http-walker.c

   1 #define USE_THE_REPOSITORY_VARIABLE
   2 #define DISABLE_SIGN_COMPARE_WARNINGS
   3
   4 #include "git-compat-util.h"
   5 #include "repository.h"
   6 #include "hex.h"
   7 #include "walker.h"
   8 #include "http.h"
   9 #include "list.h"
  10 #include "transport.h"
  11 #include "packfile.h"
  12 #include "object-store-ll.h"
  13
  14 struct alt_base {
  15         char *base;
  16         int got_indices;
  17         struct packed_git *packs;
  18         struct alt_base *next;
  19 };
  20
  21 enum object_request_state {
  22         WAITING,
  23         ABORTED,
  24         ACTIVE,
  25         COMPLETE
  26 };
  27
  28 struct object_request {
  29         struct walker *walker;
  30         struct object_id oid;
  31         struct alt_base *repo;
  32         enum object_request_state state;
  33         struct http_object_request *req;
  34         struct list_head node;
  35 };
  36
  37 struct alternates_request {
  38         struct walker *walker;
  39         const char *base;
  40         struct strbuf *url;
  41         struct strbuf *buffer;
  42         struct active_request_slot *slot;
  43         int http_specific;
  44 };
  45
  46 struct walker_data {
  47         const char *url;
  48         int got_alternates;
  49         struct alt_base *alt;
  50 };
  51
  52 static LIST_HEAD(object_queue_head);
  53
  54 static void fetch_alternates(struct walker *walker, const char *base);
  55
  56 static void process_object_response(void *callback_data);
  57
  58 static void start_object_request(struct object_request *obj_req)
  59 {
  60         struct active_request_slot *slot;
  61         struct http_object_request *req;
  62
  63         req = new_http_object_request(obj_req->repo->base, &obj_req->oid);
  64         if (!req) {
  65                 obj_req->state = ABORTED;
  66                 return;
  67         }
  68         obj_req->req = req;
  69
  70         slot = req->slot;
  71         slot->callback_func = process_object_response;
  72         slot->callback_data = obj_req;
  73
  74         /* Try to get the request started, abort the request on error */
  75         obj_req->state = ACTIVE;
  76         if (!start_active_slot(slot)) {
  77                 obj_req->state = ABORTED;
  78                 release_http_object_request(&req);
  79                 return;
  80         }
  81 }
  82
  83 static void finish_object_request(struct object_request *obj_req)
  84 {
  85         if (finish_http_object_request(obj_req->req))
  86                 return;
  87
  88         if (obj_req->req->rename == 0)
  89                 walker_say(obj_req->walker, "got %s\n", oid_to_hex(&obj_req->oid));
  90 }
  91
  92 static void process_object_response(void *callback_data)
  93 {
  94         struct object_request *obj_req =
  95                 (struct object_request *)callback_data;
  96         struct walker *walker = obj_req->walker;
  97         struct walker_data *data = walker->data;
  98         struct alt_base *alt = data->alt;
  99
 100         process_http_object_request(obj_req->req);
 101         obj_req->state = COMPLETE;
 102
 103         normalize_curl_result(&obj_req->req->curl_result,
 104                               obj_req->req->http_code,
 105                               obj_req->req->errorstr,
 106                               sizeof(obj_req->req->errorstr));
 107
 108         /* Use alternates if necessary */
 109         if (missing_target(obj_req->req)) {
 110                 fetch_alternates(walker, alt->base);
 111                 if (obj_req->repo->next) {
 112                         obj_req->repo =
 113                                 obj_req->repo->next;
 114                         release_http_object_request(&obj_req->req);
 115                         start_object_request(obj_req);
 116                         return;
 117                 }
 118         }
 119
 120         finish_object_request(obj_req);
 121 }
 122
 123 static void release_object_request(struct object_request *obj_req)
 124 {
 125         if (obj_req->req !=NULL && obj_req->req->localfile != -1)
 126                 error("fd leakage in release: %d", obj_req->req->localfile);
 127
 128         list_del(&obj_req->node);
 129         free(obj_req);
 130 }
 131
 132 static int fill_active_slot(void *data UNUSED)
 133 {
 134         struct object_request *obj_req;
 135         struct list_head *pos, *tmp, *head = &object_queue_head;
 136
 137         list_for_each_safe(pos, tmp, head) {
 138                 obj_req = list_entry(pos, struct object_request, node);
 139                 if (obj_req->state == WAITING) {
 140                         if (repo_has_object_file(the_repository, &obj_req->oid))
 141                                 obj_req->state = COMPLETE;
 142                         else {
 143                                 start_object_request(obj_req);
 144                                 return 1;
