http-walker.c

   1 #include "cache.h"
   2 #include "commit.h"
   3 #include "walker.h"
   4 #include "http.h"
   5
   6 struct alt_base
   7 {
   8         char *base;
   9         int got_indices;
  10         struct packed_git *packs;
  11         struct alt_base *next;
  12 };
  13
  14 enum object_request_state {
  15         WAITING,
  16         ABORTED,
  17         ACTIVE,
  18         COMPLETE,
  19 };
  20
  21 struct object_request
  22 {
  23         struct walker *walker;
  24         unsigned char sha1[20];
  25         struct alt_base *repo;
  26         enum object_request_state state;
  27         struct http_object_request *req;
  28         struct object_request *next;
  29 };
  30
  31 struct alternates_request {
  32         struct walker *walker;
  33         const char *base;
  34         char *url;
  35         struct strbuf *buffer;
  36         struct active_request_slot *slot;
  37         int http_specific;
  38 };
  39
  40 struct walker_data {
  41         const char *url;
  42         int got_alternates;
  43         struct alt_base *alt;
  44 };
  45
  46 static struct object_request *object_queue_head;
  47
  48 static void fetch_alternates(struct walker *walker, const char *base);
  49
  50 static void process_object_response(void *callback_data);
  51
  52 static void start_object_request(struct walker *walker,
  53                                  struct object_request *obj_req)
  54 {
  55         struct active_request_slot *slot;
  56         struct http_object_request *req;
  57
  58         req = new_http_object_request(obj_req->repo->base, obj_req->sha1);
  59         if (req == NULL) {
  60                 obj_req->state = ABORTED;
  61                 return;
  62         }
  63         obj_req->req = req;
  64
  65         slot = req->slot;
  66         slot->callback_func = process_object_response;
  67         slot->callback_data = obj_req;
  68
  69         /* Try to get the request started, abort the request on error */
  70         obj_req->state = ACTIVE;
  71         if (!start_active_slot(slot)) {
  72                 obj_req->state = ABORTED;
  73                 release_http_object_request(req);
  74                 return;
  75         }
  76 }
  77
  78 static void finish_object_request(struct object_request *obj_req)
  79 {
  80         if (finish_http_object_request(obj_req->req))
  81                 return;
  82
  83         if (obj_req->req->rename == 0)
  84                 walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
  85 }
  86
  87 static void process_object_response(void *callback_data)
  88 {
  89         struct object_request *obj_req =
  90                 (struct object_request *)callback_data;
  91         struct walker *walker = obj_req->walker;
  92         struct walker_data *data = walker->data;
  93         struct alt_base *alt = data->alt;
  94
  95         process_http_object_request(obj_req->req);
  96         obj_req->state = COMPLETE;
  97
  98         /* Use alternates if necessary */
  99         if (missing_target(obj_req->req)) {
 100                 fetch_alternates(walker, alt->base);
 101                 if (obj_req->repo->next != NULL) {
 102                         obj_req->repo =
 103                                 obj_req->repo->next;
 104                         release_http_object_request(obj_req->req);
 105                         start_object_request(walker, obj_req);
 106                         return;
 107                 }
 108         }
 109
 110         finish_object_request(obj_req);
 111 }
 112
 113 static void release_object_request(struct object_request *obj_req)
 114 {
 115         struct object_request *entry = object_queue_head;
 116
 117         if (obj_req->req !=NULL && obj_req->req->localfile != -1)
 118                 error("fd leakage in release: %d", obj_req->req->localfile);
 119         if (obj_req == object_queue_head) {
 120                 object_queue_head = obj_req->next;
 121         } else {
 122                 while (entry->next != NULL && entry->next != obj_req)
 123                         entry = entry->next;
 124                 if (entry->next == obj_req)
 125                         entry->next = entry->next->next;
 126         }
 127
 128         free(obj_req);
 129 }
 130
 131 #ifdef USE_CURL_MULTI
 132 static int fill_active_slot(struct walker *walker)
 133 {
 134         struct object_request *obj_req;
 135
