER=sccache -DCMAKE_BUILD_TYPE=Release -DCMAKE_ENABLE_ASSERTIONS=OFF -DCMAKE_POSITION_INDEPENDENT_CODE=ON -DLLVM_ENABLE_PROJECTS=clang -S llvm -B build
+        ninja -v -C build
+
+
   build-binaries:
     name: ${{ matrix.target.triple }}
     permissions:
       contents: write # To upload assets to release.
-    needs: prepare
+    needs:
+      - prepare
+      - fill-cache
     runs-on: ${{ matrix.target.runs-on }}
     strategy:
       fail-fast: false
       matrix:
         target:
           - triple: x86_64-linux-gnu-ubuntu-22.04
+            os: ubuntu-22.04
             runs-on: ubuntu-22.04-16x64
             debian-build-deps: >
               chrpath
@@ -81,6 +116,14 @@ jobs:
         ref: ${{ needs.prepare.outputs.ref }}
         path: ${{ needs.prepare.outputs.build-dir }}/llvm-project
 
+    - name: Setup sccache
+      uses: hendrikmuhs/ccache-action@v1
+      with:
+        max-size: 250M
+        key: sccache-${{ matrix.target.os }}-release
+        save: false
+        variant: sccache
+
     - name: Install Brew build dependencies
       if: matrix.target.brew-build-deps != ''
       run: brew install ${{ matrix.target.brew-build-deps }}
@@ -102,7 +145,8 @@ jobs:
         -triple ${{ matrix.target.triple }} \
         -use-ninja \
         -no-checkout \
-        -no-test-suite
+        -no-test-suite \
+        -configure-flags "-DCMAKE_C_COMPILER_LAUNCHER=sccache -DCMAKE_CXX_COMPILER_LAUNCHER=sccache"
 
     - name: Upload binaries
       if: ${{ always() && needs.prepare.outputs.upload == 'true' }}

bolt/include/bolt/Core/BinaryContext.h

@@ -1290,6 +1290,9 @@ class BinaryContext {
   /// Return true if the function should be emitted to the output file.
   bool shouldEmit(const BinaryFunction &Function) const;
 
+  /// Dump the assembly representation of MCInst to debug output.
+  void dump(const MCInst &Inst) const;
+
   /// Print the string name for a CFI operation.
   static void printCFI(raw_ostream &OS, const MCCFIInstruction &Inst);
 

bolt/lib/Core/BinaryContext.cpp

@@ -1738,6 +1738,15 @@ bool BinaryContext::shouldEmit(const BinaryFunction &Function) const {
   return HasRelocations || Function.isSimple();
 }
 
+void BinaryContext::dump(const MCInst &Inst) const {
+  if (LLVM_UNLIKELY(!InstPrinter)) {
+    dbgs() << "Cannot dump for InstPrinter is not initialized.\n";
+    return;
+  }
+  InstPrinter->printInst(&Inst, 0, "", *STI, dbgs());
+  dbgs() << "\n";
+}
+
 void BinaryContext::printCFI(raw_ostream &OS, const MCCFIInstruction &Inst) {
   uint32_t Operation = Inst.getOperation();
   switch (Operation) {

bolt/lib/Core/Exceptions.cpp

@@ -112,13 +112,18 @@ void BinaryFunction::parseLSDA(ArrayRef<uint8_t> LSDASectionData,
   uint64_t Offset = getLSDAAddress() - LSDASectionAddress;
   assert(Data.isValidOffset(Offset) && "wrong LSDA address");
 
-  uint8_t LPStartEncoding = Data.getU8(&Offset);
-  uint64_t LPStart = 0;
-  // Convert to offset if LPStartEncoding is typed absptr DW_EH_PE_absptr
-  if (std::optional<uint64_t> MaybeLPStart = Data.getEncodedPointer(
-          &Offset, LPStartEncoding, Offset + LSDASectionAddress))
-    LPStart = (LPStartEncoding && 0xFF == 0) ? *MaybeLPStart
-                                             : *MaybeLPStart - Address;
+  const uint8_t LPStartEncoding = Data.getU8(&Offset);
+  uint64_t LPStart = Address;
+  if (LPStartEncoding != dwarf::DW_EH_PE_omit) {
+    std::optional<uint64_t> MaybeLPStart = Data.getEncodedPointer(
+        &Offset, LPStartEncoding, Offset + LSDASectionAddress);
+    if (!MaybeLPStart) {
+      errs() << "BOLT-ERROR: unsupported LPStartEncoding: "
+             << (unsigned)LPStartEncoding << '\n';
+      exit(1);
+    }
+    LPStart = *MaybeLPStart;
+  }
 
