Merge remote-tracking branch 'upstream/sycl' into georgi/update-sycl-…

…supported-arch-docs

.github/CODEOWNERS

@@ -36,6 +36,7 @@ sycl/doc/extensions/ @intel/dpcpp-specification-reviewers
 
 # Unified Runtime
 sycl/cmake/modules/FetchUnifiedRuntime.cmake @intel/unified-runtime-reviewers
+sycl/cmake/modules/UnifiedRuntimeTag.cmake @intel/unified-runtime-reviewers
 sycl/include/sycl/detail/ur.hpp @intel/unified-runtime-reviewers
 sycl/source/detail/posix_ur.cpp @intel/unified-runtime-reviewers
 sycl/source/detail/ur.cpp @intel/unified-runtime-reviewers

.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/**
@@ -1008,3 +1008,8 @@ bazel:
 
 offload:
   - offload/**
+
+tablegen:
+  - llvm/include/TableGen/**
+  - llvm/lib/TableGen/**
+  - llvm/utils/TableGen/**

.github/workflows/commit-access-review.py

@@ -358,11 +358,10 @@ def main():
     gh = github.Github(login_or_token=token)
     org = gh.get_organization("llvm")
     repo = org.get_repo("llvm-project")
-    team = org.get_team_by_slug("llvm-committers")
     one_year_ago = datetime.datetime.now() - datetime.timedelta(days=365)
     triage_list = {}
-    for member in team.get_members():
-        triage_list[member.login] = User(member.login, triage_list)
+    for collaborator in repo.get_collaborators(permission="push"):
+        triage_list[collaborator.login] = User(collaborator.login, triage_list)
 
     print("Start:", len(triage_list), "triagers")
     # Step 0 Check if users have requested commit access in the last year.

.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

@@ -7,6 +7,9 @@ on:
   pull_request:
     branches:
       - main
+      - sycl
+      - sycl-devops-pr/**
+      - sycl-rel-**
       - 'users/**'
 
 jobs:

.github/workflows/release-binaries.yml

@@ -442,11 +442,22 @@ jobs:
         name: ${{ needs.prepare.outputs.release-binary-filename }}-attestation
         path: ${{ needs.prepare.outputs.release-binary-filename }}.jsonl
 
+    - name: Checkout Release Scripts
+      uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
+      with:
+        sparse-checkout: |
+          llvm/utils/release/github-upload-release.py
+          llvm/utils/git/requirements.txt
+        sparse-checkout-cone-mode: false
+
+    - name: Install Python Requirements
+      run: |
+        pip install --require-hashes -r ./llvm/utils/git/requirements.txt
+
     - name: Upload Release
       shell: bash
       run: |
-        sudo apt install python3-github
-        ./llvm-project/llvm/utils/release/github-upload-release.py \
+        ./llvm/utils/release/github-upload-release.py \
         --token ${{ github.token }} \
         --release ${{ needs.prepare.outputs.release-version }} \
         upload \

.github/workflows/sycl-containers.yaml

@@ -51,6 +51,13 @@ jobs:
             file: ubuntu2204_intel_drivers
             tag: unstable
             build_args: "use_latest=true"
+          - name: Build + Intel Drivers Ubuntu 22.04 Docker image
+            file: ubuntu2204_intel_drivers
+            tag: alldeps
+            build_args: |
+              base_image=ghcr.io/intel/llvm/ubuntu2204_build
+              base_tag=latest
+              use_latest=false
     steps:
       - name: Checkout
         uses: actions/checkout@v4

.github/workflows/sycl-linux-run-tests.yml

@@ -185,8 +185,8 @@ jobs:
       with:
         path: khronos_sycl_cts
         repository: 'KhronosGroup/SYCL-CTS'
-        ref: 'SYCL-2020'
-        default_branch: 'SYCL-2020'
+        ref: 'main'
+        default_branch: 'main'
         cache_path: "/__w/repo_cache/"
     - name: SYCL CTS GIT submodules init
       if: inputs.tests_selector == 'cts'