 145                         }
 146                 }
 147         }
 148         return 0;
 149 }
 150
 151 static void prefetch(struct walker *walker, const struct object_id *oid)
 152 {
 153         struct object_request *newreq;
 154         struct walker_data *data = walker->data;
 155
 156         newreq = xmalloc(sizeof(*newreq));
 157         newreq->walker = walker;
 158         oidcpy(&newreq->oid, oid);
 159         newreq->repo = data->alt;
 160         newreq->state = WAITING;
 161         newreq->req = NULL;
 162
 163         http_is_verbose = walker->get_verbosely;
 164         list_add_tail(&newreq->node, &object_queue_head);
 165
 166         fill_active_slots();
 167         step_active_slots();
 168 }
 169
 170 static int is_alternate_allowed(const char *url)
 171 {
 172         const char *protocols[] = {
 173                 "http", "https", "ftp", "ftps"
 174         };
 175         int i;
 176
 177         if (http_follow_config != HTTP_FOLLOW_ALWAYS) {
 178                 warning("alternate disabled by http.followRedirects: %s", url);
 179                 return 0;
 180         }
 181
 182         for (i = 0; i < ARRAY_SIZE(protocols); i++) {
 183                 const char *end;
 184                 if (skip_prefix(url, protocols[i], &end) &&
 185                     starts_with(end, "://"))
 186                         break;
 187         }
 188
 189         if (i >= ARRAY_SIZE(protocols)) {
 190                 warning("ignoring alternate with unknown protocol: %s", url);
 191                 return 0;
 192         }
 193         if (!is_transport_allowed(protocols[i], 0)) {
 194                 warning("ignoring alternate with restricted protocol: %s", url);
 195                 return 0;
 196         }
 197
 198         return 1;
 199 }
 200
 201 static void process_alternates_response(void *callback_data)
 202 {
 203         struct alternates_request *alt_req =
 204                 (struct alternates_request *)callback_data;
 205         struct walker *walker = alt_req->walker;
 206         struct walker_data *cdata = walker->data;
 207         struct active_request_slot *slot = alt_req->slot;
 208         struct alt_base *tail = cdata->alt;
 209         const char *base = alt_req->base;
 210         const char null_byte = '\0';
 211         char *data;
 212         int i = 0;
 213
 214         normalize_curl_result(&slot->curl_result, slot->http_code,
 215                               curl_errorstr, sizeof(curl_errorstr));
 216
 217         if (alt_req->http_specific) {
 218                 if (slot->curl_result != CURLE_OK ||
 219                     !alt_req->buffer->len) {
 220
 221                         /* Try reusing the slot to get non-http alternates */
 222                         alt_req->http_specific = 0;
 223                         strbuf_reset(alt_req->url);
 224                         strbuf_addf(alt_req->url, "%s/objects/info/alternates",
 225                                     base);
 226                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 227                                          alt_req->url->buf);
 228                         active_requests++;
 229                         slot->in_use = 1;
 230                         if (slot->finished)
 231                                 (*slot->finished) = 0;
 232                         if (!start_active_slot(slot)) {
 233                                 cdata->got_alternates = -1;
 234                                 slot->in_use = 0;
 235                                 if (slot->finished)
 236                                         (*slot->finished) = 1;
 237                         }
 238                         return;
 239                 }
 240         } else if (slot->curl_result != CURLE_OK) {
 241                 if (!missing_target(slot)) {
 242                         cdata->got_alternates = -1;
 243                         return;
 244                 }
 245         }
 246
 247         fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
 248         alt_req->buffer->len--;
 249         data = alt_req->buffer->buf;
 250
 251         while (i < alt_req->buffer->len) {
 252                 int posn = i;
 253                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 254                         posn++;
 255                 if (data[posn] == '\n') {
 256                         int okay = 0;
 257                         int serverlen = 0;
 258                         struct alt_base *newalt;
 259                         if (data[i] == '/') {
 260                                 /*
 261                                  * This counts
 262                                  * http://git.host/pub/scm/linux.git/
 263                                  * -----------here^
 264                                  * so memcpy(dst, base, serverlen) will
 265                                  * copy up to "...git.host".
