Merge remote-tracking branch 'origin/main' into lv-uniform-tailfold

Author: fhahn

.github/workflows/issue-write.yml

@@ -24,14 +24,21 @@ jobs:
         github.event.workflow_run.conclusion == 'failure'
       )
     steps:
+      - name: Fetch Sources
+        uses: actions/checkout@v4
+        with:
+          sparse-checkout: |
+            .github/workflows/unprivileged-download-artifact/action.yml
+          sparse-checkout-cone-mode: false
       - name: 'Download artifact'
-        uses: actions/download-artifact@6b208ae046db98c579e8a3aa621ab581ff575935 # v4.1.1
+        uses: ./.github/workflows/unprivileged-download-artifact
+        id: download-artifact
         with:
-          github-token: ${{ secrets.ISSUE_WRITE_DOWNLOAD_ARTIFACT }}
           run-id: ${{ github.event.workflow_run.id }}
-          name: workflow-args
+          artifact-name: workflow-args
 
       - name: 'Comment on PR'
+        if: steps.download-artifact.outputs.artifact-id != ''
         uses: actions/github-script@v3
         with:
           github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -144,5 +151,7 @@ jobs:
             });
 
       - name: Dump comments file
-        if: always()
+        if: >-
+          always() &&
+          steps.download-artifact.outputs.artifact-id != ''
         run: cat comments

.github/workflows/unprivileged-download-artifact/action.yml

@@ -0,0 +1,81 @@
+name: Unprivileged Download Artifact
+description: >-
+  Download artifacts from another workflow run without using an access token.
+inputs:
+  run-id:
+    description: >-
+      The run-id for the workflow run that you want to download the artifact
+      from.  If ommitted it will download the most recently created artifact
+      from the repo with the artifact-name.
+    required: false
+  artifact-name:
+    desciption: The name of the artifact to download.
+    required: true
+
+
+outputs:
+  filename:
+    description: >-
+      The filename of the downloaded artifact or the empty string if the
+      artifact was not found.
+    value: ${{ steps.download-artifact.outputs.filename }}
+  artifact-id:
+    description: "The id of the artifact being downloaded."
+    value: ${{ steps.artifact-url.outputs.id }}
+
+
+runs:
+  using: "composite"
+  steps:
+    - uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea #v7.0.1
+      id: artifact-url
+      with:
+        script: |
+          var response;
+          if (!"${{ inputs.run-id }}") {
+            response = await github.rest.actions.listArtifactsForRepo({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              name: "${{ inputs.artifact-name }}"
+            })
+          } else {
+            response = await github.rest.actions.listWorkflowRunArtifacts({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              run_id: "${{ inputs.run-id }}",
+              name: "${{ inputs.artifact-name }}"
+            })
+          }
+
+          console.log(response)
+
+          for (artifact of response.data.artifacts) {
+            console.log(artifact);
+          }
+
+          if (response.data.artifacts.length == 0) {
+            console.log("Could not find artifact ${{ inputs.artifact-name }} for workflow run ${{ inputs.run-id }}")
+            return;
+          }
+
+          const url_response = await github.rest.actions.downloadArtifact({
+            owner: context.repo.owner,
+            repo: context.repo.repo,
+            artifact_id: response.data.artifacts[0].id,
+            archive_format: "zip"
+          })
+
+          core.setOutput("url", url_response.url);
+          core.setOutput("id", response.data.artifacts[0].id);
+
+    - shell: bash
+      if: steps.artifact-url.outputs.url != ''
+      id: download-artifact
+      run: |
+        curl -L -o ${{ inputs.artifact-name }}.zip "${{ steps.artifact-url.outputs.url }}"
+        echo "filename=${{ inputs.artifact-name }}.zip" >> $GITHUB_OUTPUT
+
+    - shell: bash
+      if: steps.download-artifact.outputs.filename != ''
+      run: |
+        unzip ${{ steps.download-artifact.outputs.filename }}

a-abfdec1d.o.tmp

@@ -113,11 +113,6 @@
 
   Prints out offsets for abbrev and debug_info of Skeleton CUs that get patched.
 
-- `--deterministic-debuginfo`
-
-  Disables parallel execution of tasks that may produce nondeterministic debug
-  info
-
 - `--dot-tooltip-code`
 
   Add basic block instructions as tool tips on nodes
@@ -686,6 +681,10 @@
   threshold means fewer functions to process. E.g threshold of 90 means only top
   10 percent of functions with profile will be processed.
 
