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         char *base;
   8         int got_indices;
   9         struct packed_git *packs;
  10         struct alt_base *next;
  11 };
  12
  13 enum object_request_state {
  14         WAITING,
  15         ABORTED,
  16         ACTIVE,
  17         COMPLETE
  18 };
  19
  20 struct object_request {
  21         struct walker *walker;
  22         unsigned char sha1[20];
  23         struct alt_base *repo;
  24         enum object_request_state state;
  25         struct http_object_request *req;
  26         struct object_request *next;
  27 };
  28
  29 struct alternates_request {
  30         struct walker *walker;
  31         const char *base;
  32         struct strbuf *url;
  33         struct strbuf *buffer;
  34         struct active_request_slot *slot;
  35         int http_specific;
  36 };
  37
  38 struct walker_data {
  39         const char *url;
  40         int got_alternates;
  41         struct alt_base *alt;
  42 };
  43
  44 static struct object_request *object_queue_head;
  45
  46 static void fetch_alternates(struct walker *walker, const char *base);
  47
  48 static void process_object_response(void *callback_data);
  49
  50 static void start_object_request(struct walker *walker,
  51                                  struct object_request *obj_req)
  52 {
  53         struct active_request_slot *slot;
  54         struct http_object_request *req;
  55
  56         req = new_http_object_request(obj_req->repo->base, obj_req->sha1);
  57         if (req == NULL) {
  58                 obj_req->state = ABORTED;
  59                 return;
  60         }
  61         obj_req->req = req;
  62
  63         slot = req->slot;
  64         slot->callback_func = process_object_response;
  65         slot->callback_data = obj_req;
  66
  67         /* Try to get the request started, abort the request on error */
  68         obj_req->state = ACTIVE;
  69         if (!start_active_slot(slot)) {
  70                 obj_req->state = ABORTED;
  71                 release_http_object_request(req);
  72                 return;
  73         }
  74 }
  75
  76 static void finish_object_request(struct object_request *obj_req)
  77 {
  78         if (finish_http_object_request(obj_req->req))
  79                 return;
  80
  81         if (obj_req->req->rename == 0)
  82                 walker_say(obj_req->walker, "got %s\n", sha1_to_hex(obj_req->sha1));
  83 }
  84
  85 static void process_object_response(void *callback_data)
  86 {
  87         struct object_request *obj_req =
  88                 (struct object_request *)callback_data;
  89         struct walker *walker = obj_req->walker;
  90         struct walker_data *data = walker->data;
  91         struct alt_base *alt = data->alt;
  92
  93         process_http_object_request(obj_req->req);
  94         obj_req->state = COMPLETE;
  95
  96         /* Use alternates if necessary */
  97         if (missing_target(obj_req->req)) {
  98                 fetch_alternates(walker, alt->base);
  99                 if (obj_req->repo->next != NULL) {
 100                         obj_req->repo =
 101                                 obj_req->repo->next;
 102                         release_http_object_request(obj_req->req);
 103                         start_object_request(walker, obj_req);
 104                         return;
 105                 }
 106         }
 107
 108         finish_object_request(obj_req);
 109 }
 110
 111 static void release_object_request(struct object_request *obj_req)
 112 {
 113         struct object_request *entry = object_queue_head;
 114
 115         if (obj_req->req !=NULL && obj_req->req->localfile != -1)
 116                 error("fd leakage in release: %d", obj_req->req->localfile);
 117         if (obj_req == object_queue_head) {
 118                 object_queue_head = obj_req->next;
 119         } else {
 120                 while (entry->next != NULL && entry->next != obj_req)
 121                         entry = entry->next;
 122                 if (entry->next == obj_req)
 123                         entry->next = entry->next->next;
 124         }
 125
 126         free(obj_req);
 127 }
 128
 129 #ifdef USE_CURL_MULTI
 130 static int fill_active_slot(struct walker *walker)
 131 {
 132         struct object_request *obj_req;
 133
 134         for (obj_req = object_queue_head; obj_req; obj_req = obj_req->next) {
