merge main into amd-staging (llvm#754)

Author: ronlieb

.github/workflows/premerge.yaml

@@ -14,8 +14,6 @@ on:
       # do this is that it allows us to take advantage of concurrency groups
       # to cancel in progress CI jobs whenever the PR is closed.
       - closed
-    paths:
-      - .github/workflows/premerge.yaml
   push:
     branches:
       - 'main'

bolt/include/bolt/Core/Linker.h

@@ -46,13 +46,6 @@ class BOLTLinker {
   /// Return the address and size of a symbol or std::nullopt if it cannot be
   /// found.
   virtual std::optional<SymbolInfo> lookupSymbolInfo(StringRef Name) const = 0;
-
-  /// Return the address of a symbol or std::nullopt if it cannot be found.
-  std::optional<uint64_t> lookupSymbol(StringRef Name) const {
-    if (const auto Info = lookupSymbolInfo(Name))
-      return Info->Address;
-    return std::nullopt;
-  }
 };
 
 } // namespace bolt

bolt/lib/Core/BinaryFunction.cpp

@@ -498,6 +498,11 @@ void BinaryFunction::print(raw_ostream &OS, std::string Annotation) {
     if (!IslandOffset)
       return;
 
+    // Print label if it exists at this offset.
+    if (const BinaryData *BD =
+            BC.getBinaryDataAtAddress(getAddress() + *IslandOffset))
+      OS << BD->getName() << ":\n";
+
     const size_t IslandSize = getSizeOfDataInCodeAt(*IslandOffset);
     BC.printData(OS, BC.extractData(getAddress() + *IslandOffset, IslandSize),
                  *IslandOffset);
@@ -1066,7 +1071,7 @@ size_t BinaryFunction::getSizeOfDataInCodeAt(uint64_t Offset) const {
   auto Iter = Islands->CodeOffsets.upper_bound(Offset);
   if (Iter != Islands->CodeOffsets.end())
     return *Iter - Offset;
-  return getSize() - Offset;
+  return getMaxSize() - Offset;
 }
 
 std::optional<uint64_t>
@@ -4254,21 +4259,21 @@ void BinaryFunction::updateOutputValues(const BOLTLinker &Linker) {
 
   if (BC.HasRelocations || isInjected()) {
     if (hasConstantIsland()) {
-      const auto DataAddress =
-          Linker.lookupSymbol(getFunctionConstantIslandLabel()->getName());
-      assert(DataAddress && "Cannot find function CI symbol");
-      setOutputDataAddress(*DataAddress);
+      const auto IslandLabelSymInfo =
+          Linker.lookupSymbolInfo(getFunctionConstantIslandLabel()->getName());
+      assert(IslandLabelSymInfo && "Cannot find function CI symbol");
+      setOutputDataAddress(IslandLabelSymInfo->Address);
       for (auto It : Islands->Offsets) {
         const uint64_t OldOffset = It.first;
         BinaryData *BD = BC.getBinaryDataAtAddress(getAddress() + OldOffset);
         if (!BD)
           continue;
 
         MCSymbol *Symbol = It.second;
-        const auto NewAddress = Linker.lookupSymbol(Symbol->getName());
-        assert(NewAddress && "Cannot find CI symbol");
+        const auto SymInfo = Linker.lookupSymbolInfo(Symbol->getName());
+        assert(SymInfo && "Cannot find CI symbol");
         auto &Section = *getCodeSection();
-        const auto NewOffset = *NewAddress - Section.getOutputAddress();
+        const auto NewOffset = SymInfo->Address - Section.getOutputAddress();
         BD->setOutputLocation(Section, NewOffset);
       }
     }
@@ -4293,10 +4298,10 @@ void BinaryFunction::updateOutputValues(const BOLTLinker &Linker) {
         FF.setAddress(ColdStartSymbolInfo->Address);
         FF.setImageSize(ColdStartSymbolInfo->Size);
         if (hasConstantIsland()) {
-          const auto DataAddress = Linker.lookupSymbol(
+          const auto SymInfo = Linker.lookupSymbolInfo(
               getFunctionColdConstantIslandLabel()->getName());
-          assert(DataAddress && "Cannot find cold CI symbol");
-          setOutputColdDataAddress(*DataAddress);
+          assert(SymInfo && "Cannot find cold CI symbol");
+          setOutputColdDataAddress(SymInfo->Address);
         }
       }
     }

