http-walker.c

   1 #include "cache.h"
   2 #include "repository.h"
   3 #include "commit.h"
   4 #include "walker.h"
   5 #include "http.h"
   6 #include "list.h"
   7 #include "transport.h"
   8 #include "packfile.h"
   9 #include "object-store.h"
  10
  11 struct alt_base {
  12         char *base;
  13         int got_indices;
  14         struct packed_git *packs;
  15         struct alt_base *next;
  16 };
  17
  18 enum object_request_state {
  19         WAITING,
  20         ABORTED,
  21         ACTIVE,
  22         COMPLETE
  23 };
  24
  25 struct object_request {
  26         struct walker *walker;
  27         struct object_id oid;
  28         struct alt_base *repo;
  29         enum object_request_state state;
  30         struct http_object_request *req;
  31         struct list_head node;
  32 };
  33
  34 struct alternates_request {
  35         struct walker *walker;
  36         const char *base;
  37         struct strbuf *url;
  38         struct strbuf *buffer;
  39         struct active_request_slot *slot;
  40         int http_specific;
  41 };
  42
  43 struct walker_data {
  44         const char *url;
  45         int got_alternates;
  46         struct alt_base *alt;
  47 };
  48
  49 static LIST_HEAD(object_queue_head);
  50
  51 static void fetch_alternates(struct walker *walker, const char *base);
  52
  53 static void process_object_response(void *callback_data);
  54
  55 static void start_object_request(struct walker *walker,
  56                                  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 == NULL) {
  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 != NULL) {
 110                         obj_req->repo =
 111                                 obj_req->repo->next;
 112                         release_http_object_request(obj_req->req);
 113                         start_object_request(walker, 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 #ifdef USE_CURL_MULTI
 131 static int fill_active_slot(struct walker *walker)
 132 {
 133         struct object_request *obj_req;
 134         struct list_head *pos, *tmp, *head = &object_queue_head;
 135
 136         list_for_each_safe(pos, tmp, head) {
 137                 obj_req = list_entry(pos, struct object_request, node);
 138                 if (obj_req->state == WAITING) {
 139                         if (has_object_file(&obj_req->oid))
 140                                 obj_req->state = COMPLETE;
 141                         else {
 142                                 start_object_request(walker, obj_req);
 143                                 return 1;
 144                         }
 145                 }
 146         }
 147         return 0;
 148 }
 149 #endif
 150
 151 static void prefetch(struct walker *walker, unsigned char *sha1)
 152 {
 153         struct object_request *newreq;
 154         struct walker_data *data = walker->data;
 155
 156         newreq = xmalloc(sizeof(*newreq));
 157         newreq->walker = walker;
 158         oidread(&newreq->oid, sha1);
 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 #ifdef USE_CURL_MULTI
 167         fill_active_slots();
 168         step_active_slots();
 169 #endif
 170 }
 171
 172 static int is_alternate_allowed(const char *url)
 173 {
 174         const char *protocols[] = {
 175                 "http", "https", "ftp", "ftps"
 176         };
 177         int i;
 178
 179         if (http_follow_config != HTTP_FOLLOW_ALWAYS) {
 180                 warning("alternate disabled by http.followRedirects: %s", url);
 181                 return 0;
 182         }
 183
 184         for (i = 0; i < ARRAY_SIZE(protocols); i++) {
 185                 const char *end;
 186                 if (skip_prefix(url, protocols[i], &end) &&
 187                     starts_with(end, "://"))
 188                         break;
 189         }
 190
 191         if (i >= ARRAY_SIZE(protocols)) {
 192                 warning("ignoring alternate with unknown protocol: %s", url);
 193                 return 0;
 194         }
 195         if (!is_transport_allowed(protocols[i], 0)) {
 196                 warning("ignoring alternate with restricted protocol: %s", url);
 197                 return 0;
 198         }
 199
 200         return 1;
 201 }
 202
 203 static void process_alternates_response(void *callback_data)
 204 {
 205         struct alternates_request *alt_req =
 206                 (struct alternates_request *)callback_data;
 207         struct walker *walker = alt_req->walker;
 208         struct walker_data *cdata = walker->data;
 209         struct active_request_slot *slot = alt_req->slot;
 210         struct alt_base *tail = cdata->alt;
 211         const char *base = alt_req->base;
 212         const char null_byte = '\0';
 213         char *data;
 214         int i = 0;
 215
 216         normalize_curl_result(&slot->curl_result, slot->http_code,
 217                               curl_errorstr, sizeof(curl_errorstr));
 218
 219         if (alt_req->http_specific) {
 220                 if (slot->curl_result != CURLE_OK ||
 221                     !alt_req->buffer->len) {
 222
 223                         /* Try reusing the slot to get non-http alternates */
 224                         alt_req->http_specific = 0;
 225                         strbuf_reset(alt_req->url);
 226                         strbuf_addf(alt_req->url, "%s/objects/info/alternates",
 227                                     base);
 228                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 229                                          alt_req->url->buf);
 230                         active_requests++;
 231                         slot->in_use = 1;
 232                         if (slot->finished != NULL)
 233                                 (*slot->finished) = 0;
 234                         if (!start_active_slot(slot)) {
 235                                 cdata->got_alternates = -1;
 236                                 slot->in_use = 0;
 237                                 if (slot->finished != NULL)
