http-walker.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "walker.h"
   4 #include "http.h"
   5 #include "list.h"
   6 #include "transport.h"
   7 #include "packfile.h"
   8 #include "object-store.h"
   9
  10 struct alt_base {
  11         char *base;
  12         int got_indices;
  13         struct packed_git *packs;
  14         struct alt_base *next;
  15 };
  16
  17 enum object_request_state {
  18         WAITING,
  19         ABORTED,
  20         ACTIVE,
  21         COMPLETE
  22 };
  23
  24 struct object_request {
  25         struct walker *walker;
  26         unsigned char sha1[20];
  27         struct alt_base *repo;
  28         enum object_request_state state;
  29         struct http_object_request *req;
  30         struct list_head node;
  31 };
  32
  33 struct alternates_request {
  34         struct walker *walker;
  35         const char *base;
  36         struct strbuf *url;
  37         struct strbuf *buffer;
  38         struct active_request_slot *slot;
  39         int http_specific;
  40 };
  41
  42 struct walker_data {
  43         const char *url;
  44         int got_alternates;
  45         struct alt_base *alt;
  46 };
  47
  48 static LIST_HEAD(object_queue_head);
  49
  50 static void fetch_alternates(struct walker *walker, const char *base);
  51
  52 static void process_object_response(void *callback_data);
  53
  54 static void start_object_request(struct walker *walker,
  55                                  struct object_request *obj_req)
  56 {
  57         struct active_request_slot *slot;
  58         struct http_object_request *req;
  59
  60         req = new_http_object_request(obj_req->repo->base, obj_req->sha1);
  61         if (req == NULL) {
  62                 obj_req->state = ABORTED;
  63                 return;
  64         }
  65         obj_req->req = req;
  66
  67         slot = req->slot;
  68         slot->callback_func = process_object_response;
  69         slot->callback_data = obj_req;
  70
  71         /* Try to get the request started, abort the request on error */
  72         obj_req->state = ACTIVE;
  73         if (!start_active_slot(slot)) {
  74                 obj_req->state = ABORTED;
  75                 release_http_object_request(req);
  76                 return;
  77         }
  78 }
  79
  80 static void finish_object_request(struct object_request *obj_req)
  81 {
  82         if (finish_http_object_request(obj_req->req))
  83                 return;
  84
  85         if (obj_req->req->rename == 0)
  86                 walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
  87 }
  88
  89 static void process_object_response(void *callback_data)
  90 {
  91         struct object_request *obj_req =
  92                 (struct object_request *)callback_data;
  93         struct walker *walker = obj_req->walker;
  94         struct walker_data *data = walker->data;
  95         struct alt_base *alt = data->alt;
  96
  97         process_http_object_request(obj_req->req);
  98         obj_req->state = COMPLETE;
  99
 100         /* Use alternates if necessary */
 101         if (missing_target(obj_req->req)) {
 102                 fetch_alternates(walker, alt->base);
 103                 if (obj_req->repo->next != NULL) {
 104                         obj_req->repo =
 105                                 obj_req->repo->next;
 106                         release_http_object_request(obj_req->req);
 107                         start_object_request(walker, obj_req);
 108                         return;
 109                 }
 110         }
 111
 112         finish_object_request(obj_req);
 113 }
 114
 115 static void release_object_request(struct object_request *obj_req)
 116 {
 117         if (obj_req->req !=NULL && obj_req->req->localfile != -1)
 118                 error("fd leakage in release: %d", obj_req->req->localfile);
 119
 120         list_del(&obj_req->node);
 121         free(obj_req);
 122 }
 123
 124 #ifdef USE_CURL_MULTI
 125 static int fill_active_slot(struct walker *walker)
 126 {
 127         struct object_request *obj_req;
 128         struct list_head *pos, *tmp, *head = &object_queue_head;
 129
 130         list_for_each_safe(pos, tmp, head) {
 131                 obj_req = list_entry(pos, struct object_request, node);
 132                 if (obj_req->state == WAITING) {
 133                         if (has_sha1_file(obj_req->sha1))
 134                                 obj_req->state = COMPLETE;
 135                         else {
 136                                 start_object_request(walker, obj_req);
 137                                 return 1;
 138                         }
 139                 }
 140         }
 141         return 0;
 142 }
