http-walker.c

   1 #include "git-compat-util.h"
   2 #include "repository.h"
   3 #include "hex.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-ll.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 object_request *obj_req)
  56 {
  57         struct active_request_slot *slot;
  58         struct http_object_request *req;
  59
  60         req = new_http_object_request(obj_req->repo->base, &obj_req->oid);
  61         if (!req) {
  62                 obj_req->state = ABORTED;
  63                 return;
  64         }
  65         obj_req->req = req;
  66
  67         slot = req->slot;
  68         slot->callback_func = process_object_response;
  69         slot->callback_data = obj_req;
  70
  71         /* Try to get the request started, abort the request on error */
  72         obj_req->state = ACTIVE;
  73         if (!start_active_slot(slot)) {
  74                 obj_req->state = ABORTED;
  75                 release_http_object_request(req);
  76                 return;
  77         }
  78 }
  79
  80 static void finish_object_request(struct object_request *obj_req)
  81 {
  82         if (finish_http_object_request(obj_req->req))
  83                 return;
  84
  85         if (obj_req->req->rename == 0)
  86                 walker_say(obj_req->walker, "got %s\n", oid_to_hex(&obj_req->oid));
  87 }
  88
  89 static void process_object_response(void *callback_data)
  90 {
  91         struct object_request *obj_req =
  92                 (struct object_request *)callback_data;
  93         struct walker *walker = obj_req->walker;
  94         struct walker_data *data = walker->data;
  95         struct alt_base *alt = data->alt;
  96
  97         process_http_object_request(obj_req->req);
  98         obj_req->state = COMPLETE;
  99
 100         normalize_curl_result(&obj_req->req->curl_result,
 101                               obj_req->req->http_code,
 102                               obj_req->req->errorstr,
 103                               sizeof(obj_req->req->errorstr));
 104
 105         /* Use alternates if necessary */
 106         if (missing_target(obj_req->req)) {
 107                 fetch_alternates(walker, alt->base);
 108                 if (obj_req->repo->next) {
 109                         obj_req->repo =
 110                                 obj_req->repo->next;
 111                         release_http_object_request(obj_req->req);
 112                         start_object_request(obj_req);
 113                         return;
 114                 }
 115         }
 116
 117         finish_object_request(obj_req);
 118 }
 119
 120 static void release_object_request(struct object_request *obj_req)
 121 {
 122         if (obj_req->req !=NULL && obj_req->req->localfile != -1)
 123                 error("fd leakage in release: %d", obj_req->req->localfile);
 124
 125         list_del(&obj_req->node);
 126         free(obj_req);
 127 }
 128
 129 static int fill_active_slot(void *data UNUSED)
 130 {
 131         struct object_request *obj_req;
 132         struct list_head *pos, *tmp, *head = &object_queue_head;
 133
 134         list_for_each_safe(pos, tmp, head) {
 135                 obj_req = list_entry(pos, struct object_request, node);
 136                 if (obj_req->state == WAITING) {
 137                         if (repo_has_object_file(the_repository, &obj_req->oid))
 138                                 obj_req->state = COMPLETE;
 139                         else {
 140                                 start_object_request(obj_req);
 141                                 return 1;
 142                         }
 143                 }
 144         }
 145         return 0;
 146 }
 147
 148 static void prefetch(struct walker *walker, unsigned char *sha1)
 149 {
 150         struct object_request *newreq;
 151         struct walker_data *data = walker->data;
 152
 153         newreq = xmalloc(sizeof(*newreq));
 154         newreq->walker = walker;
 155         oidread(&newreq->oid, sha1);
 156         newreq->repo = data->alt;
 157         newreq->state = WAITING;
 158         newreq->req = NULL;
 159
 160         http_is_verbose = walker->get_verbosely;
 161         list_add_tail(&newreq->node, &object_queue_head);
 162
 163         fill_active_slots();
 164         step_active_slots();
 165 }
 166
 167 static int is_alternate_allowed(const char *url)
 168 {
 169         const char *protocols[] = {
 170                 "http", "https", "ftp", "ftps"
 171         };
 172         int i;
 173
 174         if (http_follow_config != HTTP_FOLLOW_ALWAYS) {
 175                 warning("alternate disabled by http.followRedirects: %s", url);
 176                 return 0;
 177         }
 178
 179         for (i = 0; i < ARRAY_SIZE(protocols); i++) {
 180                 const char *end;
 181                 if (skip_prefix(url, protocols[i], &end) &&
 182                     starts_with(end, "://"))
 183                         break;
 184         }
 185
 186         if (i >= ARRAY_SIZE(protocols)) {
 187                 warning("ignoring alternate with unknown protocol: %s", url);