.gitignore

@@ -30,6 +30,9 @@
 /build*
 !buildbot
 
+#pulldown temp folder
+/spirv-patches-*
+
 #==============================================================================#
 # Explicit files to ignore (only matches one).
 #==============================================================================#

bolt/include/bolt/Profile/ProfileYAMLMapping.h

@@ -95,24 +95,29 @@ template <> struct MappingTraits<bolt::SuccessorInfo> {
 
 namespace bolt {
 struct PseudoProbeInfo {
-  llvm::yaml::Hex64 GUID;
-  uint64_t Index;
-  uint8_t Type;
+  uint32_t InlineTreeIndex = 0;
+  uint64_t BlockMask = 0;            // bitset with probe indices from 1 to 64
+  std::vector<uint64_t> BlockProbes; // block probes with indices above 64
+  std::vector<uint64_t> CallProbes;
+  std::vector<uint64_t> IndCallProbes;
+  std::vector<uint32_t> InlineTreeNodes;
 
   bool operator==(const PseudoProbeInfo &Other) const {
-    return GUID == Other.GUID && Index == Other.Index;
-  }
-  bool operator!=(const PseudoProbeInfo &Other) const {
-    return !(*this == Other);
+    return InlineTreeIndex == Other.InlineTreeIndex &&
+           BlockProbes == Other.BlockProbes && CallProbes == Other.CallProbes &&
+           IndCallProbes == Other.IndCallProbes;
   }
 };
 } // end namespace bolt
 
 template <> struct MappingTraits<bolt::PseudoProbeInfo> {
   static void mapping(IO &YamlIO, bolt::PseudoProbeInfo &PI) {
-    YamlIO.mapRequired("guid", PI.GUID);
-    YamlIO.mapRequired("id", PI.Index);
-    YamlIO.mapRequired("type", PI.Type);
+    YamlIO.mapOptional("blx", PI.BlockMask, 0);
+    YamlIO.mapOptional("blk", PI.BlockProbes, std::vector<uint64_t>());
+    YamlIO.mapOptional("call", PI.CallProbes, std::vector<uint64_t>());
+    YamlIO.mapOptional("icall", PI.IndCallProbes, std::vector<uint64_t>());
+    YamlIO.mapOptional("id", PI.InlineTreeIndex, 0);
+    YamlIO.mapOptional("ids", PI.InlineTreeNodes, std::vector<uint32_t>());
   }
 
   static const bool flow = true;
@@ -158,15 +163,35 @@ template <> struct MappingTraits<bolt::BinaryBasicBlockProfile> {
                        std::vector<bolt::CallSiteInfo>());
     YamlIO.mapOptional("succ", BBP.Successors,
                        std::vector<bolt::SuccessorInfo>());
-    YamlIO.mapOptional("pseudo_probes", BBP.PseudoProbes,
+    YamlIO.mapOptional("probes", BBP.PseudoProbes,
                        std::vector<bolt::PseudoProbeInfo>());
   }
 };
 
+namespace bolt {
+struct InlineTreeNode {
+  uint32_t ParentIndexDelta;
+  uint32_t CallSiteProbe;
+  // Index in PseudoProbeDesc.GUID, UINT32_MAX for same as previous (omitted)
+  uint32_t GUIDIndex;
+  bool operator==(const InlineTreeNode &) const { return false; }
+};
+} // end namespace bolt
+
+template <> struct MappingTraits<bolt::InlineTreeNode> {
+  static void mapping(IO &YamlIO, bolt::InlineTreeNode &ITI) {
+    YamlIO.mapOptional("g", ITI.GUIDIndex, UINT32_MAX);
+    YamlIO.mapOptional("p", ITI.ParentIndexDelta, 0);
+    YamlIO.mapOptional("cs", ITI.CallSiteProbe, 0);
+  }
+
+  static const bool flow = true;
+};
 } // end namespace yaml
 } // end namespace llvm
 