 136         for (obj_req = object_queue_head; obj_req; obj_req = obj_req->next) {
 137                 if (obj_req->state == WAITING) {
 138                         if (has_sha1_file(obj_req->sha1))
 139                                 obj_req->state = COMPLETE;
 140                         else {
 141                                 start_object_request(walker, obj_req);
 142                                 return 1;
 143                         }
 144                 }
 145         }
 146         return 0;
 147 }
 148 #endif
 149
 150 static void prefetch(struct walker *walker, unsigned char *sha1)
 151 {
 152         struct object_request *newreq;
 153         struct object_request *tail;
 154         struct walker_data *data = walker->data;
 155
 156         newreq = xmalloc(sizeof(*newreq));
 157         newreq->walker = walker;
 158         hashcpy(newreq->sha1, sha1);
 159         newreq->repo = data->alt;
 160         newreq->state = WAITING;
 161         newreq->req = NULL;
 162         newreq->next = NULL;
 163
 164         http_is_verbose = walker->get_verbosely;
 165
 166         if (object_queue_head == NULL) {
 167                 object_queue_head = newreq;
 168         } else {
 169                 tail = object_queue_head;
 170                 while (tail->next != NULL)
 171                         tail = tail->next;
 172                 tail->next = newreq;
 173         }
 174
 175 #ifdef USE_CURL_MULTI
 176         fill_active_slots();
 177         step_active_slots();
 178 #endif
 179 }
 180
 181 static void process_alternates_response(void *callback_data)
 182 {
 183         struct alternates_request *alt_req =
 184                 (struct alternates_request *)callback_data;
 185         struct walker *walker = alt_req->walker;
 186         struct walker_data *cdata = walker->data;
 187         struct active_request_slot *slot = alt_req->slot;
 188         struct alt_base *tail = cdata->alt;
 189         const char *base = alt_req->base;
 190         static const char null_byte = '\0';
 191         char *data;
 192         int i = 0;
 193
 194         if (alt_req->http_specific) {
 195                 if (slot->curl_result != CURLE_OK ||
 196                     !alt_req->buffer->len) {
 197
 198                         /* Try reusing the slot to get non-http alternates */
 199                         alt_req->http_specific = 0;
 200                         sprintf(alt_req->url, "%s/objects/info/alternates",
 201                                 base);
 202                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 203                                          alt_req->url);
 204                         active_requests++;
 205                         slot->in_use = 1;
 206                         if (slot->finished != NULL)
 207                                 (*slot->finished) = 0;
 208                         if (!start_active_slot(slot)) {
 209                                 cdata->got_alternates = -1;
 210                                 slot->in_use = 0;
 211                                 if (slot->finished != NULL)
 212                                         (*slot->finished) = 1;
 213                         }
 214                         return;
 215                 }
 216         } else if (slot->curl_result != CURLE_OK) {
 217                 if (!missing_target(slot)) {
 218                         cdata->got_alternates = -1;
 219                         return;
 220                 }
 221         }
 222
 223         fwrite_buffer(&null_byte, 1, 1, alt_req->buffer);
 224         alt_req->buffer->len--;
 225         data = alt_req->buffer->buf;
 226
 227         while (i < alt_req->buffer->len) {
 228                 int posn = i;
 229                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 230                         posn++;
 231                 if (data[posn] == '\n') {
 232                         int okay = 0;
 233                         int serverlen = 0;
 234                         struct alt_base *newalt;
 235                         char *target = NULL;
 236                         if (data[i] == '/') {
 237                                 /*
 238                                  * This counts
 239                                  * http://git.host/pub/scm/linux.git/
 240                                  * -----------here^
 241                                  * so memcpy(dst, base, serverlen) will
 242                                  * copy up to "...git.host".