 188                 return 0;
 189         }
 190         if (!is_transport_allowed(protocols[i], 0)) {
 191                 warning("ignoring alternate with restricted protocol: %s", url);
 192                 return 0;
 193         }
 194
 195         return 1;
 196 }
 197
 198 static void process_alternates_response(void *callback_data)
 199 {
 200         struct alternates_request *alt_req =
 201                 (struct alternates_request *)callback_data;
 202         struct walker *walker = alt_req->walker;
 203         struct walker_data *cdata = walker->data;
 204         struct active_request_slot *slot = alt_req->slot;
 205         struct alt_base *tail = cdata->alt;
 206         const char *base = alt_req->base;
 207         const char null_byte = '\0';
 208         char *data;
 209         int i = 0;
 210
 211         normalize_curl_result(&slot->curl_result, slot->http_code,
 212                               curl_errorstr, sizeof(curl_errorstr));
 213
 214         if (alt_req->http_specific) {
 215                 if (slot->curl_result != CURLE_OK ||
 216                     !alt_req->buffer->len) {
 217
 218                         /* Try reusing the slot to get non-http alternates */
 219                         alt_req->http_specific = 0;
 220                         strbuf_reset(alt_req->url);
 221                         strbuf_addf(alt_req->url, "%s/objects/info/alternates",
 222                                     base);
 223                         curl_easy_setopt(slot->curl, CURLOPT_URL,
 224                                          alt_req->url->buf);
 225                         active_requests++;
 226                         slot->in_use = 1;
 227                         if (slot->finished)
 228                                 (*slot->finished) = 0;
 229                         if (!start_active_slot(slot)) {
 230                                 cdata->got_alternates = -1;
 231                                 slot->in_use = 0;
 232                                 if (slot->finished)
 233                                         (*slot->finished) = 1;
 234                         }
 235                         return;
 236                 }
 237         } else if (slot->curl_result != CURLE_OK) {
 238                 if (!missing_target(slot)) {
 239                         cdata->got_alternates = -1;
 240                         return;
 241                 }
 242         }
 243
 244         fwrite_buffer((char *)&null_byte, 1, 1, alt_req->buffer);
 245         alt_req->buffer->len--;
 246         data = alt_req->buffer->buf;
 247
 248         while (i < alt_req->buffer->len) {
 249                 int posn = i;
 250                 while (posn < alt_req->buffer->len && data[posn] != '\n')
 251                         posn++;
 252                 if (data[posn] == '\n') {
 253                         int okay = 0;
 254                         int serverlen = 0;
 255                         struct alt_base *newalt;
 256                         if (data[i] == '/') {
 257                                 /*
 258                                  * This counts
 259                                  * http://git.host/pub/scm/linux.git/
 260                                  * -----------here^
 261                                  * so memcpy(dst, base, serverlen) will
 262                                  * copy up to "...git.host".
 263                                  */
 264                                 const char *colon_ss = strstr(base,"://");
 265                                 if (colon_ss) {
 266                                         serverlen = (strchr(colon_ss + 3, '/')
 267                                                      - base);
 268                                         okay = 1;
 269                                 }
 270                         } else if (!memcmp(data + i, "../", 3)) {
 271                                 /*
 272                                  * Relative URL; chop the corresponding
 273                                  * number of subpath from base (and ../
 274                                  * from data), and concatenate the result.
 275                                  *
 276                                  * The code first drops ../ from data, and
 277                                  * then drops one ../ from data and one path
 278                                  * from base.  IOW, one extra ../ is dropped
 279                                  * from data than path is dropped from base.
 280                                  *
 281                                  * This is not wrong.  The alternate in
 282                                  *     http://git.host/pub/scm/linux.git/
 283                                  * to borrow from
 284                                  *     http://git.host/pub/scm/linus.git/
 285                                  * is ../../linus.git/objects/.  You need
 286                                  * two ../../ to borrow from your direct
 287                                  * neighbour.
