[Cherrypick] Introduce options for remote cache key scrubbing.

Introduce an experimental flag for remote cache key scrubbing.

This PR introduces the --experimental_remote_scrubbing_config flag, which can
be used to scrub platform-dependent data from the key used to retrieve and
store action results from a remote or disk cache This way, actions executing
on different platforms but targeting the same platform may be able to share
cache entries.

To avoid flag proliferation and make for a nicer API, the scrubbing parameters
are supplied via an external file in text proto format.

This is a simplified implementation of one of the ideas described in [1],
highly influenced by Olivier Notteghem's earlier proposal [2], intended to
provide a simple yet flexible API to enable further experimentation. It must
be used with care, as incorrect settings can compromise build correctness.

[1] https://docs.google.com/document/d/1uMPj2s0TlHSIKSngqOkWJoeqOtKzaxQLtBrRfYif3Lo/edit?usp=sharing
[2] bazelbuild#18669

Closes bazelbuild#19523.

PiperOrigin-RevId: 571947013
Change-Id: Ic8da59946ea6d6811b840aee01548746aab2eba0

Fix previous commit

Resolve conflicts

src/main/java/com/google/devtools/build/lib/actions/cache/PersistentStringIndexer.java

@@ -66,7 +66,7 @@ public PersistentIndexMap(Path mapFile, Path journalFile, Clock clock) throws IO
     @Override
     protected boolean updateJournal() {
       long time = clock.nanoTime();
-      if (SAVE_INTERVAL_NS == 0 || time > nextUpdate) {
+      if (time > nextUpdate) {
         nextUpdate = time + SAVE_INTERVAL_NS;
         return true;
       }

src/main/java/com/google/devtools/build/lib/remote/BUILD

@@ -35,6 +35,7 @@ java_library(
             "AbstractActionInputPrefetcher.java",
             "ToplevelArtifactsDownloader.java",
             "LeaseService.java",
+            "Scrubber.java",
         ],
     ),
     exports = [
@@ -84,6 +85,7 @@ java_library(
         "//src/main/java/com/google/devtools/build/lib/exec/local",
         "//src/main/java/com/google/devtools/build/lib/packages/semantics",
         "//src/main/java/com/google/devtools/build/lib/profiler",
+        "//src/main/java/com/google/devtools/build/lib/remote:scrubber",
         "//src/main/java/com/google/devtools/build/lib/remote/common",
         "//src/main/java/com/google/devtools/build/lib/remote/common:cache_not_found_exception",
         "//src/main/java/com/google/devtools/build/lib/remote/disk",
@@ -239,3 +241,16 @@ java_library(
         "//third_party:jsr305",
     ],
 )
+
+java_library(
+    name = "scrubber",
+    srcs = ["Scrubber.java"],
+    deps = [
+        "//src/main/java/com/google/devtools/build/lib/actions",
+        "//src/main/java/com/google/devtools/build/lib/actions:artifacts",
+        "//src/main/protobuf:remote_scrubbing_java_proto",
+        "//third_party:guava",
+        "//third_party:jsr305",
+        "//third_party/protobuf:protobuf_java",
+    ],
+)

src/main/java/com/google/devtools/build/lib/remote/RemoteExecutionService.java

@@ -86,6 +86,7 @@
 import com.google.devtools.build.lib.remote.RemoteExecutionService.ActionResultMetadata.DirectoryMetadata;
 import com.google.devtools.build.lib.remote.RemoteExecutionService.ActionResultMetadata.FileMetadata;
 import com.google.devtools.build.lib.remote.RemoteExecutionService.ActionResultMetadata.SymlinkMetadata;
+import com.google.devtools.build.lib.remote.Scrubber.SpawnScrubber;
 import com.google.devtools.build.lib.remote.common.BulkTransferException;
 import com.google.devtools.build.lib.remote.common.OperationObserver;
 import com.google.devtools.build.lib.remote.common.OutputDigestMismatchException;
@@ -175,6 +176,8 @@ public class RemoteExecutionService {
   private final AtomicBoolean shutdown = new AtomicBoolean(false);
   private final AtomicBoolean buildInterrupted = new AtomicBoolean(false);
 
+  @Nullable private final Scrubber scrubber;
+
   public RemoteExecutionService(
       Executor executor,
       Reporter reporter,
@@ -205,6 +208,7 @@ public RemoteExecutionService(
     if (remoteOptions.remoteMerkleTreeCacheSize != 0) {
       merkleTreeCacheBuilder.maximumSize(remoteOptions.remoteMerkleTreeCacheSize);
     }
+    this.scrubber = remoteOptions.scrubber;
     this.merkleTreeCache = merkleTreeCacheBuilder.build();
 
     this.tempPathGenerator = tempPathGenerator;
@@ -213,12 +217,13 @@ public RemoteExecutionService(
     this.scheduler = Schedulers.from(executor, /*interruptibleWorker=*/ true);
   }
 
