[Caching] Add new CacheReplay APIs to libSwiftScan

Add new APIs libSwiftScan that can be used for cache query and cache
replay. This enables swift-driver or build system to query the cache and
replay the compilation results without invocation swift-frontend for
better scheduling.

Author: cachemeifyoucan

include/swift-c/DependencyScan/DependencyScan.h

@@ -25,7 +25,7 @@
 /// SWIFTSCAN_VERSION_MINOR should increase when there are API additions.
 /// SWIFTSCAN_VERSION_MAJOR is intended for "major" source/ABI breaking changes.
 #define SWIFTSCAN_VERSION_MAJOR 0
-#define SWIFTSCAN_VERSION_MINOR 5
+#define SWIFTSCAN_VERSION_MINOR 6
 
 SWIFTSCAN_BEGIN_DECLS
 
@@ -441,6 +441,52 @@ typedef struct swiftscan_cas_options_s *swiftscan_cas_options_t;
 /// ActionCache.
 typedef struct swiftscan_cas_s *swiftscan_cas_t;
 
+/// Opaque container for a cached compilation.
+typedef struct swiftscan_cached_compilation_s *swiftscan_cached_compilation_t;
+
+/// Opaque container for a cached compilation output.
+typedef struct swiftscan_cached_output_s *swiftscan_cached_output_t;
+
+/// Opaque type for a cache replay instance.
+typedef struct swiftscan_cache_replay_instance_s
+    *swiftscan_cache_replay_instance_t;
+
+/// Opaque container for a cached compilation replay result.
+typedef struct swiftscan_cache_replay_result_s *swiftscan_cache_replay_result_t;
+
+/// Opaque type for a cancellation token for async cache operations.
+typedef struct swiftscan_cache_cancellation_token_s
+    *swiftscan_cache_cancellation_token_t;
+
+/// Enum types for output types for cache key computation.
+/// List should contain all the output file types, including supplemententary
+/// outputs, except diagnostics outputs, which are covered by cached diagnostic
+/// entry.
+typedef enum {
+  SWIFTSCAN_OUTPUT_TYPE_OBJECT = 0,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFTMODULE = 1,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFTINTERFACE = 2,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFTPRIVATEINTERFACE = 3,
+  SWIFTSCAN_OUTPUT_TYPE_CLANG_MODULE = 4,
+  SWIFTSCAN_OUTPUT_TYPE_CLANG_PCH = 5,
+  SWIFTSCAN_OUTPUT_TYPE_CLANG_HEADER = 6,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFT_SOURCE_INFO = 7,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFT_MODULE_DOC = 8,
+  SWIFTSCAN_OUTPUT_TYPE_DEPENDENCIES = 9,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFT_DEPS = 10,
+  SWIFTSCAN_OUTPUT_TYPE_MODULE_TRACE = 11,
+  SWIFTSCAN_OUTPUT_TYPE_TBD = 12,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFT_MODULE_SUMMARY = 13,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFT_ABI_DESCRIPTOR = 14,
+  SWIFTSCAN_OUTPUT_TYPE_SWIFT_API_DESCRIPTOR = 15,
+  SWIFTSCAN_OUTPUT_TYPE_CONST_VALUE = 16,
+  SWIFTSCAN_OUTPUT_TYPE_MODULE_SEMANTIC_INFO = 17,
+  SWIFTSCAN_OUTPUT_TYPE_YAML_OPT_RECORD = 18,
+  SWIFTSCAN_OUTPUT_TYPE_BITSTREAM_OPT_RECORD = 19,
+  SWIFTSCAN_OUTPUT_TYPE_CACHED_DIAGNOSTICS = 20,
+  SWIFTSCAN_OUTPUT_TYPE_LAST = SWIFTSCAN_OUTPUT_TYPE_CACHED_DIAGNOSTICS
+} swiftscan_output_kind_t;
+
 /// Create a \c CASOptions for creating CAS inside scanner specified.
 SWIFTSCAN_PUBLIC swiftscan_cas_options_t swiftscan_cas_options_create(void);
 
@@ -472,16 +518,16 @@ swiftscan_cas_options_set_option(swiftscan_cas_options_t options,
 SWIFTSCAN_PUBLIC swiftscan_cas_t swiftscan_cas_create_from_options(
     swiftscan_cas_options_t options, swiftscan_string_ref_t *error);
 
