Merge pull request #451 from Xilinx/bump_to_0a17bdfc

[AutoBump] Merge with fixes of 0a17bdf (Oct 15) (14) (Needs ONNX Bump)(Needs downstream changes)(Needs Torch Bump)

Author: jorickert

.github/workflows/containers/github-action-ci/stage1.Dockerfile

@@ -2,7 +2,7 @@ FROM docker.io/library/ubuntu:22.04 as base
 ENV LLVM_SYSROOT=/opt/llvm
 
 FROM base as stage1-toolchain
-ENV LLVM_VERSION=18.1.8
+ENV LLVM_VERSION=19.1.2
 
 RUN apt-get update && \
     apt-get install -y \

bolt/include/bolt/Core/DIEBuilder.h

@@ -314,7 +314,7 @@ class DIEBuilder {
 
     BC.errs()
         << "BOLT-ERROR: unable to find TypeUnit for Type Unit at offset 0x"
-        << DU.getOffset() << "\n";
+        << Twine::utohexstr(DU.getOffset()) << "\n";
     return nullptr;
   }
 

bolt/lib/Core/BinaryContext.cpp

@@ -1294,8 +1294,8 @@ bool BinaryContext::handleAArch64Veneer(uint64_t Address, bool MatchOnly) {
     Veneer->getOrCreateLocalLabel(Address);
     Veneer->setMaxSize(TotalSize);
     Veneer->updateState(BinaryFunction::State::Disassembled);
-    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: handling veneer function at 0x" << Address
-                      << "\n");
+    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: handling veneer function at 0x"
+                      << Twine::utohexstr(Address) << "\n");
     return true;
   };
 

bolt/lib/Core/DIEBuilder.cpp

@@ -281,8 +281,7 @@ void DIEBuilder::buildTypeUnits(DebugStrOffsetsWriter *StrOffsetWriter,
     for (auto &Row : TUIndex.getRows()) {
       uint64_t Signature = Row.getSignature();
       // manually populate TypeUnit to UnitVector
-      DwarfContext->getTypeUnitForHash(DwarfContext->getMaxVersion(), Signature,
-                                       true);
+      DwarfContext->getTypeUnitForHash(Signature, true);
     }
   }
   const unsigned int CUNum = getCUNum(DwarfContext, isDWO());

bolt/lib/Passes/LongJmp.cpp

@@ -324,9 +324,8 @@ uint64_t LongJmpPass::tentativeLayoutRelocColdPart(
 uint64_t LongJmpPass::tentativeLayoutRelocMode(
     const BinaryContext &BC, std::vector<BinaryFunction *> &SortedFunctions,
     uint64_t DotAddress) {
-
   // Compute hot cold frontier
-  uint32_t LastHotIndex = -1u;
+  int64_t LastHotIndex = -1u;
   uint32_t CurrentIndex = 0;
   if (opts::HotFunctionsAtEnd) {
     for (BinaryFunction *BF : SortedFunctions) {
@@ -351,19 +350,20 @@ uint64_t LongJmpPass::tentativeLayoutRelocMode(
   // Hot
   CurrentIndex = 0;
   bool ColdLayoutDone = false;
+  auto runColdLayout = [&]() {
+    DotAddress = tentativeLayoutRelocColdPart(BC, SortedFunctions, DotAddress);
+    ColdLayoutDone = true;
+    if (opts::HotFunctionsAtEnd)
+      DotAddress = alignTo(DotAddress, opts::AlignText);
+  };
   for (BinaryFunction *Func : SortedFunctions) {
     if (!BC.shouldEmit(*Func)) {
       HotAddresses[Func] = Func->getAddress();
       continue;
     }
 
-    if (!ColdLayoutDone && CurrentIndex >= LastHotIndex) {
-      DotAddress =
-          tentativeLayoutRelocColdPart(BC, SortedFunctions, DotAddress);
-      ColdLayoutDone = true;
-      if (opts::HotFunctionsAtEnd)
-        DotAddress = alignTo(DotAddress, opts::AlignText);
-    }
+    if (!ColdLayoutDone && CurrentIndex >= LastHotIndex)
+      runColdLayout();
 
