Merge branch 'main' into reverse-null

Author: Discookie

.github/new-prs-labeler.yml

@@ -668,7 +668,7 @@ mlgo:
   - llvm/lib/CodeGen/ML*
   - llvm/unittests/CodeGen/ML*
   - llvm/test/CodeGen/MLRegAlloc/**
-  - llvm/utils/mlgo-utils/*
+  - llvm/utils/mlgo-utils/**
 
 tools:llvm-exegesis:
   - llvm/tools/llvm-exegesis/**

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

@@ -242,6 +242,7 @@ jobs:
         - { config: mingw-dll, mingw: true }
         - { config: mingw-static, mingw: true }
         - { config: mingw-dll-i686, mingw: true }
+        - { config: mingw-incomplete-sysroot, mingw: true }
     steps:
       - uses: actions/checkout@v4
       - name: Install dependencies
@@ -260,6 +261,12 @@ jobs:
           del llvm-mingw*.zip
           mv llvm-mingw* c:\llvm-mingw
           echo "c:\llvm-mingw\bin" | Out-File -FilePath $Env:GITHUB_PATH -Encoding utf8 -Append
+      - name: Simulate a from-scratch build of llvm-mingw
+        if: ${{ matrix.config == 'mingw-incomplete-sysroot' }}
+        run: |
+          rm -r c:\llvm-mingw\include\c++
+          rm -r c:\llvm-mingw\*-w64-mingw32\lib\libc++*
+          rm -r c:\llvm-mingw\*-w64-mingw32\lib\libunwind*
       - name: Add Git Bash to the path
         run: |
           echo "c:\Program Files\Git\usr\bin" | Out-File -FilePath $Env:GITHUB_PATH -Encoding utf8 -Append

.github/workflows/pr-code-format.yml

@@ -12,6 +12,10 @@ on:
 jobs:
   code_formatter:
     runs-on: ubuntu-latest
+    timeout-minutes: 30
+    concurrency:
+      group: ${{ github.workflow }}-${{ github.event.pull_request.number }}
+      cancel-in-progress: true
     if: github.repository == 'llvm/llvm-project'
     steps:
       - name: Fetch LLVM sources

.github/workflows/release-binaries-save-stage/action.yml

@@ -10,6 +10,9 @@ inputs:
     required: true
     type: 'string'
 
+permissions:
+  contents: read
+
 runs:
   using: "composite"
   steps:
@@ -18,6 +21,9 @@ runs:
     - name: Package Build and Source Directories
       shell: bash
       run: |
+        # Remove .git/config to avoid leaking GITHUB_TOKEN stored there.
+        # See https://unit42.paloaltonetworks.com/github-repo-artifacts-leak-tokens/
+        rm -Rf .git/config
         # Windows does not support symlinks, so we need to dereference them.
         tar --exclude build/ ${{ (runner.os == 'Windows' && '-h') || '' }} -c . | zstd -T0 -c > ../llvm-project.tar.zst
         mv ../llvm-project.tar.zst .

bolt/include/bolt/Utils/Utils.h

@@ -41,6 +41,11 @@ std::string getEscapedName(const StringRef &Name);
 /// Return the unescaped name
 std::string getUnescapedName(const StringRef &Name);
 
+/// Return a common part for a given \p Name wrt a given \p Suffixes list.
+/// Preserve the suffix if \p KeepSuffix is set, only dropping characters
+/// following it, otherwise drop the suffix as well.
+std::optional<StringRef> getCommonName(const StringRef Name, bool KeepSuffix,
+                                       ArrayRef<StringRef> Suffixes);
 /// LTO-generated function names take a form:
 ///
 ///   <function_name>.lto_priv.<decimal_number>/...

