outputs,
TimestampGranularityMonitor tsgm) {
this.inputArtifactData = Preconditions.checkNotNull(inputArtifactData);
this.outputs = ImmutableSet.copyOf(outputs);
this.tsgm = tsgm;
}
@Override
public Metadata getMetadataMaybe(Artifact artifact) {
try {
return getMetadata(artifact);
} catch (IOException e) {
return null;
}
}
/**
* Gets the {@link TimestampGranularityMonitor} to use for a given artifact.
*
* If the artifact is of type "constant metadata", this returns null so that changes to such
* artifacts do not tickle the timestamp granularity monitor, delaying the build for no reason.
*
* @param artifact the artifact for which to fetch the timestamp granularity monitor
* @return the timestamp granularity monitor to use, which may be null
*/
private TimestampGranularityMonitor getTimestampGranularityMonitor(Artifact artifact) {
return artifact.isConstantMetadata() ? null : tsgm;
}
private static Metadata metadataFromValue(FileArtifactValue value) throws FileNotFoundException {
if (value == FileArtifactValue.MISSING_FILE_MARKER
|| value == FileArtifactValue.OMITTED_FILE_MARKER) {
throw new FileNotFoundException();
}
// If the file is a directory, we need to return the mtime because the action cache uses mtime
// to determine if this artifact has changed. We want this code path to go away somehow
// for directories (maybe by implementing FileSet in Skyframe).
return value.isFile() ? new Metadata(value.getDigest()) : new Metadata(value.getModifiedTime());
}
@Override
public Metadata getMetadata(Artifact artifact) throws IOException {
Metadata metadata = getRealMetadata(artifact);
return artifact.isConstantMetadata() ? Metadata.CONSTANT_METADATA : metadata;
}
@Nullable
private FileArtifactValue getInputFileArtifactValue(Artifact input) {
if (outputs.contains(input)) {
return null;
}
if (input.hasParent() && outputs.contains(input.getParent())) {
return null;
}
return Preconditions.checkNotNull(inputArtifactData.get(input), input);
}
/**
* Get the real (viz. on-disk) metadata for an Artifact.
* A key assumption is that getRealMetadata() will be called for every Artifact in this
* ActionMetadataHandler, to populate additionalOutputData and outputTreeArtifactData.
*
*
We cache data for constant-metadata artifacts, even though it is technically unnecessary,
* because the data stored in this cache is consumed by various parts of Blaze via the {@link
* ActionExecutionValue} (for now, {@link FilesystemValueChecker} and {@link ArtifactFunction}).
* It is simpler for those parts if every output of the action is present in the cache. However,
* we must not return the actual metadata for a constant-metadata artifact.
*/
private Metadata getRealMetadata(Artifact artifact) throws IOException {
FileArtifactValue value = getInputFileArtifactValue(artifact);
if (value != null) {
return metadataFromValue(value);
}
if (artifact.isSourceArtifact()) {
// A discovered input we didn't have data for.
// TODO(bazel-team): Change this to an assertion once Skyframe has native input discovery, so
// all inputs will already have metadata known.
return null;
} else if (artifact.isMiddlemanArtifact()) {
// A middleman artifact's data was either already injected from the action cache checker using
// #setDigestForVirtualArtifact, or it has the default middleman value.
value = additionalOutputData.get(artifact);
if (value != null) {
return metadataFromValue(value);
}
value = FileArtifactValue.DEFAULT_MIDDLEMAN;
FileArtifactValue oldValue = additionalOutputData.putIfAbsent(artifact, value);
checkInconsistentData(artifact, oldValue, value);
return metadataFromValue(value);
} else if (artifact.isTreeArtifact()) {
TreeArtifactValue setValue = getTreeArtifactValue(artifact);
if (setValue != null && setValue != TreeArtifactValue.MISSING_TREE_ARTIFACT) {
return setValue.getMetadata();
}
// We use FileNotFoundExceptions to determine if an Artifact was or wasn't found.
