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