 238                                         (*slot->finished) = 1;
 239                         }
 240                         return;
 241                 }
 242         } else if (slot->curl_result != CURLE_OK) {
 243                 if (!missing_target(slot)) {
 244                         cdata->got_alternates = -1;
 245                         return;
 246                 }
 247         }
 248
 249         fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
 250         alt_req->buffer->len--;
 251         data = alt_req->buffer->buf;
 252
 253         while (i < alt_req->buffer->len) {
 254                 int posn = i;
 255                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 256                         posn++;
 257                 if (data[posn] == '\n') {
 258                         int okay = 0;
 259                         int serverlen = 0;
 260                         struct alt_base *newalt;
 261                         if (data[i] == '/') {
 262                                 /*
 263                                  * This counts
 264                                  * http://git.host/pub/scm/linux.git/
 265                                  * -----------here^
 266                                  * so memcpy(dst, base, serverlen) will
 267                                  * copy up to "...git.host".
 268                                  */
 269                                 const char *colon_ss = strstr(base,"://");
 270                                 if (colon_ss) {
 271                                         serverlen = (strchr(colon_ss + 3, '/')
 272                                                      - base);
 273                                         okay = 1;
 274                                 }
 275                         } else if (!memcmp(data + i, "../", 3)) {
 276                                 /*
 277                                  * Relative URL; chop the corresponding
 278                                  * number of subpath from base (and ../
 279                                  * from data), and concatenate the result.
 280                                  *
 281                                  * The code first drops ../ from data, and
 282                                  * then drops one ../ from data and one path
 283                                  * from base.  IOW, one extra ../ is dropped
 284                                  * from data than path is dropped from base.
 285                                  *
 286                                  * This is not wrong.  The alternate in
 287                                  *     http://git.host/pub/scm/linux.git/
 288                                  * to borrow from
 289                                  *     http://git.host/pub/scm/linus.git/
 290                                  * is ../../linus.git/objects/.  You need
 291                                  * two ../../ to borrow from your direct
 292                                  * neighbour.
 293                                  */
 294                                 i += 3;
 295                                 serverlen = strlen(base);
 296                                 while (i + 2 < posn &&
 297                                        !memcmp(data + i, "../", 3)) {
 298                                         do {
 299                                                 serverlen--;
 300                                         } while (serverlen &&
 301                                                  base[serverlen - 1] != '/');
 302                                         i += 3;
 303                                 }
 304                                 /* If the server got removed, give up. */
 305                                 okay = strchr(base, ':') - base + 3 <
 306                                        serverlen;
 307                         } else if (alt_req->http_specific) {
 308                                 char *colon = strchr(data + i, ':');
 309                                 char *slash = strchr(data + i, '/');
 310                                 if (colon && slash && colon < data + posn &&
 311                                     slash < data + posn && colon < slash) {
 312                                         okay = 1;
 313                                 }
 314                         }
 315                         if (okay) {
 316                                 struct strbuf target = STRBUF_INIT;
 317                                 strbuf_add(&target, base, serverlen);
 318                                 strbuf_add(&target, data + i, posn - i);
 319                                 if (!strbuf_strip_suffix(&target, "objects")) {
 320                                         warning("ignoring alternate that does"
 321                                                 " not end in 'objects': %s",
 322                                                 target.buf);
 323                                         strbuf_release(&target);
 324                                 } else if (is_alternate_allowed(target.buf)) {
 325                                         warning("adding alternate object store: %s",
 326                                                 target.buf);
 327                                         newalt = xmalloc(sizeof(*newalt));
 328                                         newalt->next = NULL;
 329                                         newalt->base = strbuf_detach(&target, NULL);
 330                                         newalt->got_indices = 0;
 331                                         newalt->packs = NULL;
 332
 333                                         while (tail->next != NULL)
 334                                                 tail = tail->next;
 335                                         tail->next = newalt;
 336                                 } else {
 337                                         strbuf_release(&target);
 338                                 }
 339                         }
 340                 }
 341                 i = posn + 1;
 342         }
