| 1 | package blob |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "crypto/sha256" |
| 6 | "database/sql" |
| 7 | "encoding/hex" |
| 8 | "errors" |
| 9 | "io" |
| 10 | "net/http" |
| 11 | "net/http/httptest" |
| 12 | "os" |
| 13 | "path/filepath" |
| 14 | "strings" |
| 15 | "sync" |
| 16 | "testing" |
| 17 | "time" |
| 18 | ) |
| 19 | |
| 20 | func multipartReviewRequest(sha string, length, offset int64, body string) *http.Request { |
| 21 | r := httptest.NewRequest(http.MethodPatch, "https://relay.test/" +sha, strings.NewReader(body)) |
| 22 | r.Header.Set("Content-Type" , "application/octet-stream" ) |
| 23 | r.Header.Set("Upload-Type" , "application/octet-stream" ) |
| 24 | r.Header.Set("Upload-Length" , formatInt(length)) |
| 25 | r.Header.Set("Upload-Offset" , formatInt(offset)) |
| 26 | return r |
| 27 | } |
| 28 | |
| 29 | func TestMultipartReviewConcurrentChunksProduceOneBlob(t *testing.T) { |
| 30 | s := testService(t) |
| 31 | body := []byte(strings.Repeat("a" , 70000) + strings.Repeat("b" , 70000)) |
| 32 | digest := sha256.Sum256(body) |
| 33 | sha := hex.EncodeToString(digest[:]) |
| 34 | parts := []struct { |
| 35 | offset int64 |
| 36 | body string |
| 37 | }{ |
| 38 | {70000, string(body[70000:])}, |
| 39 | {0, string(body[:70000])}, |
| 40 | } |
| 41 | responses := make([]*httptest.ResponseRecorder, len(parts)) |
| 42 | var wg sync.WaitGroup |
| 43 | for i, part := range parts { |
| 44 | wg.Add(1) |
| 45 | go func() { |
| 46 | defer wg.Done() |
| 47 | responses[i] = record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), part.offset, part.body)) |
| 48 | }() |
| 49 | } |
| 50 | wg.Wait() |
| 51 | var created, partial int |
| 52 | for _, response := range responses { |
| 53 | switch response.Code { |
| 54 | case http.StatusCreated: |
| 55 | created++ |
| 56 | case http.StatusNoContent: |
| 57 | partial++ |
| 58 | default: |
| 59 | t.Fatalf("concurrent chunk status = %d: %s" , response.Code, response.Body.String()) |
| 60 | } |
| 61 | } |
| 62 | if created != 1 || partial != 1 { |
| 63 | t.Fatalf("concurrent chunk results created=%d partial=%d" , created, partial) |
| 64 | } |
| 65 | entry, file, err := s.Get(context.Background(), sha) |
| 66 | if err != nil { |
| 67 | t.Fatal(err) |
| 68 | } |
| 69 | defer file.Close() |
| 70 | if entry.Size != int64(len(body)) { |
| 71 | t.Fatalf("stored size = %d, want %d" , entry.Size, len(body)) |
| 72 | } |
| 73 | got, err := io.ReadAll(file) |
| 74 | if err != nil || string(got) != string(body) { |
| 75 | t.Fatalf("stored body mismatch: %v" , err) |
| 76 | } |
| 77 | } |
| 78 | |
| 79 | func TestMultipartReviewStalledBodyDoesNotBlockOrdinaryUpload(t *testing.T) { |
| 80 | s := testService(t) |
| 81 | body := "stalled" |
| 82 | digest := sha256.Sum256([]byte(body)) |
| 83 | sha := hex.EncodeToString(digest[:]) |
| 84 | started := make(chan struct{}) |
| 85 | release := make(chan struct{}) |
| 86 | r := multipartReviewRequest(sha, int64(len(body)), 0, body) |
| 87 | r.Body = &multipartBlockingBody{Reader: strings.NewReader(body), started: started, release: release} |
| 88 | r.ContentLength = int64(len(body)) |
| 89 | result := make(chan *httptest.ResponseRecorder, 1) |
| 90 | go func() { result <- record(s.Handler(), r) }() |
| 91 | select { |
| 92 | case <-started: |
| 93 | case <-time.After(time.Second): |
| 94 | t.Fatal("multipart body was not read" ) |
| 95 | } |
| 96 | if _, err := s.Put(context.Background(), PutOptions{Reader: strings.NewReader("ordinary" ), Type: "text/plain" , Uploader: "uploader" }); err != nil { |
| 97 | t.Fatalf("ordinary upload blocked by stalled multipart body: %v" , err) |
