src/import/pull-job.c

   1 /***
   2   This file is part of systemd.
   3
   4   Copyright 2015 Lennart Poettering
   5
   6   systemd is free software; you can redistribute it and/or modify it
   7   under the terms of the GNU Lesser General Public License as published by
   8   the Free Software Foundation; either version 2.1 of the License, or
   9   (at your option) any later version.
  10
  11   systemd is distributed in the hope that it will be useful, but
  12   WITHOUT ANY WARRANTY; without even the implied warranty of
  13   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  14   Lesser General Public License for more details.
  15
  16   You should have received a copy of the GNU Lesser General Public License
  17   along with systemd; If not, see <http://www.gnu.org/licenses/>.
  18 ***/
  19
  20 #include <sys/xattr.h>
  21
  22 #include "alloc-util.h"
  23 #include "fd-util.h"
  24 #include "hexdecoct.h"
  25 #include "io-util.h"
  26 #include "machine-pool.h"
  27 #include "parse-util.h"
  28 #include "pull-job.h"
  29 #include "string-util.h"
  30 #include "strv.h"
  31 #include "xattr-util.h"
  32
  33 PullJob* pull_job_unref(PullJob *j) {
  34         if (!j)
  35                 return NULL;
  36
  37         curl_glue_remove_and_free(j->glue, j->curl);
  38         curl_slist_free_all(j->request_header);
  39
  40         safe_close(j->disk_fd);
  41
  42         import_compress_free(&j->compress);
  43
  44         if (j->checksum_context)
  45                 gcry_md_close(j->checksum_context);
  46
  47         free(j->url);
  48         free(j->etag);
  49         strv_free(j->old_etags);
  50         free(j->payload);
  51         free(j->checksum);
  52
  53         return mfree(j);
  54 }
  55
  56 static void pull_job_finish(PullJob *j, int ret) {
  57         assert(j);
  58
  59         if (j->state == PULL_JOB_DONE ||
  60             j->state == PULL_JOB_FAILED)
  61                 return;
  62
  63         if (ret == 0) {
  64                 j->state = PULL_JOB_DONE;
  65                 j->progress_percent = 100;
  66                 log_info("Download of %s complete.", j->url);
  67         } else {
  68                 j->state = PULL_JOB_FAILED;
  69                 j->error = ret;
  70         }
  71
  72         if (j->on_finished)
  73                 j->on_finished(j);
  74 }
  75
  76 void pull_job_curl_on_finished(CurlGlue *g, CURL *curl, CURLcode result) {
  77         PullJob *j = NULL;
  78         CURLcode code;
  79         long status;
  80         int r;
  81
  82         if (curl_easy_getinfo(curl, CURLINFO_PRIVATE, (char **)&j) != CURLE_OK)
  83                 return;
  84
  85         if (!j || j->state == PULL_JOB_DONE || j->state == PULL_JOB_FAILED)
  86                 return;
  87
  88         if (result != CURLE_OK) {
  89                 log_error("Transfer failed: %s", curl_easy_strerror(result));
  90                 r = -EIO;
  91                 goto finish;
  92         }
  93
  94         code = curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &status);
  95         if (code != CURLE_OK) {
  96                 log_error("Failed to retrieve response code: %s", curl_easy_strerror(code));
  97                 r = -EIO;
  98                 goto finish;
  99         } else if (status == 304) {
 100                 log_info("Image already downloaded. Skipping download.");
 101                 j->etag_exists = true;
 102                 r = 0;
 103                 goto finish;
 104         } else if (status >= 300) {
 105                 log_error("HTTP request to %s failed with code %li.", j->url, status);
 106                 r = -EIO;
 107                 goto finish;
 108         } else if (status < 200) {
 109                 log_error("HTTP request to %s finished with unexpected code %li.", j->url, status);
 110                 r = -EIO;
 111                 goto finish;
 112         }
 113
 114         if (j->state != PULL_JOB_RUNNING) {
 115                 log_error("Premature connection termination.");
 116                 r = -EIO;
 117                 goto finish;
 118         }
 119
 120         if (j->content_length != (uint64_t) -1 &&