 135                 if (obj_req->state == WAITING) {
 136                         if (has_sha1_file(obj_req->sha1))
 137                                 obj_req->state = COMPLETE;
 138                         else {
 139                                 start_object_request(walker, obj_req);
 140                                 return 1;
 141                         }
 142                 }
 143         }
 144         return 0;
 145 }
 146 #endif
 147
 148 static void prefetch(struct walker *walker, unsigned char *sha1)
 149 {
 150         struct object_request *newreq;
 151         struct object_request *tail;
 152         struct walker_data *data = walker->data;
 153
 154         newreq = xmalloc(sizeof(*newreq));
 155         newreq->walker = walker;
 156         hashcpy(newreq->sha1, sha1);
 157         newreq->repo = data->alt;
 158         newreq->state = WAITING;
 159         newreq->req = NULL;
 160         newreq->next = NULL;
 161
 162         http_is_verbose = walker->get_verbosely;
 163
 164         if (object_queue_head == NULL) {
 165                 object_queue_head = newreq;
 166         } else {
 167                 tail = object_queue_head;
 168                 while (tail->next != NULL)
 169                         tail = tail->next;
 170                 tail->next = newreq;
 171         }
 172
 173 #ifdef USE_CURL_MULTI
 174         fill_active_slots();
 175         step_active_slots();
 176 #endif
 177 }
 178
 179 static void process_alternates_response(void *callback_data)
 180 {
 181         struct alternates_request *alt_req =
 182                 (struct alternates_request *)callback_data;
 183         struct walker *walker = alt_req->walker;
 184         struct walker_data *cdata = walker->data;
 185         struct active_request_slot *slot = alt_req->slot;
 186         struct alt_base *tail = cdata->alt;
 187         const char *base = alt_req->base;
 188         const char null_byte = '\0';
 189         char *data;
 190         int i = 0;
 191
 192         if (alt_req->http_specific) {
 193                 if (slot->curl_result != CURLE_OK ||
 194                     !alt_req->buffer->len) {
 195
 196                         /* Try reusing the slot to get non-http alternates */
 197                         alt_req->http_specific = 0;
 198                         strbuf_reset(alt_req->url);
 199                         strbuf_addf(alt_req->url, "%s/objects/info/alternates",
 200                                     base);
 201                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 202                                          alt_req->url->buf);
 203                         active_requests++;
 204                         slot->in_use = 1;
 205                         if (slot->finished != NULL)
 206                                 (*slot->finished) = 0;
 207                         if (!start_active_slot(slot)) {
 208                                 cdata->got_alternates = -1;
 209                                 slot->in_use = 0;
 210                                 if (slot->finished != NULL)
 211                                         (*slot->finished) = 1;
 212                         }
 213                         return;
 214                 }
 215         } else if (slot->curl_result != CURLE_OK) {
 216                 if (!missing_target(slot)) {
 217                         cdata->got_alternates = -1;
 218                         return;
 219                 }
 220         }
 221
 222         fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
 223         alt_req->buffer->len--;
 224         data = alt_req->buffer->buf;
 225
 226         while (i < alt_req->buffer->len) {
 227                 int posn = i;
 228                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 229                         posn++;
 230                 if (data[posn] == '\n') {
 231                         int okay = 0;
 232                         int serverlen = 0;
 233                         struct alt_base *newalt;
 234                         if (data[i] == '/') {
 235                                 /*
 236                                  * This counts
 237                                  * http://git.host/pub/scm/linux.git/
 238                                  * -----------here^
 239                                  * so memcpy(dst, base, serverlen) will
 240                                  * copy up to "...git.host".
