src/import/pull-job.c

   1 /* SPDX-License-Identifier: LGPL-2.1+ */
   2
   3 #include <fcntl.h>
   4 #include <sys/stat.h>
   5 #include <sys/xattr.h>
   6
   7 #include "alloc-util.h"
   8 #include "fd-util.h"
   9 #include "gcrypt-util.h"
  10 #include "hexdecoct.h"
  11 #include "import-util.h"
  12 #include "io-util.h"
  13 #include "machine-pool.h"
  14 #include "parse-util.h"
  15 #include "pull-common.h"
  16 #include "pull-job.h"
  17 #include "string-util.h"
  18 #include "strv.h"
  19 #include "xattr-util.h"
  20
  21 PullJob* pull_job_unref(PullJob *j) {
  22         if (!j)
  23                 return NULL;
  24
  25         curl_glue_remove_and_free(j->glue, j->curl);
  26         curl_slist_free_all(j->request_header);
  27
  28         safe_close(j->disk_fd);
  29
  30         import_compress_free(&j->compress);
  31
  32         if (j->checksum_context)
  33                 gcry_md_close(j->checksum_context);
  34
  35         free(j->url);
  36         free(j->etag);
  37         strv_free(j->old_etags);
  38         free(j->payload);
  39         free(j->checksum);
  40
  41         return mfree(j);
  42 }
  43
  44 static void pull_job_finish(PullJob *j, int ret) {
  45         assert(j);
  46
  47         if (IN_SET(j->state, PULL_JOB_DONE, PULL_JOB_FAILED))
  48                 return;
  49
  50         if (ret == 0) {
  51                 j->state = PULL_JOB_DONE;
  52                 j->progress_percent = 100;
  53                 log_info("Download of %s complete.", j->url);
  54         } else {
  55                 j->state = PULL_JOB_FAILED;
  56                 j->error = ret;
  57         }
  58
  59         if (j->on_finished)
  60                 j->on_finished(j);
  61 }
  62
  63 static int pull_job_restart(PullJob *j) {
  64         int r;
  65         char *chksum_url = NULL;
  66
  67         r = import_url_change_last_component(j->url, "SHA256SUMS", &chksum_url);
  68         if (r < 0)
  69                 return r;
  70
  71         free(j->url);
  72         j->url = chksum_url;
  73         j->state = PULL_JOB_INIT;
  74         j->payload = mfree(j->payload);
  75         j->payload_size = 0;
  76         j->payload_allocated = 0;
  77         j->written_compressed = 0;
  78         j->written_uncompressed = 0;
  79
  80         r = pull_job_begin(j);
  81         if (r < 0)
  82                 return r;
  83
  84         return 0;
  85 }
  86
  87 void pull_job_curl_on_finished(CurlGlue *g, CURL *curl, CURLcode result) {
  88         PullJob *j = NULL;
  89         CURLcode code;
  90         long status;
  91         int r;
  92
  93         if (curl_easy_getinfo(curl, CURLINFO_PRIVATE, (char **)&j) != CURLE_OK)
  94                 return;
  95
  96         if (!j || IN_SET(j->state, PULL_JOB_DONE, PULL_JOB_FAILED))
  97                 return;
  98
  99         if (result != CURLE_OK) {
 100                 log_error("Transfer failed: %s", curl_easy_strerror(result));
 101                 r = -EIO;
 102                 goto finish;
 103         }
 104
 105         code = curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &status);
 106         if (code != CURLE_OK) {
 107                 log_error("Failed to retrieve response code: %s", curl_easy_strerror(code));
 108                 r = -EIO;
 109                 goto finish;
 110         } else if (status == 304) {
 111                 log_info("Image already downloaded. Skipping download.");
 112                 j->etag_exists = true;
 113                 r = 0;
 114                 goto finish;
 115         } else if (status >= 300) {
 116                 if (status == 404 && j->style == VERIFICATION_PER_FILE) {
 117
 118                         /* retry pull job with SHA256SUMS file */
 119                         r = pull_job_restart(j);
 120                         if (r < 0)
 121                                 goto finish;
 122
 123                         code = curl_easy_getinfo(j->curl, CURLINFO_RESPONSE_CODE, &status);
 124                         if (code != CURLE_OK) {
 125                                 log_error("Failed to retrieve response code: %s", curl_easy_strerror(code));
 126                                 r = -EIO;
 127                                 goto finish;
 128                         }
 129
 130                         if (status == 0) {
 131                                 j->style = VERIFICATION_PER_DIRECTORY;