| 98 | } |
| 99 | close(release) |
| 100 | select { |
| 101 | case response := <-result: |
| 102 | if response.Code != http.StatusCreated { |
| 103 | t.Fatalf("multipart completion status = %d" , response.Code) |
| 104 | } |
| 105 | case <-time.After(time.Second): |
| 106 | t.Fatal("stalled multipart request did not finish" ) |
| 107 | } |
| 108 | } |
| 109 | |
| 110 | func TestMultipartReviewStalledHashDoesNotBlockOrdinaryUpload(t *testing.T) { |
| 111 | s := testService(t) |
| 112 | body := "hash-paused" |
| 113 | digest := sha256.Sum256([]byte(body)) |
| 114 | sha := hex.EncodeToString(digest[:]) |
| 115 | started := make(chan struct{}) |
| 116 | release := make(chan struct{}) |
| 117 | s.verifyMultipartHook = func(ctx context.Context, path, hash string, size int64) error { |
| 118 | close(started) |
| 119 | <-release |
| 120 | return verifyMultipart(ctx, path, hash, size) |
| 121 | } |
| 122 | result := make(chan *httptest.ResponseRecorder, 1) |
| 123 | r := multipartReviewRequest(sha, int64(len(body)), 0, body) |
| 124 | go func() { result <- record(s.Handler(), r) }() |
| 125 | select { |
| 126 | case <-started: |
| 127 | case <-time.After(time.Second): |
| 128 | t.Fatal("multipart hash was not started" ) |
| 129 | } |
| 130 | ordinary := make(chan error, 1) |
| 131 | go func() { |
| 132 | _, err := s.Put(context.Background(), PutOptions{Reader: strings.NewReader("ordinary" ), Type: "text/plain" , Uploader: "other" }) |
| 133 | ordinary <- err |
| 134 | }() |
| 135 | select { |
| 136 | case err := <-ordinary: |
| 137 | if err != nil { |
| 138 | t.Fatalf("ordinary upload while multipart hash paused: %v" , err) |
| 139 | } |
| 140 | case <-time.After(time.Second): |
| 141 | t.Fatal("ordinary upload blocked by multipart hashing" ) |
| 142 | } |
| 143 | close(release) |
| 144 | select { |
| 145 | case response := <-result: |
| 146 | if response.Code != http.StatusCreated { |
| 147 | t.Fatalf("multipart completion status = %d: %s" , response.Code, response.Body.String()) |
| 148 | } |
| 149 | case <-time.After(time.Second): |
| 150 | t.Fatal("multipart hash did not finish" ) |
| 151 | } |
| 152 | } |
| 153 | |
| 154 | func TestMultipartReviewStalledHashDoesNotBlockAnotherSession(t *testing.T) { |
| 155 | s := testService(t) |
| 156 | body := "hash-paused" |
| 157 | digest := sha256.Sum256([]byte(body)) |
| 158 | sha := hex.EncodeToString(digest[:]) |
| 159 | started := make(chan struct{}) |
| 160 | release := make(chan struct{}) |
| 161 | var unblock sync.Once |
| 162 | defer unblock.Do(func() { close(release) }) |
| 163 | s.verifyMultipartHook = func(ctx context.Context, path, hash string, size int64) error { |
| 164 | close(started) |
| 165 | <-release |
| 166 | return verifyMultipart(ctx, path, hash, size) |
| 167 | } |
| 168 | first := make(chan *httptest.ResponseRecorder, 1) |
| 169 | go func() { first <- record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), 0, body)) }() |
| 170 | select { |
| 171 | case <-started: |
| 172 | case <-time.After(5 * time.Second): |
| 173 | t.Fatal("multipart hash did not start" ) |
| 174 | } |
| 175 | // A different session must reserve space and commit a partial chunk while |
| 176 | // the first session hashes. Ordinary PUT does not take multipartMu and |
| 177 | // would not detect the tenant-wide finalization lock regression. |
| 178 | secondBody := strings.Repeat("b" , 70000) |
| 179 | secondDigest := sha256.Sum256([]byte(secondBody + "remaining" )) |
| 180 | second := make(chan *httptest.ResponseRecorder, 1) |
| 181 | go func() { |