+- `--match-with-call-graph`
+
+  Match functions with call graph
+
 - `--memcpy1-spec=<func1,func2:cs1:cs2,func3:cs1,...>`
 
   List of functions with call sites for which to specialize memcpy() for size 1

bolt/docs/CommandLineArgumentReference.md

@@ -49,6 +49,7 @@ $ cd ${TOPLEV}/stage3
 $ CPATH=${TOPLEV}/stage2-prof-use-lto/install/bin/
 $ cmake -G Ninja ${TOPLEV}/llvm -DLLVM_TARGETS_TO_BUILD=X86 -DCMAKE_BUILD_TYPE=Release \
     -DCMAKE_C_COMPILER=$CPATH/clang -DCMAKE_CXX_COMPILER=$CPATH/clang++ \
+    -DLLVM_ENABLE_PROJECTS="clang" \
     -DLLVM_USE_LINKER=lld -DCMAKE_INSTALL_PREFIX=${TOPLEV}/stage3/install
 $ perf record -e cycles:u -j any,u -- ninja clang
 ```

bolt/docs/OptimizingClang.md

@@ -45,7 +45,7 @@ def parse_bolt_options(output):
         cleaned_line = line.strip()
 
         if cleaned_line.casefold() in map(str.casefold, section_headers):
-            if prev_section != None:  # Save last option from prev section
+            if prev_section is not None:  # Save last option from prev section
                 add_info(sections, current_section, option, description)
                 option, description = None, []
 
@@ -76,7 +76,7 @@ def parse_bolt_options(output):
                 description = [descr]
                 if option.startswith("--print") or option.startswith("--time"):
                     current_section = "BOLT printing options:"
-                elif prev_section != None:
+                elif prev_section is not None:
                     current_section = prev_section
             continue
 

bolt/docs/generate_doc.py

@@ -43,6 +43,59 @@ class YAMLProfileReader : public ProfileReaderBase {
   using ProfileLookupMap =
       DenseMap<uint32_t, yaml::bolt::BinaryFunctionProfile *>;
 
+  /// A class for matching binary functions in functions in the YAML profile.
+  /// First, a call graph is constructed for both profiled and binary functions.
+  /// Then functions are hashed based on the names of their callee/caller
+  /// functions. Finally, functions are matched based on these neighbor hashes.
+  class CallGraphMatcher {
+  public:
+    /// Constructs the call graphs for binary and profiled functions and
+    /// computes neighbor hashes for binary functions.
+    CallGraphMatcher(BinaryContext &BC, yaml::bolt::BinaryProfile &YamlBP,
+                     ProfileLookupMap &IdToYAMLBF);
+
+    /// Returns the YamlBFs adjacent to the parameter YamlBF in the call graph.
+    std::optional<std::set<yaml::bolt::BinaryFunctionProfile *>>
+    getAdjacentYamlBFs(yaml::bolt::BinaryFunctionProfile &YamlBF) {
+      auto It = YamlBFAdjacencyMap.find(&YamlBF);
+      return It == YamlBFAdjacencyMap.end() ? std::nullopt
+                                            : std::make_optional(It->second);
+    }
+
+    /// Returns the binary functions with the parameter neighbor hash.
+    std::optional<std::vector<BinaryFunction *>>
+    getBFsWithNeighborHash(uint64_t NeighborHash) {
+      auto It = NeighborHashToBFs.find(NeighborHash);
+      return It == NeighborHashToBFs.end() ? std::nullopt
+                                           : std::make_optional(It->second);
+    }
+
+  private:
+    /// Adds edges to the binary function call graph given the callsites of the
+    /// parameter function.
+    void constructBFCG(BinaryContext &BC, yaml::bolt::BinaryProfile &YamlBP);
+
+    /// Using the constructed binary function call graph, computes and creates
+    /// mappings from "neighbor hash" (composed of the function names of callee
+    /// and caller functions of a function) to binary functions.
+    void computeBFNeighborHashes(BinaryContext &BC);
+
+    /// Constructs the call graph for profile functions.
+    void constructYAMLFCG(yaml::bolt::BinaryProfile &YamlBP,
+                          ProfileLookupMap &IdToYAMLBF);
+
+    /// Adjacency map for binary functions in the call graph.
+    DenseMap<BinaryFunction *, std::set<BinaryFunction *>> BFAdjacencyMap;
+
+    /// Maps neighbor hashes to binary functions.
+    DenseMap<uint64_t, std::vector<BinaryFunction *>> NeighborHashToBFs;
+
+    /// Adjacency map for profile functions in the call graph.
+    DenseMap<yaml::bolt::BinaryFunctionProfile *,
+             std::set<yaml::bolt::BinaryFunctionProfile *>>
+        YamlBFAdjacencyMap;
+  };
+
 private:
   /// Adjustments for basic samples profiles (without LBR).
   bool NormalizeByInsnCount{false};
@@ -100,6 +153,9 @@ class YAMLProfileReader : public ProfileReaderBase {
   /// Matches functions using exact hash.
   size_t matchWithHash(BinaryContext &BC);
 
+  /// Matches functions using the call graph.
+  size_t matchWithCallGraph(BinaryContext &BC);
+
   /// Matches functions with similarly named profiled functions.
   size_t matchWithNameSimilarity(BinaryContext &BC);