 121             j->content_length != j->written_compressed) {
 122                 log_error("Download truncated.");
 123                 r = -EIO;
 124                 goto finish;
 125         }
 126
 127         if (j->checksum_context) {
 128                 uint8_t *k;
 129
 130                 k = gcry_md_read(j->checksum_context, GCRY_MD_SHA256);
 131                 if (!k) {
 132                         log_error("Failed to get checksum.");
 133                         r = -EIO;
 134                         goto finish;
 135                 }
 136
 137                 j->checksum = hexmem(k, gcry_md_get_algo_dlen(GCRY_MD_SHA256));
 138                 if (!j->checksum) {
 139                         r = log_oom();
 140                         goto finish;
 141                 }
 142
 143                 log_debug("SHA256 of %s is %s.", j->url, j->checksum);
 144         }
 145
 146         if (j->disk_fd >= 0 && j->allow_sparse) {
 147                 /* Make sure the file size is right, in case the file was
 148                  * sparse and we just seeked for the last part */
 149
 150                 if (ftruncate(j->disk_fd, j->written_uncompressed) < 0) {
 151                         r = log_error_errno(errno, "Failed to truncate file: %m");
 152                         goto finish;
 153                 }
 154
 155                 if (j->etag)
 156                         (void) fsetxattr(j->disk_fd, "user.source_etag", j->etag, strlen(j->etag), 0);
 157                 if (j->url)
 158                         (void) fsetxattr(j->disk_fd, "user.source_url", j->url, strlen(j->url), 0);
 159
 160                 if (j->mtime != 0) {
 161                         struct timespec ut[2];
 162
 163                         timespec_store(&ut[0], j->mtime);
 164                         ut[1] = ut[0];
 165                         (void) futimens(j->disk_fd, ut);
 166
 167                         (void) fd_setcrtime(j->disk_fd, j->mtime);
 168                 }
 169         }
 170
 171         r = 0;
 172
 173 finish:
 174         pull_job_finish(j, r);
 175 }
 176
 177 static int pull_job_write_uncompressed(const void *p, size_t sz, void *userdata) {
 178         PullJob *j = userdata;
 179         ssize_t n;
 180
 181         assert(j);
 182         assert(p);
 183
 184         if (sz <= 0)
 185                 return 0;
 186
 187         if (j->written_uncompressed + sz < j->written_uncompressed) {
 188                 log_error("File too large, overflow");
 189                 return -EOVERFLOW;
 190         }
 191
 192         if (j->written_uncompressed + sz > j->uncompressed_max) {
 193                 log_error("File overly large, refusing");
 194                 return -EFBIG;
 195         }
 196
 197         if (j->disk_fd >= 0) {
 198
 199                 if (j->grow_machine_directory && j->written_since_last_grow >= GROW_INTERVAL_BYTES) {
 200                         j->written_since_last_grow = 0;
 201                         grow_machine_directory();
 202                 }
 203
 204                 if (j->allow_sparse)
 205                         n = sparse_write(j->disk_fd, p, sz, 64);
 206                 else
 207                         n = write(j->disk_fd, p, sz);
 208                 if (n < 0)
 209                         return log_error_errno(errno, "Failed to write file: %m");
 210                 if ((size_t) n < sz) {
 211                         log_error("Short write");
 212                         return -EIO;
 213                 }
 214         } else {
 215
 216                 if (!GREEDY_REALLOC(j->payload, j->payload_allocated, j->payload_size + sz))
 217                         return log_oom();
 218
 219                 memcpy(j->payload + j->payload_size, p, sz);
 220                 j->payload_size += sz;
 221         }
 222
 223         j->written_uncompressed += sz;
 224         j->written_since_last_grow += sz;
 225
 226         return 0;
 227 }
 228
 229 static int pull_job_write_compressed(PullJob *j, void *p, size_t sz) {
 230         int r;
 231
 232         assert(j);
 233         assert(p);
 234
 235         if (sz <= 0)
 236                 return 0;
 237
 238         if (j->written_compressed + sz < j->written_compressed) {