 243                                  */
 244                                 const char *colon_ss = strstr(base,"://");
 245                                 if (colon_ss) {
 246                                         serverlen = (strchr(colon_ss + 3, '/')
 247                                                      - base);
 248                                         okay = 1;
 249                                 }
 250                         } else if (!memcmp(data + i, "../", 3)) {
 251                                 /*
 252                                  * Relative URL; chop the corresponding
 253                                  * number of subpath from base (and ../
 254                                  * from data), and concatenate the result.
 255                                  *
 256                                  * The code first drops ../ from data, and
 257                                  * then drops one ../ from data and one path
 258                                  * from base.  IOW, one extra ../ is dropped
 259                                  * from data than path is dropped from base.
 260                                  *
 261                                  * This is not wrong.  The alternate in
 262                                  *     http://git.host/pub/scm/linux.git/
 263                                  * to borrow from
 264                                  *     http://git.host/pub/scm/linus.git/
 265                                  * is ../../linus.git/objects/.  You need
 266                                  * two ../../ to borrow from your direct
 267                                  * neighbour.
 268                                  */
 269                                 i += 3;
 270                                 serverlen = strlen(base);
 271                                 while (i + 2 < posn &&
 272                                        !memcmp(data + i, "../", 3)) {
 273                                         do {
 274                                                 serverlen--;
 275                                         } while (serverlen &&
 276                                                  base[serverlen - 1] != '/');
 277                                         i += 3;
 278                                 }
 279                                 /* If the server got removed, give up. */
 280                                 okay = strchr(base, ':') - base + 3 <
 281                                        serverlen;
 282                         } else if (alt_req->http_specific) {
 283                                 char *colon = strchr(data + i, ':');
 284                                 char *slash = strchr(data + i, '/');
 285                                 if (colon && slash && colon < data + posn &&
 286                                     slash < data + posn && colon < slash) {
 287                                         okay = 1;
 288                                 }
 289                         }
 290                         /* skip "objects\n" at end */
 291                         if (okay) {
 292                                 target = xmalloc(serverlen + posn - i - 6);
 293                                 memcpy(target, base, serverlen);
 294                                 memcpy(target + serverlen, data + i,
 295                                        posn - i - 7);
 296                                 target[serverlen + posn - i - 7] = 0;
 297                                 if (walker->get_verbosely)
 298                                         fprintf(stderr,
 299                                                 "Also look at %s\n", target);
 300                                 newalt = xmalloc(sizeof(*newalt));
 301                                 newalt->next = NULL;
 302                                 newalt->base = target;
 303                                 newalt->got_indices = 0;
 304                                 newalt->packs = NULL;
 305
 306                                 while (tail->next != NULL)
 307                                         tail = tail->next;
 308                                 tail->next = newalt;
 309                         }
 310                 }
 311                 i = posn + 1;
 312         }
 313
 314         cdata->got_alternates = 1;
 315 }
 316
 317 static void fetch_alternates(struct walker *walker, const char *base)
 318 {
 319         struct strbuf buffer = STRBUF_INIT;
 320         char *url;
 321         struct active_request_slot *slot;
 322         struct alternates_request alt_req;
 323         struct walker_data *cdata = walker->data;
 324
 325         /*
 326          * If another request has already started fetching alternates,
 327          * wait for them to arrive and return to processing this request's
 328          * curl message
 329          */
 330 #ifdef USE_CURL_MULTI
 331         while (cdata->got_alternates == 0) {
 332                 step_active_slots();
 333         }
 334 #endif
 335
 336         /* Nothing to do if they've already been fetched */