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::yaml::bolt::BinaryBasicBlockProfile)
+LLVM_YAML_IS_FLOW_SEQUENCE_VECTOR(llvm::yaml::bolt::InlineTreeNode)
 
 namespace llvm {
 namespace yaml {
@@ -179,8 +204,7 @@ struct BinaryFunctionProfile {
   llvm::yaml::Hex64 Hash{0};
   uint64_t ExecCount{0};
   std::vector<BinaryBasicBlockProfile> Blocks;
-  llvm::yaml::Hex64 GUID{0};
-  llvm::yaml::Hex64 PseudoProbeDescHash{0};
+  std::vector<InlineTreeNode> InlineTree;
   bool Used{false};
 };
 } // end namespace bolt
@@ -194,9 +218,8 @@ template <> struct MappingTraits<bolt::BinaryFunctionProfile> {
     YamlIO.mapRequired("nblocks", BFP.NumBasicBlocks);
     YamlIO.mapOptional("blocks", BFP.Blocks,
                        std::vector<bolt::BinaryBasicBlockProfile>());
-    YamlIO.mapOptional("guid", BFP.GUID, (uint64_t)0);
-    YamlIO.mapOptional("pseudo_probe_desc_hash", BFP.PseudoProbeDescHash,
-                       (uint64_t)0);
+    YamlIO.mapOptional("inline_tree", BFP.InlineTree,
+                       std::vector<bolt::InlineTreeNode>());
   }
 };
 
@@ -246,10 +269,33 @@ template <> struct MappingTraits<bolt::BinaryProfileHeader> {
   }
 };
 
+namespace bolt {
+struct ProfilePseudoProbeDesc {
+  std::vector<Hex64> GUID;
+  std::vector<Hex64> Hash;
+  std::vector<uint32_t> GUIDHashIdx; // Index of hash for that GUID in Hash
+
+  bool operator==(const ProfilePseudoProbeDesc &Other) const {
+    // Only treat empty Desc as equal
+    return GUID.empty() && Other.GUID.empty() && Hash.empty() &&
+           Other.Hash.empty() && GUIDHashIdx.empty() &&
+           Other.GUIDHashIdx.empty();
+  }
+};
+} // end namespace bolt
+
+template <> struct MappingTraits<bolt::ProfilePseudoProbeDesc> {
+  static void mapping(IO &YamlIO, bolt::ProfilePseudoProbeDesc &PD) {
+    YamlIO.mapRequired("gs", PD.GUID);
+    YamlIO.mapRequired("gh", PD.GUIDHashIdx);
+    YamlIO.mapRequired("hs", PD.Hash);
+  }
+};
 } // end namespace yaml
 } // end namespace llvm
 
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::yaml::bolt::BinaryFunctionProfile)
+LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::yaml::bolt::ProfilePseudoProbeDesc)
 
 namespace llvm {
 namespace yaml {
@@ -258,13 +304,16 @@ namespace bolt {
 struct BinaryProfile {
   BinaryProfileHeader Header;
   std::vector<BinaryFunctionProfile> Functions;
+  ProfilePseudoProbeDesc PseudoProbeDesc;
 };
 } // namespace bolt
 
 template <> struct MappingTraits<bolt::BinaryProfile> {
   static void mapping(IO &YamlIO, bolt::BinaryProfile &BP) {
     YamlIO.mapRequired("header", BP.Header);
     YamlIO.mapRequired("functions", BP.Functions);
+    YamlIO.mapOptional("pseudo_probe_desc", BP.PseudoProbeDesc,
+                       bolt::ProfilePseudoProbeDesc());
   }
 };
 

bolt/include/bolt/Profile/YAMLProfileWriter.h

@@ -32,8 +32,27 @@ class YAMLProfileWriter {
   /// Save execution profile for that instance.
   std::error_code writeProfile(const RewriteInstance &RI);
 