 132                                 return;
 133                         }
 134                 }
 135
 136                 log_error("HTTP request to %s failed with code %li.", j->url, status);
 137                 r = -EIO;
 138                 goto finish;
 139         } else if (status < 200) {
 140                 log_error("HTTP request to %s finished with unexpected code %li.", j->url, status);
 141                 r = -EIO;
 142                 goto finish;
 143         }
 144
 145         if (j->state != PULL_JOB_RUNNING) {
 146                 log_error("Premature connection termination.");
 147                 r = -EIO;
 148                 goto finish;
 149         }
 150
 151         if (j->content_length != (uint64_t) -1 &&
 152             j->content_length != j->written_compressed) {
 153                 log_error("Download truncated.");
 154                 r = -EIO;
 155                 goto finish;
 156         }
 157
 158         if (j->checksum_context) {
 159                 uint8_t *k;
 160
 161                 k = gcry_md_read(j->checksum_context, GCRY_MD_SHA256);
 162                 if (!k) {
 163                         log_error("Failed to get checksum.");
 164                         r = -EIO;
 165                         goto finish;
 166                 }
 167
 168                 j->checksum = hexmem(k, gcry_md_get_algo_dlen(GCRY_MD_SHA256));
 169                 if (!j->checksum) {
 170                         r = log_oom();
 171                         goto finish;
 172                 }
 173
 174                 log_debug("SHA256 of %s is %s.", j->url, j->checksum);
 175         }
 176
 177         if (j->disk_fd >= 0 && j->allow_sparse) {
 178                 /* Make sure the file size is right, in case the file was
 179                  * sparse and we just seeked for the last part */
 180
 181                 if (ftruncate(j->disk_fd, j->written_uncompressed) < 0) {
 182                         r = log_error_errno(errno, "Failed to truncate file: %m");
 183                         goto finish;
 184                 }
 185
 186                 if (j->etag)
 187                         (void) fsetxattr(j->disk_fd, "user.source_etag", j->etag, strlen(j->etag), 0);
 188                 if (j->url)
 189                         (void) fsetxattr(j->disk_fd, "user.source_url", j->url, strlen(j->url), 0);
 190
 191                 if (j->mtime != 0) {
 192                         struct timespec ut[2];
 193
 194                         timespec_store(&ut[0], j->mtime);
 195                         ut[1] = ut[0];
 196                         (void) futimens(j->disk_fd, ut);
 197
 198                         (void) fd_setcrtime(j->disk_fd, j->mtime);
 199                 }
 200         }
 201
 202         r = 0;
 203
 204 finish:
 205         pull_job_finish(j, r);
 206 }
 207
 208 static int pull_job_write_uncompressed(const void *p, size_t sz, void *userdata) {
 209         PullJob *j = userdata;
 210         ssize_t n;
 211
 212         assert(j);
 213         assert(p);
 214
 215         if (sz <= 0)
 216                 return 0;
 217
 218         if (j->written_uncompressed + sz < j->written_uncompressed)
 219                 return log_error_errno(SYNTHETIC_ERRNO(EOVERFLOW),
 220                                        "File too large, overflow");
 221
 222         if (j->written_uncompressed + sz > j->uncompressed_max)
 223                 return log_error_errno(SYNTHETIC_ERRNO(EFBIG),
 224                                        "File overly large, refusing");
 225
 226         if (j->disk_fd >= 0) {
 227
 228                 if (j->allow_sparse)
 229                         n = sparse_write(j->disk_fd, p, sz, 64);
 230                 else {
 231                         n = write(j->disk_fd, p, sz);
 232                         if (n < 0)
 233                                 n = -errno;
 234                 }
 235                 if (n < 0)
 236                         return log_error_errno((int) n, "Failed to write file: %m");
 237                 if ((size_t) n < sz)
 238                         return log_error_errno(SYNTHETIC_ERRNO(EIO), "Short write");
 239         } else {
 240
 241                 if (!GREEDY_REALLOC(j->payload, j->payload_allocated, j->payload_size + sz))
 242                         return log_oom();
 243
 244                 memcpy(j->payload + j->payload_size, p, sz);