 143 #endif
 144
 145 static void prefetch(struct walker *walker, unsigned char *sha1)
 146 {
 147         struct object_request *newreq;
 148         struct walker_data *data = walker->data;
 149
 150         newreq = xmalloc(sizeof(*newreq));
 151         newreq->walker = walker;
 152         hashcpy(newreq->sha1, sha1);
 153         newreq->repo = data->alt;
 154         newreq->state = WAITING;
 155         newreq->req = NULL;
 156
 157         http_is_verbose = walker->get_verbosely;
 158         list_add_tail(&newreq->node, &object_queue_head);
 159
 160 #ifdef USE_CURL_MULTI
 161         fill_active_slots();
 162         step_active_slots();
 163 #endif
 164 }
 165
 166 static int is_alternate_allowed(const char *url)
 167 {
 168         const char *protocols[] = {
 169                 "http", "https", "ftp", "ftps"
 170         };
 171         int i;
 172
 173         if (http_follow_config != HTTP_FOLLOW_ALWAYS) {
 174                 warning("alternate disabled by http.followRedirects: %s", url);
 175                 return 0;
 176         }
 177
 178         for (i = 0; i < ARRAY_SIZE(protocols); i++) {
 179                 const char *end;
 180                 if (skip_prefix(url, protocols[i], &end) &&
 181                     starts_with(end, "://"))
 182                         break;
 183         }
 184
 185         if (i >= ARRAY_SIZE(protocols)) {
 186                 warning("ignoring alternate with unknown protocol: %s", url);
 187                 return 0;
 188         }
 189         if (!is_transport_allowed(protocols[i], 0)) {
 190                 warning("ignoring alternate with restricted protocol: %s", url);
 191                 return 0;
 192         }
 193
 194         return 1;
 195 }
 196
 197 static void process_alternates_response(void *callback_data)
 198 {
 199         struct alternates_request *alt_req =
 200                 (struct alternates_request *)callback_data;
 201         struct walker *walker = alt_req->walker;
 202         struct walker_data *cdata = walker->data;
 203         struct active_request_slot *slot = alt_req->slot;
 204         struct alt_base *tail = cdata->alt;
 205         const char *base = alt_req->base;
 206         const char null_byte = '\0';
 207         char *data;
 208         int i = 0;
 209
 210         if (alt_req->http_specific) {
 211                 if (slot->curl_result != CURLE_OK ||
 212                     !alt_req->buffer->len) {
 213
 214                         /* Try reusing the slot to get non-http alternates */
 215                         alt_req->http_specific = 0;
 216                         strbuf_reset(alt_req->url);
 217                         strbuf_addf(alt_req->url, "%s/objects/info/alternates",
 218                                     base);
 219                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 220                                          alt_req->url->buf);
 221                         active_requests++;
 222                         slot->in_use = 1;
 223                         if (slot->finished != NULL)
 224                                 (*slot->finished) = 0;
 225                         if (!start_active_slot(slot)) {
 226                                 cdata->got_alternates = -1;
 227                                 slot->in_use = 0;
 228                                 if (slot->finished != NULL)
 229                                         (*slot->finished) = 1;
 230                         }
 231                         return;
 232                 }
 233         } else if (slot->curl_result != CURLE_OK) {
 234                 if (!missing_target(slot)) {
 235                         cdata->got_alternates = -1;
 236                         return;
 237                 }
 238         }
 239
 240         fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
 241         alt_req->buffer->len--;
 242         data = alt_req->buffer->buf;
 243
 244         while (i < alt_req->buffer->len) {
 245                 int posn = i;
 246                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 247                         posn++;
 248                 if (data[posn] == '\n') {
 249                         int okay = 0;
 250                         int serverlen = 0;
 251                         struct alt_base *newalt;
 252                         if (data[i] == '/') {
 253                                 /*
 254                                  * This counts
 255                                  * http://git.host/pub/scm/linux.git/
 256                                  * -----------here^
 257                                  * so memcpy(dst, base, serverlen) will
 258                                  * copy up to "...git.host".