     DotAddress = alignTo(DotAddress, Func->getMinAlignment());
     uint64_t Pad =
@@ -382,6 +382,11 @@ uint64_t LongJmpPass::tentativeLayoutRelocMode(
     DotAddress += Func->estimateConstantIslandSize();
     ++CurrentIndex;
   }
+
+  // Ensure that tentative code layout always runs for cold blocks.
+  if (!ColdLayoutDone)
+    runColdLayout();
+
   // BBs
   for (BinaryFunction *Func : SortedFunctions)
     tentativeBBLayout(*Func);

bolt/lib/Passes/VeneerElimination.cpp

@@ -73,12 +73,12 @@ Error VeneerElimination::runOnFunctions(BinaryContext &BC) {
           continue;
 
         const MCSymbol *TargetSymbol = BC.MIB->getTargetSymbol(Instr, 0);
-        if (VeneerDestinations.find(TargetSymbol) == VeneerDestinations.end())
+        auto It = VeneerDestinations.find(TargetSymbol);
+        if (It == VeneerDestinations.end())
           continue;
 
         VeneerCallers++;
-        BC.MIB->replaceBranchTarget(Instr, VeneerDestinations[TargetSymbol],
-                                    BC.Ctx.get());
+        BC.MIB->replaceBranchTarget(Instr, It->second, BC.Ctx.get());
       }
     }
   }

bolt/lib/Rewrite/DWARFRewriter.cpp

@@ -1362,7 +1362,7 @@ void DWARFRewriter::updateDWARFObjectAddressRanges(
                Die.getTag() == dwarf::DW_TAG_compile_unit)) {
     if (opts::Verbosity >= 1)
       errs() << "BOLT-WARNING: cannot update ranges for DIE in Unit offset 0x"
-             << Unit.getOffset() << '\n';
+             << Twine::utohexstr(Unit.getOffset()) << '\n';
   }
 }
 

bolt/test/AArch64/split-funcs-lite.s

@@ -0,0 +1,27 @@
+# This test checks that tentative code layout for cold blocks always runs.
+# It commonly happens when using lite mode with split functions.
+
+# REQUIRES: system-linux, asserts
+
+# RUN: %clang %cflags -o %t %s
+# RUN: %clang %s %cflags -Wl,-q -o %t
+# RUN: link_fdata --no-lbr %s %t %t.fdata
+# RUN: llvm-bolt %t -o %t.bolt --data %t.fdata -split-functions \
+# RUN:   -debug 2>&1 | FileCheck %s
+
+  .text
+  .globl  foo
+  .type foo, %function
+foo:
+.entry_bb:
+# FDATA: 1 foo #.entry_bb# 10
+    cmp x0, #0
+    b.eq .Lcold_bb1
+    ret
+.Lcold_bb1:
+    ret
+
+## Force relocation mode.
+.reloc 0, R_AARCH64_NONE
+
+# CHECK: foo{{.*}} cold tentative: {{.*}}

clang-tools-extra/clang-doc/Generators.h

@@ -55,4 +55,8 @@ std::string getTagType(TagTypeKind AS);
 } // namespace doc
 } // namespace clang
 
+namespace llvm {
+extern template class Registry<clang::doc::Generator>;
+} // namespace llvm
+
 #endif // LLVM_CLANG_TOOLS_EXTRA_CLANG_DOC_GENERATOR_H

clang-tools-extra/clang-doc/tool/ClangDocMain.cpp

@@ -300,8 +300,7 @@ Example usage for a project using a compile commands database:
   llvm::StringMap<std::vector<StringRef>> USRToBitcode;
   Executor->get()->getToolResults()->forEachResult(
       [&](StringRef Key, StringRef Value) {
-        auto R = USRToBitcode.try_emplace(Key, std::vector<StringRef>());
-        R.first->second.emplace_back(Value);
+        USRToBitcode[Key].emplace_back(Value);
       });
 
   // Collects all Infos according to their unique USR value. This map is added