bolt/lib/Profile/DataAggregator.cpp

@@ -88,7 +88,7 @@ MaxSamples("max-samples",
   cl::cat(AggregatorCategory));
 
 extern cl::opt<opts::ProfileFormatKind> ProfileFormat;
-extern cl::opt<bool> ProfileUsePseudoProbes;
+extern cl::opt<bool> ProfileWritePseudoProbes;
 extern cl::opt<std::string> SaveProfile;
 
 cl::opt<bool> ReadPreAggregated(
@@ -2300,7 +2300,7 @@ std::error_code DataAggregator::writeBATYAML(BinaryContext &BC,
   yaml::bolt::BinaryProfile BP;
 
   const MCPseudoProbeDecoder *PseudoProbeDecoder =
-      opts::ProfileUsePseudoProbes ? BC.getPseudoProbeDecoder() : nullptr;
+      opts::ProfileWritePseudoProbes ? BC.getPseudoProbeDecoder() : nullptr;
 
   // Fill out the header info.
   BP.Header.Version = 1;
@@ -2427,11 +2427,15 @@ std::error_code DataAggregator::writeBATYAML(BinaryContext &BC,
           }
         }
       }
-      // Drop blocks without a hash, won't be useful for stale matching.
-      llvm::erase_if(YamlBF.Blocks,
-                     [](const yaml::bolt::BinaryBasicBlockProfile &YamlBB) {
-                       return YamlBB.Hash == (yaml::Hex64)0;
-                     });
+      // Skip printing if there's no profile data
+      llvm::erase_if(
+          YamlBF.Blocks, [](const yaml::bolt::BinaryBasicBlockProfile &YamlBB) {
+            auto HasCount = [](const auto &SI) { return SI.Count; };
+            bool HasAnyCount = YamlBB.ExecCount ||
+                               llvm::any_of(YamlBB.Successors, HasCount) ||
+                               llvm::any_of(YamlBB.CallSites, HasCount);
+            return !HasAnyCount;
+          });
       BP.Functions.emplace_back(YamlBF);
     }
   }

bolt/lib/Profile/YAMLProfileReader.cpp

@@ -49,11 +49,6 @@ 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));
-
-llvm::cl::opt<bool> ProfileUsePseudoProbes(
-    "profile-use-pseudo-probes",
-    cl::desc("Use pseudo probes for profile generation and matching"),
-    cl::Hidden, cl::cat(BoltOptCategory));
 } // namespace opts
 
 namespace llvm {
@@ -373,6 +368,7 @@ Error YAMLProfileReader::preprocessProfile(BinaryContext &BC) {
     return errorCodeToError(EC);
   }
   yaml::Input YamlInput(MB.get()->getBuffer());
+  YamlInput.setAllowUnknownKeys(true);
 
   // Consume YAML file.
   YamlInput >> YamlBP;

bolt/lib/Profile/YAMLProfileWriter.cpp

@@ -13,6 +13,7 @@
 #include "bolt/Profile/DataAggregator.h"
 #include "bolt/Profile/ProfileReaderBase.h"
 #include "bolt/Rewrite/RewriteInstance.h"
+#include "bolt/Utils/CommandLineOpts.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/FileSystem.h"
 #include "llvm/Support/raw_ostream.h"
@@ -21,8 +22,12 @@
 #define DEBUG_TYPE "bolt-prof"
 
 namespace opts {
-extern llvm::cl::opt<bool> ProfileUseDFS;
-extern llvm::cl::opt<bool> ProfileUsePseudoProbes;
+using namespace llvm;
+extern cl::opt<bool> ProfileUseDFS;
+cl::opt<bool> ProfileWritePseudoProbes(
+    "profile-write-pseudo-probes",
+    cl::desc("Use pseudo probes in profile generation"), cl::Hidden,
+    cl::cat(BoltOptCategory));
 } // namespace opts
 
 namespace llvm {
@@ -59,7 +64,7 @@ YAMLProfileWriter::convert(const BinaryFunction &BF, bool UseDFS,
   yaml::bolt::BinaryFunctionProfile YamlBF;
   const BinaryContext &BC = BF.getBinaryContext();
   const MCPseudoProbeDecoder *PseudoProbeDecoder =
-      opts::ProfileUsePseudoProbes ? BC.getPseudoProbeDecoder() : nullptr;
+      opts::ProfileWritePseudoProbes ? BC.getPseudoProbeDecoder() : nullptr;
 