 259                                  */
 260                                 const char *colon_ss = strstr(base,"://");
 261                                 if (colon_ss) {
 262                                         serverlen = (strchr(colon_ss + 3, '/')
 263                                                      - base);
 264                                         okay = 1;
 265                                 }
 266                         } else if (!memcmp(data + i, "../", 3)) {
 267                                 /*
 268                                  * Relative URL; chop the corresponding
 269                                  * number of subpath from base (and ../
 270                                  * from data), and concatenate the result.
 271                                  *
 272                                  * The code first drops ../ from data, and
 273                                  * then drops one ../ from data and one path
 274                                  * from base.  IOW, one extra ../ is dropped
 275                                  * from data than path is dropped from base.
 276                                  *
 277                                  * This is not wrong.  The alternate in
 278                                  *     http://git.host/pub/scm/linux.git/
 279                                  * to borrow from
 280                                  *     http://git.host/pub/scm/linus.git/
 281                                  * is ../../linus.git/objects/.  You need
 282                                  * two ../../ to borrow from your direct
 283                                  * neighbour.
 284                                  */
 285                                 i += 3;
 286                                 serverlen = strlen(base);
 287                                 while (i + 2 < posn &&
 288                                        !memcmp(data + i, "../", 3)) {
 289                                         do {
 290                                                 serverlen--;
 291                                         } while (serverlen &&
 292                                                  base[serverlen - 1] != '/');
 293                                         i += 3;
 294                                 }
 295                                 /* If the server got removed, give up. */
 296                                 okay = strchr(base, ':') - base + 3 <
 297                                        serverlen;
 298                         } else if (alt_req->http_specific) {
 299                                 char *colon = strchr(data + i, ':');
 300                                 char *slash = strchr(data + i, '/');
 301                                 if (colon && slash && colon < data + posn &&
 302                                     slash < data + posn && colon < slash) {
 303                                         okay = 1;
 304                                 }
 305                         }
 306                         if (okay) {
 307                                 struct strbuf target = STRBUF_INIT;
 308                                 strbuf_add(&target, base, serverlen);
 309                                 strbuf_add(&target, data + i, posn - i);
 310                                 if (!strbuf_strip_suffix(&target, "objects")) {
 311                                         warning("ignoring alternate that does"
 312                                                 " not end in 'objects': %s",
 313                                                 target.buf);
 314                                         strbuf_release(&target);
 315                                 } else if (is_alternate_allowed(target.buf)) {
 316                                         warning("adding alternate object store: %s",
 317                                                 target.buf);
 318                                         newalt = xmalloc(sizeof(*newalt));
 319                                         newalt->next = NULL;
 320                                         newalt->base = strbuf_detach(&target, NULL);
 321                                         newalt->got_indices = 0;
 322                                         newalt->packs = NULL;
 323
 324                                         while (tail->next != NULL)
 325                                                 tail = tail->next;
 326                                         tail->next = newalt;
 327                                 } else {
 328                                         strbuf_release(&target);
 329                                 }
 330                         }
 331                 }
 332                 i = posn + 1;
 333         }
 334
 335         cdata->got_alternates = 1;
 336 }
 337
 338 static void fetch_alternates(struct walker *walker, const char *base)
 339 {
 340         struct strbuf buffer = STRBUF_INIT;
 341         struct strbuf url = STRBUF_INIT;
 342         struct active_request_slot *slot;
 343         struct alternates_request alt_req;
 344         struct walker_data *cdata = walker->data;
 345
 346         /*
 347          * If another request has already started fetching alternates,
 348          * wait for them to arrive and return to processing this request's
 349          * curl message
 350          */
 351 #ifdef USE_CURL_MULTI
