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                         /* skip "objects\n" at end */
 300                         if (okay) {
 301                                 struct strbuf target = STRBUF_INIT;
 302                                 strbuf_add(&target, base, serverlen);
 303                                 strbuf_add(&target, data + i, posn - i - 7);
 304
 305                                 if (is_alternate_allowed(target.buf)) {
 306                                         warning("adding alternate object store: %s",
 307                                                 target.buf);
 308                                         newalt = xmalloc(sizeof(*newalt));
 309                                         newalt->next = NULL;
 310                                         newalt->base = strbuf_detach(&target, NULL);
 311                                         newalt->got_indices = 0;
 312                                         newalt->packs = NULL;
 313
 314                                         while (tail->next != NULL)
 315                                                 tail = tail->next;
 316                                         tail->next = newalt;
 317                                 }
 318                         }
 319                 }
 320                 i = posn + 1;
 321         }
 322
 323         cdata->got_alternates = 1;
 324 }
 325
 326 static void fetch_alternates(struct walker *walker, const char *base)
 327 {
 328         struct strbuf buffer = STRBUF_INIT;
 329         struct strbuf url = STRBUF_INIT;
 330         struct active_request_slot *slot;
 331         struct alternates_request alt_req;
 332         struct walker_data *cdata = walker->data;
 333
 334         if (http_follow_config != HTTP_FOLLOW_ALWAYS)
 335                 return;
 336
 337         /*
 338          * If another request has already started fetching alternates,
 339          * wait for them to arrive and return to processing this request's
 340          * curl message
 341          */
 342 #ifdef USE_CURL_MULTI
 343         while (cdata->got_alternates == 0) {
 344                 step_active_slots();
 345         }
 346 #endif
 347
 348         /* Nothing to do if they've already been fetched */
 349         if (cdata->got_alternates == 1)
 350                 return;
 351
 352         /* Start the fetch */
 353         cdata->got_alternates = 0;
 354
 355         if (walker->get_verbosely)
 356                 fprintf(stderr, "Getting alternates list for %s\n", base);
 357
 358         strbuf_addf(&url, "%s/objects/info/http-alternates", base);
 359
 360         /*
 361          * Use a callback to process the result, since another request
 362          * may fail and need to have alternates loaded before continuing
 363          */
 364         slot = get_active_slot();
 365         slot->callback_func = process_alternates_response;
 366         alt_req.walker = walker;
 367         slot->callback_data = &alt_req;
 368
 369         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 370         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 371         curl_easy_setopt(slot->curl, CURLOPT_URL, url.buf);
 372
 373         alt_req.base = base;
 374         alt_req.url = &url;
 375         alt_req.buffer = &buffer;
 376         alt_req.http_specific = 1;
 377         alt_req.slot = slot;
 378
 379         if (start_active_slot(slot))
 380                 run_active_slot(slot);
 381         else
 382                 cdata->got_alternates = -1;
 383
 384         strbuf_release(&buffer);
 385         strbuf_release(&url);
 386 }
 387
 388 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 389 {
 390         int ret;
 391
 392         if (repo->got_indices)
 393                 return 0;
 394
 395         if (walker->get_verbosely)
 396                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 397
 398         switch (http_get_info_packs(repo->base, &repo->packs)) {
 399         case HTTP_OK:
 400         case HTTP_MISSING_TARGET:
 401                 repo->got_indices = 1;
 402                 ret = 0;
 403                 break;
 404         default:
 405                 repo->got_indices = 0;
 406                 ret = -1;
 407         }
 408
 409         return ret;
 410 }
 411
 412 static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 413 {
 414         struct packed_git *target;
 415         int ret;
 416         struct slot_results results;
 417         struct http_pack_request *preq;
 418
 419         if (fetch_indices(walker, repo))
 420                 return -1;
 421         target = find_sha1_pack(sha1, repo->packs);
 422         if (!target)
 423                 return -1;
 424
 425         if (walker->get_verbosely) {
 426                 fprintf(stderr, "Getting pack %s\n",
 427                         sha1_to_hex(target->sha1));
 428                 fprintf(stderr, " which contains %s\n",
 429                         sha1_to_hex(sha1));
 430         }
 431
 432         preq = new_http_pack_request(target, repo->base);
 433         if (preq == NULL)
 434                 goto abort;
 435         preq->lst = &repo->packs;
 436         preq->slot->results = &results;
 437
 438         if (start_active_slot(preq->slot)) {
 439                 run_active_slot(preq->slot);
 440                 if (results.curl_result != CURLE_OK) {
 441                         error("Unable to get pack file %s\n%s", preq->url,
 442                               curl_errorstr);
 443                         goto abort;
 444                 }
 445         } else {
 446                 error("Unable to start request");
 447                 goto abort;
 448         }
 449
 450         ret = finish_http_pack_request(preq);
 451         release_http_pack_request(preq);
 452         if (ret)
 453                 return ret;
 454
 455         return 0;
 456
 457 abort:
 458         return -1;
 459 }
 460
 461 static void abort_object_request(struct object_request *obj_req)
 462 {
 463         release_object_request(obj_req);
 464 }
 465
 466 static int fetch_object(struct walker *walker, unsigned char *sha1)
 467 {
 468         char *hex = sha1_to_hex(sha1);
 469         int ret = 0;
 470         struct object_request *obj_req = NULL;
 471         struct http_object_request *req;
 472         struct list_head *pos, *head = &object_queue_head;
 473
 474         list_for_each(pos, head) {
 475                 obj_req = list_entry(pos, struct object_request, node);
 476                 if (!hashcmp(obj_req->sha1, sha1))
 477                         break;
 478         }
 479         if (obj_req == NULL)
 480                 return error("Couldn't find request for %s in the queue", hex);
 481
 482         if (has_sha1_file(obj_req->sha1)) {
 483                 if (obj_req->req != NULL)
 484                         abort_http_object_request(obj_req->req);
 485                 abort_object_request(obj_req);
 486                 return 0;
 487         }
 488
 489 #ifdef USE_CURL_MULTI
 490         while (obj_req->state == WAITING)
 491                 step_active_slots();
 492 #else
 493         start_object_request(walker, obj_req);
 494 #endif
 495
 496         /*
 497          * obj_req->req might change when fetching alternates in the callback
 498          * process_object_response; therefore, the "shortcut" variable, req,
 499          * is used only after we're done with slots.