   const uint16_t LBRProfile = BF.getProfileFlags() & BinaryFunction::PF_LBR;
 

bolt/lib/Rewrite/PseudoProbeRewriter.cpp

@@ -14,6 +14,7 @@
 #include "bolt/Rewrite/MetadataRewriter.h"
 #include "bolt/Rewrite/MetadataRewriters.h"
 #include "bolt/Utils/CommandLineOpts.h"
+#include "bolt/Utils/Utils.h"
 #include "llvm/IR/Function.h"
 #include "llvm/MC/MCPseudoProbe.h"
 #include "llvm/Support/CommandLine.h"
@@ -49,7 +50,7 @@ static cl::opt<PrintPseudoProbesOptions> PrintPseudoProbes(
                clEnumValN(PPP_All, "all", "enable all debugging printout")),
     cl::Hidden, cl::cat(BoltCategory));
 
-extern cl::opt<bool> ProfileUsePseudoProbes;
+extern cl::opt<bool> ProfileWritePseudoProbes;
 } // namespace opts
 
 namespace {
@@ -71,7 +72,8 @@ class PseudoProbeRewriter final : public MetadataRewriter {
 
   /// Parse .pseudo_probe_desc section and .pseudo_probe section
   /// Setup Pseudo probe decoder
-  void parsePseudoProbe();
+  /// If \p ProfiledOnly is set, only parse records for functions with profile.
+  void parsePseudoProbe(bool ProfiledOnly = false);
 
   /// PseudoProbe decoder
   std::shared_ptr<MCPseudoProbeDecoder> ProbeDecoderPtr;
@@ -90,21 +92,21 @@ class PseudoProbeRewriter final : public MetadataRewriter {
 };
 
 Error PseudoProbeRewriter::preCFGInitializer() {
-  if (opts::ProfileUsePseudoProbes)
-    parsePseudoProbe();
+  if (opts::ProfileWritePseudoProbes)
+    parsePseudoProbe(true);
 
   return Error::success();
 }
 
 Error PseudoProbeRewriter::postEmitFinalizer() {
-  if (!opts::ProfileUsePseudoProbes)
+  if (!opts::ProfileWritePseudoProbes)
     parsePseudoProbe();
   updatePseudoProbes();
 
   return Error::success();
 }
 
-void PseudoProbeRewriter::parsePseudoProbe() {
+void PseudoProbeRewriter::parsePseudoProbe(bool ProfiledOnly) {
   MCPseudoProbeDecoder &ProbeDecoder(*ProbeDecoderPtr);
   PseudoProbeDescSection = BC.getUniqueSectionByName(".pseudo_probe_desc");
   PseudoProbeSection = BC.getUniqueSectionByName(".pseudo_probe");
@@ -133,10 +135,22 @@ void PseudoProbeRewriter::parsePseudoProbe() {
 
   MCPseudoProbeDecoder::Uint64Set GuidFilter;
   MCPseudoProbeDecoder::Uint64Map FuncStartAddrs;
+  SmallVector<StringRef, 0> Suffixes(
+      {".destroy", ".resume", ".llvm.", ".cold", ".warm"});
   for (const BinaryFunction *F : BC.getAllBinaryFunctions()) {
+    bool HasProfile = F->hasProfileAvailable();
     for (const MCSymbol *Sym : F->getSymbols()) {
-      FuncStartAddrs[Function::getGUID(NameResolver::restore(Sym->getName()))] =
-          F->getAddress();
+      StringRef SymName = Sym->getName();
+      for (auto Name : {std::optional(NameResolver::restore(SymName)),
+                        getCommonName(SymName, false, Suffixes)}) {
+        if (!Name)
+          continue;
+        SymName = *Name;
+        uint64_t GUID = Function::getGUID(SymName);
+        FuncStartAddrs[GUID] = F->getAddress();
+        if (ProfiledOnly && HasProfile)
+          GuidFilter.insert(GUID);
+      }
     }
   }
   Contents = PseudoProbeSection->getContents();
@@ -155,13 +169,25 @@ void PseudoProbeRewriter::parsePseudoProbe() {
     ProbeDecoder.printProbesForAllAddresses(outs());
   }
 