 352         while (cdata->got_alternates == 0) {
 353                 step_active_slots();
 354         }
 355 #endif
 356
 357         /* Nothing to do if they've already been fetched */
 358         if (cdata->got_alternates == 1)
 359                 return;
 360
 361         /* Start the fetch */
 362         cdata->got_alternates = 0;
 363
 364         if (walker->get_verbosely)
 365                 fprintf(stderr, "Getting alternates list for %s\n", base);
 366
 367         strbuf_addf(&url, "%s/objects/info/http-alternates", base);
 368
 369         /*
 370          * Use a callback to process the result, since another request
 371          * may fail and need to have alternates loaded before continuing
 372          */
 373         slot = get_active_slot();
 374         slot->callback_func = process_alternates_response;
 375         alt_req.walker = walker;
 376         slot->callback_data = &alt_req;
 377
 378         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 379         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 380         curl_easy_setopt(slot->curl, CURLOPT_URL, url.buf);
 381
 382         alt_req.base = base;
 383         alt_req.url = &url;
 384         alt_req.buffer = &buffer;
 385         alt_req.http_specific = 1;
 386         alt_req.slot = slot;
 387
 388         if (start_active_slot(slot))
 389                 run_active_slot(slot);
 390         else
 391                 cdata->got_alternates = -1;
 392
 393         strbuf_release(&buffer);
 394         strbuf_release(&url);
 395 }
 396
 397 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 398 {
 399         int ret;
 400
 401         if (repo->got_indices)
 402                 return 0;
 403
 404         if (walker->get_verbosely)
 405                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 406
 407         switch (http_get_info_packs(repo->base, &repo->packs)) {
 408         case HTTP_OK:
 409         case HTTP_MISSING_TARGET:
 410                 repo->got_indices = 1;
 411                 ret = 0;
 412                 break;
 413         default:
 414                 repo->got_indices = 0;
 415                 ret = -1;
 416         }
 417
 418         return ret;
 419 }
 420
 421 static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 422 {
 423         struct packed_git *target;
 424         int ret;
 425         struct slot_results results;
 426         struct http_pack_request *preq;
 427
 428         if (fetch_indices(walker, repo))
 429                 return -1;
 430         target = find_sha1_pack(sha1, repo->packs);
 431         if (!target)
 432                 return -1;
 433
 434         if (walker->get_verbosely) {
 435                 fprintf(stderr, "Getting pack %s\n",
 436                         sha1_to_hex(target->sha1));
 437                 fprintf(stderr, " which contains %s\n",
 438                         sha1_to_hex(sha1));
 439         }
 440
 441         preq = new_http_pack_request(target, repo->base);
 442         if (preq == NULL)
 443                 goto abort;
 444         preq->lst = &repo->packs;
 445         preq->slot->results = &results;
 446
 447         if (start_active_slot(preq->slot)) {
 448                 run_active_slot(preq->slot);
 449                 if (results.curl_result != CURLE_OK) {
 450                         error("Unable to get pack file %s\n%s", preq->url,
 451                               curl_errorstr);
 452                         goto abort;
 453                 }
 454         } else {
 455                 error("Unable to start request");
 456                 goto abort;
 457         }
 458
 459         ret = finish_http_pack_request(preq);
 460         release_http_pack_request(preq);
 461         if (ret)
 462                 return ret;
 463
 464         return 0;
 465
 466 abort:
 467         return -1;
 468 }
 469
 470 static void abort_object_request(struct object_request *obj_req)
 471 {
 472         release_object_request(obj_req);
 473 }
 474
 475 static int fetch_object(struct walker *walker, unsigned char *sha1)
 476 {
 477         char *hex = sha1_to_hex(sha1);
 478         int ret = 0;
 479         struct object_request *obj_req = NULL;
 480         struct http_object_request *req;
 481         struct list_head *pos, *head = &object_queue_head;
 482
 483         list_for_each(pos, head) {
 484                 obj_req = list_entry(pos, struct object_request, node);
 485                 if (!hashcmp(obj_req->sha1, sha1))
 486                         break;
 487         }
 488         if (obj_req == NULL)
 489                 return error("Couldn't find request for %s in the queue", hex);
 490
 491         if (has_sha1_file(obj_req->sha1)) {
 492                 if (obj_req->req != NULL)
 493                         abort_http_object_request(obj_req->req);
 494                 abort_object_request(obj_req);
 495                 return 0;
 496         }
 497
 498 #ifdef USE_CURL_MULTI
 499         while (obj_req->state == WAITING)
 500                 step_active_slots();
 501 #else
 502         start_object_request(walker, obj_req);
 503 #endif
 504
 505         /*
 506          * obj_req->req might change when fetching alternates in the callback
 507          * process_object_response; therefore, the "shortcut" variable, req,
 508          * is used only after we're done with slots.