   const uint8_t TTypeEncoding = Data.getU8(&Offset);
   LSDATypeEncoding = TTypeEncoding;
@@ -175,38 +180,38 @@ void BinaryFunction::parseLSDA(ArrayRef<uint8_t> LSDASectionData,
     uint64_t LandingPad = *Data.getEncodedPointer(
         &CallSitePtr, CallSiteEncoding, CallSitePtr + LSDASectionAddress);
     uint64_t ActionEntry = Data.getULEB128(&CallSitePtr);
-
-    uint64_t LPOffset = LPStart + LandingPad;
-    uint64_t LPAddress = Address + LPOffset;
-
-    // Verify if landing pad code is located outside current function
-    // Support landing pad to builtin_unreachable
-    if (LPAddress < Address || LPAddress > Address + getSize()) {
-      BinaryFunction *Fragment =
-          BC.getBinaryFunctionContainingAddress(LPAddress);
-      assert(Fragment != nullptr &&
-             "BOLT-ERROR: cannot find landing pad fragment");
-      BC.addInterproceduralReference(this, Fragment->getAddress());
-      BC.processInterproceduralReferences();
-      assert(isParentOrChildOf(*Fragment) &&
-             "BOLT-ERROR: cannot have landing pads in different functions");
-      setHasIndirectTargetToSplitFragment(true);
-      BC.addFragmentsToSkip(this);
-      return;
-    }
+    if (LandingPad)
+      LandingPad += LPStart;
 
     if (opts::PrintExceptions) {
       outs() << "Call Site: [0x" << Twine::utohexstr(RangeBase + Start)
              << ", 0x" << Twine::utohexstr(RangeBase + Start + Length)
-             << "); landing pad: 0x" << Twine::utohexstr(LPOffset)
+             << "); landing pad: 0x" << Twine::utohexstr(LandingPad)
              << "; action entry: 0x" << Twine::utohexstr(ActionEntry) << "\n";
       outs() << "  current offset is " << (CallSitePtr - CallSiteTableStart)
              << '\n';
     }
 
     // Create a handler entry if necessary.
     MCSymbol *LPSymbol = nullptr;
-    if (LPOffset) {
+    if (LandingPad) {
+      // Verify if landing pad code is located outside current function
+      // Support landing pad to builtin_unreachable
+      if (LandingPad < Address || LandingPad > Address + getSize()) {
+        BinaryFunction *Fragment =
+            BC.getBinaryFunctionContainingAddress(LandingPad);
+        assert(Fragment != nullptr &&
+               "BOLT-ERROR: cannot find landing pad fragment");
+        BC.addInterproceduralReference(this, Fragment->getAddress());
+        BC.processInterproceduralReferences();
+        assert(isParentOrChildOf(*Fragment) &&
+               "BOLT-ERROR: cannot have landing pads in different functions");
+        setHasIndirectTargetToSplitFragment(true);
+        BC.addFragmentsToSkip(this);
+        return;
+      }
+
+      const uint64_t LPOffset = LandingPad - getAddress();
       if (!getInstructionAtOffset(LPOffset)) {
         if (opts::Verbosity >= 1)
           errs() << "BOLT-WARNING: landing pad " << Twine::utohexstr(LPOffset)

bolt/lib/Passes/ValidateInternalCalls.cpp

@@ -281,18 +281,16 @@ bool ValidateInternalCalls::analyzeFunction(BinaryFunction &Function) const {
           LLVM_DEBUG({
             dbgs() << "Detected out-of-range PIC reference in " << Function
                    << "\nReturn address load: ";
-            BC.InstPrinter->printInst(TargetInst, 0, "", *BC.STI, dbgs());
-            dbgs() << "\nUse: ";
-            BC.InstPrinter->printInst(&Use, 0, "", *BC.STI, dbgs());
-            dbgs() << "\n";
+            BC.dump(*TargetInst);
+            dbgs() << "Use: ";
+            BC.dump(Use);
             Function.dump();
           });
           return false;
         }
         LLVM_DEBUG({
           dbgs() << "Validated access: ";
-          BC.InstPrinter->printInst(&Use, 0, "", *BC.STI, dbgs());
-          dbgs() << "\n";
+          BC.dump(Use);
         });
       }
       if (!UseDetected) {

bolt/lib/Profile/YAMLProfileReader.cpp

@@ -31,7 +31,7 @@ static llvm::cl::opt<bool>
 llvm::cl::opt<bool> ProfileUseDFS("profile-use-dfs",
                                   cl::desc("use DFS order for YAML profile"),
                                   cl::Hidden, cl::cat(BoltOptCategory));
-}
+} // namespace opts
 
 namespace llvm {
 namespace bolt {
@@ -354,7 +354,7 @@ Error YAMLProfileReader::readProfile(BinaryContext &BC) {
       matchProfileToFunction(YamlBF, Function);
   }
 
-  for (auto &[CommonName, LTOProfiles]: LTOCommonNameMap) {
+  for (const auto &[CommonName, LTOProfiles] : LTOCommonNameMap) {
     if (!LTOCommonNameFunctionMap.contains(CommonName))
       continue;
     std::unordered_set<BinaryFunction *> &Functions =

bolt/test/lsda.cpp renamed to bolt/test/lsda-section-name.cpp

@@ -1,8 +1,8 @@
 // This test check that LSDA section named by .gcc_except_table.main is
 // disassembled by BOLT.
 
