Handle multiple llvm-cas execs for validation

While we intentionally wanto to ignore minor differences in CASOptions
for validation, the llvm-cas executable path is significant because
there could be differences in format version between tools. If the
llvm-cas binaries end up using the same format version only one of them
will perform the real validation and the rest will skip.

Author: benlangmuir

Sources/SWBBuildSystem/BuildOperation.swift

@@ -846,13 +846,15 @@ package final class BuildOperation: BuildSystemOperation {
         assert(UserDefaults.enableCASValidation)
 
         let casPath = info.options.casPath
+        let ruleInfo = "ValidateCAS \(casPath.str) \(info.llvmCasExec.str)"
 
         let signatureCtx = InsecureHashContext()
         signatureCtx.add(string: "ValidateCAS")
         signatureCtx.add(string: casPath.str)
+        signatureCtx.add(string: info.llvmCasExec.str)
         let signature = signatureCtx.signature
 
-        let activityId = delegate.beginActivity(self, ruleInfo: "ValidateCAS \(casPath.str)", executionDescription: "Validate CAS contents at \(casPath.str)", signature: signature, target: nil, parentActivity: nil)
+        let activityId = delegate.beginActivity(self, ruleInfo: ruleInfo, executionDescription: "Validate CAS contents at \(casPath.str)", signature: signature, target: nil, parentActivity: nil)
         var status: BuildOperationTaskEnded.Status = .failed
         defer {
             delegate.endActivity(self, id: activityId, signature: signature, status: status)

Sources/SWBTaskExecution/BuildDescription.swift

@@ -1511,3 +1511,15 @@ extension BuildDescription.CASValidationInfo: Serializable {
     }
 }
 
+// Note: for the purposes of validation we intentionally ignore irrelevant
+// differences in CASOptions. However, we need to keep the llvm-cas executable
+// in case there are multiple cas format versions sharing the path.
+extension BuildDescription.CASValidationInfo: Hashable {
+    package func hash(into hasher: inout Hasher) {
+        hasher.combine(options.casPath)
+        hasher.combine(llvmCasExec)
+    }
+    static package func ==(lhs: Self, rhs: Self) -> Bool {
+        return lhs.options.casPath == rhs.options.casPath && lhs.llvmCasExec == rhs.llvmCasExec
+    }
+}

Sources/SWBTaskExecution/BuildDescriptionManager.swift

@@ -180,7 +180,8 @@ package final class BuildDescriptionManager: Sendable {
         var settingsPerTarget = [ConfiguredTarget:Settings]()
         var rootPathsPerTarget = [ConfiguredTarget:[Path]]()
         var moduleCachePathsPerTarget = [ConfiguredTarget: [Path]]()
-        var casValidationInfos: [Path: BuildDescription.CASValidationInfo] = [:]
+
+        var casValidationInfos: Set<BuildDescription.CASValidationInfo> = []
         let buildGraph = planRequest.buildGraph
         let shouldValidateCAS = Settings.supportsCompilationCaching(plan.workspaceContext.core) && UserDefaults.enableCASValidation
 
@@ -204,11 +205,10 @@ package final class BuildDescriptionManager: Sendable {
             if shouldValidateCAS, settings.globalScope.evaluate(BuiltinMacros.CLANG_ENABLE_COMPILE_CACHE) || settings.globalScope.evaluate(BuiltinMacros.SWIFT_ENABLE_COMPILE_CACHE) {
                 // FIXME: currently we only handle the compiler cache here, because the plugin configuration for the generic CAS is not configured by build settings.
                 for purpose in [CASOptions.Purpose.compiler(.c)] {
-                    if let casOpts = try? CASOptions.create(settings.globalScope, purpose),
-                       !casValidationInfos.contains(casOpts.casPath) {
+                    if let casOpts = try? CASOptions.create(settings.globalScope, purpose) {
                         let execName = settings.globalScope.evaluate(BuiltinMacros.VALIDATE_CAS_EXEC).nilIfEmpty ?? "llvm-cas"
                         if let execPath = settings.executableSearchPaths.lookup(Path(execName)) {
-                            casValidationInfos[casOpts.casPath] = .init(options: casOpts, llvmCasExec: execPath)
+                            casValidationInfos.insert(.init(options: casOpts, llvmCasExec: execPath))
                         }
                     }
                 }
@@ -246,7 +246,7 @@ package final class BuildDescriptionManager: Sendable {
         }
 
