http-walker.c

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