-/// Dispose the \c cas instance.
-SWIFTSCAN_PUBLIC void swiftscan_cas_dispose(swiftscan_cas_t cas);
-
 /// Store content into CAS. Return \c CASID as string. Return NULL on error.
 /// If error happens, the error message is returned via `error` parameter, and
 /// caller needs to free the error message via `swiftscan_string_dispose`.
 SWIFTSCAN_PUBLIC swiftscan_string_ref_t
 swiftscan_cas_store(swiftscan_cas_t cas, uint8_t *data, unsigned size,
                     swiftscan_string_ref_t *error);
 
+/// Dispose the \c cas instance.
+SWIFTSCAN_PUBLIC void swiftscan_cas_dispose(swiftscan_cas_t cas);
+
 /// Compute \c CacheKey for the outputs of a primary input file from a compiler
 /// invocation with command-line \c argc and \c argv. When primary input file
 /// is not available for compilation, e.g., using WMO, primary file is the first
@@ -492,6 +538,151 @@ SWIFTSCAN_PUBLIC swiftscan_string_ref_t
 swiftscan_cache_compute_key(swiftscan_cas_t cas, int argc, const char **argv,
                             const char *input, swiftscan_string_ref_t *error);
 
+/// Query the result of the compilation using the output cache key. \c globally
+/// suggests if the lookup should check remote cache if such operation exists.
+/// Returns the cached compilation of the result if found, or nullptr if output
+/// is not found or an error occurs. When an error occurs, the error message is
+/// returned via \c error parameter and its caller needs to free the message
+/// using `swiftscan_string_dispose`. The returned cached compilation needs to
+/// be freed via `swiftscan_cached_compilation_dispose`.
+SWIFTSCAN_PUBLIC swiftscan_cached_compilation_t
+swiftscan_cache_query(swiftscan_cas_t cas, const char *key, bool globally,
+                      swiftscan_string_ref_t *error);
+
+/// Async version of `swiftscan_cache_query` where result is returned via
+/// callback. Both cache_result enum and cached compilation will be provided to
+/// callback. \c ctx is an opaque value that passed to the callback and \c
+/// swiftscan_cache_cancellation_token_t will return an token that can be used
+/// to cancel the async operation. The token needs to be freed by caller using
+/// `swiftscan_cache_cancellation_token_dispose`. If no token is needed, nullptr
+/// can be passed and no token will be returned.
+SWIFTSCAN_PUBLIC void swiftscan_cache_query_async(
+    swiftscan_cas_t cas, const char *key, bool globally, void *ctx,
+    void (*callback)(void *ctx, swiftscan_cached_compilation_t,
+                     swiftscan_string_ref_t error),
+    swiftscan_cache_cancellation_token_t *);
+
+/// Query the number of outputs from a cached compilation.
+SWIFTSCAN_PUBLIC unsigned swiftscan_cached_compilation_get_num_outputs(
+    swiftscan_cached_compilation_t);
+
+/// Get the output kind for the given index in the cached compilation.
+SWIFTSCAN_PUBLIC swiftscan_output_kind_t
+swiftscan_cached_compilation_get_output_kind(swiftscan_cached_compilation_t,
+                                             unsigned idx);
+
+/// Get the cached output for the given index in the cached compilation.
+SWIFTSCAN_PUBLIC swiftscan_cached_output_t
+swiftscan_cached_compilation_get_output(swiftscan_cached_compilation_t,
+                                        unsigned idx);
+
+/// Check if the requested cached compilation is uncacheable. That means the
+/// compiler decides to skip caching its output even the compilation is
+/// successful.
+SWIFTSCAN_PUBLIC bool
+    swiftscan_cached_compilation_is_uncacheable(swiftscan_cached_compilation_t);
+
+/// Make the cache compilation available globally. \c callback will be called
+/// on completion.
+/// \c swiftscan_cache_cancellation_token_t will return an token that can be
+/// used to cancel the async operation. The token needs to be freed by caller
+/// using `swiftscan_cache_cancellation_token_dispose`. If no token is needed,
+/// nullptr can be passed and no token will be returned.
+SWIFTSCAN_PUBLIC void swiftscan_cached_compilation_make_global_async(
+    swiftscan_cas_t cas, swiftscan_cached_compilation_t, void *ctx,
+    void (*callback)(void *ctx, swiftscan_string_ref_t error),
+    swiftscan_cache_cancellation_token_t *);
+
+/// Dispose a cached compilation.
+SWIFTSCAN_PUBLIC
+void swiftscan_cached_compilation_dispose(swiftscan_cached_compilation_t);