 288                                  */
 289                                 i += 3;
 290                                 serverlen = strlen(base);
 291                                 while (i + 2 < posn &&
 292                                        !memcmp(data + i, "../", 3)) {
 293                                         do {
 294                                                 serverlen--;
 295                                         } while (serverlen &&
 296                                                  base[serverlen - 1] != '/');
 297                                         i += 3;
 298                                 }
 299                                 /* If the server got removed, give up. */
 300                                 okay = strchr(base, ':') - base + 3 <
 301                                        serverlen;
 302                         } else if (alt_req->http_specific) {
 303                                 char *colon = strchr(data + i, ':');
 304                                 char *slash = strchr(data + i, '/');
 305                                 if (colon && slash && colon < data + posn &&
 306                                     slash < data + posn && colon < slash) {
 307                                         okay = 1;
 308                                 }
 309                         }
 310                         if (okay) {
 311                                 struct strbuf target = STRBUF_INIT;
 312                                 strbuf_add(&target, base, serverlen);
 313                                 strbuf_add(&target, data + i, posn - i);
 314                                 if (!strbuf_strip_suffix(&target, "objects")) {
 315                                         warning("ignoring alternate that does"
 316                                                 " not end in 'objects': %s",
 317                                                 target.buf);
 318                                         strbuf_release(&target);
 319                                 } else if (is_alternate_allowed(target.buf)) {
 320                                         warning("adding alternate object store: %s",
 321                                                 target.buf);
 322                                         newalt = xmalloc(sizeof(*newalt));
 323                                         newalt->next = NULL;
 324                                         newalt->base = strbuf_detach(&target, NULL);
 325                                         newalt->got_indices = 0;
 326                                         newalt->packs = NULL;
 327
 328                                         while (tail->next != NULL)
 329                                                 tail = tail->next;
 330                                         tail->next = newalt;
 331                                 } else {
 332                                         strbuf_release(&target);
 333                                 }
 334                         }
 335                 }
 336                 i = posn + 1;
 337         }
 338
 339         cdata->got_alternates = 1;
 340 }
 341
 342 static void fetch_alternates(struct walker *walker, const char *base)
 343 {
 344         struct strbuf buffer = STRBUF_INIT;
 345         struct strbuf url = STRBUF_INIT;
 346         struct active_request_slot *slot;
 347         struct alternates_request alt_req;
 348         struct walker_data *cdata = walker->data;
 349
 350         /*
 351          * If another request has already started fetching alternates,
 352          * wait for them to arrive and return to processing this request's
 353          * curl message
 354          */
 355         while (cdata->got_alternates == 0) {
 356                 step_active_slots();
 357         }
 358
 359         /* Nothing to do if they've already been fetched */
 360         if (cdata->got_alternates == 1)
 361                 return;
 362
 363         /* Start the fetch */
 364         cdata->got_alternates = 0;
 365
 366         if (walker->get_verbosely)
 367                 fprintf(stderr, "Getting alternates list for %s\n", base);
 368
 369         strbuf_addf(&url, "%s/objects/info/http-alternates", base);
 370
 371         /*
 372          * Use a callback to process the result, since another request
 373          * may fail and need to have alternates loaded before continuing
 374          */
 375         slot = get_active_slot();
 376         slot->callback_func = process_alternates_response;
 377         alt_req.walker = walker;
 378         slot->callback_data = &alt_req;
 379
 380         curl_easy_setopt(slot->curl, CURLOPT_WRITEDATA, &buffer);
 381         curl_easy_setopt(slot->curl, CURLOPT_WRITEFUNCTION, fwrite_buffer);
 382         curl_easy_setopt(slot->curl, CURLOPT_URL, url.buf);
 383
 384         alt_req.base = base;
 385         alt_req.url = &url;
 386         alt_req.buffer = &buffer;
 387         alt_req.http_specific = 1;
 388         alt_req.slot = slot;
 389
 390         if (start_active_slot(slot))
 391                 run_active_slot(slot);
 392         else
 393                 cdata->got_alternates = -1;
 394
 395         strbuf_release(&buffer);
 396         strbuf_release(&url);
 397 }
 398
 399 static int fetch_indices(struct walker *walker, struct alt_base *repo)
 400 {
 401         int ret;
 402
 403         if (repo->got_indices)
 404                 return 0;
 405
 406         if (walker->get_verbosely)
 407                 fprintf(stderr, "Getting pack list for %s\n", repo->base);
 408