 343
 344         cdata->got_alternates = 1;
 345 }
 346
 347 static void fetch_alternates(struct walker *walker, const char *base)
 348 {
 349         struct strbuf buffer = STRBUF_INIT;
 350         struct strbuf url = STRBUF_INIT;
 351         struct active_request_slot *slot;
 352         struct alternates_request alt_req;
 353         struct walker_data *cdata = walker->data;
 354
 355         /*
 356          * If another request has already started fetching alternates,
 357          * wait for them to arrive and return to processing this request's
 358          * curl message
 359          */
 360 #ifdef USE_CURL_MULTI
 361         while (cdata->got_alternates == 0) {
 362                 step_active_slots();
 363         }
 364 #endif
 365
 366         /* Nothing to do if they've already been fetched */
 367         if (cdata->got_alternates == 1)
 368                 return;
 369
 370         /* Start the fetch */
 371         cdata->got_alternates = 0;
 372
 373         if (walker->get_verbosely)
 374                 fprintf(stderr, "Getting alternates list for %s\n", base);
 375
 376         strbuf_addf(&url, "%s/objects/info/http-alternates", base);
 377
 378         /*
 379          * Use a callback to process the result, since another request
 380          * may fail and need to have alternates loaded before continuing
 381          */
 382         slot = get_active_slot();
 383         slot->callback_func = process_alternates_response;
 384         alt_req.walker = walker;
 385         slot->callback_data = &alt_req;
 386
 387         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 388         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 389         curl_easy_setopt(slot->curl, CURLOPT_URL, url.buf);
 390
 391         alt_req.base = base;
 392         alt_req.url = &url;
 393         alt_req.buffer = &buffer;
 394         alt_req.http_specific = 1;
 395         alt_req.slot = slot;
 396
 397         if (start_active_slot(slot))
 398                 run_active_slot(slot);
 399         else
 400                 cdata->got_alternates = -1;
 401
 402         strbuf_release(&buffer);
 403         strbuf_release(&url);
 404 }
 405
 406 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 407 {
 408         int ret;
 409
 410         if (repo->got_indices)
 411                 return 0;
 412
 413         if (walker->get_verbosely)
 414                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 415
 416         switch (http_get_info_packs(repo->base, &repo->packs)) {
 417         case HTTP_OK:
 418         case HTTP_MISSING_TARGET:
 419                 repo->got_indices = 1;
 420                 ret = 0;
 421                 break;
 422         default:
 423                 repo->got_indices = 0;
 424                 ret = -1;
 425         }
 426
 427         return ret;
 428 }
 429
 430 static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 431 {
 432         struct packed_git *target;
 433         int ret;
 434         struct slot_results results;
 435         struct http_pack_request *preq;
 436
 437         if (fetch_indices(walker, repo))
 438                 return -1;
 439         target = find_sha1_pack(sha1, repo->packs);
 440         if (!target)
 441                 return -1;
 442         close_pack_index(target);
 443
 444         if (walker->get_verbosely) {
 445                 fprintf(stderr, "Getting pack %s\n",
 446                         hash_to_hex(target->hash));
 447                 fprintf(stderr, " which contains %s\n",
 448                         hash_to_hex(sha1));
 449         }
 450
 451         preq = new_http_pack_request(target->hash, repo->base);
 452         if (preq == NULL)
 453                 goto abort;
 454         preq->slot->results = &results;
 455
 456         if (start_active_slot(preq->slot)) {
 457                 run_active_slot(preq->slot);
 458                 if (results.curl_result != CURLE_OK) {
 459                         error("Unable to get pack file %s\n%s", preq->url,
 460                               curl_errorstr);
 461                         goto abort;
 462                 }
 463         } else {
 464                 error("Unable to start request");
 465                 goto abort;
 466         }
 467
 468         ret = finish_http_pack_request(preq);
 469         release_http_pack_request(preq);
 470         if (ret)
 471                 return ret;
 472         http_install_packfile(target, &repo->packs);
 473
 474         return 0;
 475
 476 abort:
 477         return -1;
 478 }
 479
 480 static void abort_object_request(struct object_request *obj_req)
 481 {
 482         release_object_request(obj_req);
 483 }
 484
 485 static int fetch_object(struct walker *walker, unsigned char *hash)
 486 {
 487         char *hex = hash_to_hex(hash);
 488         int ret = 0;
 489         struct object_request *obj_req = NULL;
 490         struct http_object_request *req;
 491         struct list_head *pos, *head = &object_queue_head;
 492
 493         list_for_each(pos, head) {
 494                 obj_req = list_entry(pos, struct object_request, node);
 495                 if (hasheq(obj_req->oid.hash, hash))
 496                         break;
 497         }
 498         if (obj_req == NULL)
 499                 return error("Couldn't find request for %s in the queue", hex);
 500
 501         if (has_object_file(&obj_req->oid)) {
 502                 if (obj_req->req != NULL)
 503                         abort_http_object_request(obj_req->req);
 504                 abort_object_request(obj_req);
 505                 return 0;
 506         }
 507
 508 #ifdef USE_CURL_MULTI
 509         while (obj_req->state == WAITING)
 510                 step_active_slots();
 511 #else
 512         start_object_request(walker, obj_req);
 513 #endif
 514
 515         /*
 516          * obj_req->req might change when fetching alternates in the callback
 517          * process_object_response; therefore, the "shortcut" variable, req,
 518          * is used only after we're done with slots.