 239                 log_error("File too large, overflow");
 240                 return -EOVERFLOW;
 241         }
 242
 243         if (j->written_compressed + sz > j->compressed_max) {
 244                 log_error("File overly large, refusing.");
 245                 return -EFBIG;
 246         }
 247
 248         if (j->content_length != (uint64_t) -1 &&
 249             j->written_compressed + sz > j->content_length) {
 250                 log_error("Content length incorrect.");
 251                 return -EFBIG;
 252         }
 253
 254         if (j->checksum_context)
 255                 gcry_md_write(j->checksum_context, p, sz);
 256
 257         r = import_uncompress(&j->compress, p, sz, pull_job_write_uncompressed, j);
 258         if (r < 0)
 259                 return r;
 260
 261         j->written_compressed += sz;
 262
 263         return 0;
 264 }
 265
 266 static int pull_job_open_disk(PullJob *j) {
 267         int r;
 268
 269         assert(j);
 270
 271         if (j->on_open_disk) {
 272                 r = j->on_open_disk(j);
 273                 if (r < 0)
 274                         return r;
 275         }
 276
 277         if (j->disk_fd >= 0) {
 278                 /* Check if we can do sparse files */
 279
 280                 if (lseek(j->disk_fd, SEEK_SET, 0) == 0)
 281                         j->allow_sparse = true;
 282                 else {
 283                         if (errno != ESPIPE)
 284                                 return log_error_errno(errno, "Failed to seek on file descriptor: %m");
 285
 286                         j->allow_sparse = false;
 287                 }
 288         }
 289
 290         if (j->calc_checksum) {
 291                 if (gcry_md_open(&j->checksum_context, GCRY_MD_SHA256, 0) != 0) {
 292                         log_error("Failed to initialize hash context.");
 293                         return -EIO;
 294                 }
 295         }
 296
 297         return 0;
 298 }
 299
 300 static int pull_job_detect_compression(PullJob *j) {
 301         _cleanup_free_ uint8_t *stub = NULL;
 302         size_t stub_size;
 303
 304         int r;
 305
 306         assert(j);
 307
 308         r = import_uncompress_detect(&j->compress, j->payload, j->payload_size);
 309         if (r < 0)
 310                 return log_error_errno(r, "Failed to initialize compressor: %m");
 311         if (r == 0)
 312                 return 0;
 313
 314         log_debug("Stream is compressed: %s", import_compress_type_to_string(j->compress.type));
 315
 316         r = pull_job_open_disk(j);
 317         if (r < 0)
 318                 return r;
 319
 320         /* Now, take the payload we read so far, and decompress it */
 321         stub = j->payload;
 322         stub_size = j->payload_size;
 323
 324         j->payload = NULL;
 325         j->payload_size = 0;
 326         j->payload_allocated = 0;
 327
 328         j->state = PULL_JOB_RUNNING;
 329
 330         r = pull_job_write_compressed(j, stub, stub_size);
 331         if (r < 0)
 332                 return r;
 333
 334         return 0;
 335 }
 336
 337 static size_t pull_job_write_callback(void *contents, size_t size, size_t nmemb, void *userdata) {
 338         PullJob *j = userdata;
 339         size_t sz = size * nmemb;
 340         int r;
 341
 342         assert(contents);
 343         assert(j);
 344
 345         switch (j->state) {
 346
 347         case PULL_JOB_ANALYZING:
 348                 /* Let's first check what it actually is */
 349
 350                 if (!GREEDY_REALLOC(j->payload, j->payload_allocated, j->payload_size + sz)) {
 351                         r = log_oom();
 352                         goto fail;
 353                 }
 354
 355                 memcpy(j->payload + j->payload_size, contents, sz);
 356                 j->payload_size += sz;
 357
 358                 r = pull_job_detect_compression(j);
 359                 if (r < 0)
 360                         goto fail;
 361
 362                 break;
 363
 364         case PULL_JOB_RUNNING:
 365
 366                 r = pull_job_write_compressed(j, contents, sz);
 367                 if (r < 0)