 245                 j->payload_size += sz;
 246         }
 247
 248         j->written_uncompressed += sz;
 249
 250         return 0;
 251 }
 252
 253 static int pull_job_write_compressed(PullJob *j, void *p, size_t sz) {
 254         int r;
 255
 256         assert(j);
 257         assert(p);
 258
 259         if (sz <= 0)
 260                 return 0;
 261
 262         if (j->written_compressed + sz < j->written_compressed)
 263                 return log_error_errno(SYNTHETIC_ERRNO(EOVERFLOW), "File too large, overflow");
 264
 265         if (j->written_compressed + sz > j->compressed_max)
 266                 return log_error_errno(SYNTHETIC_ERRNO(EFBIG), "File overly large, refusing.");
 267
 268         if (j->content_length != (uint64_t) -1 &&
 269             j->written_compressed + sz > j->content_length)
 270                 return log_error_errno(SYNTHETIC_ERRNO(EFBIG),
 271                                        "Content length incorrect.");
 272
 273         if (j->checksum_context)
 274                 gcry_md_write(j->checksum_context, p, sz);
 275
 276         r = import_uncompress(&j->compress, p, sz, pull_job_write_uncompressed, j);
 277         if (r < 0)
 278                 return r;
 279
 280         j->written_compressed += sz;
 281
 282         return 0;
 283 }
 284
 285 static int pull_job_open_disk(PullJob *j) {
 286         int r;
 287
 288         assert(j);
 289
 290         if (j->on_open_disk) {
 291                 r = j->on_open_disk(j);
 292                 if (r < 0)
 293                         return r;
 294         }
 295
 296         if (j->disk_fd >= 0) {
 297                 /* Check if we can do sparse files */
 298
 299                 if (lseek(j->disk_fd, SEEK_SET, 0) == 0)
 300                         j->allow_sparse = true;
 301                 else {
 302                         if (errno != ESPIPE)
 303                                 return log_error_errno(errno, "Failed to seek on file descriptor: %m");
 304
 305                         j->allow_sparse = false;
 306                 }
 307         }
 308
 309         if (j->calc_checksum) {
 310                 initialize_libgcrypt(false);
 311
 312                 if (gcry_md_open(&j->checksum_context, GCRY_MD_SHA256, 0) != 0)
 313                         return log_error_errno(SYNTHETIC_ERRNO(EIO),
 314                                                "Failed to initialize hash context.");
 315         }
 316
 317         return 0;
 318 }
 319
 320 static int pull_job_detect_compression(PullJob *j) {
 321         _cleanup_free_ uint8_t *stub = NULL;
 322         size_t stub_size;
 323
 324         int r;
 325
 326         assert(j);
 327
 328         r = import_uncompress_detect(&j->compress, j->payload, j->payload_size);
 329         if (r < 0)
 330                 return log_error_errno(r, "Failed to initialize compressor: %m");
 331         if (r == 0)
 332                 return 0;
 333
 334         log_debug("Stream is compressed: %s", import_compress_type_to_string(j->compress.type));
 335
 336         r = pull_job_open_disk(j);
 337         if (r < 0)
 338                 return r;
 339
 340         /* Now, take the payload we read so far, and decompress it */
 341         stub = j->payload;
 342         stub_size = j->payload_size;
 343
 344         j->payload = NULL;
 345         j->payload_size = 0;
 346         j->payload_allocated = 0;
 347
 348         j->state = PULL_JOB_RUNNING;
 349
 350         r = pull_job_write_compressed(j, stub, stub_size);
 351         if (r < 0)
 352                 return r;
 353
 354         return 0;
 355 }
 356
 357 static size_t pull_job_write_callback(void *contents, size_t size, size_t nmemb, void *userdata) {
 358         PullJob *j = userdata;
 359         size_t sz = size * nmemb;
 360         int r;
 361
 362         assert(contents);
 363         assert(j);
 364
 365         switch (j->state) {
 366
 367         case PULL_JOB_ANALYZING:
 368                 /* Let's first check what it actually is */
 369
 370                 if (!GREEDY_REALLOC(j->payload, j->payload_allocated, j->payload_size + sz)) {
 371                         r = log_oom();
 372                         goto fail;
 373                 }
 374
 375                 memcpy(j->payload + j->payload_size, contents, sz);
 376                 j->payload_size += sz;
 377
 378                 r = pull_job_detect_compression(j);
 379                 if (r < 0)