 241                                  */
 242                                 const char *colon_ss = strstr(base,"://");
 243                                 if (colon_ss) {
 244                                         serverlen = (strchr(colon_ss + 3, '/')
 245                                                      - base);
 246                                         okay = 1;
 247                                 }
 248                         } else if (!memcmp(data + i, "../", 3)) {
 249                                 /*
 250                                  * Relative URL; chop the corresponding
 251                                  * number of subpath from base (and ../
 252                                  * from data), and concatenate the result.
 253                                  *
 254                                  * The code first drops ../ from data, and
 255                                  * then drops one ../ from data and one path
 256                                  * from base.  IOW, one extra ../ is dropped
 257                                  * from data than path is dropped from base.
 258                                  *
 259                                  * This is not wrong.  The alternate in
 260                                  *     http://git.host/pub/scm/linux.git/
 261                                  * to borrow from
 262                                  *     http://git.host/pub/scm/linus.git/
 263                                  * is ../../linus.git/objects/.  You need
 264                                  * two ../../ to borrow from your direct
 265                                  * neighbour.
 266                                  */
 267                                 i += 3;
 268                                 serverlen = strlen(base);
 269                                 while (i + 2 < posn &&
 270                                        !memcmp(data + i, "../", 3)) {
 271                                         do {
 272                                                 serverlen--;
 273                                         } while (serverlen &&
 274                                                  base[serverlen - 1] != '/');
 275                                         i += 3;
 276                                 }
 277                                 /* If the server got removed, give up. */
 278                                 okay = strchr(base, ':') - base + 3 <
 279                                        serverlen;
 280                         } else if (alt_req->http_specific) {
 281                                 char *colon = strchr(data + i, ':');
 282                                 char *slash = strchr(data + i, '/');
 283                                 if (colon && slash && colon < data + posn &&
 284                                     slash < data + posn && colon < slash) {
 285                                         okay = 1;
 286                                 }
 287                         }
 288                         /* skip "objects\n" at end */
 289                         if (okay) {
 290                                 struct strbuf target = STRBUF_INIT;
 291                                 strbuf_add(&target, base, serverlen);
 292                                 strbuf_add(&target, data + i, posn - i - 7);
 293                                 if (walker->get_verbosely)
 294                                         fprintf(stderr, "Also look at %s\n",
 295                                                 target.buf);
 296                                 newalt = xmalloc(sizeof(*newalt));
 297                                 newalt->next = NULL;
 298                                 newalt->base = strbuf_detach(&target, NULL);
 299                                 newalt->got_indices = 0;
 300                                 newalt->packs = NULL;
 301
 302                                 while (tail->next != NULL)
 303                                         tail = tail->next;
 304                                 tail->next = newalt;
 305                         }
 306                 }
 307                 i = posn + 1;
 308         }
 309
 310         cdata->got_alternates = 1;
 311 }
 312
 313 static void fetch_alternates(struct walker *walker, const char *base)
 314 {
 315         struct strbuf buffer = STRBUF_INIT;
 316         struct strbuf url = STRBUF_INIT;
 317         struct active_request_slot *slot;
 318         struct alternates_request alt_req;
 319         struct walker_data *cdata = walker->data;
 320
 321         /*
 322          * If another request has already started fetching alternates,
 323          * wait for them to arrive and return to processing this request's
 324          * curl message
 325          */
 326 #ifdef USE_CURL_MULTI
 327         while (cdata->got_alternates == 0) {
 328                 step_active_slots();
 329         }
 330 #endif
 331
 332         /* Nothing to do if they've already been fetched */