-  static Command buildCommand(
+  private Command buildCommand(
       Collection<? extends ActionInput> outputs,
       List<String> arguments,
       ImmutableMap<String, String> env,
       @Nullable Platform platform,
-      RemotePathResolver remotePathResolver) {
+      RemotePathResolver remotePathResolver,
+      @Nullable SpawnScrubber spawnScrubber) {
     Command.Builder command = Command.newBuilder();
     ArrayList<String> outputFiles = new ArrayList<>();
     ArrayList<String> outputDirectories = new ArrayList<>();
@@ -239,6 +244,9 @@ static Command buildCommand(
       command.setPlatform(platform);
     }
     for (String arg : arguments) {
+      if (spawnScrubber != null) {
+        arg = spawnScrubber.transformArgument(arg);
+      }
       command.addArguments(decodeBytestringUtf8(arg));
     }
     // Sorting the environment pairs by variable name.
@@ -360,15 +368,18 @@ private SortedMap<PathFragment, ActionInput> buildOutputDirMap(Spawn spawn) {
   }
 
   private MerkleTree buildInputMerkleTree(
-      Spawn spawn, SpawnExecutionContext context, ToolSignature toolSignature)
+      Spawn spawn,
+      SpawnExecutionContext context,
+      ToolSignature toolSignature,
+      @Nullable SpawnScrubber spawnScrubber)
       throws IOException, ForbiddenActionInputException {
     // Add output directories to inputs so that they are created as empty directories by the
     // executor. The spec only requires the executor to create the parent directory of an output
     // directory, which differs from the behavior of both local and sandboxed execution.
     SortedMap<PathFragment, ActionInput> outputDirMap = buildOutputDirMap(spawn);
     boolean useMerkleTreeCache = remoteOptions.remoteMerkleTreeCache;
-    if (toolSignature != null) {
-      // Marking tool files is not yet supported in conjunction with the merkle tree cache.
+    if (toolSignature != null || spawnScrubber != null) {
+      // The Merkle tree cache is not yet compatible with scrubbing or marking tool files.
       useMerkleTreeCache = false;
     }
     if (useMerkleTreeCache) {
@@ -377,15 +388,23 @@ private MerkleTree buildInputMerkleTree(
       remotePathResolver.walkInputs(
           spawn,
           context,
-          (Object nodeKey, InputWalker walker) -> {
-            subMerkleTrees.add(
-                buildMerkleTreeVisitor(
-                    nodeKey, walker, metadataProvider, context.getPathResolver()));
-          });
+          (Object nodeKey, InputWalker walker) ->
+              subMerkleTrees.add(
+                  buildMerkleTreeVisitor(
+                      nodeKey,
+                      walker,
+                      metadataProvider,
+                      context.getPathResolver(),
+                      spawnScrubber)));
       if (!outputDirMap.isEmpty()) {
         subMerkleTrees.add(
             MerkleTree.build(
-                outputDirMap, metadataProvider, execRoot, context.getPathResolver(), digestUtil));
+                outputDirMap,
+                metadataProvider,
+                execRoot,
+                context.getPathResolver(),
+                /* spawnScrubber= */ null,
+                digestUtil));
       }
       return MerkleTree.merge(subMerkleTrees, digestUtil);
     } else {
@@ -406,6 +425,7 @@ private MerkleTree buildInputMerkleTree(
           context.getMetadataProvider(),
           execRoot,
           context.getPathResolver(),
+          spawnScrubber,
           digestUtil);
     }
   }
@@ -414,7 +434,8 @@ private MerkleTree buildMerkleTreeVisitor(
       Object nodeKey,
       InputWalker walker,
       MetadataProvider metadataProvider,
-      ArtifactPathResolver artifactPathResolver)
+      ArtifactPathResolver artifactPathResolver,
+      @Nullable SpawnScrubber spawnScrubber)
       throws IOException, ForbiddenActionInputException {
     // Deduplicate concurrent computations for the same node. It's not possible to use
     // MerkleTreeCache#get(key, loader) because the loading computation may cause other nodes to be
@@ -426,7 +447,8 @@ private MerkleTree buildMerkleTreeVisitor(
       // No preexisting cache entry, so we must do the computation ourselves.
       try {
         freshFuture.complete(
-            uncachedBuildMerkleTreeVisitor(walker, metadataProvider, artifactPathResolver));
+            uncachedBuildMerkleTreeVisitor(
+                walker, metadataProvider, artifactPathResolver, spawnScrubber));
       } catch (Exception e) {
         freshFuture.completeExceptionally(e);
       }
@@ -450,7 +472,8 @@ private MerkleTree buildMerkleTreeVisitor(
   public MerkleTree uncachedBuildMerkleTreeVisitor(
       InputWalker walker,
       MetadataProvider metadataProvider,
-      ArtifactPathResolver artifactPathResolver)
+      ArtifactPathResolver artifactPathResolver,
+      @Nullable SpawnScrubber scrubber)
       throws IOException, ForbiddenActionInputException {
     ConcurrentLinkedQueue<MerkleTree> subMerkleTrees = new ConcurrentLinkedQueue<>();
     subMerkleTrees.add(
@@ -459,18 +482,19 @@ public MerkleTree uncachedBuildMerkleTreeVisitor(
             metadataProvider,
             execRoot,
             artifactPathResolver,
+            scrubber,
             digestUtil));
     walker.visitNonLeaves(
         (Object subNodeKey, InputWalker subWalker) -> {
           subMerkleTrees.add(
               buildMerkleTreeVisitor(
-                  subNodeKey, subWalker, metadataProvider, artifactPathResolver));
+                  subNodeKey, subWalker, metadataProvider, artifactPathResolver, scrubber));
         });
     return MerkleTree.merge(subMerkleTrees, digestUtil);
   }
 