| 182 | second <- record(s.Handler(), multipartReviewRequest(hex.EncodeToString(secondDigest[:]), int64(len(secondBody)+len("remaining" )), 0, secondBody)) |
| 183 | }() |
| 184 | select { |
| 185 | case response := <-second: |
| 186 | if response.Code != http.StatusNoContent { |
| 187 | t.Fatalf("second session = %d: %s" , response.Code, response.Body.String()) |
| 188 | } |
| 189 | case <-time.After(5 * time.Second): |
| 190 | t.Fatal("another multipart session blocked by hashing" ) |
| 191 | } |
| 192 | unblock.Do(func() { close(release) }) |
| 193 | select { |
| 194 | case response := <-first: |
| 195 | if response.Code != http.StatusCreated { |
| 196 | t.Fatalf("finalizing session = %d: %s" , response.Code, response.Body.String()) |
| 197 | } |
| 198 | case <-time.After(5 * time.Second): |
| 199 | t.Fatal("finalizing session did not finish" ) |
| 200 | } |
| 201 | } |
| 202 | |
| 203 | func TestMultipartReviewFinalizingSessionRejectsConcurrentChunk(t *testing.T) { |
| 204 | s := testService(t) |
| 205 | body := "same-session" |
| 206 | digest := sha256.Sum256([]byte(body)) |
| 207 | sha := hex.EncodeToString(digest[:]) |
| 208 | started := make(chan struct{}) |
| 209 | release := make(chan struct{}) |
| 210 | s.verifyMultipartHook = func(ctx context.Context, path, hash string, size int64) error { |
| 211 | close(started) |
| 212 | <-release |
| 213 | return verifyMultipart(ctx, path, hash, size) |
| 214 | } |
| 215 | first := make(chan *httptest.ResponseRecorder, 1) |
| 216 | go func() { first <- record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), 0, body)) }() |
| 217 | select { |
| 218 | case <-started: |
| 219 | case <-time.After(time.Second): |
| 220 | t.Fatal("multipart hash was not started" ) |
| 221 | } |
| 222 | second := record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), 0, body)) |
| 223 | if second.Code != http.StatusConflict { |
| 224 | t.Fatalf("concurrent same-session chunk status = %d: %s" , second.Code, second.Body.String()) |
| 225 | } |
| 226 | close(release) |
| 227 | if response := <-first; response.Code != http.StatusCreated { |
| 228 | t.Fatalf("finalizing upload status = %d: %s" , response.Code, response.Body.String()) |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | func TestMultipartReviewCleanupDoesNotRemoveActiveFinalization(t *testing.T) { |
| 233 | s := testService(t) |
| 234 | body := "cleanup-paused" |
| 235 | digest := sha256.Sum256([]byte(body)) |
| 236 | sha := hex.EncodeToString(digest[:]) |
| 237 | started := make(chan struct{}) |
| 238 | release := make(chan struct{}) |
| 239 | s.verifyMultipartHook = func(ctx context.Context, path, hash string, size int64) error { |
| 240 | close(started) |
| 241 | <-release |
| 242 | return verifyMultipart(ctx, path, hash, size) |
| 243 | } |
| 244 | result := make(chan *httptest.ResponseRecorder, 1) |
| 245 | go func() { result <- record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), 0, body)) }() |
| 246 | select { |
| 247 | case <-started: |
| 248 | case <-time.After(time.Second): |
| 249 | t.Fatal("multipart hash was not started" ) |
| 250 | } |
| 251 | if _, err := s.store.DB().Exec("UPDATE multipart_uploads SET last_seen=0" ); err != nil { |
| 252 | t.Fatal(err) |
| 253 | } |
| 254 | if err := s.CleanupMultipart(context.Background()); err != nil { |
| 255 | t.Fatal(err) |
| 256 | } |
| 257 | id := multipartID(sha, "uploader" ) |
| 258 | if _, err := os.Stat(s.multipartPath(id)); err != nil { |
| 259 | t.Fatalf("active multipart file removed during cleanup: %v" , err) |
| 260 | } |
| 261 | var found int |