 368                         goto fail;
 369
 370                 break;
 371
 372         case PULL_JOB_DONE:
 373         case PULL_JOB_FAILED:
 374                 r = -ESTALE;
 375                 goto fail;
 376
 377         default:
 378                 assert_not_reached("Impossible state.");
 379         }
 380
 381         return sz;
 382
 383 fail:
 384         pull_job_finish(j, r);
 385         return 0;
 386 }
 387
 388 static size_t pull_job_header_callback(void *contents, size_t size, size_t nmemb, void *userdata) {
 389         PullJob *j = userdata;
 390         size_t sz = size * nmemb;
 391         _cleanup_free_ char *length = NULL, *last_modified = NULL;
 392         char *etag;
 393         int r;
 394
 395         assert(contents);
 396         assert(j);
 397
 398         if (j->state == PULL_JOB_DONE || j->state == PULL_JOB_FAILED) {
 399                 r = -ESTALE;
 400                 goto fail;
 401         }
 402
 403         assert(j->state == PULL_JOB_ANALYZING);
 404
 405         r = curl_header_strdup(contents, sz, "ETag:", &etag);
 406         if (r < 0) {
 407                 log_oom();
 408                 goto fail;
 409         }
 410         if (r > 0) {
 411                 free(j->etag);
 412                 j->etag = etag;
 413
 414                 if (strv_contains(j->old_etags, j->etag)) {
 415                         log_info("Image already downloaded. Skipping download.");
 416                         j->etag_exists = true;
 417                         pull_job_finish(j, 0);
 418                         return sz;
 419                 }
 420
 421                 return sz;
 422         }
 423
 424         r = curl_header_strdup(contents, sz, "Content-Length:", &length);
 425         if (r < 0) {
 426                 log_oom();
 427                 goto fail;
 428         }
 429         if (r > 0) {
 430                 (void) safe_atou64(length, &j->content_length);
 431
 432                 if (j->content_length != (uint64_t) -1) {
 433                         char bytes[FORMAT_BYTES_MAX];
 434
 435                         if (j->content_length > j->compressed_max) {
 436                                 log_error("Content too large.");
 437                                 r = -EFBIG;
 438                                 goto fail;
 439                         }
 440
 441                         log_info("Downloading %s for %s.", format_bytes(bytes, sizeof(bytes), j->content_length), j->url);
 442                 }
 443
 444                 return sz;
 445         }
 446
 447         r = curl_header_strdup(contents, sz, "Last-Modified:", &last_modified);
 448         if (r < 0) {
 449                 log_oom();
 450                 goto fail;
 451         }
 452         if (r > 0) {
 453                 (void) curl_parse_http_time(last_modified, &j->mtime);
 454                 return sz;
 455         }
 456
 457         if (j->on_header) {
 458                 r = j->on_header(j, contents, sz);
 459                 if (r < 0)
 460                         goto fail;
 461         }
 462
 463         return sz;
 464
 465 fail:
 466         pull_job_finish(j, r);
 467         return 0;
 468 }
 469
 470 static int pull_job_progress_callback(void *userdata, curl_off_t dltotal, curl_off_t dlnow, curl_off_t ultotal, curl_off_t ulnow) {
 471         PullJob *j = userdata;
 472         unsigned percent;
 473         usec_t n;
 474
 475         assert(j);
 476
 477         if (dltotal <= 0)
 478                 return 0;
 479
 480         percent = ((100 * dlnow) / dltotal);
 481         n = now(CLOCK_MONOTONIC);
 482
 483         if (n > j->last_status_usec + USEC_PER_SEC &&
 484             percent != j->progress_percent &&
 485             dlnow < dltotal) {
 486                 char buf[FORMAT_TIMESPAN_MAX];
 487
 488                 if (n - j->start_usec > USEC_PER_SEC && dlnow > 0) {
 489                         char y[FORMAT_BYTES_MAX];
 490                         usec_t left, done;
 491
 492                         done = n - j->start_usec;
 493                         left = (usec_t) (((double) done * (double) dltotal) / dlnow) - done;
 494
 495                         log_info("Got %u%% of %s. %s left at %s/s.",
