coverage: restore the scripts .gitignore swallowed

.github/workflows/coverage-nightly.yaml

@@ -101,7 +101,8 @@ jobs:
         if: always()
         with:
           name: coverage-raw-${{ matrix.name }}
-          path: coverage-raw/*.profraw
+          # pyln-testing puts the profiles in per-test subdirectories
+          path: coverage-raw/
           if-no-files-found: error
 
   report:
@@ -114,13 +115,13 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v6
 
+      # The raw profile format has no compatibility guarantees across LLVM
+      # releases, so llvm-profdata has to come from the same release as the
+      # clang that produced the profiles: the distro default in both cases.
       - name: Install LLVM tools
         run: |
-          wget https://apt.llvm.org/llvm.sh
-          chmod +x llvm.sh
-          sudo ./llvm.sh 18
-          sudo ln -sf /usr/bin/llvm-profdata-18 /usr/bin/llvm-profdata
-          sudo ln -sf /usr/bin/llvm-cov-18 /usr/bin/llvm-cov
+          sudo apt-get update
+          sudo apt-get install -y llvm
 
       - name: Download build artifact
         uses: actions/download-artifact@v8
@@ -136,28 +137,18 @@ jobs:
           pattern: coverage-raw-*
           path: coverage-artifacts
 
+      # collect-coverage.sh walks the tree itself and rejects the profiles
+      # that crashed tests left half-written, so it gets the whole download.
       - name: Merge coverage data
-        run: |
-          mkdir -p coverage-raw coverage
-          find coverage-artifacts -name "*.profraw" -exec cp {} coverage-raw/ \;
-          PROFRAW_COUNT=$(ls -1 coverage-raw/*.profraw 2>/dev/null | wc -l)
-          echo "Found $PROFRAW_COUNT profile files"
-          if [ "$PROFRAW_COUNT" -eq 0 ]; then
-            echo "ERROR: No coverage data found"
-            exit 1
-          fi
-          chmod +x contrib/coverage/collect-coverage.sh
-          CLN_COVERAGE_DIR=coverage-raw ./contrib/coverage/collect-coverage.sh
+        run: ./contrib/coverage/collect-coverage.sh coverage-artifacts
 
       - name: Generate HTML report
-        run: |
-          chmod +x contrib/coverage/generate-coverage-report.sh
-          ./contrib/coverage/generate-coverage-report.sh
+        run: ./contrib/coverage/generate-coverage-report.sh
 
       - name: Upload to Codecov
         uses: codecov/codecov-action@v6
         with:
-          files: coverage/merged.profdata
+          files: coverage/coverage.lcov
           flags: integration-tests
           name: cln-nightly-coverage
           token: ${{ secrets.CODECOV_TOKEN }}

.gitignore

@@ -23,7 +23,7 @@ gen_*.c
 gen_*.h
 wire/gen_*_csv
 cli/lightning-cli
-coverage
+/coverage
 # Coverage profiling data files
 *.profraw
 *.profdata

contrib/coverage/clang-coverage-report.sh

@@ -0,0 +1,27 @@
+#!/bin/bash -eu
+#
+# Generates an HTML coverage report from a raw Clang coverage profile. See
+# https://clang.llvm.org/docs/SourceBasedCodeCoverage.html for more details.
+# 
+# Example usage to create full_channel.html from full_channel.profraw for the
+# run-full_channel unit test:
+#   ./contrib/clang-coverage-report.sh channeld/test/run-full_channel \
+#       full_channel.profraw full_channel.html
+
+if [[ "$#" -ne 3 ]]; then
+	echo "Usage: $0 BINARY RAW_PROFILE_FILE TARGET_HTML_FILE"
+	exit 1
+fi
+
+readonly BINARY="$1"
+readonly RAW_PROFILE_FILE="$2"
+readonly TARGET_HTML_FILE="$3"
+
+MERGED_PROFILE_FILE=$(mktemp)
+readonly MERGED_PROFILE_FILE
+
+llvm-profdata merge -sparse "${RAW_PROFILE_FILE}" -o "${MERGED_PROFILE_FILE}"
+llvm-cov show "${BINARY}" -instr-profile="${MERGED_PROFILE_FILE}" -format=html \
+	> "${TARGET_HTML_FILE}"
+
+rm "${MERGED_PROFILE_FILE}"