// Calling code depends on this particular exception.
throw new FileNotFoundException(artifact + " not found");
}
// It's an ordinary artifact.
FileValue fileValue = outputArtifactData.get(artifact);
if (fileValue != null) {
// Non-middleman artifacts should only have additionalOutputData if they have
// outputArtifactData. We don't assert this because of concurrency possibilities, but at least
// we don't check additionalOutputData unless we expect that we might see the artifact there.
value = additionalOutputData.get(artifact);
// If additional output data is present for this artifact, we use it in preference to the
// usual calculation.
if (value != null) {
return metadataFromValue(value);
}
if (!fileValue.exists()) {
throw new FileNotFoundException(artifact.prettyPrint() + " does not exist");
}
return new Metadata(Preconditions.checkNotNull(fileValue.getDigest(), artifact));
}
// We do not cache exceptions besides nonexistence here, because it is unlikely that the file
// will be requested from this cache too many times.
fileValue = constructFileValue(artifact, null);
return maybeStoreAdditionalData(artifact, fileValue, null);
}
/**
* Check that the new {@code data} we just calculated for an {@link Artifact} agrees with the
* {@code oldData} (presumably calculated concurrently), if it was present.
*/
// Not private only because used by SkyframeActionExecutor's metadata handler.
static void checkInconsistentData(Artifact artifact,
@Nullable Object oldData, Object data) throws IOException {
if (oldData != null && !oldData.equals(data)) {
// Another thread checked this file since we looked at the map, and got a different answer
// than we did. Presumably the user modified the file between reads.
throw new IOException("Data for " + artifact.prettyPrint() + " changed to " + data
+ " after it was calculated as " + oldData);
}
}
/**
* See {@link #getAdditionalOutputData} for why we sometimes need to store additional data, even
* for normal (non-middleman) artifacts.
*/
@Nullable
private Metadata maybeStoreAdditionalData(Artifact artifact, FileValue data,
@Nullable byte[] injectedDigest) throws IOException {
if (!data.exists()) {
// Nonexistent files should only occur before executing an action.
throw new FileNotFoundException(artifact.prettyPrint() + " does not exist");
}
boolean isFile = data.isFile();
if (isFile && !artifact.hasParent() && data.getDigest() != null) {
// We do not need to store the FileArtifactValue separately -- the digest is in the file value
// and that is all that is needed for this file's metadata.
return new Metadata(data.getDigest());
}
// Unfortunately, the FileValue does not contain enough information for us to calculate the
// corresponding FileArtifactValue -- either the metadata must use the modified time, which we
// do not expose in the FileValue, or the FileValue didn't store the digest So we store the
// metadata separately.
// Use the FileValue's digest if no digest was injected, or if the file can't be digested.
injectedDigest = injectedDigest != null || !isFile ? injectedDigest : data.getDigest();
FileArtifactValue value =
FileArtifactValue.create(artifact, isFile, isFile ? data.getSize() : 0, injectedDigest);
FileArtifactValue oldValue = additionalOutputData.putIfAbsent(artifact, value);
checkInconsistentData(artifact, oldValue, value);
return metadataFromValue(value);
}
@Override
public void setDigestForVirtualArtifact(Artifact artifact, Md5Digest md5Digest) {
Preconditions.checkArgument(artifact.isMiddlemanArtifact(), artifact);
Preconditions.checkNotNull(md5Digest, artifact);
additionalOutputData.put(
artifact, FileArtifactValue.createProxy(md5Digest.getDigestBytesUnsafe()));
}
private Set getTreeArtifactContents(Artifact artifact) {
Preconditions.checkArgument(artifact.isTreeArtifact(), artifact);
Set contents = outputDirectoryListings.get(artifact);
if (contents == null) {
// Unfortunately, there is no such thing as a ConcurrentHashSet.
contents = Collections.newSetFromMap(new ConcurrentHashMap());
Set oldContents = outputDirectoryListings.putIfAbsent(artifact, contents);
// Avoid a race condition.
if (oldContents != null) {
contents = oldContents;
}
}
return contents;
}
private TreeArtifactValue getTreeArtifactValue(Artifact artifact) throws IOException {
TreeArtifactValue value = outputTreeArtifactData.get(artifact);
if (value != null) {
return value;
}
Set registeredContents = outputDirectoryListings.get(artifact);
if (registeredContents != null) {
// Check that our registered outputs matches on-disk outputs. Only perform this check
// when contents were explicitly registered.