 333         if (cdata->got_alternates == 1)
 334                 return;
 335
 336         /* Start the fetch */
 337         cdata->got_alternates = 0;
 338
 339         if (walker->get_verbosely)
 340                 fprintf(stderr, "Getting alternates list for %s\n", base);
 341
 342         strbuf_addf(&url, "%s/objects/info/http-alternates", base);
 343
 344         /*
 345          * Use a callback to process the result, since another request
 346          * may fail and need to have alternates loaded before continuing
 347          */
 348         slot = get_active_slot();
 349         slot->callback_func = process_alternates_response;
 350         alt_req.walker = walker;
 351         slot->callback_data = &alt_req;
 352
 353         curl_easy_setopt(slot->curl, CURLOPT_FILE, &buffer);
 354         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 355         curl_easy_setopt(slot->curl, CURLOPT_URL, url.buf);
 356
 357         alt_req.base = base;
 358         alt_req.url = &url;
 359         alt_req.buffer = &buffer;
 360         alt_req.http_specific = 1;
 361         alt_req.slot = slot;
 362
 363         if (start_active_slot(slot))
 364                 run_active_slot(slot);
 365         else
 366                 cdata->got_alternates = -1;
 367
 368         strbuf_release(&buffer);
 369         strbuf_release(&url);
 370 }
 371
 372 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 373 {
 374         int ret;
 375
 376         if (repo->got_indices)
 377                 return 0;
 378
 379         if (walker->get_verbosely)
 380                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 381
 382         switch (http_get_info_packs(repo->base, &repo->packs)) {
 383         case HTTP_OK:
 384         case HTTP_MISSING_TARGET:
 385                 repo->got_indices = 1;
 386                 ret = 0;
 387                 break;
 388         default:
 389                 repo->got_indices = 0;
 390                 ret = -1;
 391         }
 392
 393         return ret;
 394 }
 395
 396 static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 397 {
 398         struct packed_git *target;
 399         int ret;
 400         struct slot_results results;
 401         struct http_pack_request *preq;
 402
 403         if (fetch_indices(walker, repo))
 404                 return -1;
 405         target = find_sha1_pack(sha1, repo->packs);
 406         if (!target)
 407                 return -1;
 408
 409         if (walker->get_verbosely) {
 410                 fprintf(stderr, "Getting pack %s\n",
 411                         sha1_to_hex(target->sha1));
 412                 fprintf(stderr, " which contains %s\n",
 413                         sha1_to_hex(sha1));
 414         }
 415
 416         preq = new_http_pack_request(target, repo->base);
 417         if (preq == NULL)
 418                 goto abort;
 419         preq->lst = &repo->packs;
 420         preq->slot->results = &results;
 421
 422         if (start_active_slot(preq->slot)) {
 423                 run_active_slot(preq->slot);
 424                 if (results.curl_result != CURLE_OK) {
 425                         error("Unable to get pack file %s\n%s", preq->url,
 426                               curl_errorstr);
 427                         goto abort;
 428                 }
 429         } else {
 430                 error("Unable to start request");
 431                 goto abort;
 432         }
 433
 434         ret = finish_http_pack_request(preq);
 435         release_http_pack_request(preq);
 436         if (ret)
 437                 return ret;
 438
 439         return 0;
 440
 441 abort:
 442         return -1;
 443 }
 444
 445 static void abort_object_request(struct object_request *obj_req)
 446 {
 447         release_object_request(obj_req);
 448 }
 449
 450 static int fetch_object(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 451 {
 452         char *hex = sha1_to_hex(sha1);
 453         int ret = 0;
 454         struct object_request *obj_req = object_queue_head;
 455         struct http_object_request *req;
 456
 457         while (obj_req != NULL && hashcmp(obj_req->sha1, sha1))
 458                 obj_req = obj_req->next;
 459         if (obj_req == NULL)
 460                 return error("Couldn't find request for %s in the queue", hex);
 461
 462         if (has_sha1_file(obj_req->sha1)) {
 463                 if (obj_req->req != NULL)
 464                         abort_http_object_request(obj_req->req);
 465                 abort_object_request(obj_req);
 466                 return 0;
 467         }
 468
 469 #ifdef USE_CURL_MULTI
 470         while (obj_req->state == WAITING)
 471                 step_active_slots();
 472 #else
 473         start_object_request(walker, obj_req);
 474 #endif
 475
 476         /*
 477          * obj_req->req might change when fetching alternates in the callback
 478          * process_object_response; therefore, the "shortcut" variable, req,
 479          * is used only after we're done with slots.