-  for (const auto &FuncDesc : ProbeDecoder.getGUID2FuncDescMap()) {
-    uint64_t GUID = FuncDesc.FuncGUID;
-    if (!FuncStartAddrs.contains(GUID))
-      continue;
-    BinaryFunction *BF = BC.getBinaryFunctionAtAddress(FuncStartAddrs[GUID]);
-    assert(BF);
-    BF->setGUID(GUID);
+  const GUIDProbeFunctionMap &GUID2Func = ProbeDecoder.getGUID2FuncDescMap();
+  // Checks GUID in GUID2Func and returns it if it's present or null otherwise.
+  auto checkGUID = [&](StringRef SymName) -> uint64_t {
+    uint64_t GUID = Function::getGUID(SymName);
+    if (GUID2Func.find(GUID) == GUID2Func.end())
+      return 0;
+    return GUID;
+  };
+  for (BinaryFunction *F : BC.getAllBinaryFunctions()) {
+    for (const MCSymbol *Sym : F->getSymbols()) {
+      StringRef SymName = NameResolver::restore(Sym->getName());
+      uint64_t GUID = checkGUID(SymName);
+      std::optional<StringRef> CommonName =
+          getCommonName(SymName, false, Suffixes);
+      if (!GUID && CommonName)
+        GUID = checkGUID(*CommonName);
+      if (GUID)
+        F->setGUID(GUID);
+    }
   }
 }
 

bolt/lib/Utils/Utils.cpp

@@ -66,15 +66,21 @@ std::string getUnescapedName(const StringRef &Name) {
   return Output;
 }
 
-std::optional<StringRef> getLTOCommonName(const StringRef Name) {
-  for (StringRef Suffix : {".__uniq.", ".lto_priv.", ".constprop.", ".llvm."}) {
+std::optional<StringRef> getCommonName(const StringRef Name, bool KeepSuffix,
+                                       ArrayRef<StringRef> Suffixes) {
+  for (StringRef Suffix : Suffixes) {
     size_t LTOSuffixPos = Name.find(Suffix);
     if (LTOSuffixPos != StringRef::npos)
-      return Name.substr(0, LTOSuffixPos + Suffix.size());
+      return Name.substr(0, LTOSuffixPos + (KeepSuffix ? Suffix.size() : 0));
   }
   return std::nullopt;
 }
 
+std::optional<StringRef> getLTOCommonName(const StringRef Name) {
+  return getCommonName(Name, true,
+                       {".__uniq.", ".lto_priv.", ".constprop.", ".llvm."});
+}
+
 std::optional<uint8_t> readDWARFExpressionTargetReg(StringRef ExprBytes) {
   uint8_t Opcode = ExprBytes[0];
   if (Opcode == dwarf::DW_CFA_def_cfa_expression)