// TODO(bazel-team): Provide a way for actions to register empty TreeArtifacts.
// By the time we're constructing TreeArtifactValues, use of the metadata handler
// should be single threaded and there should be no race condition.
// The current design of ActionMetadataHandler makes this hard to enforce.
Set paths = null;
paths = TreeArtifactValue.explodeDirectory(artifact);
Set diskFiles = ActionInputHelper.asTreeFileArtifacts(artifact, paths);
if (!diskFiles.equals(registeredContents)) {
// There might be more than one error here. We first look for missing output files.
Set missingFiles = Sets.difference(registeredContents, diskFiles);
if (!missingFiles.isEmpty()) {
// Don't throw IOException--getMetadataMaybe() eats them.
// TODO(bazel-team): Report this error in a better way when called by checkOutputs()
// Currently it's hard to report this error without refactoring, since checkOutputs()
// likes to substitute its own error messages upon catching IOException, and falls
// through to unrecoverable error behavior on any other exception.
throw new IOException("Output file " + missingFiles.iterator().next()
+ " was registered, but not present on disk");
}
Set extraFiles = Sets.difference(diskFiles, registeredContents);
// extraFiles cannot be empty
throw new IOException(
"File " + extraFiles.iterator().next().getParentRelativePath()
+ ", present in TreeArtifact " + artifact + ", was not registered");
}
value = constructTreeArtifactValue(registeredContents);
} else {
value = constructTreeArtifactValueFromFilesystem(artifact);
}
TreeArtifactValue oldValue = outputTreeArtifactData.putIfAbsent(artifact, value);
checkInconsistentData(artifact, oldValue, value);
return value;
}
private TreeArtifactValue constructTreeArtifactValue(Collection contents)
throws IOException {
Map values =
Maps.newHashMapWithExpectedSize(contents.size());
for (TreeFileArtifact treeFileArtifact : contents) {
FileArtifactValue cachedValue = additionalOutputData.get(treeFileArtifact);
if (cachedValue == null) {
FileValue fileValue = outputArtifactData.get(treeFileArtifact);
// This is similar to what's present in getRealMetadataForArtifact, except
// we get back the FileValue, not the metadata.
// We do not cache exceptions besides nonexistence here, because it is unlikely that the
// file will be requested from this cache too many times.
if (fileValue == null) {
try {
fileValue = constructFileValue(treeFileArtifact, /*statNoFollow=*/ null);
} catch (FileNotFoundException e) {
String errorMessage = String.format(
"Failed to resolve relative path %s inside TreeArtifact %s. "
+ "The associated file is either missing or is an invalid symlink.",
treeFileArtifact.getParentRelativePath(),
treeFileArtifact.getParent().getExecPathString());
throw new IOException(errorMessage, e);
}
}
// A minor hack: maybeStoreAdditionalData will force the data to be stored
// in additionalOutputData.
maybeStoreAdditionalData(treeFileArtifact, fileValue, null);
cachedValue = Preconditions.checkNotNull(
additionalOutputData.get(treeFileArtifact), treeFileArtifact);
}
values.put(treeFileArtifact, cachedValue);
}
return TreeArtifactValue.create(values);
}
private TreeArtifactValue constructTreeArtifactValueFromFilesystem(Artifact artifact)
throws IOException {
Preconditions.checkState(artifact.isTreeArtifact(), artifact);
if (!artifact.getPath().isDirectory() || artifact.getPath().isSymbolicLink()) {
return TreeArtifactValue.MISSING_TREE_ARTIFACT;
}
Set paths = null;
paths = TreeArtifactValue.explodeDirectory(artifact);
// If you're reading tree artifacts from disk while outputDirectoryListings are being injected,
// something has gone terribly wrong.