   @Nullable
-  private static ByteString buildSalt(Spawn spawn) {
+  private static ByteString buildSalt(Spawn spawn, @Nullable SpawnScrubber spawnScrubber) {
     String workspace =
         spawn.getExecutionInfo().get(ExecutionRequirements.DIFFERENTIATE_WORKSPACE_CACHE);
     if (workspace != null) {
@@ -479,7 +503,15 @@ private static ByteString buildSalt(Spawn spawn) {
               .addProperties(
                   Platform.Property.newBuilder().setName("workspace").setValue(workspace).build())
               .build();
-      return platform.toByteString();
+      ByteString value = platform.toByteString();
+      if (spawnScrubber != null && spawnScrubber.getSalt() != null) {
+        value.concat(ByteString.copyFromUtf8(spawnScrubber.getSalt()));
+      }
+      return value;
+    }
+
+    if (spawnScrubber != null && spawnScrubber.getSalt() != null) {
+        return ByteString.copyFromUtf8(spawnScrubber.getSalt());
     }
 
     return null;
@@ -517,7 +549,9 @@ public RemoteAction buildRemoteAction(Spawn spawn, SpawnExecutionContext context
     remoteActionBuildingSemaphore.acquire();
     try {
       ToolSignature toolSignature = getToolSignature(spawn, context);
-      final MerkleTree merkleTree = buildInputMerkleTree(spawn, context, toolSignature);
+      SpawnScrubber spawnScrubber = scrubber != null ? scrubber.forSpawn(spawn) : null;
+      final MerkleTree merkleTree =
+          buildInputMerkleTree(spawn, context, toolSignature, spawnScrubber);
 
       // Get the remote platform properties.
       Platform platform = PlatformUtils.getPlatformProto(spawn, remoteOptions);
@@ -535,7 +569,8 @@ public RemoteAction buildRemoteAction(Spawn spawn, SpawnExecutionContext context
               spawn.getArguments(),
               spawn.getEnvironment(),
               platform,
-              remotePathResolver);
+              remotePathResolver,
+              spawnScrubber);
       Digest commandHash = digestUtil.compute(command);
       Action action =
           Utils.buildAction(
@@ -544,7 +579,7 @@ public RemoteAction buildRemoteAction(Spawn spawn, SpawnExecutionContext context
               platform,
               context.getTimeout(),
               Spawns.mayBeCachedRemotely(spawn),
-              buildSalt(spawn));
+              buildSalt(spawn, spawnScrubber));
 
       ActionKey actionKey = digestUtil.computeActionKey(action);
 
@@ -1449,7 +1484,8 @@ public void uploadInputsIfNotPresent(RemoteAction action, boolean force)
         Spawn spawn = action.getSpawn();
         SpawnExecutionContext context = action.getSpawnExecutionContext();
         ToolSignature toolSignature = getToolSignature(spawn, context);
-        merkleTree = buildInputMerkleTree(spawn, context, toolSignature);
+        SpawnScrubber spawnScrubber = scrubber != null ? scrubber.forSpawn(spawn) : null;
+        merkleTree = buildInputMerkleTree(spawn, context, toolSignature, spawnScrubber);
       }
 
       remoteExecutionCache.ensureInputsPresent(

src/main/java/com/google/devtools/build/lib/remote/RemoteModule.java

@@ -311,6 +311,14 @@ public void beforeCommand(CommandEnvironment env) throws AbruptExitException {
           FailureDetails.RemoteOptions.Code.EXECUTION_WITH_INVALID_CACHE);
     }
 
+    boolean enableScrubbing = remoteOptions.scrubber != null;
+    if (enableScrubbing && enableRemoteExecution) {
+
+      throw createOptionsExitException(
+          "Cannot combine remote cache key scrubbing with remote execution",
+          FailureDetails.RemoteOptions.Code.EXECUTION_WITH_SCRUBBING);
+    }
+
     // TODO(bazel-team): Consider adding a warning or more validation if the remoteDownloadRegex is
     // used without Build without the Bytes.
     if (!remoteOptions.remoteOutputsMode.downloadAllOutputs()) {