contrib/coverage/cleanup-corrupt-profraw.sh

@@ -0,0 +1,55 @@
+#!/bin/bash -eu
+# Remove corrupt .profraw files from the coverage directory
+# Usage: ./cleanup-corrupt-profraw.sh [COVERAGE_DIR]
+
+COVERAGE_DIR="${1:-${CLN_COVERAGE_DIR:-/tmp/cln-coverage}}"
+
+if [ ! -d "$COVERAGE_DIR" ]; then
+    echo "Coverage directory not found: $COVERAGE_DIR"
+    exit 1
+fi
+
+echo "Scanning for corrupt profraw files in: $COVERAGE_DIR"
+
+# Find all profraw files
+mapfile -t PROFRAW_FILES < <(find "$COVERAGE_DIR" -name "*.profraw" 2>/dev/null || true)
+
+if [ ${#PROFRAW_FILES[@]} -eq 0 ]; then
+    echo "No .profraw files found"
+    exit 0
+fi
+
+echo "Found ${#PROFRAW_FILES[@]} profile files"
+
+CORRUPT_FILES=()
+for profraw in "${PROFRAW_FILES[@]}"; do
+    # Try to validate the file
+    if ! llvm-profdata show "$profraw" >/dev/null 2>&1; then
+        CORRUPT_FILES+=("$profraw")
+    fi
+done
+
+if [ ${#CORRUPT_FILES[@]} -eq 0 ]; then
+    echo "✓ No corrupt files found"
+    exit 0
+fi
+
+echo ""
+echo "Found ${#CORRUPT_FILES[@]} corrupt file(s):"
+for corrupt in "${CORRUPT_FILES[@]}"; do
+    echo "  - $corrupt"
+done
+
+echo ""
+read -p "Delete these corrupt files? [y/N] " -n 1 -r
+echo
+
+if [[ $REPLY =~ ^[Yy]$ ]]; then
+    for corrupt in "${CORRUPT_FILES[@]}"; do
+        rm -f "$corrupt"
+        echo "  Deleted: $corrupt"
+    done
+    echo "✓ Removed ${#CORRUPT_FILES[@]} corrupt file(s)"
+else
+    echo "No files were deleted"
+fi

contrib/coverage/collect-coverage.sh

@@ -0,0 +1,116 @@
+#!/bin/bash -eu
+# Merge all .profraw files into a single .profdata file
+# Usage: ./collect-coverage.sh [COVERAGE_DIR] [OUTPUT_FILE]
+
+COVERAGE_DIR="${1:-${CLN_COVERAGE_DIR:-/tmp/cln-coverage}}"
+OUTPUT="${2:-coverage/merged.profdata}"
+
+echo "Collecting coverage from: $COVERAGE_DIR"
+
+# Find all profraw files
+mapfile -t PROFRAW_FILES < <(find "$COVERAGE_DIR" -name "*.profraw" 2>/dev/null || true)
+
+if [ ${#PROFRAW_FILES[@]} -eq 0 ]; then
+    echo "ERROR: No .profraw files found in $COVERAGE_DIR"
+    exit 1
+fi
+
+echo "Found ${#PROFRAW_FILES[@]} profile files"
+
+# Validate each profraw file and filter out corrupt/incomplete ones
+# Define validation function for parallel execution
+validate_file() {
+    local profraw="$1"
+
+    # Check if file is empty
+    if [ ! -s "$profraw" ]; then
+        return 1  # Empty
+    fi
+
+    # Check if file is suspiciously small (likely incomplete write)
+    # Valid profraw files are typically > 1KB
+    filesize=$(stat -c%s "$profraw" 2>/dev/null || stat -f%z "$profraw" 2>/dev/null)
+    if [ "$filesize" -lt 1024 ]; then
+        return 2  # Too small
+    fi
+
+    # Try to validate the file by checking if llvm-profdata can read it
+    if llvm-profdata show "$profraw" >/dev/null 2>&1; then
+        echo "$profraw"  # Valid - output to stdout
+        return 0
+    else
+        return 3  # Corrupt
+    fi
+}
+
+# Export function for parallel execution
+export -f validate_file
+
+TOTAL=${#PROFRAW_FILES[@]}
+NPROC=$(nproc 2>/dev/null || echo 4)
+echo "Validating ${TOTAL} files in parallel (using ${NPROC} cores)..."
+
+# Run validation in parallel and collect valid files
+mapfile -t VALID_FILES < <(
+    printf '%s\n' "${PROFRAW_FILES[@]}" | \
+    xargs -P "$NPROC" -I {} bash -c 'validate_file "$@"' _ {}
+)
+
+# Calculate error counts
+CORRUPT_COUNT=$((TOTAL - ${#VALID_FILES[@]}))
+
+if [ ${#VALID_FILES[@]} -eq 0 ]; then
+    echo "ERROR: No valid .profraw files found (all $CORRUPT_COUNT files were corrupt/incomplete)"
+    exit 1
+fi
+
+echo "Valid files: ${#VALID_FILES[@]}"
+if [ $CORRUPT_COUNT -gt 0 ]; then
+    echo "Filtered out: $CORRUPT_COUNT files (empty/small/corrupt)"
+fi
+mkdir -p "$(dirname "$OUTPUT")"
+
+# Merge with -sparse flag for efficiency
+# Use batched merging to avoid "Argument list too long" errors
+BATCH_SIZE=500
+TOTAL_FILES=${#VALID_FILES[@]}
+
+if [ "$TOTAL_FILES" -le "$BATCH_SIZE" ]; then
+    # Small enough to merge in one go
+    echo "Merging ${TOTAL_FILES} files..."
+    llvm-profdata merge -sparse "${VALID_FILES[@]}" -o "$OUTPUT"
+else
+    # Need to merge in batches
+    echo "Merging ${TOTAL_FILES} files in batches of ${BATCH_SIZE}..."
+
+    # Create temp directory for intermediate files
+    TEMP_DIR=$(mktemp -d "${TMPDIR:-/tmp}/profdata-merge.XXXXXX")
+    trap 'rm -rf "$TEMP_DIR"' EXIT
+
+    BATCH_NUM=0
+    INTERMEDIATE_FILES=()
+
+    # Merge files in batches
+    for ((i=0; i<TOTAL_FILES; i+=BATCH_SIZE)); do
+        BATCH_NUM=$((BATCH_NUM + 1))
+        END=$((i + BATCH_SIZE))
+        if [ "$END" -gt "$TOTAL_FILES" ]; then
+            END=$TOTAL_FILES
+        fi
+
+        BATCH_FILES=("${VALID_FILES[@]:$i:$BATCH_SIZE}")
+        INTERMEDIATE="$TEMP_DIR/batch-$BATCH_NUM.profdata"
+
+        echo "  Batch $BATCH_NUM: merging files $((i+1))-$END..."
+        llvm-profdata merge -sparse "${BATCH_FILES[@]}" -o "$INTERMEDIATE"
+        INTERMEDIATE_FILES+=("$INTERMEDIATE")
+    done
+
+    # Merge all intermediate files into final output
+    echo "Merging ${#INTERMEDIATE_FILES[@]} intermediate files into final output..."
+    llvm-profdata merge -sparse "${INTERMEDIATE_FILES[@]}" -o "$OUTPUT"
+
+    # Cleanup handled by trap
+fi
+
+echo "✓ Merged profile: $OUTPUT"