Object previousDirectoryListing =
outputDirectoryListings.put(artifact,
Collections.newSetFromMap(new ConcurrentHashMap()));
Preconditions.checkState(previousDirectoryListing == null,
"Race condition while constructing TreArtifactValue: %s, %s",
artifact, previousDirectoryListing);
return constructTreeArtifactValue(ActionInputHelper.asTreeFileArtifacts(artifact, paths));
}
@Override
public void addExpandedTreeOutput(TreeFileArtifact output) {
Set values = getTreeArtifactContents(output.getParent());
values.add(output);
}
@Override
public Iterable getExpandedOutputs(Artifact artifact) {
return ImmutableSet.copyOf(getTreeArtifactContents(artifact));
}
@Override
public void injectDigest(ActionInput output, FileStatus statNoFollow, byte[] digest) {
// Assumption: any non-Artifact output is 'virtual' and should be ignored here.
if (output instanceof Artifact) {
final Artifact artifact = (Artifact) output;
Preconditions.checkState(injectedFiles.add(artifact), artifact);
FileValue fileValue;
try {
// This call may do an unnecessary call to Path#getFastDigest to see if the digest is
// readily available. We cannot pass the digest in, though, because if it is not available
// from the filesystem, this FileValue will not compare equal to another one created for the
// same file, because the other one will be missing its digest.
fileValue = fileValueFromArtifact(artifact,
FileStatusWithDigestAdapter.adapt(statNoFollow),
getTimestampGranularityMonitor(artifact));
// Ensure the digest supplied matches the actual digest if it exists.
byte[] fileDigest = fileValue.getDigest();
if (fileDigest != null && !Arrays.equals(digest, fileDigest)) {
BaseEncoding base16 = BaseEncoding.base16();
String digestString = (digest != null) ? base16.encode(digest) : "null";
String fileDigestString = base16.encode(fileDigest);
throw new IllegalStateException("Expected digest " + digestString + " for artifact "
+ artifact + ", but got " + fileDigestString + " (" + fileValue + ")");
}
outputArtifactData.put(artifact, fileValue);
} catch (IOException e) {
// Do nothing - we just failed to inject metadata. Real error handling will be done later,
// when somebody will try to access that file.
return;
}
// If needed, insert additional data. Note that this can only be true if the file is empty or
// the filesystem does not support fast digests. Since we usually only inject digests when
// running with a filesystem that supports fast digests, this is fairly unlikely.
try {
maybeStoreAdditionalData(artifact, fileValue, digest);
} catch (IOException e) {
if (fileValue.getSize() != 0) {
// Empty files currently have their mtimes examined, and so could throw. No other files
// should throw, since all filesystem access has already been done.
throw new IllegalStateException(
"Filesystem should not have been accessed while injecting data for "
+ artifact.prettyPrint(), e);
}
// Ignore exceptions for empty files, as above.
}
}
}
@Override
public void markOmitted(ActionInput output) {
if (output instanceof Artifact) {
Artifact artifact = (Artifact) output;
Preconditions.checkState(omittedOutputs.add(artifact), artifact);
additionalOutputData.put(artifact, FileArtifactValue.OMITTED_FILE_MARKER);
}
}
@Override
public boolean artifactOmitted(Artifact artifact) {
return omittedOutputs.contains(artifact);
}
@Override
public void discardOutputMetadata() {
Preconditions.checkState(injectedFiles.isEmpty(),
"Files cannot be injected before action execution: %s", injectedFiles);
Preconditions.checkState(omittedOutputs.isEmpty(),
"Artifacts cannot be marked omitted before action execution: %s", omittedOutputs);
outputArtifactData.clear();
outputDirectoryListings.clear();
outputTreeArtifactData.clear();
additionalOutputData.clear();
}
@Override
public boolean isRegularFile(Artifact artifact) {
// Currently this method is used only for genrule input directory checks. If we need to call
// this on output artifacts too, this could be more efficient.