         // Create the build description.
-        return try await BuildDescription.construct(workspace: buildGraph.workspaceContext.workspace, tasks: plan.tasks, path: path, signature: signature, buildCommand: planRequest.buildRequest.buildCommand, diagnostics: planningDiagnostics, indexingInfo: [], fs: fs, bypassActualTasks: bypassActualTasks, targetsBuildInParallel: buildGraph.targetsBuildInParallel, emitFrontendCommandLines: plan.emitFrontendCommandLines, moduleSessionFilePath: planRequest.workspaceContext.getModuleSessionFilePath(planRequest.buildRequest.parameters), invalidationPaths: plan.invalidationPaths, recursiveSearchPathResults: plan.recursiveSearchPathResults, copiedPathMap: plan.copiedPathMap, rootPathsPerTarget: rootPathsPerTarget, moduleCachePathsPerTarget: moduleCachePathsPerTarget, casValidationInfos: casValidationInfos.values.sorted(by: \.options.casPath), staleFileRemovalIdentifierPerTarget: staleFileRemovalIdentifierPerTarget, settingsPerTarget: settingsPerTarget, delegate: delegate, targetDependencies: buildGraph.targetDependenciesByGuid, definingTargetsByModuleName: definingTargetsByModuleName, capturedBuildInfo: capturedBuildInfo, userPreferences: buildGraph.workspaceContext.userPreferences)
+        return try await BuildDescription.construct(workspace: buildGraph.workspaceContext.workspace, tasks: plan.tasks, path: path, signature: signature, buildCommand: planRequest.buildRequest.buildCommand, diagnostics: planningDiagnostics, indexingInfo: [], fs: fs, bypassActualTasks: bypassActualTasks, targetsBuildInParallel: buildGraph.targetsBuildInParallel, emitFrontendCommandLines: plan.emitFrontendCommandLines, moduleSessionFilePath: planRequest.workspaceContext.getModuleSessionFilePath(planRequest.buildRequest.parameters), invalidationPaths: plan.invalidationPaths, recursiveSearchPathResults: plan.recursiveSearchPathResults, copiedPathMap: plan.copiedPathMap, rootPathsPerTarget: rootPathsPerTarget, moduleCachePathsPerTarget: moduleCachePathsPerTarget, casValidationInfos: casValidationInfos.sorted(by: \.options.casPath), staleFileRemovalIdentifierPerTarget: staleFileRemovalIdentifierPerTarget, settingsPerTarget: settingsPerTarget, delegate: delegate, targetDependencies: buildGraph.targetDependenciesByGuid, definingTargetsByModuleName: definingTargetsByModuleName, capturedBuildInfo: capturedBuildInfo, userPreferences: buildGraph.workspaceContext.userPreferences)
     }
 
     /// Encapsulates the two ways `getNewOrCachedBuildDescription` can be called, whether we want to retrieve or create a build description based on a plan or whether we have an explicit build description ID that we want to retrieve and we don't need to create a new one.

Tests/SWBBuildSystemTests/ClangCompilationCachingTests.swift

@@ -1879,17 +1879,20 @@ fileprivate struct ClangCompilationCachingTests: CoreBasedTests {
                 """
             }
 
+            let specificCAS = casPath.join(usePlugin ? "plugin" : "builtin")
+            let ruleInfo = "ValidateCAS \(specificCAS.str) \(try await ConditionTraitContext.shared.llvmCasToolPath.str)"
+
             let checkBuild = { (expectedOutput: ByteString?) in
                 try await tester.checkBuild(runDestination: .macOS, persistent: true) { results in
-                    let specificCAS = casPath.join(usePlugin ? "plugin" : "builtin")
+
                     if enableCaching {
-                        results.check(contains: .activityStarted(ruleInfo: "ValidateCAS \(specificCAS.str)"))
+                        results.check(contains: .activityStarted(ruleInfo: ruleInfo))
                         if let expectedOutput {
-                            results.check(contains: .activityEmittedData(ruleInfo: "ValidateCAS \(specificCAS.str)", expectedOutput.bytes))
+                            results.check(contains: .activityEmittedData(ruleInfo: ruleInfo, expectedOutput.bytes))
                         }
-                        results.check(contains: .activityEnded(ruleInfo: "ValidateCAS \(specificCAS.str)", status: .succeeded))
+                        results.check(contains: .activityEnded(ruleInfo: ruleInfo, status: .succeeded))
                     } else {
-                        results.check(notContains: .activityStarted(ruleInfo: "ValidateCAS \(specificCAS.str)"))
+                        results.check(notContains: .activityStarted(ruleInfo: ruleInfo))
                     }
                     results.checkNoDiagnostics()
                 }
@@ -1947,14 +1950,16 @@ fileprivate struct ClangCompilationCachingTests: CoreBasedTests {
                 """
             }
 