| 262 | if err := s.store.DB().QueryRow("SELECT COUNT(*) FROM multipart_uploads WHERE id=?" , id).Scan(&found); err != nil { |
| 263 | t.Fatal(err) |
| 264 | } |
| 265 | if found != 1 { |
| 266 | t.Fatal("active multipart reservation removed during cleanup" ) |
| 267 | } |
| 268 | close(release) |
| 269 | if response := <-result; response.Code != http.StatusCreated { |
| 270 | t.Fatalf("finalizing upload status = %d: %s" , response.Code, response.Body.String()) |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | func TestMultipartReviewPolicyRecheckedAfterHash(t *testing.T) { |
| 275 | s := testService(t) |
| 276 | body := "policy-after-hash" |
| 277 | digest := sha256.Sum256([]byte(body)) |
| 278 | sha := hex.EncodeToString(digest[:]) |
| 279 | started := make(chan struct{}) |
| 280 | release := make(chan struct{}) |
| 281 | s.verifyMultipartHook = func(ctx context.Context, path, hash string, size int64) error { |
| 282 | close(started) |
| 283 | <-release |
| 284 | return verifyMultipart(ctx, path, hash, size) |
| 285 | } |
| 286 | s.config.Limits = func() Limits { return Limits{MaxFileBytes: int64(len(body))} } |
| 287 | result := make(chan *httptest.ResponseRecorder, 1) |
| 288 | go func() { result <- record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), 0, body)) }() |
| 289 | <-started |
| 290 | s.config.Limits = func() Limits { return Limits{MaxFileBytes: int64(len(body) - 1)} } |
| 291 | close(release) |
| 292 | if response := <-result; response.Code != http.StatusRequestEntityTooLarge { |
| 293 | t.Fatalf("policy-changed upload status = %d: %s" , response.Code, response.Body.String()) |
| 294 | } |
| 295 | if _, _, err := s.Get(context.Background(), sha); !errors.Is(err, sql.ErrNoRows) { |
| 296 | t.Fatalf("policy-rejected upload lookup = %v" , err) |
| 297 | } |
| 298 | } |
| 299 | |
| 300 | func TestMultipartReviewCanceledHashCanBeRetried(t *testing.T) { |
| 301 | s := testService(t) |
| 302 | body := "cancel-and-retry" |
| 303 | digest := sha256.Sum256([]byte(body)) |
| 304 | sha := hex.EncodeToString(digest[:]) |
| 305 | started := make(chan struct{}) |
| 306 | release := make(chan struct{}) |
| 307 | s.verifyMultipartHook = func(ctx context.Context, path, hash string, size int64) error { |
| 308 | close(started) |
| 309 | <-release |
| 310 | return verifyMultipart(ctx, path, hash, size) |
| 311 | } |
| 312 | ctx, cancel := context.WithCancel(context.Background()) |
| 313 | r := multipartReviewRequest(sha, int64(len(body)), 0, body).WithContext(ctx) |
| 314 | result := make(chan *httptest.ResponseRecorder, 1) |
| 315 | go func() { result <- record(s.Handler(), r) }() |
| 316 | <-started |
| 317 | cancel() |
| 318 | close(release) |
| 319 | <-result |
| 320 | s.verifyMultipartHook = nil |
| 321 | if response := record(s.Handler(), multipartReviewRequest(sha, int64(len(body)), 0, body)); response.Code != http.StatusCreated { |
| 322 | t.Fatalf("retry after canceled hash status = %d: %s" , response.Code, response.Body.String()) |
| 323 | } |
| 324 | } |
| 325 | |
| 326 | type multipartBlockingBody struct { |
| 327 | io.Reader |
| 328 | started chan<- struct{} |
| 329 | release <-chan struct{} |
| 330 | once sync.Once |
| 331 | } |
| 332 | |
| 333 | func (b *multipartBlockingBody) Read(p []byte) (int, error) { |
| 334 | b.once.Do(func() { close(b.started) }) |
| 335 | <-b.release |
| 336 | return b.Reader.Read(p) |
| 337 | } |
| 338 | |
| 339 | func (b *multipartBlockingBody) Close() error { return nil } |
| 340 | |
| 341 | func TestMultipartReviewReservationsBlockUploadsAndReleaseOnExpiry(t *testing.T) { |
| 342 | s := testService(t) |
| 343 | s.config.Limits = func() Limits { return Limits{UserStorageBytes: 8} } |