+
+/// Download and materialize the cached output if needed from a remote CAS.
+/// Return true if load is successful, else false if not found or error. If
+/// error, the error message is returned via \c error parameter and its caller
+/// needs to free the message using `swiftscan_string_dispose`.
+SWIFTSCAN_PUBLIC bool
+swiftscan_cached_output_load(swiftscan_cached_output_t,
+                             swiftscan_string_ref_t *error);
+
+/// Async version of `swiftscan_cached_output_load` where result is
+/// returned via callback. \c ctx is an opaque value that passed to the callback
+/// and \c swiftscan_cache_cancellation_token_t will return an token that can be
+/// used to cancel the async operation. The token needs to be freed by caller
+/// using `swiftscan_cache_cancellation_token_dispose`. If no token is needed,
+/// nullptr can be passed and no token will be returned.
+SWIFTSCAN_PUBLIC void swiftscan_cached_output_load_async(
+    swiftscan_cached_output_t, void *ctx,
+    void (*callback)(void *ctx, bool success, swiftscan_string_ref_t error),
+    swiftscan_cache_cancellation_token_t *);
+
+/// Check if cached output is materialized locally and can be accessed
+/// without downloading.
+SWIFTSCAN_PUBLIC bool
+    swiftscan_cached_output_is_materialized(swiftscan_cached_output_t);
+
+/// Return the casid for the cached output as \c swiftscan_string_ref_t and the
+/// returned string needs to be freed using `swiftscan_string_dispose`. CASID
+/// can be requested before load/materializaing.
+SWIFTSCAN_PUBLIC swiftscan_string_ref_t
+    swiftscan_cached_output_get_casid(swiftscan_cached_output_t);
+
+/// Dispose a cached output.
+SWIFTSCAN_PUBLIC
+void swiftscan_cached_output_dispose(swiftscan_cached_output_t);
+
+/// Cancel the async cache action that is associated with token.
+SWIFTSCAN_PUBLIC void
+    swiftscan_cache_action_cancel(swiftscan_cache_cancellation_token_t);
+
+/// Dispose the cancellation token.
+SWIFTSCAN_PUBLIC void swiftscan_cache_cancellation_token_dispose(
+    swiftscan_cache_cancellation_token_t);
+
+/// Create a swift cached compilation replay instance with its command-line
+/// invocation. Return nullptr when errors occurs and the error message is
+/// returned via \c error parameter and its caller needs to free the message
+/// using `swiftscan_string_dispose`.
+SWIFTSCAN_PUBLIC swiftscan_cache_replay_instance_t
+swiftscan_cache_replay_instance_create(int argc, const char **argv,
+                                       swiftscan_string_ref_t *error);
+
+/// Dispose swift cached compilation replay instance.
+SWIFTSCAN_PUBLIC void
+    swiftscan_cache_replay_instance_dispose(swiftscan_cache_replay_instance_t);
+
+/// Replay the cached compilation using cached compliation replay instance.
+/// Returns replay result or nullptr if output not found or error occurs. If
+/// error, the error message is returned via \c error parameter and its caller
+/// needs to free the message using `swiftscan_string_dispose`.
+SWIFTSCAN_PUBLIC swiftscan_cache_replay_result_t
+swiftscan_cache_replay_compilation(swiftscan_cache_replay_instance_t,
+                                   swiftscan_cached_compilation_t,
+                                   swiftscan_string_ref_t *error);
+
+/// Get return code from replay which should matches the returnCode from the
+/// corresponding swift-frontend invocation. When the replay is sucessful, this
+/// should always be 0.
+SWIFTSCAN_PUBLIC
+int swiftscan_cache_replay_result_get_return_code(
+    swiftscan_cache_replay_result_t);
+
+/// Get stdout from cached replay result. The returning swiftscan_string_ref_t
+/// is owned by replay result and should not be disposed.
+SWIFTSCAN_PUBLIC
+swiftscan_string_ref_t
+    swiftscan_cache_replay_result_get_stdout(swiftscan_cache_replay_result_t);
+
+/// Get stderr from cached replay result. The returning swiftscan_string_ref_t
+/// is owned by replay result and should not be disposed.
+SWIFTSCAN_PUBLIC
+swiftscan_string_ref_t
+    swiftscan_cache_replay_result_get_stderr(swiftscan_cache_replay_result_t);
+
+/// Dispose a cached replay result.
+SWIFTSCAN_PUBLIC
+void swiftscan_cache_replay_result_dispose(swiftscan_cache_replay_result_t);
+
 //===----------------------------------------------------------------------===//
 