FileArtifactValue value = getInputFileArtifactValue(artifact);
if (value != null && value.isFile()) {
return true;
}
return artifact.getPath().isFile();
}
@Override
public boolean isInjected(Artifact file) {
return injectedFiles.contains(file);
}
/** @return data for output files that was computed during execution. */
Map getOutputArtifactData() {
return outputArtifactData;
}
/**
* @return data for TreeArtifacts that was computed during execution. May contain copies of
* {@link TreeArtifactValue#MISSING_TREE_ARTIFACT}.
*/
Map getOutputTreeArtifactData() {
return outputTreeArtifactData;
}
/**
* Returns data for any output files whose metadata was not computable from the corresponding
* entry in {@link #getOutputArtifactData}.
*
* There are three reasons why we might not be able to compute metadata for an artifact from
* the FileValue. First, middleman artifacts have no corresponding FileValues. Second, if
* computing a file's digest is not fast, the FileValue does not do so, so a file on a filesystem
* without fast digests has to have its metadata stored separately. Third, some files' metadata
* (directories, empty files) contain their mtimes, which the FileValue does not expose, so that
* has to be stored separately.
*
*
Note that for files that need digests, we can't easily inject the digest in the FileValue
* because it would complicate equality-checking on subsequent builds -- if our filesystem doesn't
* do fast digests, the comparison value would not have a digest.
*/
Map getAdditionalOutputData() {
return additionalOutputData;
}
/** Constructs a new FileValue, saves it, and checks inconsistent data. */
FileValue constructFileValue(Artifact artifact, @Nullable FileStatusWithDigest statNoFollow)
throws IOException {
FileValue value = fileValueFromArtifact(artifact, statNoFollow,
getTimestampGranularityMonitor(artifact));
FileValue oldFsValue = outputArtifactData.putIfAbsent(artifact, value);
checkInconsistentData(artifact, oldFsValue, null);
return value;
}
@VisibleForTesting
static FileValue fileValueFromArtifact(Artifact artifact,
@Nullable FileStatusWithDigest statNoFollow, @Nullable TimestampGranularityMonitor tsgm)
throws IOException {
Path path = artifact.getPath();
RootedPath rootedPath =
RootedPath.toRootedPath(artifact.getRoot().getPath(), artifact.getRootRelativePath());
if (statNoFollow == null) {
statNoFollow = FileStatusWithDigestAdapter.adapt(path.statIfFound(Symlinks.NOFOLLOW));
if (statNoFollow == null) {
return FileValue.value(rootedPath, FileStateValue.NONEXISTENT_FILE_STATE_NODE,
rootedPath, FileStateValue.NONEXISTENT_FILE_STATE_NODE);
}
}
Path realPath = path;
// We use FileStatus#isSymbolicLink over Path#isSymbolicLink to avoid the unnecessary stat
// done by the latter.
if (statNoFollow.isSymbolicLink()) {
realPath = path.resolveSymbolicLinks();
// We need to protect against symlink cycles since FileValue#value assumes it's dealing with a
// file that's not in a symlink cycle.
if (realPath.equals(path)) {
throw new IOException("symlink cycle");
}
}
RootedPath realRootedPath = RootedPath.toRootedPathMaybeUnderRoot(realPath,
ImmutableList.of(artifact.getRoot().getPath()));
FileStateValue fileStateValue;
FileStateValue realFileStateValue;
try {
fileStateValue = FileStateValue.createWithStatNoFollow(rootedPath, statNoFollow, tsgm);
// TODO(bazel-team): consider avoiding a 'stat' here when the symlink target hasn't changed
// and is a source file (since changes to those are checked separately).
realFileStateValue = realPath.equals(path) ? fileStateValue
: FileStateValue.create(realRootedPath, tsgm);
} catch (InconsistentFilesystemException e) {
throw new IOException(e);
}
return FileValue.value(rootedPath, fileStateValue, realRootedPath, realFileStateValue);
}
}