bolt/lib/Rewrite/JITLinkLinker.cpp

@@ -125,11 +125,11 @@ struct JITLinkLinker::Context : jitlink::JITLinkContext {
       std::string SymName = (*Symbol.first).str();
       LLVM_DEBUG(dbgs() << "BOLT: looking for " << SymName << "\n");
 
-      if (auto Address = Linker.lookupSymbol(SymName)) {
+      if (auto SymInfo = Linker.lookupSymbolInfo(SymName)) {
         LLVM_DEBUG(dbgs() << "Resolved to address 0x"
-                          << Twine::utohexstr(*Address) << "\n");
+                          << Twine::utohexstr(SymInfo->Address) << "\n");
         AllResults[Symbol.first] = orc::ExecutorSymbolDef(
-            orc::ExecutorAddr(*Address), JITSymbolFlags());
+            orc::ExecutorAddr(SymInfo->Address), JITSymbolFlags());
         continue;
       }
 

bolt/lib/Rewrite/RewriteInstance.cpp

@@ -5907,9 +5907,9 @@ void RewriteInstance::writeEHFrameHeader() {
 }
 
 uint64_t RewriteInstance::getNewValueForSymbol(const StringRef Name) {
-  auto Value = Linker->lookupSymbol(Name);
+  auto Value = Linker->lookupSymbolInfo(Name);
   if (Value)
-    return *Value;
+    return Value->Address;
 
   // Return the original value if we haven't emitted the symbol.
   BinaryData *BD = BC->getBinaryDataByName(Name);

bolt/lib/RuntimeLibs/HugifyRuntimeLibrary.cpp

@@ -68,10 +68,11 @@ void HugifyRuntimeLibrary::link(BinaryContext &BC, StringRef ToolPath,
 
   assert(!RuntimeStartAddress &&
          "We don't currently support linking multiple runtime libraries");
-  RuntimeStartAddress = Linker.lookupSymbol("__bolt_hugify_self").value_or(0);
-  if (!RuntimeStartAddress) {
+  auto StartSymInfo = Linker.lookupSymbolInfo("__bolt_hugify_self");
+  if (!StartSymInfo) {
     errs() << "BOLT-ERROR: hugify library does not define __bolt_hugify_self: "
            << LibPath << "\n";
     exit(1);
   }
+  RuntimeStartAddress = StartSymInfo->Address;
 }

bolt/lib/RuntimeLibs/InstrumentationRuntimeLibrary.cpp

@@ -203,27 +203,35 @@ void InstrumentationRuntimeLibrary::link(
   if (BC.isMachO())
     return;
 
-  RuntimeFiniAddress = Linker.lookupSymbol("__bolt_instr_fini").value_or(0);
-  if (!RuntimeFiniAddress) {
+  std::optional<BOLTLinker::SymbolInfo> FiniSymInfo =
+      Linker.lookupSymbolInfo("__bolt_instr_fini");
+  if (!FiniSymInfo) {
     errs() << "BOLT-ERROR: instrumentation library does not define "
               "__bolt_instr_fini: "
            << LibPath << "\n";
     exit(1);
   }
-  RuntimeStartAddress = Linker.lookupSymbol("__bolt_instr_start").value_or(0);
-  if (!RuntimeStartAddress) {
+  RuntimeFiniAddress = FiniSymInfo->Address;
+
+  std::optional<BOLTLinker::SymbolInfo> StartSymInfo =
+      Linker.lookupSymbolInfo("__bolt_instr_start");
+  if (!StartSymInfo) {
     errs() << "BOLT-ERROR: instrumentation library does not define "
               "__bolt_instr_start: "
            << LibPath << "\n";
     exit(1);
   }
+  RuntimeStartAddress = StartSymInfo->Address;
+
   outs() << "BOLT-INFO: output linked against instrumentation runtime "
             "library, lib entry point is 0x"
          << Twine::utohexstr(RuntimeStartAddress) << "\n";
+
+  std::optional<BOLTLinker::SymbolInfo> ClearSymInfo =
+      Linker.lookupSymbolInfo("__bolt_instr_clear_counters");
+  const uint64_t ClearSymAddress = ClearSymInfo ? ClearSymInfo->Address : 0;
   outs() << "BOLT-INFO: clear procedure is 0x"
-         << Twine::utohexstr(
-                Linker.lookupSymbol("__bolt_instr_clear_counters").value_or(0))
-         << "\n";
+         << Twine::utohexstr(ClearSymAddress) << "\n";
 
