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