 519          */
 520         while (obj_req->state == ACTIVE)
 521                 run_active_slot(obj_req->req->slot);
 522
 523         req = obj_req->req;
 524
 525         if (req->localfile != -1) {
 526                 close(req->localfile);
 527                 req->localfile = -1;
 528         }
 529
 530         normalize_curl_result(&req->curl_result, req->http_code,
 531                               req->errorstr, sizeof(req->errorstr));
 532
 533         if (obj_req->state == ABORTED) {
 534                 ret = error("Request for %s aborted", hex);
 535         } else if (req->curl_result != CURLE_OK &&
 536                    req->http_code != 416) {
 537                 if (missing_target(req))
 538                         ret = -1; /* Be silent, it is probably in a pack. */
 539                 else
 540                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 541                                     req->errorstr, req->curl_result,
 542                                     req->http_code, hex);
 543         } else if (req->zret != Z_STREAM_END) {
 544                 walker->corrupt_object_found++;
 545                 ret = error("File %s (%s) corrupt", hex, req->url);
 546         } else if (!oideq(&obj_req->oid, &req->real_oid)) {
 547                 ret = error("File %s has bad hash", hex);
 548         } else if (req->rename < 0) {
 549                 struct strbuf buf = STRBUF_INIT;
 550                 loose_object_path(the_repository, &buf, &req->oid);
 551                 ret = error("unable to write sha1 filename %s", buf.buf);
 552                 strbuf_release(&buf);
 553         }
 554
 555         release_http_object_request(req);
 556         release_object_request(obj_req);
 557         return ret;
 558 }
 559
 560 static int fetch(struct walker *walker, unsigned char *hash)
 561 {
 562         struct walker_data *data = walker->data;
 563         struct alt_base *altbase = data->alt;
 564
 565         if (!fetch_object(walker, hash))
 566                 return 0;
 567         while (altbase) {
 568                 if (!http_fetch_pack(walker, altbase, hash))
 569                         return 0;
 570                 fetch_alternates(walker, data->alt->base);
 571                 altbase = altbase->next;
 572         }
 573         return error("Unable to find %s under %s", hash_to_hex(hash),
 574                      data->alt->base);
 575 }
 576
 577 static int fetch_ref(struct walker *walker, struct ref *ref)
 578 {
 579         struct walker_data *data = walker->data;
 580         return http_fetch_ref(data->alt->base, ref);
 581 }
 582
 583 static void cleanup(struct walker *walker)
 584 {
 585         struct walker_data *data = walker->data;
 586         struct alt_base *alt, *alt_next;
 587
 588         if (data) {
 589                 alt = data->alt;
 590                 while (alt) {
 591                         alt_next = alt->next;
 592
 593                         free(alt->base);
 594                         free(alt);
 595
 596                         alt = alt_next;
 597                 }
 598                 free(data);
 599                 walker->data = NULL;
 600         }
 601 }
 602
 603 struct walker *get_http_walker(const char *url)
 604 {
 605         char *s;
 606         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 607         struct walker *walker = xmalloc(sizeof(struct walker));
 608
 609         data->alt = xmalloc(sizeof(*data->alt));
 610         data->alt->base = xstrdup(url);
 611         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 612                 *s = 0;
 613
 614         data->alt->got_indices = 0;
 615         data->alt->packs = NULL;
 616         data->alt->next = NULL;
 617         data->got_alternates = -1;
 618
 619         walker->corrupt_object_found = 0;
 620         walker->fetch = fetch;
 621         walker->fetch_ref = fetch_ref;
 622         walker->prefetch = prefetch;
 623         walker->cleanup = cleanup;
 624         walker->data = data;
 625
 626 #ifdef USE_CURL_MULTI
 627         add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 628 #endif
 629
 630         return walker;
 631 }