 337         if (cdata->got_alternates == 1)
 338                 return;
 339
 340         /* Start the fetch */
 341         cdata->got_alternates = 0;
 342
 343         if (walker->get_verbosely)
 344                 fprintf(stderr, "Getting alternates list for %s\n", base);
 345
 346         url = xmalloc(strlen(base) + 31);
 347         sprintf(url, "%s/objects/info/http-alternates", base);
 348
 349         /*
 350          * Use a callback to process the result, since another request
 351          * may fail and need to have alternates loaded before continuing
 352          */
 353         slot = get_active_slot();
 354         slot->callback_func = process_alternates_response;
 355         alt_req.walker = walker;
 356         slot->callback_data = &alt_req;
 357
 358         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 359         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 360         curl_easy_setopt(slot->curl, CURLOPT_URL, url);
 361
 362         alt_req.base = base;
 363         alt_req.url = url;
 364         alt_req.buffer = &buffer;
 365         alt_req.http_specific = 1;
 366         alt_req.slot = slot;
 367
 368         if (start_active_slot(slot))
 369                 run_active_slot(slot);
 370         else
 371                 cdata->got_alternates = -1;
 372
 373         strbuf_release(&buffer);
 374         free(url);
 375 }
 376
 377 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 378 {
 379         int ret;
 380
 381         if (repo->got_indices)
 382                 return 0;
 383
 384         if (walker->get_verbosely)
 385                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 386
 387         switch (http_get_info_packs(repo->base, &repo->packs)) {
 388         case HTTP_OK:
 389         case HTTP_MISSING_TARGET:
 390                 repo->got_indices = 1;
 391                 ret = 0;
 392                 break;
 393         default:
 394                 repo->got_indices = 0;
 395                 ret = -1;
 396         }
 397
 398         return ret;
 399 }
 400
 401 static int fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 402 {
 403         struct packed_git *target;
 404         int ret;
 405         struct slot_results results;
 406         struct http_pack_request *preq;
 407
 408         if (fetch_indices(walker, repo))
 409                 return -1;
 410         target = find_sha1_pack(sha1, repo->packs);
 411         if (!target)
 412                 return -1;
 413
 414         if (walker->get_verbosely) {
 415                 fprintf(stderr, "Getting pack %s\n",
 416                         sha1_to_hex(target->sha1));
 417                 fprintf(stderr, " which contains %s\n",
 418                         sha1_to_hex(sha1));
 419         }
 420
 421         preq = new_http_pack_request(target, repo->base);
 422         if (preq == NULL)
 423                 goto abort;
 424         preq->lst = &repo->packs;
 425         preq->slot->results = &results;
 426
 427         if (start_active_slot(preq->slot)) {
 428                 run_active_slot(preq->slot);
 429                 if (results.curl_result != CURLE_OK) {
 430                         error("Unable to get pack file %s\n%s", preq->url,
 431                               curl_errorstr);
 432                         goto abort;
 433                 }
 434         } else {
 435                 error("Unable to start request");
 436                 goto abort;
 437         }
 438
 439         ret = finish_http_pack_request(preq);
 440         release_http_pack_request(preq);
 441         if (ret)
 442                 return ret;
 443
 444         return 0;
 445
 446 abort:
 447         return -1;
 448 }
 449
 450 static void abort_object_request(struct object_request *obj_req)
 451 {
 452         release_object_request(obj_req);
 453 }
 454
 455 static int fetch_object(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 456 {
 457         char *hex = sha1_to_hex(sha1);
 458         int ret = 0;
 459         struct object_request *obj_req = object_queue_head;
 460         struct http_object_request *req;
 461
 462         while (obj_req != NULL && hashcmp(obj_req->sha1, sha1))
 463                 obj_req = obj_req->next;
 464         if (obj_req == NULL)
 465                 return error("Couldn't find request for %s in the queue", hex);
 466
 467         if (has_sha1_file(obj_req->sha1)) {
 468                 if (obj_req->req != NULL)
 469                         abort_http_object_request(obj_req->req);
 470                 abort_object_request(obj_req);
 471                 return 0;
 472         }
 473
 474 #ifdef USE_CURL_MULTI
 475         while (obj_req->state == WAITING)
 476                 step_active_slots();
 477 #else
 478         start_object_request(walker, obj_req);
 479 #endif
 480
 481         /*
 482          * obj_req->req might change when fetching alternates in the callback
 483          * process_object_response; therefore, the "shortcut" variable, req,
 484          * is used only after we're done with slots.