 480          */
 481         while (obj_req->state == ACTIVE)
 482                 run_active_slot(obj_req->req->slot);
 483
 484         req = obj_req->req;
 485
 486         if (req->localfile != -1) {
 487                 close(req->localfile);
 488                 req->localfile = -1;
 489         }
 490
 491         if (obj_req->state == ABORTED) {
 492                 ret = error("Request for %s aborted", hex);
 493         } else if (req->curl_result != CURLE_OK &&
 494                    req->http_code != 416) {
 495                 if (missing_target(req))
 496                         ret = -1; /* Be silent, it is probably in a pack. */
 497                 else
 498                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 499                                     req->errorstr, req->curl_result,
 500                                     req->http_code, hex);
 501         } else if (req->zret != Z_STREAM_END) {
 502                 walker->corrupt_object_found++;
 503                 ret = error("File %s (%s) corrupt", hex, req->url);
 504         } else if (hashcmp(obj_req->sha1, req->real_sha1)) {
 505                 ret = error("File %s has bad hash", hex);
 506         } else if (req->rename < 0) {
 507                 ret = error("unable to write sha1 filename %s",
 508                             sha1_file_name(req->sha1));
 509         }
 510
 511         release_http_object_request(req);
 512         release_object_request(obj_req);
 513         return ret;
 514 }
 515
 516 static int fetch(struct walker *walker, unsigned char *sha1)
 517 {
 518         struct walker_data *data = walker->data;
 519         struct alt_base *altbase = data->alt;
 520
 521         if (!fetch_object(walker, altbase, sha1))
 522                 return 0;
 523         while (altbase) {
 524                 if (!http_fetch_pack(walker, altbase, sha1))
 525                         return 0;
 526                 fetch_alternates(walker, data->alt->base);
 527                 altbase = altbase->next;
 528         }
 529         return error("Unable to find %s under %s", sha1_to_hex(sha1),
 530                      data->alt->base);
 531 }
 532
 533 static int fetch_ref(struct walker *walker, struct ref *ref)
 534 {
 535         struct walker_data *data = walker->data;
 536         return http_fetch_ref(data->alt->base, ref);
 537 }
 538
 539 static void cleanup(struct walker *walker)
 540 {
 541         struct walker_data *data = walker->data;
 542         struct alt_base *alt, *alt_next;
 543
 544         if (data) {
 545                 alt = data->alt;
 546                 while (alt) {
 547                         alt_next = alt->next;
 548
 549                         free(alt->base);
 550                         free(alt);
 551
 552                         alt = alt_next;
 553                 }
 554                 free(data);
 555                 walker->data = NULL;
 556         }
 557 }
 558
 559 struct walker *get_http_walker(const char *url)
 560 {
 561         char *s;
 562         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 563         struct walker *walker = xmalloc(sizeof(struct walker));
 564
 565         data->alt = xmalloc(sizeof(*data->alt));
 566         data->alt->base = xstrdup(url);
 567         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 568                 *s = 0;
 569
 570         data->alt->got_indices = 0;
 571         data->alt->packs = NULL;
 572         data->alt->next = NULL;
 573         data->got_alternates = -1;
 574
 575         walker->corrupt_object_found = 0;
 576         walker->fetch = fetch;
 577         walker->fetch_ref = fetch_ref;
 578         walker->prefetch = prefetch;
 579         walker->cleanup = cleanup;
 580         walker->data = data;
 581
 582 #ifdef USE_CURL_MULTI
 583         add_fill_function(walker, (int (*)(void *)) fill_active_slot);
 584 #endif
 585
 586         return walker;
 587 }