Fix missing basedirs implementation for multilevel, add tests

Author: Felixoid

src/cache/multilevel.rs

@@ -286,31 +286,42 @@ pub struct MultiLevelStorage {
     write_policy: WritePolicy,
     /// Lock-free atomic statistics per level
     atomic_stats: Vec<Arc<AtomicLevelStats>>,
+    /// Base directories for path normalization, propagated to compiler pipeline
+    basedirs: Vec<Vec<u8>>,
 }
 
 impl MultiLevelStorage {
+    /// Collect and deduplicate basedirs from all cache levels.
+    fn collect_basedirs(levels: &[Arc<dyn Storage>]) -> Vec<Vec<u8>> {
+        let mut seen = Vec::new();
+        for level in levels {
+            for basedir in level.basedirs() {
+                if !seen.contains(basedir) {
+                    seen.push(basedir.clone());
+                }
+            }
+        }
+        seen
+    }
+
     /// Create a new multi-level storage from a list of storage backends.
     ///
     /// Levels are checked in order (L0, L1, L2, ...) during reads.
     /// All levels receive writes in parallel.
     pub fn new(levels: Vec<Arc<dyn Storage>>) -> Self {
-        let atomic_stats = AtomicLevelStats::from_levels(&levels);
-
-        MultiLevelStorage {
-            levels,
-            write_policy: WritePolicy::default(),
-            atomic_stats,
-        }
+        Self::with_write_policy(levels, WritePolicy::default())
     }
 
     /// Create a new multi-level storage with explicit write policy.
     pub fn with_write_policy(levels: Vec<Arc<dyn Storage>>, write_policy: WritePolicy) -> Self {
         let atomic_stats = AtomicLevelStats::from_levels(&levels);
+        let basedirs = Self::collect_basedirs(&levels);
 
         MultiLevelStorage {
             levels,
             write_policy,
             atomic_stats,
+            basedirs,
         }
     }
 
@@ -837,6 +848,10 @@ impl Storage for MultiLevelStorage {
             .unwrap_or_default()
     }
 
+    fn basedirs(&self) -> &[Vec<u8>] {
+        &self.basedirs
+    }
+
     async fn get_preprocessor_cache_entry(
         &self,
         key: &str,

tests/integration/scripts/test-basedirs.sh

@@ -116,6 +116,129 @@ test_backend "webdav" \
     "SCCACHE_WEBDAV_USERNAME=bar" \
     "SCCACHE_WEBDAV_PASSWORD=baz"
 