-// RUN: %clang++ %cxxflags -O3 -flto=thin -no-pie -c %s -o %t.o
-// RUN: %clang++ %cxxflags -flto=thin -no-pie -fuse-ld=lld %t.o -o %t.exe \
+// RUN: %clang++ %cxxflags -O3 -no-pie -c %s -o %t.o
+// RUN: %clang++ %cxxflags -no-pie -fuse-ld=lld %t.o -o %t.exe \
 // RUN:   -Wl,-q -Wl,--script=%S/Inputs/lsda.ldscript
 // RUN: llvm-readelf -SW %t.exe | FileCheck %s
 // RUN: llvm-bolt %t.exe -o %t.bolt

bolt/test/runtime/X86/exceptions-lpstart-zero.s

@@ -0,0 +1,91 @@
+# RUN: %clangxx %cflags -no-pie %s -o %t.exe -Wl,-q
+# RUN: llvm-bolt %t.exe -o %t.exe.bolt
+# RUN: %t.exe.bolt
+
+# REQUIRES: system-linux
+
+## Test that BOLT properly handles LPStart when LPStartEncoding is different
+## from DW_EH_PE_omit.
+
+# The test case compiled with -O1 from:
+#
+# int main() {
+#   try {
+#     throw 42;
+#   } catch (...) {
+#     return 0;
+#   }
+#   return 1;
+# }
+#
+# The exception table was modified with udata4 LPStartEncoding and sdata4
+# CallSiteEncoding.
+
+	.text
+	.globl	main                            # -- Begin function main
+	.p2align	4, 0x90
+	.type	main,@function
+main:                                   # @main
+.Lfunc_begin0:
+	.cfi_startproc
+	.cfi_personality 3, __gxx_personality_v0
+	.cfi_lsda 3, .Lexception0
+# %bb.0:
+	pushq	%rax
+	.cfi_def_cfa_offset 16
+	movl	$4, %edi
+	callq	__cxa_allocate_exception
+	movl	$42, (%rax)
+.Ltmp0:
+	movl	$_ZTIi, %esi
+	movq	%rax, %rdi
+	xorl	%edx, %edx
+	callq	__cxa_throw
+.Ltmp1:
+# %bb.1:
+.LBB0_2:
+.Ltmp2:
+	movq	%rax, %rdi
+	callq	__cxa_begin_catch
+	callq	__cxa_end_catch
+	xorl	%eax, %eax
+	popq	%rcx
+	.cfi_def_cfa_offset 8
+	retq
+.Lfunc_end0:
+	.size	main, .Lfunc_end0-main
+	.cfi_endproc
+	.section	.gcc_except_table,"a",@progbits
+	.p2align	2
+GCC_except_table0:
+.Lexception0:
+	.byte	3                               # @LPStart Encoding = udata4
+  .long 0
+	.byte	3                               # @TType Encoding = udata4
+	.uleb128 .Lttbase0-.Lttbaseref0
+.Lttbaseref0:
+	.byte	11                              # Call site Encoding = sdata4
+	.uleb128 .Lcst_end0-.Lcst_begin0
+.Lcst_begin0:
+	.long .Lfunc_begin0-.Lfunc_begin0    # >> Call Site 1 <<
+	.long .Ltmp0-.Lfunc_begin0           #   Call between .Lfunc_begin0 and .Ltmp0
+	.long	0                               #     has no landing pad
+	.byte	0                               #   On action: cleanup
+	.long .Ltmp0-.Lfunc_begin0           # >> Call Site 2 <<
+	.long .Ltmp1-.Ltmp0                  #   Call between .Ltmp0 and .Ltmp1
+	.long .Ltmp2                         #     jumps to .Ltmp2
+	.byte	1                               #   On action: 1
+	.long .Ltmp1-.Lfunc_begin0           # >> Call Site 3 <<
+	.long .Lfunc_end0-.Ltmp1             #   Call between .Ltmp1 and .Lfunc_end0
+	.long	0                               #     has no landing pad
+	.byte	0                               #   On action: cleanup
+.Lcst_end0:
+	.byte	1                               # >> Action Record 1 <<
+                                        #   Catch TypeInfo 1
+	.byte	0                               #   No further actions
+	.p2align	2
+                                        # >> Catch TypeInfos <<
+	.long	0                               # TypeInfo 1
+.Lttbase0:
+	.p2align	2
+                                        # -- End function