| 344 | full := "12345678" |
| 345 | digest := sha256.Sum256([]byte(full)) |
| 346 | sha := hex.EncodeToString(digest[:]) |
| 347 | partial := multipartReviewRequest(sha, int64(len(full)), 0, full[:4]) |
| 348 | partial.ContentLength = 4 |
| 349 | if response := record(s.Handler(), partial); response.Code != http.StatusNoContent { |
| 350 | t.Fatalf("partial reservation status = %d: %s" , response.Code, response.Body.String()) |
| 351 | } |
| 352 | if _, err := s.Put(context.Background(), PutOptions{Reader: strings.NewReader("1234" ), Type: "text/plain" , Uploader: "uploader" }); !errors.Is(err, ErrQuotaExceeded) { |
| 353 | t.Fatalf("ordinary upload after reservation = %v" , err) |
| 354 | } |
| 355 | other := "abcdefgh" |
| 356 | otherDigest := sha256.Sum256([]byte(other)) |
| 357 | second := multipartReviewRequest(hex.EncodeToString(otherDigest[:]), int64(len(other)), 0, other[:4]) |
| 358 | second.ContentLength = 4 |
| 359 | if response := record(s.Handler(), second); response.Code != http.StatusForbidden { |
| 360 | t.Fatalf("second reserved upload status = %d" , response.Code) |
| 361 | } |
| 362 | if _, err := s.store.DB().Exec("UPDATE multipart_uploads SET last_seen=0" ); err != nil { |
| 363 | t.Fatal(err) |
| 364 | } |
| 365 | if err := s.CleanupMultipart(context.Background()); err != nil { |
| 366 | t.Fatal(err) |
| 367 | } |
| 368 | matches, err := filepath.Glob(filepath.Join(s.root, ".multipart-*" )) |
| 369 | if err != nil { |
| 370 | t.Fatal(err) |
| 371 | } |
| 372 | if len(matches) != 0 { |
| 373 | t.Fatalf("expired multipart files remain: %v" , matches) |
| 374 | } |
| 375 | if _, err := s.Put(context.Background(), PutOptions{Reader: strings.NewReader("1234" ), Type: "text/plain" , Uploader: "uploader" }); err != nil { |
| 376 | t.Fatalf("ordinary upload after expiry: %v" , err) |
| 377 | } |
| 378 | } |
| 379 | |
| 380 | func TestMultipartReviewHashMismatchReleasesReservation(t *testing.T) { |
| 381 | s := testService(t) |
| 382 | s.config.Limits = func() Limits { return Limits{UserStorageBytes: 5} } |
| 383 | wrong := strings.Repeat("a" , 64) |
| 384 | r := multipartReviewRequest(wrong, 5, 0, "wrong" ) |
| 385 | r.ContentLength = 5 |
| 386 | if response := record(s.Handler(), r); response.Code != http.StatusConflict { |
| 387 | t.Fatalf("hash mismatch status = %d: %s" , response.Code, response.Body.String()) |
| 388 | } |
| 389 | if _, err := s.Put(context.Background(), PutOptions{Reader: strings.NewReader("right" ), Type: "text/plain" , Uploader: "uploader" }); err != nil { |
| 390 | t.Fatalf("upload after mismatch reservation: %v" , err) |
| 391 | } |
| 392 | } |
| 393 | |
| 394 | func TestMultipartReviewExistingBlobCannotBypassModerationBlock(t *testing.T) { |
| 395 | s := testService(t) |
| 396 | body := "blocked" |
| 397 | digest := sha256.Sum256([]byte(body)) |
| 398 | sha := hex.EncodeToString(digest[:]) |
| 399 | if _, err := s.Put(context.Background(), PutOptions{Reader: strings.NewReader(body), Type: "text/plain" }); err != nil { |
| 400 | t.Fatal(err) |
| 401 | } |
| 402 | if _, err := s.store.DB().Exec("INSERT INTO blob_blocks(sha256,reason,blocked_at) VALUES(?,?,?)" , sha, "review" , time.Now().Unix()); err != nil { |
| 403 | t.Fatal(err) |
| 404 | } |
| 405 | r := multipartReviewRequest(sha, int64(len(body)), 0, body) |
| 406 | r.ContentLength = int64(len(body)) |
| 407 | if response := record(s.Handler(), r); response.Code != http.StatusForbidden { |
| 408 | t.Fatalf("blocked existing blob status = %d: %s" , response.Code, response.Body.String()) |
| 409 | } |
| 410 | } |