 380                         goto fail;
 381
 382                 break;
 383
 384         case PULL_JOB_RUNNING:
 385
 386                 r = pull_job_write_compressed(j, contents, sz);
 387                 if (r < 0)
 388                         goto fail;
 389
 390                 break;
 391
 392         case PULL_JOB_DONE:
 393         case PULL_JOB_FAILED:
 394                 r = -ESTALE;
 395                 goto fail;
 396
 397         default:
 398                 assert_not_reached("Impossible state.");
 399         }
 400
 401         return sz;
 402
 403 fail:
 404         pull_job_finish(j, r);
 405         return 0;
 406 }
 407
 408 static size_t pull_job_header_callback(void *contents, size_t size, size_t nmemb, void *userdata) {
 409         PullJob *j = userdata;
 410         size_t sz = size * nmemb;
 411         _cleanup_free_ char *length = NULL, *last_modified = NULL;
 412         char *etag;
 413         int r;
 414
 415         assert(contents);
 416         assert(j);
 417
 418         if (IN_SET(j->state, PULL_JOB_DONE, PULL_JOB_FAILED)) {
 419                 r = -ESTALE;
 420                 goto fail;
 421         }
 422
 423         assert(j->state == PULL_JOB_ANALYZING);
 424
 425         r = curl_header_strdup(contents, sz, "ETag:", &etag);
 426         if (r < 0) {
 427                 log_oom();
 428                 goto fail;
 429         }
 430         if (r > 0) {
 431                 free(j->etag);
 432                 j->etag = etag;
 433
 434                 if (strv_contains(j->old_etags, j->etag)) {
 435                         log_info("Image already downloaded. Skipping download.");
 436                         j->etag_exists = true;
 437                         pull_job_finish(j, 0);
 438                         return sz;
 439                 }
 440
 441                 return sz;
 442         }
 443
 444         r = curl_header_strdup(contents, sz, "Content-Length:", &length);
 445         if (r < 0) {
 446                 log_oom();
 447                 goto fail;
 448         }
 449         if (r > 0) {
 450                 (void) safe_atou64(length, &j->content_length);
 451
 452                 if (j->content_length != (uint64_t) -1) {
 453                         char bytes[FORMAT_BYTES_MAX];
 454
 455                         if (j->content_length > j->compressed_max) {
 456                                 log_error("Content too large.");
 457                                 r = -EFBIG;
 458                                 goto fail;
 459                         }
 460
 461                         log_info("Downloading %s for %s.", format_bytes(bytes, sizeof(bytes), j->content_length), j->url);
 462                 }
 463
 464                 return sz;
 465         }
 466
 467         r = curl_header_strdup(contents, sz, "Last-Modified:", &last_modified);
 468         if (r < 0) {
 469                 log_oom();
 470                 goto fail;
 471         }
 472         if (r > 0) {
 473                 (void) curl_parse_http_time(last_modified, &j->mtime);
 474                 return sz;
 475         }
 476
 477         if (j->on_header) {
 478                 r = j->on_header(j, contents, sz);
 479                 if (r < 0)
 480                         goto fail;
 481         }
 482
 483         return sz;
 484
 485 fail:
 486         pull_job_finish(j, r);
 487         return 0;
 488 }
 489
 490 static int pull_job_progress_callback(void *userdata, curl_off_t dltotal, curl_off_t dlnow, curl_off_t ultotal, curl_off_t ulnow) {
 491         PullJob *j = userdata;
 492         unsigned percent;
 493         usec_t n;
 494
 495         assert(j);
 496
 497         if (dltotal <= 0)
 498                 return 0;
 499
 500         percent = ((100 * dlnow) / dltotal);
 501         n = now(CLOCK_MONOTONIC);
 502
 503         if (n > j->last_status_usec + USEC_PER_SEC &&
 504             percent != j->progress_percent &&
 505             dlnow < dltotal) {
 506                 char buf[FORMAT_TIMESPAN_MAX];
 507
 508                 if (n - j->start_usec > USEC_PER_SEC && dlnow > 0) {
 509                         char y[FORMAT_BYTES_MAX];
 510                         usec_t left, done;
 511
 512                         done = n - j->start_usec;
 513                         left = (usec_t) (((double) done * (double) dltotal) / dlnow) - done;
 514
 515                         log_info("Got %u%% of %s. %s left at %s/s.",