 SWIFTSCAN_END_DECLS

include/swift/Frontend/CachingUtils.h

@@ -63,19 +63,19 @@ std::vector<std::string> remapPathsFromCommandLine(
 namespace cas {
 class CachedResultLoader {
 public:
-  CachedResultLoader(llvm::cas::ObjectStore &CAS, llvm::cas::ActionCache &Cache,
-                     llvm::cas::ObjectRef CacheKey)
-      : CAS(CAS), Cache(Cache), CacheKey(CacheKey) {}
+  CachedResultLoader(llvm::cas::ObjectStore &CAS,
+                     llvm::cas::ObjectRef OutputRef)
+      : CAS(CAS), OutputRef(OutputRef) {}
 
   using CallbackTy =
       llvm::function_ref<llvm::Error(file_types::ID, llvm::cas::ObjectRef)>;
-  // Replay the cached result, return false if a cache miss happened.
-  llvm::Expected<bool> replay(CallbackTy Callback);
+
+  /// Replay the cached result.
+  llvm::Error replay(CallbackTy Callback);
 
 private:
   llvm::cas::ObjectStore &CAS;
-  llvm::cas::ActionCache &Cache;
-  llvm::cas::ObjectRef CacheKey;
+  llvm::cas::ObjectRef OutputRef;
 };
 } // end namespace cas
 } // end namespace swift

lib/DriverTool/swift_cache_tool_main.cpp

@@ -369,10 +369,21 @@ int SwiftCacheToolInvocation::validateOutputs() {
     report_fatal_error(DB.takeError());
 
   auto &CAS = *DB->first;
+  auto &Cache = *DB->second;
 
   PrintingDiagnosticConsumer PDC;
   Instance.getDiags().addConsumer(PDC);
 
+  auto lookupFailed = [&](StringRef Key) {
+    llvm::errs() << "failed to find output for cache key " << Key << "\n";
+    return true;
+  };
+  auto lookupError = [&](llvm::Error Err, StringRef Key) {
+    llvm::errs() << "failed to find output for cache key " << Key << ": "
+                 << toString(std::move(Err)) << "\n";
+    return true;
+  };
+
   auto validateCacheKeysFromFile = [&](const std::string &Path) {
     auto Keys = readOutputEntriesFromFile(Path);
     if (!Keys) {
@@ -391,28 +402,30 @@ int SwiftCacheToolInvocation::validateOutputs() {
             return true;
           }
           auto Ref = CAS.getReference(*ID);
-          if (!Ref) {
-            llvm::errs() << "failed to find output for cache key " << Key
-                         << "\n";
-            return true;
-          }
-          cas::CachedResultLoader Loader(*DB->first, *DB->second, *Ref);
-          auto Result = Loader.replay(
+          if (!Ref)
+            return lookupFailed(*Key);
+          auto KeyID = CAS.getID(*Ref);
+          auto Lookup = Cache.get(KeyID);
+          if (!Lookup)
+            return lookupError(Lookup.takeError(), *Key);
+
+          if (!*Lookup)
+            return lookupFailed(*Key);
+
+          auto OutputRef = CAS.getReference(**Lookup);
+          if (!OutputRef)
+            return lookupFailed(*Key);
+
+          cas::CachedResultLoader Loader(CAS, *OutputRef);
+          if (auto Err = Loader.replay(
                   [&](file_types::ID Kind, ObjectRef Ref) -> llvm::Error {
                     auto Proxy = CAS.getProxy(Ref);
                     if (!Proxy)
                       return Proxy.takeError();
                     return llvm::Error::success();
-                  });
-
-          if (!Result) {
+                  })) {
             llvm::errs() << "failed to find output for cache key " << *Key
-                         << ": " << toString(Result.takeError()) << "\n";
-            return true;
-          }
-          if (!*Result) {
-            llvm::errs() << "failed to load output for cache key " << *Key
-                         << "\n";
+                         << ": " << toString(std::move(Err)) << "\n";
             return true;
           }
           continue;