   emitTablesAsELFNote(BC);
 }

bolt/test/AArch64/data-in-code.s

@@ -7,7 +7,8 @@
 ## Check disassembly of BOLT input.
 # RUN: llvm-objdump %t.exe -d | FileCheck %s
 
-# RUN: llvm-bolt %t.exe -o %t.bolt --print-disasm | FileCheck %s
+# RUN: llvm-bolt %t.exe -o %t.bolt --print-disasm \
+# RUN:   | FileCheck %s --check-prefixes CHECK,CHECK-BOLT-ONLY
 
 .text
 .balign 4
@@ -16,16 +17,21 @@
 .type _start, %function
 _start:
   mov x0, #0x0
+  ldr x1, .L1
   .word 0x4f82e010
   ret
+.size _start, .-_start
+.L1:
   .byte 0x0, 0xff, 0x42
 # CHECK-LABEL: _start
 # CHECK:        mov x0, #0x0
+# CHECK-NEXT:   ldr x1
+# CHECK-BOLT-ONLY-SAME: ISLANDat[[ADDR:]]
 # CHECK-NEXT:   .word 0x4f82e010
 # CHECK-NEXT:   ret
+# CHECK-BOLT-ONLY-NEXT: ISLANDat[[ADDR]]
 # CHECK-NEXT:   .short 0xff00
 # CHECK-NEXT:   .byte 0x42
-.size _start, .-_start
 
 ## Force relocation mode.
   .reloc 0, R_AARCH64_NONE

clang-tools-extra/clangd/unittests/ASTTests.cpp

@@ -329,7 +329,7 @@ TEST(ClangdAST, GetContainedAutoParamType) {
        auto &&d,
        auto *&e,
        auto (*f)(int)
-    ){};
+    ){ return 0; };
 
     int withoutAuto(
       int a,
@@ -338,7 +338,7 @@ TEST(ClangdAST, GetContainedAutoParamType) {
       int &&d,
       int *&e,
       int (*f)(int)
-    ){};
+    ){ return 0; };
   )cpp");
   TU.ExtraArgs.push_back("-std=c++20");
   auto AST = TU.build();

clang-tools-extra/clangd/unittests/FindSymbolsTests.cpp

@@ -113,7 +113,7 @@ TEST(WorkspaceSymbols, Unnamed) {
 TEST(WorkspaceSymbols, InMainFile) {
   TestTU TU;
   TU.Code = R"cpp(
-      int test() {}
+      int test() { return 0; }
       static void test2() {}
       )cpp";
   EXPECT_THAT(getSymbols(TU, "test"),
@@ -537,12 +537,14 @@ TEST(DocumentSymbols, InHeaderFile) {
   TestTU TU;
   TU.AdditionalFiles["bar.h"] = R"cpp(
       int foo() {
+        return 0;
       }
       )cpp";
   TU.Code = R"cpp(
       int i; // declaration to finish preamble
       #include "bar.h"
       int test() {
+        return 0;
       }
       )cpp";
   EXPECT_THAT(getSymbols(TU.build()),
@@ -780,7 +782,7 @@ TEST(DocumentSymbols, FuncTemplates) {
   TestTU TU;
   Annotations Source(R"cpp(
     template <class T>
-    T foo() {}
+    T foo() { return T{}; }
 
     auto x = foo<int>();
     auto y = foo<double>();