 516                                  percent,
 517                                  j->url,
 518                                  format_timespan(buf, sizeof(buf), left, USEC_PER_SEC),
 519                                  format_bytes(y, sizeof(y), (uint64_t) ((double) dlnow / ((double) done / (double) USEC_PER_SEC))));
 520                 } else
 521                         log_info("Got %u%% of %s.", percent, j->url);
 522
 523                 j->progress_percent = percent;
 524                 j->last_status_usec = n;
 525
 526                 if (j->on_progress)
 527                         j->on_progress(j);
 528         }
 529
 530         return 0;
 531 }
 532
 533 int pull_job_new(PullJob **ret, const char *url, CurlGlue *glue, void *userdata) {
 534         _cleanup_(pull_job_unrefp) PullJob *j = NULL;
 535         _cleanup_free_ char *u = NULL;
 536
 537         assert(url);
 538         assert(glue);
 539         assert(ret);
 540
 541         u = strdup(url);
 542         if (!u)
 543                 return -ENOMEM;
 544
 545         j = new(PullJob, 1);
 546         if (!j)
 547                 return -ENOMEM;
 548
 549         *j = (PullJob) {
 550                 .state = PULL_JOB_INIT,
 551                 .disk_fd = -1,
 552                 .userdata = userdata,
 553                 .glue = glue,
 554                 .content_length = (uint64_t) -1,
 555                 .start_usec = now(CLOCK_MONOTONIC),
 556                 .compressed_max = 64LLU * 1024LLU * 1024LLU * 1024LLU, /* 64GB safety limit */
 557                 .uncompressed_max = 64LLU * 1024LLU * 1024LLU * 1024LLU, /* 64GB safety limit */
 558                 .style = VERIFICATION_STYLE_UNSET,
 559                 .url = TAKE_PTR(u),
 560         };
 561
 562         *ret = TAKE_PTR(j);
 563
 564         return 0;
 565 }
 566
 567 int pull_job_begin(PullJob *j) {
 568         int r;
 569
 570         assert(j);
 571
 572         if (j->state != PULL_JOB_INIT)
 573                 return -EBUSY;
 574
 575         r = curl_glue_make(&j->curl, j->url, j);
 576         if (r < 0)
 577                 return r;
 578
 579         if (!strv_isempty(j->old_etags)) {
 580                 _cleanup_free_ char *cc = NULL, *hdr = NULL;
 581
 582                 cc = strv_join(j->old_etags, ", ");
 583                 if (!cc)
 584                         return -ENOMEM;
 585
 586                 hdr = strappend("If-None-Match: ", cc);
 587                 if (!hdr)
 588                         return -ENOMEM;
 589
 590                 if (!j->request_header) {
 591                         j->request_header = curl_slist_new(hdr, NULL);
 592                         if (!j->request_header)
 593                                 return -ENOMEM;
 594                 } else {
 595                         struct curl_slist *l;
 596
 597                         l = curl_slist_append(j->request_header, hdr);
 598                         if (!l)
 599                                 return -ENOMEM;
 600
 601                         j->request_header = l;
 602                 }
 603         }
 604
 605         if (j->request_header) {
 606                 if (curl_easy_setopt(j->curl, CURLOPT_HTTPHEADER, j->request_header) != CURLE_OK)
 607                         return -EIO;
 608         }
 609
 610         if (curl_easy_setopt(j->curl, CURLOPT_WRITEFUNCTION, pull_job_write_callback) != CURLE_OK)
 611                 return -EIO;
 612
 613         if (curl_easy_setopt(j->curl, CURLOPT_WRITEDATA, j) != CURLE_OK)
 614                 return -EIO;
 615
 616         if (curl_easy_setopt(j->curl, CURLOPT_HEADERFUNCTION, pull_job_header_callback) != CURLE_OK)
 617                 return -EIO;
 618
 619         if (curl_easy_setopt(j->curl, CURLOPT_HEADERDATA, j) != CURLE_OK)
 620                 return -EIO;
 621
 622         if (curl_easy_setopt(j->curl, CURLOPT_XFERINFOFUNCTION, pull_job_progress_callback) != CURLE_OK)
 623                 return -EIO;
 624
 625         if (curl_easy_setopt(j->curl, CURLOPT_XFERINFODATA, j) != CURLE_OK)
 626                 return -EIO;
 627
 628         if (curl_easy_setopt(j->curl, CURLOPT_NOPROGRESS, 0) != CURLE_OK)
 629                 return -EIO;
 630
 631         r = curl_glue_add(j->glue, j->curl);
 632         if (r < 0)
 633                 return r;
 634
 635         j->state = PULL_JOB_ANALYZING;
 636
 637         return 0;
 638 }