 485          */
 486         while (obj_req->state == ACTIVE)
 487                 run_active_slot(obj_req->req->slot);
 488
 489         req = obj_req->req;
 490
 491         if (req->localfile != -1) {
 492                 close(req->localfile);
 493                 req->localfile = -1;
 494         }
 495
 496         if (obj_req->state == ABORTED) {
 497                 ret = error("Request for %s aborted", hex);
 498         } else if (req->curl_result != CURLE_OK &&
 499                    req->http_code != 416) {
 500                 if (missing_target(req))
 501                         ret = -1; /* Be silent, it is probably in a pack. */
 502                 else
 503                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 504                                     req->errorstr, req->curl_result,
 505                                     req->http_code, hex);
 506         } else if (req->zret != Z_STREAM_END) {
 507                 walker->corrupt_object_found++;
 508                 ret = error("File %s (%s) corrupt", hex, req->url);
 509         } else if (hashcmp(obj_req->sha1, req->real_sha1)) {
 510                 ret = error("File %s has bad hash", hex);
 511         } else if (req->rename < 0) {
 512                 ret = error("unable to write sha1 filename %s",
 513                             req->filename);
 514         }
 515
 516         release_http_object_request(req);
 517         release_object_request(obj_req);
 518         return ret;
 519 }
 520
 521 static int fetch(struct walker *walker, unsigned char *sha1)
 522 {
 523         struct walker_data *data = walker->data;
 524         struct alt_base *altbase = data->alt;
 525
 526         if (!fetch_object(walker, altbase, sha1))
 527                 return 0;
 528         while (altbase) {
 529                 if (!fetch_pack(walker, altbase, sha1))
 530                         return 0;
 531                 fetch_alternates(walker, data->alt->base);
 532                 altbase = altbase->next;
 533         }
 534         return error("Unable to find %s under %s", sha1_to_hex(sha1),
 535                      data->alt->base);
 536 }
 537
 538 static int fetch_ref(struct walker *walker, struct ref *ref)
 539 {
 540         struct walker_data *data = walker->data;
 541         return http_fetch_ref(data->alt->base, ref);
 542 }
 543
 544 static void cleanup(struct walker *walker)
 545 {
 546         struct walker_data *data = walker->data;
 547         struct alt_base *alt, *alt_next;
 548
 549         if (data) {
 550                 alt = data->alt;
 551                 while (alt) {
 552                         alt_next = alt->next;
 553
 554                         free(alt->base);
 555                         free(alt);
 556
 557                         alt = alt_next;
 558                 }
 559                 free(data);
 560                 walker->data = NULL;
 561         }
 562 }
 563
 564 struct walker *get_http_walker(const char *url)
 565 {
 566         char *s;
 567         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 568         struct walker *walker = xmalloc(sizeof(struct walker));
 569
 570         data->alt = xmalloc(sizeof(*data->alt));
 571         data->alt->base = xmalloc(strlen(url) + 1);
 572         strcpy(data->alt->base, url);
 573         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 574                 *s = 0;
 575
 576         data->alt->got_indices = 0;
 577         data->alt->packs = NULL;
 578         data->alt->next = NULL;
 579         data->got_alternates = -1;
 580
 581         walker->corrupt_object_found = 0;
 582         walker->fetch = fetch;
 583         walker->fetch_ref = fetch_ref;
 584         walker->prefetch = prefetch;
 585         walker->cleanup = cleanup;
 586         walker->data = data;
 587
 588 #ifdef USE_CURL_MULTI
 589         add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 590 #endif
 591
 592         return walker;
 593 }