 266                                  */
 267                                 const char *colon_ss = strstr(base,"://");
 268                                 if (colon_ss) {
 269                                         serverlen = (strchr(colon_ss + 3, '/')
 270                                                      - base);
 271                                         okay = 1;
 272                                 }
 273                         } else if (!memcmp(data + i, "../", 3)) {
 274                                 /*
 275                                  * Relative URL; chop the corresponding
 276                                  * number of subpath from base (and ../
 277                                  * from data), and concatenate the result.
 278                                  *
 279                                  * The code first drops ../ from data, and
 280                                  * then drops one ../ from data and one path
 281                                  * from base.  IOW, one extra ../ is dropped
 282                                  * from data than path is dropped from base.
 283                                  *
 284                                  * This is not wrong.  The alternate in
 285                                  *     http://git.host/pub/scm/linux.git/
 286                                  * to borrow from
 287                                  *     http://git.host/pub/scm/linus.git/
 288                                  * is ../../linus.git/objects/.  You need
 289                                  * two ../../ to borrow from your direct
 290                                  * neighbour.
 291                                  */
 292                                 i += 3;
 293                                 serverlen = strlen(base);
 294                                 while (i + 2 < posn &&
 295                                        !memcmp(data + i, "../", 3)) {
 296                                         do {
 297                                                 serverlen--;
 298                                         } while (serverlen &&
 299                                                  base[serverlen - 1] != '/');
 300                                         i += 3;
 301                                 }
 302                                 /* If the server got removed, give up. */
 303                                 okay = strchr(base, ':') - base + 3 <
 304                                        serverlen;
 305                         } else if (alt_req->http_specific) {
 306                                 char *colon = strchr(data + i, ':');
 307                                 char *slash = strchr(data + i, '/');
 308                                 if (colon && slash && colon < data + posn &&
 309                                     slash < data + posn && colon < slash) {
 310                                         okay = 1;
 311                                 }
 312                         }
 313                         if (okay) {
 314                                 struct strbuf target = STRBUF_INIT;
 315                                 strbuf_add(&target, base, serverlen);
 316                                 strbuf_add(&target, data + i, posn - i);
 317                                 if (!strbuf_strip_suffix(&target, "objects")) {
 318                                         warning("ignoring alternate that does"
 319                                                 " not end in 'objects': %s",
 320                                                 target.buf);
 321                                         strbuf_release(&target);
 322                                 } else if (is_alternate_allowed(target.buf)) {
 323                                         warning("adding alternate object store: %s",
 324                                                 target.buf);
 325                                         newalt = xmalloc(sizeof(*newalt));
 326                                         newalt->next = NULL;
 327                                         newalt->base = strbuf_detach(&target, NULL);
 328                                         newalt->got_indices = 0;
 329                                         newalt->packs = NULL;
 330
 331                                         while (tail->next != NULL)
 332                                                 tail = tail->next;
 333                                         tail->next = newalt;
 334                                 } else {
 335                                         strbuf_release(&target);
 336                                 }
 337                         }
 338                 }
 339                 i = posn + 1;
 340         }
 341
 342         cdata->got_alternates = 1;
 343 }
 344
 345 static void fetch_alternates(struct walker *walker, const char *base)
 346 {
 347         struct strbuf buffer = STRBUF_INIT;
 348         struct strbuf url = STRBUF_INIT;
 349         struct active_request_slot *slot;
 350         struct alternates_request alt_req;
 351         struct walker_data *cdata = walker->data;
 352
 353         /*
 354          * If another request has already started fetching alternates,
 355          * wait for them to arrive and return to processing this request's
 356          * curl message
 357          */