contrib/coverage/generate-coverage-report.sh

@@ -0,0 +1,70 @@
+#!/bin/bash -eu
+# Generate HTML and text coverage reports from merged profile data
+# Usage: ./generate-coverage-report.sh [PROFDATA_FILE] [OUTPUT_DIR]
+
+PROFDATA="${1:-coverage/merged.profdata}"
+OUTPUT_DIR="${2:-coverage/html}"
+
+if [ ! -f "$PROFDATA" ]; then
+    echo "ERROR: Profile not found: $PROFDATA"
+    echo "Run collect-coverage.sh first to create the merged profile"
+    exit 1
+fi
+
+# Get all binaries from Makefile (includes plugins, tools, test binaries)
+echo "Discovering instrumented binaries from Makefile..."
+mapfile -t BINARIES < <(make -qp 2>/dev/null | awk '/^ALL_PROGRAMS :=/ {$1=$2=""; print}' | tr ' ' '\n' | grep -v '^$')
+mapfile -t TEST_BINARIES < <(make -qp 2>/dev/null | awk '/^ALL_TEST_PROGRAMS :=/ {$1=$2=""; print}' | tr ' ' '\n' | grep -v '^$')
+
+# Combine all binaries
+ALL_BINARIES=("${BINARIES[@]}" "${TEST_BINARIES[@]}")
+
+# Build llvm-cov arguments
+ARGS=()
+for bin in "${ALL_BINARIES[@]}"; do
+    if [ -f "$bin" ]; then
+        if [ ${#ARGS[@]} -eq 0 ]; then
+            ARGS+=("$bin")  # First binary is primary
+        else
+            ARGS+=("-object=$bin")  # Others use -object=
+        fi
+    fi
+done
+
+if [ ${#ARGS[@]} -eq 0 ]; then
+    echo "ERROR: No instrumented binaries found"
+    echo "Make sure you've built with --enable-coverage"
+    exit 1
+fi
+
+echo "Generating coverage report for ${#ARGS[@]} binaries..."
+
+# Generate HTML report
+llvm-cov show "${ARGS[@]}" \
+    -instr-profile="$PROFDATA" \
+    -format=html \
+    -output-dir="$OUTPUT_DIR" \
+    -show-line-counts-or-regions \
+    -show-instantiations=false
+
+echo "✓ HTML report: $OUTPUT_DIR/index.html"
+
+# Generate text summary
+mkdir -p coverage
+llvm-cov report "${ARGS[@]}" \
+    -instr-profile="$PROFDATA" \
+    | tee coverage/summary.txt
+
+echo "✓ Summary: coverage/summary.txt"
+
+# Machine-readable export for tools that ingest lcov: Codecov, and
+# lcov_cobertura for GitLab's coverage visualization. Both match the paths in
+# the report against the repository tree, so strip the build directory off.
+llvm-cov export "${ARGS[@]}" \
+    -instr-profile="$PROFDATA" \
+    -format=lcov > coverage/coverage.lcov.tmp
+
+sed "s|^SF:$PWD/|SF:|" coverage/coverage.lcov.tmp > coverage/coverage.lcov
+rm -f coverage/coverage.lcov.tmp
+
+echo "✓ LCOV: coverage/coverage.lcov"