+            let specificCAS = casPath.join("builtin")
+            let ruleInfo = "ValidateCAS \(specificCAS.str) \(try await ConditionTraitContext.shared.llvmCasToolPath.str)"
+
             let checkBuild = { (expectedOutput: ByteString?) in
                 try await tester.checkBuild(runDestination: .macOS, persistent: true) { results in
-                    let specificCAS = casPath.join("builtin")
-                    results.check(contains: .activityStarted(ruleInfo: "ValidateCAS \(specificCAS.str)"))
+                    results.check(contains: .activityStarted(ruleInfo: ruleInfo))
                     if let expectedOutput {
-                        results.check(contains: .activityEmittedData(ruleInfo: "ValidateCAS \(specificCAS.str)", expectedOutput.bytes))
+                        results.check(contains: .activityEmittedData(ruleInfo: ruleInfo, expectedOutput.bytes))
                     }
-                    results.check(contains: .activityEnded(ruleInfo: "ValidateCAS \(specificCAS.str)", status: .succeeded))
+                    results.check(contains: .activityEnded(ruleInfo: ruleInfo, status: .succeeded))
                     results.checkNoDiagnostics()
                 }
             }
@@ -1969,6 +1974,76 @@ fileprivate struct ClangCompilationCachingTests: CoreBasedTests {
             try await checkBuild("recovered from invalid data\n")
         }
     }
+
+    @Test(.requireCASValidation, .requireSDKs(.macOS))
+    func validateCASMultipleExec() async throws {
+        try await withTemporaryDirectory { (tmpDirPath: Path) in
+            let casPath = tmpDirPath.join("CompilationCache")
+            let buildSettings: [String: String] = [
+                "PRODUCT_NAME": "$(TARGET_NAME)",
+                "CLANG_ENABLE_COMPILE_CACHE": "YES",
+                "CLANG_ENABLE_MODULES": "NO",
+                "COMPILATION_CACHE_CAS_PATH": casPath.str,
+            ]
+            let llvmCasExec = try await ConditionTraitContext.shared.llvmCasToolPath
+            // Create a trivially different path. If we ever canonicalize the path it will be harder to test this.
+            let llvmCasExec2 = Path("\(llvmCasExec.dirname.str)\(Path.pathSeparator)\(Path.pathSeparator)\(llvmCasExec.basename)")
+            let testWorkspace = TestWorkspace(
+                "Test",
+                sourceRoot: tmpDirPath.join("Test"),
+                projects: [
+                    TestProject(
+                        "aProject",
+                        groupTree: TestGroup(
+                            "Sources",
+                            children: [
+                                TestFile("file.c"),
+                            ]),
+                        buildConfigurations: [TestBuildConfiguration(
+                            "Debug",
+                            buildSettings: buildSettings)],
+                        targets: [
+                            TestStandardTarget(
+                                "Library1",
+                                type: .staticLibrary,
+                                buildPhases: [
+                                    TestSourcesBuildPhase(["file.c"]),
+                                ]),
+                            TestStandardTarget(
+                                "Library2",
+                                type: .staticLibrary,
+                                buildConfigurations: [TestBuildConfiguration(
+                                    "Debug",
+                                    buildSettings: [
+                                        "VALIDATE_CAS_EXEC": llvmCasExec2.str,
+                                    ])],
+                                buildPhases: [
+                                    TestSourcesBuildPhase(["file.c"]),
+                                ]),
+                        ])])
+
+            let tester = try await BuildOperationTester(getCore(), testWorkspace, simulated: false)
+            try await tester.fs.writeFileContents(testWorkspace.sourceRoot.join("aProject/file.c")) { stream in
+                stream <<<
+                """
+                #include <stdio.h>
+                int something = 1;
+                """
+            }
+
+            let specificCAS = casPath.join("builtin")
+            let parameters = BuildParameters(configuration: "Debug", activeRunDestination: .macOS)
+            let targets = tester.workspace.allTargets.map({ BuildRequest.BuildTargetInfo(parameters: parameters, target: $0) })
+
+            try await tester.checkBuild(runDestination: .macOS, buildRequest: BuildRequest(parameters: parameters, buildTargets: targets, continueBuildingAfterErrors: false, useParallelTargets: true, useImplicitDependencies: true, useDryRun: false), persistent: true) { results in
+                for ruleInfo in ["ValidateCAS \(specificCAS.str) \(llvmCasExec.str)", "ValidateCAS \(specificCAS.str) \(llvmCasExec2.str)"] {
+                    results.check(contains: .activityStarted(ruleInfo: ruleInfo))
+                    results.check(contains: .activityEnded(ruleInfo: ruleInfo, status: .succeeded))
+                }
+                results.checkNoDiagnostics()
+            }
+        }
+    }
 }
 
 extension BuildOperationTester.BuildResults {

Tests/SWBBuildSystemTests/SwiftCompilationCachingTests.swift

@@ -247,14 +247,16 @@ fileprivate struct SwiftCompilationCachingTests: CoreBasedTests {
                 """
             }
 
+            let specificCAS = casPath.join("builtin")
+            let ruleInfo = "ValidateCAS \(specificCAS.str) \(try await ConditionTraitContext.shared.llvmCasToolPath.str)"
+
             let checkBuild = { (expectedOutput: ByteString?) in
                 try await tester.checkBuild(runDestination: .macOS, persistent: true) { results in
-                    let specificCAS = casPath.join("builtin")
-                    results.check(contains: .activityStarted(ruleInfo: "ValidateCAS \(specificCAS.str)"))
+                    results.check(contains: .activityStarted(ruleInfo: ruleInfo))
                     if let expectedOutput {
-                        results.check(contains: .activityEmittedData(ruleInfo: "ValidateCAS \(specificCAS.str)", expectedOutput.bytes))
+                        results.check(contains: .activityEmittedData(ruleInfo: ruleInfo, expectedOutput.bytes))
                     }
-                    results.check(contains: .activityEnded(ruleInfo: "ValidateCAS \(specificCAS.str)", status: .succeeded))
+                    results.check(contains: .activityEnded(ruleInfo: ruleInfo, status: .succeeded))
                     results.checkNoDiagnostics()
                 }
             }