bolt/test/X86/pseudoprobe-decoding-inline.test

@@ -6,11 +6,11 @@
 # PREAGG: B X:0 #main# 1 0
 ## Check pseudo-probes in regular YAML profile (non-BOLTed binary)
 # RUN: link_fdata %s %S/../../../llvm/test/tools/llvm-profgen/Inputs/inline-cs-pseudoprobe.perfbin %t.preagg PREAGG
-# RUN: perf2bolt %S/../../../llvm/test/tools/llvm-profgen/Inputs/inline-cs-pseudoprobe.perfbin -p %t.preagg --pa -w %t.yaml -o %t.fdata --profile-use-pseudo-probes
+# RUN: perf2bolt %S/../../../llvm/test/tools/llvm-profgen/Inputs/inline-cs-pseudoprobe.perfbin -p %t.preagg --pa -w %t.yaml -o %t.fdata --profile-write-pseudo-probes
 # RUN: FileCheck --input-file %t.yaml %s --check-prefix CHECK-YAML
 ## Check pseudo-probes in BAT YAML profile (BOLTed binary)
 # RUN: link_fdata %s %t.bolt %t.preagg2 PREAGG
-# RUN: perf2bolt %t.bolt -p %t.preagg2 --pa -w %t.yaml2 -o %t.fdata2 --profile-use-pseudo-probes
+# RUN: perf2bolt %t.bolt -p %t.preagg2 --pa -w %t.yaml2 -o %t.fdata2 --profile-write-pseudo-probes
 # RUN: FileCheck --input-file %t.yaml2 %s --check-prefix CHECK-YAML
 # CHECK-YAML: name: bar
 # CHECK-YAML: - bid: 0
@@ -30,7 +30,7 @@
 # CHECK-YAML: guid: 0xDB956436E78DD5FA
 # CHECK-YAML: pseudo_probe_desc_hash: 0x10000FFFFFFFF
 #
-## Check that without --profile-use-pseudo-probes option, no pseudo probes are
+## Check that without --profile-write-pseudo-probes option, no pseudo probes are
 ## generated
 # RUN: perf2bolt %S/../../../llvm/test/tools/llvm-profgen/Inputs/inline-cs-pseudoprobe.perfbin -p %t.preagg --pa -w %t.yaml -o %t.fdata
 # RUN: FileCheck --input-file %t.yaml %s --check-prefix CHECK-NO-OPT

bolt/test/X86/yaml-unknown-keys.test

@@ -0,0 +1,50 @@
+## Test that BOLT gracefully handles a YAML profile with unknown keys.
+
+# REQUIRES: system-linux
+# RUN: split-file %s %t
+# RUN: llvm-mc -filetype=obj -triple x86_64-unknown-unknown %t/main.s -o %t.o
+# RUN: %clang %cflags %t.o -o %t.exe -Wl,-q -nostdlib
+# RUN: llvm-bolt %t.exe -o %t.null --data %t/profile.yaml \
+# RUN:   --profile-ignore-hash -v=1 2>&1 | FileCheck %s
+# CHECK: warning: unknown key 'unknown-header-key'
+# CHECK: warning: unknown key 'unknown_succ_key'
+# CHECK: warning: unknown key 'unknown_block_key'
+# CHECK: warning: unknown key 'unknown_function_key'
+# CHECK: warning: unknown key 'unknown_toplev_key'
+#--- main.s
+  .globl main
+  .type main, %function
+main:
+  .cfi_startproc
+  cmpl	$0x0, %eax
+  jne .LBB0
+.LBB0:
+  retq
+  .cfi_endproc
+.size main, .-main
+#--- profile.yaml
+---
+header:
+  profile-version: 1
+  binary-name:     'yaml-multiple-profiles.test.tmp.exe'
+  binary-build-id: '<unknown>'
+  profile-flags:   [ lbr ]
+  profile-origin:  branch profile reader
+  profile-events:  ''
+  dfs-order:       false
+  unknown-header-key: true
+functions:
+  - name:            'main'
+    fid:             1
+    hash:            0x50BBA3441D436491
+    exec:            1
+    nblocks:         1
+    blocks:
+      - bid:             0
+        insns:           2
+        hash:            0x4D4D8FAF7D4C0000
+        succ:            [ { bid: 1, cnt: 0, unknown_succ_key: 0x10 } ]
+        unknown_block_key: [ ]
+    unknown_function_key: 1
+unknown_toplev_key: ''
+...

bolt/tools/merge-fdata/merge-fdata.cpp

@@ -392,6 +392,7 @@ int main(int argc, char **argv) {
     if (std::error_code EC = MB.getError())
       report_error(InputDataFilename, EC);
     yaml::Input YamlInput(MB.get()->getBuffer());
+    YamlInput.setAllowUnknownKeys(true);
 
     errs() << "Merging data from " << InputDataFilename << "...\n";