 496                                  percent,
 497                                  j->url,
 498                                  format_timespan(buf, sizeof(buf), left, USEC_PER_SEC),
 499                                  format_bytes(y, sizeof(y), (uint64_t) ((double) dlnow / ((double) done / (double) USEC_PER_SEC))));
 500                 } else
 501                         log_info("Got %u%% of %s.", percent, j->url);
 502
 503                 j->progress_percent = percent;
 504                 j->last_status_usec = n;
 505
 506                 if (j->on_progress)
 507                         j->on_progress(j);
 508         }
 509
 510         return 0;
 511 }
 512
 513 int pull_job_new(PullJob **ret, const char *url, CurlGlue *glue, void *userdata) {
 514         _cleanup_(pull_job_unrefp) PullJob *j = NULL;
 515
 516         assert(url);
 517         assert(glue);
 518         assert(ret);
 519
 520         j = new0(PullJob, 1);
 521         if (!j)
 522                 return -ENOMEM;
 523
 524         j->state = PULL_JOB_INIT;
 525         j->disk_fd = -1;
 526         j->userdata = userdata;
 527         j->glue = glue;
 528         j->content_length = (uint64_t) -1;
 529         j->start_usec = now(CLOCK_MONOTONIC);
 530         j->compressed_max = j->uncompressed_max = 8LLU * 1024LLU * 1024LLU * 1024LLU; /* 8GB */
 531
 532         j->url = strdup(url);
 533         if (!j->url)
 534                 return -ENOMEM;
 535
 536         *ret = j;
 537         j = NULL;
 538
 539         return 0;
 540 }
 541
 542 int pull_job_begin(PullJob *j) {
 543         int r;
 544
 545         assert(j);
 546
 547         if (j->state != PULL_JOB_INIT)
 548                 return -EBUSY;
 549
 550         if (j->grow_machine_directory)
 551                 grow_machine_directory();
 552
 553         r = curl_glue_make(&j->curl, j->url, j);
 554         if (r < 0)
 555                 return r;
 556
 557         if (!strv_isempty(j->old_etags)) {
 558                 _cleanup_free_ char *cc = NULL, *hdr = NULL;
 559
 560                 cc = strv_join(j->old_etags, ", ");
 561                 if (!cc)
 562                         return -ENOMEM;
 563
 564                 hdr = strappend("If-None-Match: ", cc);
 565                 if (!hdr)
 566                         return -ENOMEM;
 567
 568                 if (!j->request_header) {
 569                         j->request_header = curl_slist_new(hdr, NULL);
 570                         if (!j->request_header)
 571                                 return -ENOMEM;
 572                 } else {
 573                         struct curl_slist *l;
 574
 575                         l = curl_slist_append(j->request_header, hdr);
 576                         if (!l)
 577                                 return -ENOMEM;
 578
 579                         j->request_header = l;
 580                 }
 581         }
 582
 583         if (j->request_header) {
 584                 if (curl_easy_setopt(j->curl, CURLOPT_HTTPHEADER, j->request_header) != CURLE_OK)
 585                         return -EIO;
 586         }
 587
 588         if (curl_easy_setopt(j->curl, CURLOPT_WRITEFUNCTION, pull_job_write_callback) != CURLE_OK)
 589                 return -EIO;
 590
 591         if (curl_easy_setopt(j->curl, CURLOPT_WRITEDATA, j) != CURLE_OK)
 592                 return -EIO;
 593
 594         if (curl_easy_setopt(j->curl, CURLOPT_HEADERFUNCTION, pull_job_header_callback) != CURLE_OK)
 595                 return -EIO;
 596
 597         if (curl_easy_setopt(j->curl, CURLOPT_HEADERDATA, j) != CURLE_OK)
 598                 return -EIO;
 599
 600         if (curl_easy_setopt(j->curl, CURLOPT_XFERINFOFUNCTION, pull_job_progress_callback) != CURLE_OK)
 601                 return -EIO;
 602
 603         if (curl_easy_setopt(j->curl, CURLOPT_XFERINFODATA, j) != CURLE_OK)
 604                 return -EIO;
 605
 606         if (curl_easy_setopt(j->curl, CURLOPT_NOPROGRESS, 0) != CURLE_OK)
 607                 return -EIO;
 608
 609         r = curl_glue_add(j->glue, j->curl);
 610         if (r < 0)
 611                 return r;
 612
 613         j->state = PULL_JOB_ANALYZING;
 614
 615         return 0;
 616 }