 358         while (cdata->got_alternates == 0) {
 359                 step_active_slots();
 360         }
 361
 362         /* Nothing to do if they've already been fetched */
 363         if (cdata->got_alternates == 1)
 364                 return;
 365
 366         /* Start the fetch */
 367         cdata->got_alternates = 0;
 368
 369         if (walker->get_verbosely)
 370                 fprintf(stderr, "Getting alternates list for %s\n", base);
 371
 372         strbuf_addf(&url, "%s/objects/info/http-alternates", base);
 373
 374         /*
 375          * Use a callback to process the result, since another request
 376          * may fail and need to have alternates loaded before continuing
 377          */
 378         slot = get_active_slot();
 379         slot->callback_func = process_alternates_response;
 380         alt_req.walker = walker;
 381         slot->callback_data = &alt_req;
 382
 383         curl_easy_setopt(slot->curl, CURLOPT_WRITEDATA, &buffer);
 384         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 385         curl_easy_setopt(slot->curl, CURLOPT_URL, url.buf);
 386
 387         alt_req.base = base;
 388         alt_req.url = &url;
 389         alt_req.buffer = &buffer;
 390         alt_req.http_specific = 1;
 391         alt_req.slot = slot;
 392
 393         if (start_active_slot(slot))
 394                 run_active_slot(slot);
 395         else
 396                 cdata->got_alternates = -1;
 397
 398         strbuf_release(&buffer);
 399         strbuf_release(&url);
 400 }
 401
 402 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 403 {
 404         int ret;
 405
 406         if (repo->got_indices)
 407                 return 0;
 408
 409         if (walker->get_verbosely)
 410                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 411
 412         switch (http_get_info_packs(repo->base, &repo->packs)) {
 413         case HTTP_OK:
 414         case HTTP_MISSING_TARGET:
 415                 repo->got_indices = 1;
 416                 ret = 0;
 417                 break;
 418         default:
 419                 repo->got_indices = 0;
 420                 ret = -1;
 421         }
 422
 423         return ret;
 424 }
 425
 426 static int http_fetch_pack(struct walker *walker, struct alt_base *repo,
 427                            const struct object_id *oid)
 428 {
 429         struct packed_git *target;
 430         int ret;
 431         struct slot_results results;
 432         struct http_pack_request *preq;
 433
 434         if (fetch_indices(walker, repo))
 435                 return -1;
 436         target = find_oid_pack(oid, repo->packs);
 437         if (!target)
 438                 return -1;
 439         close_pack_index(target);
 440
 441         if (walker->get_verbosely) {
 442                 fprintf(stderr, "Getting pack %s\n",
 443                         hash_to_hex(target->hash));
 444                 fprintf(stderr, " which contains %s\n",
 445                         oid_to_hex(oid));
 446         }
 447
 448         preq = new_http_pack_request(target->hash, repo->base);
 449         if (!preq)
 450                 goto abort;
 451         preq->slot->results = &results;
 452
 453         if (start_active_slot(preq->slot)) {
 454                 run_active_slot(preq->slot);
 455                 if (results.curl_result != CURLE_OK) {
 456                         error("Unable to get pack file %s\n%s", preq->url,
 457                               curl_errorstr);
 458                         goto abort;
 459                 }
 460         } else {
 461                 error("Unable to start request");
 462                 goto abort;
 463         }
 464
 465         ret = finish_http_pack_request(preq);
 466         release_http_pack_request(preq);
 467         if (ret)
 468                 return ret;
 469         http_install_packfile(target, &repo->packs);
 470
 471         return 0;
 472
 473 abort:
 474         return -1;
 475 }
 476
 477 static void abort_object_request(struct object_request *obj_req)
 478 {
 479         release_object_request(obj_req);
 480 }
 481
 482 static int fetch_object(struct walker *walker, const struct object_id *oid)
 483 {
 484         char *hex = oid_to_hex(oid);
 485         int ret = 0;
 486         struct object_request *obj_req = NULL;
 487         struct http_object_request *req;
 488         struct list_head *pos, *head = &object_queue_head;
 489
 490         list_for_each(pos, head) {
 491                 obj_req = list_entry(pos, struct object_request, node);
 492                 if (oideq(&obj_req->oid, oid))
 493                         break;
 494         }
 495         if (!obj_req)
 496                 return error("Couldn't find request for %s in the queue", hex);
 497
 498         if (repo_has_object_file(the_repository, &obj_req->oid)) {
 499                 if (obj_req->req)
 500                         abort_http_object_request(&obj_req->req);
 501                 abort_object_request(obj_req);
 502                 return 0;
 503         }
 504
 505         while (obj_req->state == WAITING)
 506                 step_active_slots();
 507
 508         /*
 509          * obj_req->req might change when fetching alternates in the callback
 510          * process_object_response; therefore, the "shortcut" variable, req,
 511          * is used only after we're done with slots.