+  using InlineTreeMapTy =
+      DenseMap<const MCDecodedPseudoProbeInlineTree *, uint32_t>;
+  struct InlineTreeDesc {
+    template <typename T> using GUIDMapTy = std::unordered_map<uint64_t, T>;
+    using GUIDNodeMap = GUIDMapTy<const MCDecodedPseudoProbeInlineTree *>;
+    using GUIDNumMap = GUIDMapTy<uint32_t>;
+    GUIDNodeMap TopLevelGUIDToInlineTree;
+    GUIDNumMap GUIDIdxMap;
+    GUIDNumMap HashIdxMap;
+  };
+
+  static std::tuple<std::vector<yaml::bolt::InlineTreeNode>, InlineTreeMapTy>
+  convertBFInlineTree(const MCPseudoProbeDecoder &Decoder,
+                      const InlineTreeDesc &InlineTree, uint64_t GUID);
+
+  static std::tuple<yaml::bolt::ProfilePseudoProbeDesc, InlineTreeDesc>
+  convertPseudoProbeDesc(const MCPseudoProbeDecoder &PseudoProbeDecoder);
+
   static yaml::bolt::BinaryFunctionProfile
   convert(const BinaryFunction &BF, bool UseDFS,
+          const InlineTreeDesc &InlineTree,
           const BoltAddressTranslation *BAT = nullptr);
 
   /// Set CallSiteInfo destination fields from \p Symbol and return a target
@@ -42,8 +61,39 @@ class YAMLProfileWriter {
   setCSIDestination(const BinaryContext &BC, yaml::bolt::CallSiteInfo &CSI,
                     const MCSymbol *Symbol, const BoltAddressTranslation *BAT,
                     uint32_t Offset = 0);
-};
 
+private:
+  struct InlineTreeNode {
+    const MCDecodedPseudoProbeInlineTree *InlineTree;
+    uint64_t GUID;
+    uint64_t Hash;
+    uint32_t ParentId;
+    uint32_t InlineSite;
+  };
+  static std::vector<InlineTreeNode>
+  collectInlineTree(const MCPseudoProbeDecoder &Decoder,
+                    const MCDecodedPseudoProbeInlineTree &Root);
+
+  // 0 - block probe, 1 - indirect call, 2 - direct call
+  using ProbeList = std::array<SmallVector<uint64_t, 0>, 3>;
+  using NodeIdToProbes = DenseMap<uint32_t, ProbeList>;
+  static std::vector<yaml::bolt::PseudoProbeInfo>
+  convertNodeProbes(NodeIdToProbes &NodeProbes);
+
+public:
+  template <typename T>
+  static std::vector<yaml::bolt::PseudoProbeInfo>
+  writeBlockProbes(T Probes, const InlineTreeMapTy &InlineTreeNodeId) {
+    NodeIdToProbes NodeProbes;
+    for (const MCDecodedPseudoProbe &Probe : Probes) {
+      auto It = InlineTreeNodeId.find(Probe.getInlineTreeNode());
+      if (It == InlineTreeNodeId.end())
+        continue;
+      NodeProbes[It->second][Probe.getType()].emplace_back(Probe.getIndex());
+    }
+    return convertNodeProbes(NodeProbes);
+  }
+};
 } // namespace bolt
 } // namespace llvm
 

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/Passes/ADRRelaxationPass.cpp

@@ -59,10 +59,15 @@ void ADRRelaxationPass::runOnFunction(BinaryFunction &BF) {
       // Don't relax adr if it points to the same function and it is not split
       // and BF initial size is < 1MB.
       const unsigned OneMB = 0x100000;
-      if (!BF.isSplit() && BF.getSize() < OneMB) {
+      if (BF.getSize() < OneMB) {
         BinaryFunction *TargetBF = BC.getFunctionForSymbol(Symbol);
-        if (TargetBF && TargetBF == &BF)
+        if (TargetBF == &BF && !BF.isSplit())
           continue;
+        // No relaxation needed if ADR references a basic block in the same
+        // fragment.
+        if (BinaryBasicBlock *TargetBB = BF.getBasicBlockForLabel(Symbol))
+          if (BB.getFragmentNum() == TargetBB->getFragmentNum())
+            continue;
       }
 
       MCPhysReg Reg;