 500          */
 501         while (obj_req->state == ACTIVE)
 502                 run_active_slot(obj_req->req->slot);
 503
 504         req = obj_req->req;
 505
 506         if (req->localfile != -1) {
 507                 close(req->localfile);
 508                 req->localfile = -1;
 509         }
 510
 511         /*
 512          * we turned off CURLOPT_FAILONERROR to avoid losing a
 513          * persistent connection and got CURLE_OK.
 514          */
 515         if (req->http_code >= 300 && req->curl_result == CURLE_OK &&
 516                         (starts_with(req->url, "http://") ||
 517                          starts_with(req->url, "https://"))) {
 518                 req->curl_result = CURLE_HTTP_RETURNED_ERROR;
 519                 xsnprintf(req->errorstr, sizeof(req->errorstr),
 520                           "HTTP request failed");
 521         }
 522
 523         if (obj_req->state == ABORTED) {
 524                 ret = error("Request for %s aborted", hex);
 525         } else if (req->curl_result != CURLE_OK &&
 526                    req->http_code != 416) {
 527                 if (missing_target(req))
 528                         ret = -1; /* Be silent, it is probably in a pack. */
 529                 else
 530                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 531                                     req->errorstr, req->curl_result,
 532                                     req->http_code, hex);
 533         } else if (req->zret != Z_STREAM_END) {
 534                 walker->corrupt_object_found++;
 535                 ret = error("File %s (%s) corrupt", hex, req->url);
 536         } else if (hashcmp(obj_req->sha1, req->real_sha1)) {
 537                 ret = error("File %s has bad hash", hex);
 538         } else if (req->rename < 0) {
 539                 ret = error("unable to write sha1 filename %s",
 540                             sha1_file_name(req->sha1));
 541         }
 542
 543         release_http_object_request(req);
 544         release_object_request(obj_req);
 545         return ret;
 546 }
 547
 548 static int fetch(struct walker *walker, unsigned char *sha1)
 549 {
 550         struct walker_data *data = walker->data;
 551         struct alt_base *altbase = data->alt;
 552
 553         if (!fetch_object(walker, sha1))
 554                 return 0;
 555         while (altbase) {
 556                 if (!http_fetch_pack(walker, altbase, sha1))
 557                         return 0;
 558                 fetch_alternates(walker, data->alt->base);
 559                 altbase = altbase->next;
 560         }
 561         return error("Unable to find %s under %s", sha1_to_hex(sha1),
 562                      data->alt->base);
 563 }
 564
 565 static int fetch_ref(struct walker *walker, struct ref *ref)
 566 {
 567         struct walker_data *data = walker->data;
 568         return http_fetch_ref(data->alt->base, ref);
 569 }
 570
 571 static void cleanup(struct walker *walker)
 572 {
 573         struct walker_data *data = walker->data;
 574         struct alt_base *alt, *alt_next;
 575
 576         if (data) {
 577                 alt = data->alt;
 578                 while (alt) {
 579                         alt_next = alt->next;
 580
 581                         free(alt->base);
 582                         free(alt);
 583
 584                         alt = alt_next;
 585                 }
 586                 free(data);
 587                 walker->data = NULL;
 588         }
 589 }
 590
 591 struct walker *get_http_walker(const char *url)
 592 {
 593         char *s;
 594         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 595         struct walker *walker = xmalloc(sizeof(struct walker));
 596
 597         data->alt = xmalloc(sizeof(*data->alt));
 598         data->alt->base = xstrdup(url);
 599         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 600                 *s = 0;
 601
 602         data->alt->got_indices = 0;
 603         data->alt->packs = NULL;
 604         data->alt->next = NULL;
 605         data->got_alternates = -1;
 606
 607         walker->corrupt_object_found = 0;
 608         walker->fetch = fetch;
 609         walker->fetch_ref = fetch_ref;
 610         walker->prefetch = prefetch;
 611         walker->cleanup = cleanup;
 612         walker->data = data;
 613
 614 #ifdef USE_CURL_MULTI
 615         add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 616 #endif
 617
 618         return walker;
 619 }