+# Function to test basedirs with multi-level cache (disk + remote)
+# Tests that basedirs normalization works across cache levels and backfill
+test_multilevel_backend() {
+    local backend_name="$1"
+    local level_name="$backend_name"
+    if [ "$backend_name" = "azblob" ]; then
+        level_name="azure"
+    fi
+    shift
+    cp -r /sccache/tests/integration/basedirs-autotools /build/dir1
+    cp -r /sccache/tests/integration/basedirs-autotools /build/dir2
+
+    echo ""
+    echo "=========================================="
+    echo "Testing multilevel basedirs: disk + $backend_name"
+    echo "=========================================="
+
+    # Stop any running sccache server
+    "$SCCACHE" --stop-server 2>/dev/null || true
+
+    # Set backend-specific environment variables (passed as arguments)
+    for env_var in "$@"; do
+        export "${env_var?}"
+    done
+
+    # Configure basedirs and multi-level cache
+    export SCCACHE_BASEDIRS="/build/dir1:/build/dir2"
+    export SCCACHE_MULTILEVEL_CHAIN="disk,$level_name"
+    export SCCACHE_DIR="/build/sccache-ml-basedirs"
+    rm -rf /build/sccache-ml-basedirs
+    mkdir -p /build/sccache-ml-basedirs
+
+    # Start sccache server
+    "$SCCACHE" --start-server
+
+    # Verify multi-level is active
+    STATS_JSON=$("$SCCACHE" --show-stats --stats-format=json)
+    CACHE_LOCATION=$(echo "$STATS_JSON" | python3 -c "import sys, json; print(json.load(sys.stdin).get('cache_location', ''))" || echo "unknown")
+    echo "Cache location: $CACHE_LOCATION"
+
+    if ! echo "$CACHE_LOCATION" | grep -qi "Multi-level"; then
+        echo "✗ FAIL: Multi-level cache not detected in cache_location: $CACHE_LOCATION"
+        exit 1
+    fi
+
+    echo "Test 1: Compile from first directory (cache miss, populates L0 disk + L1 $backend_name)"
+    autotools /build/dir1
+
+    STATS_JSON=$("$SCCACHE" --show-stats --stats-format=json)
+    FIRST_MISSES=$(echo "$STATS_JSON" | python3 -c "import sys, json; stats = json.load(sys.stdin).get('stats', {}); print(stats.get('cache_misses', {}).get('counts', {}).get('C/C++', 0))")
+    echo "Cache misses after first build: $FIRST_MISSES"
+
+    echo ""
+    echo "Test 2: Compile from second directory (cache hit expected via basedirs)"
+    autotools /build/dir2
+
+    STATS_JSON=$("$SCCACHE" --show-stats --stats-format=json)
+    CACHE_HITS=$(echo "$STATS_JSON" | python3 -c "import sys, json; stats = json.load(sys.stdin).get('stats', {}); print(stats.get('cache_hits', {}).get('counts', {}).get('C/C++', 0))")
+    SECOND_MISSES=$(echo "$STATS_JSON" | python3 -c "import sys, json; stats = json.load(sys.stdin).get('stats', {}); print(stats.get('cache_misses', {}).get('counts', {}).get('C/C++', 0))")
+    echo "Cache hits: $CACHE_HITS, misses: $SECOND_MISSES (first build: $FIRST_MISSES)"
+
+    if [ "$FIRST_MISSES" != "$SECOND_MISSES" ]; then
+        echo "✗ FAIL: multilevel disk+$backend_name - Cache misses increased from $FIRST_MISSES to $SECOND_MISSES"
+        echo "$STATS_JSON" | python3 -m json.tool
+        exit 1
+    fi
+
+    echo ""
+    echo "Test 3: Clear L0 (disk), rebuild from dir1 (should hit L1 $backend_name and backfill)"
+    "$SCCACHE" --stop-server 2>/dev/null || true
+    rm -rf /build/sccache-ml-basedirs
+    mkdir -p /build/sccache-ml-basedirs
+    rm -rf /build/dir1
+    cp -r /sccache/tests/integration/basedirs-autotools /build/dir1
+    "$SCCACHE" --start-server
+
+    autotools /build/dir1
+
+    STATS_JSON=$("$SCCACHE" --show-stats --stats-format=json)
+    THIRD_MISSES=$(echo "$STATS_JSON" | python3 -c "import sys, json; stats = json.load(sys.stdin).get('stats', {}); print(stats.get('cache_misses', {}).get('counts', {}).get('C/C++', 0))")
+    echo "Cache misses after L0 clear and rebuild: $THIRD_MISSES (should be 0)"
+
+    if [ "$THIRD_MISSES" -gt 0 ]; then
+        echo "✗ FAIL: multilevel disk+$backend_name - Misses after L0 clear ($THIRD_MISSES), L1 should have served data"
+        echo "$STATS_JSON" | python3 -m json.tool
+        exit 1
+    fi
+
+    # Give backfill time to complete
+    sleep 2
+
+    echo ""
+    echo "Test 4: Rebuild from dir2 (should hit backfilled L0 via basedirs)"
+    rm -rf /build/dir2
+    cp -r /sccache/tests/integration/basedirs-autotools /build/dir2
+    autotools /build/dir2
+
+    STATS_JSON=$("$SCCACHE" --show-stats --stats-format=json)
+    FOURTH_MISSES=$(echo "$STATS_JSON" | python3 -c "import sys, json; stats = json.load(sys.stdin).get('stats', {}); print(stats.get('cache_misses', {}).get('counts', {}).get('C/C++', 0))")
+
+    if [ "$FOURTH_MISSES" -gt 0 ]; then
+        echo "✗ FAIL: multilevel disk+$backend_name - Misses on build 4, basedirs + backfill should provide hits"
+        echo "$STATS_JSON" | python3 -m json.tool
+        exit 1
+    fi
+
+    echo "✓ PASS: multilevel disk+$backend_name - Basedirs + multilevel + backfill all working"
+
+    # Clean up for next test
+    rm -rf /build/dir1 /build/dir2 /build/sccache-ml-basedirs
+    "$SCCACHE" --stop-server &>/dev/null || true
+
+    # Unset environment variables
+    for env_var in "$@"; do
+        VAR_NAME="${env_var%%=*}"
+        unset "$VAR_NAME"
+    done
+    unset SCCACHE_BASEDIRS SCCACHE_MULTILEVEL_CHAIN SCCACHE_DIR
+}
+
+# Test multilevel basedirs with redis
+test_multilevel_backend "redis" "SCCACHE_REDIS_ENDPOINT=tcp://redis:6379"
+
 echo ""
 echo "=========================================="
 echo "All basedir tests completed successfully!"