Implement garbage collection when deleting cached OCI VMs (#185)

* Properly calculate remotely-retrieved manifest digests

* Implement garbage collection when deleting cached OCI VMs

Sources/tart/VMDirectory.swift

@@ -20,6 +20,10 @@ struct VMDirectory: Prunable {
     baseURL.appendingPathComponent("nvram.bin")
   }
 
+  var explicitlyPulledMark: URL {
+    baseURL.appendingPathComponent(".explicitly-pulled")
+  }
+
   var name: String {
     baseURL.lastPathComponent
   }
@@ -89,4 +93,12 @@ struct VMDirectory: Prunable {
   func sizeBytes() throws -> Int {
     try configURL.sizeBytes() + diskURL.sizeBytes() + nvramURL.sizeBytes()
   }
+
+  func markExplicitlyPulled() {
+    FileManager.default.createFile(atPath: explicitlyPulledMark.path, contents: nil)
+  }
+
+  func isExplicitlyPulled() -> Bool {
+    FileManager.default.fileExists(atPath: explicitlyPulledMark.path)
+  }
 }

Sources/tart/VMStorageOCI.swift

@@ -42,6 +42,44 @@ class VMStorageOCI: PrunableStorage {
 
   func delete(_ name: RemoteName) throws {
     try FileManager.default.removeItem(at: vmURL(name))
+    try gc()
+  }
+
+  func gc() throws {
+    var refCounts = Dictionary<URL, UInt>()
+
+    guard let enumerator = FileManager.default.enumerator(at: baseURL,
+            includingPropertiesForKeys: [.isSymbolicLinkKey]) else {
+      return
+    }
+
+    for case let foundURL as URL in enumerator {
+      let isSymlink = try foundURL.resourceValues(forKeys: [.isSymbolicLinkKey]).isSymbolicLink!
+
+      // Perform garbage collection for tag-based images
+      // with broken outgoing references
+      if isSymlink && foundURL == foundURL.resolvingSymlinksInPath() {
+        try FileManager.default.removeItem(at: foundURL)
+        continue
+      }
+
+      let vmDir = VMDirectory(baseURL: foundURL.resolvingSymlinksInPath())
+      if !vmDir.initialized {
+        continue
+      }
+
+      refCounts[vmDir.baseURL] = (refCounts[vmDir.baseURL] ?? 0) + (isSymlink ? 1 : 0)
+    }
+
+    // Perform garbage collection for digest-based images
+    // with no incoming references
+    for (baseURL, incRefCount) in refCounts {
+      let vmDir = VMDirectory(baseURL: baseURL)
+
+      if !vmDir.isExplicitlyPulled() && incRefCount == 0 {
+        try FileManager.default.removeItem(at: baseURL)
+      }
+    }
   }
 
   func list() throws -> [(String, VMDirectory, Bool)] {
@@ -82,10 +120,10 @@ class VMStorageOCI: PrunableStorage {
   func pull(_ name: RemoteName, registry: Registry) async throws {
     defaultLogger.appendNewLine("pulling manifest...")
 
-    let (manifest, _) = try await registry.pullManifest(reference: name.reference.value)
+    let (manifest, manifestData) = try await registry.pullManifest(reference: name.reference.value)
 
-    var digestName = RemoteName(host: name.host, namespace: name.namespace,
-            reference: Reference(digest: try manifest.digest()))
+    let digestName = RemoteName(host: name.host, namespace: name.namespace,
+            reference: Reference(digest: Digest.hash(manifestData)))
 
     if !exists(digestName) {
       let tmpVMDir = try VMDirectory.temporary()
@@ -113,8 +151,12 @@ class VMStorageOCI: PrunableStorage {
     }
 
     if name != digestName {
-      // Overwrite the old symbolic link
+      // Create new or overwrite the old symbolic link
       try link(from: digestName, to: name)
+    } else {
+      // Ensure that images pulled by content digest
+      // are excluded from garbage collection
+      VMDirectory(baseURL: vmURL(name)).markExplicitlyPulled()
     }
   }