 512          */
 513         while (obj_req->state == ACTIVE)
 514                 run_active_slot(obj_req->req->slot);
 515
 516         req = obj_req->req;
 517
 518         if (req->localfile != -1) {
 519                 close(req->localfile);
 520                 req->localfile = -1;
 521         }
 522
 523         normalize_curl_result(&req->curl_result, req->http_code,
 524                               req->errorstr, sizeof(req->errorstr));
 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 (!oideq(&obj_req->oid, &req->real_oid)) {
 540                 ret = error("File %s has bad hash", hex);
 541         } else if (req->rename < 0) {
 542                 struct strbuf buf = STRBUF_INIT;
 543                 loose_object_path(the_repository, &buf, &req->oid);
 544                 ret = error("unable to write sha1 filename %s", buf.buf);
 545                 strbuf_release(&buf);
 546         }
 547
 548         release_http_object_request(&obj_req->req);
 549         release_object_request(obj_req);
 550         return ret;
 551 }
 552
 553 static int fetch(struct walker *walker, const struct object_id *oid)
 554 {
 555         struct walker_data *data = walker->data;
 556         struct alt_base *altbase = data->alt;
 557
 558         if (!fetch_object(walker, oid))
 559                 return 0;
 560         while (altbase) {
 561                 if (!http_fetch_pack(walker, altbase, oid))
 562                         return 0;
 563                 fetch_alternates(walker, data->alt->base);
 564                 altbase = altbase->next;
 565         }
 566         return error("Unable to find %s under %s", oid_to_hex(oid),
 567                      data->alt->base);
 568 }
 569
 570 static int fetch_ref(struct walker *walker, struct ref *ref)
 571 {
 572         struct walker_data *data = walker->data;
 573         return http_fetch_ref(data->alt->base, ref);
 574 }
 575
 576 static void cleanup(struct walker *walker)
 577 {
 578         struct walker_data *data = walker->data;
 579         struct alt_base *alt, *alt_next;
 580
 581         if (data) {
 582                 alt = data->alt;
 583                 while (alt) {
 584                         struct packed_git *pack;
 585
 586                         alt_next = alt->next;
 587
 588                         pack = alt->packs;
 589                         while (pack) {
 590                                 struct packed_git *pack_next = pack->next;
 591                                 close_pack(pack);
 592                                 free(pack);
 593                                 pack = pack_next;
 594                         }
 595
 596                         free(alt->base);
 597                         free(alt);
 598
 599                         alt = alt_next;
 600                 }
 601                 free(data);
 602                 walker->data = NULL;
 603         }
 604 }
 605
 606 struct walker *get_http_walker(const char *url)
 607 {
 608         char *s;
 609         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 610         struct walker *walker = xmalloc(sizeof(struct walker));
 611
 612         data->alt = xmalloc(sizeof(*data->alt));
 613         data->alt->base = xstrdup(url);
 614         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 615                 *s = 0;
 616
 617         data->alt->got_indices = 0;
 618         data->alt->packs = NULL;
 619         data->alt->next = NULL;
 620         data->got_alternates = -1;
 621
 622         walker->corrupt_object_found = 0;
 623         walker->fetch = fetch;
 624         walker->fetch_ref = fetch_ref;
 625         walker->prefetch = prefetch;
 626         walker->cleanup = cleanup;
 627         walker->data = data;
 628
 629         add_fill_function(NULL, fill_active_slot);
 630
 631         return walker;
 632 }