 509          */
 510         while (obj_req->state == ACTIVE)
 511                 run_active_slot(obj_req->req->slot);
 512
 513         req = obj_req->req;
 514
 515         if (req->localfile != -1) {
 516                 close(req->localfile);
 517                 req->localfile = -1;
 518         }
 519
 520         /*
 521          * we turned off CURLOPT_FAILONERROR to avoid losing a
 522          * persistent connection and got CURLE_OK.
 523          */
 524         if (req->http_code >= 300 && req->curl_result == CURLE_OK &&
 525                         (starts_with(req->url, "http://") ||
 526                          starts_with(req->url, "https://"))) {
 527                 req->curl_result = CURLE_HTTP_RETURNED_ERROR;
 528                 xsnprintf(req->errorstr, sizeof(req->errorstr),
 529                           "HTTP request failed");
 530         }
 531
 532         if (obj_req->state == ABORTED) {
 533                 ret = error("Request for %s aborted", hex);
 534         } else if (req->curl_result != CURLE_OK &&
 535                    req->http_code != 416) {
 536                 if (missing_target(req))
 537                         ret = -1; /* Be silent, it is probably in a pack. */
 538                 else
 539                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 540                                     req->errorstr, req->curl_result,
 541                                     req->http_code, hex);
 542         } else if (req->zret != Z_STREAM_END) {
 543                 walker->corrupt_object_found++;
 544                 ret = error("File %s (%s) corrupt", hex, req->url);
 545         } else if (hashcmp(obj_req->sha1, req->real_sha1)) {
 546                 ret = error("File %s has bad hash", hex);
 547         } else if (req->rename < 0) {
 548                 struct strbuf buf = STRBUF_INIT;
 549                 sha1_file_name(&buf, req->sha1);
 550                 ret = error("unable to write sha1 filename %s", buf.buf);
 551                 strbuf_release(&buf);
 552         }
 553
 554         release_http_object_request(req);
 555         release_object_request(obj_req);
 556         return ret;
 557 }
 558
 559 static int fetch(struct walker *walker, unsigned char *sha1)
 560 {
 561         struct walker_data *data = walker->data;
 562         struct alt_base *altbase = data->alt;
 563
 564         if (!fetch_object(walker, sha1))
 565                 return 0;
 566         while (altbase) {
 567                 if (!http_fetch_pack(walker, altbase, sha1))
 568                         return 0;
 569                 fetch_alternates(walker, data->alt->base);
 570                 altbase = altbase->next;
 571         }
 572         return error("Unable to find %s under %s", sha1_to_hex(sha1),
 573                      data->alt->base);
 574 }
 575
 576 static int fetch_ref(struct walker *walker, struct ref *ref)
 577 {
 578         struct walker_data *data = walker->data;
 579         return http_fetch_ref(data->alt->base, ref);
 580 }
 581
 582 static void cleanup(struct walker *walker)
 583 {
 584         struct walker_data *data = walker->data;
 585         struct alt_base *alt, *alt_next;
 586
 587         if (data) {
 588                 alt = data->alt;
 589                 while (alt) {
 590                         alt_next = alt->next;
 591
 592                         free(alt->base);
 593                         free(alt);
 594
 595                         alt = alt_next;
 596                 }
 597                 free(data);
 598                 walker->data = NULL;
 599         }
 600 }
 601
 602 struct walker *get_http_walker(const char *url)
 603 {
 604         char *s;
 605         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 606         struct walker *walker = xmalloc(sizeof(struct walker));
 607
 608         data->alt = xmalloc(sizeof(*data->alt));
 609         data->alt->base = xstrdup(url);
 610         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 611                 *s = 0;
 612
 613         data->alt->got_indices = 0;
 614         data->alt->packs = NULL;
 615         data->alt->next = NULL;
 616         data->got_alternates = -1;
 617
 618         walker->corrupt_object_found = 0;
 619         walker->fetch = fetch;
 620         walker->fetch_ref = fetch_ref;
 621         walker->prefetch = prefetch;
 622         walker->cleanup = cleanup;
 623         walker->data = data;
 624
 625 #ifdef USE_CURL_MULTI
 626         add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 627 #endif
 628
 629         return walker;
 630 }