 409         switch (http_get_info_packs(repo->base, &repo->packs)) {
 410         case HTTP_OK:
 411         case HTTP_MISSING_TARGET:
 412                 repo->got_indices = 1;
 413                 ret = 0;
 414                 break;
 415         default:
 416                 repo->got_indices = 0;
 417                 ret = -1;
 418         }
 419
 420         return ret;
 421 }
 422
 423 static int http_fetch_pack(struct walker *walker, struct alt_base *repo, unsigned char *sha1)
 424 {
 425         struct packed_git *target;
 426         int ret;
 427         struct slot_results results;
 428         struct http_pack_request *preq;
 429
 430         if (fetch_indices(walker, repo))
 431                 return -1;
 432         target = find_sha1_pack(sha1, repo->packs);
 433         if (!target)
 434                 return -1;
 435         close_pack_index(target);
 436
 437         if (walker->get_verbosely) {
 438                 fprintf(stderr, "Getting pack %s\n",
 439                         hash_to_hex(target->hash));
 440                 fprintf(stderr, " which contains %s\n",
 441                         hash_to_hex(sha1));
 442         }
 443
 444         preq = new_http_pack_request(target->hash, repo->base);
 445         if (!preq)
 446                 goto abort;
 447         preq->slot->results = &results;
 448
 449         if (start_active_slot(preq->slot)) {
 450                 run_active_slot(preq->slot);
 451                 if (results.curl_result != CURLE_OK) {
 452                         error("Unable to get pack file %s\n%s", preq->url,
 453                               curl_errorstr);
 454                         goto abort;
 455                 }
 456         } else {
 457                 error("Unable to start request");
 458                 goto abort;
 459         }
 460
 461         ret = finish_http_pack_request(preq);
 462         release_http_pack_request(preq);
 463         if (ret)
 464                 return ret;
 465         http_install_packfile(target, &repo->packs);
 466
 467         return 0;
 468
 469 abort:
 470         return -1;
 471 }
 472
 473 static void abort_object_request(struct object_request *obj_req)
 474 {
 475         release_object_request(obj_req);
 476 }
 477
 478 static int fetch_object(struct walker *walker, unsigned char *hash)
 479 {
 480         char *hex = hash_to_hex(hash);
 481         int ret = 0;
 482         struct object_request *obj_req = NULL;
 483         struct http_object_request *req;
 484         struct list_head *pos, *head = &object_queue_head;
 485
 486         list_for_each(pos, head) {
 487                 obj_req = list_entry(pos, struct object_request, node);
 488                 if (hasheq(obj_req->oid.hash, hash))
 489                         break;
 490         }
 491         if (!obj_req)
 492                 return error("Couldn't find request for %s in the queue", hex);
 493
 494         if (repo_has_object_file(the_repository, &obj_req->oid)) {
 495                 if (obj_req->req)
 496                         abort_http_object_request(obj_req->req);
 497                 abort_object_request(obj_req);
 498                 return 0;
 499         }
 500
 501         while (obj_req->state == WAITING)
 502                 step_active_slots();
 503
 504         /*
 505          * obj_req->req might change when fetching alternates in the callback
 506          * process_object_response; therefore, the "shortcut" variable, req,
 507          * is used only after we're done with slots.
 508          */
 509         while (obj_req->state == ACTIVE)
 510                 run_active_slot(obj_req->req->slot);
 511
 512         req = obj_req->req;
 513
 514         if (req->localfile != -1) {
 515                 close(req->localfile);
 516                 req->localfile = -1;
 517         }
 518
 519         normalize_curl_result(&req->curl_result, req->http_code,
 520                               req->errorstr, sizeof(req->errorstr));
 521
 522         if (obj_req->state == ABORTED) {
 523                 ret = error("Request for %s aborted", hex);
 524         } else if (req->curl_result != CURLE_OK &&
 525                    req->http_code != 416) {
 526                 if (missing_target(req))
 527                         ret = -1; /* Be silent, it is probably in a pack. */
 528                 else
 529                         ret = error("%s (curl_result = %d, http_code = %ld, sha1 = %s)",
 530                                     req->errorstr, req->curl_result,
 531                                     req->http_code, hex);
 532         } else if (req->zret != Z_STREAM_END) {
 533                 walker->corrupt_object_found++;
 534                 ret = error("File %s (%s) corrupt", hex, req->url);
 535         } else if (!oideq(&obj_req->oid, &req->real_oid)) {
 536                 ret = error("File %s has bad hash", hex);
 537         } else if (req->rename < 0) {
 538                 struct strbuf buf = STRBUF_INIT;
 539                 loose_object_path(the_repository, &buf, &req->oid);
 540                 ret = error("unable to write sha1 filename %s", buf.buf);
 541                 strbuf_release(&buf);
 542         }
 543
 544         release_http_object_request(req);
 545         release_object_request(obj_req);
 546         return ret;
 547 }
 548
 549 static int fetch(struct walker *walker, unsigned char *hash)
 550 {
 551         struct walker_data *data = walker->data;
 552         struct alt_base *altbase = data->alt;
 553
 554         if (!fetch_object(walker, hash))
 555                 return 0;
 556         while (altbase) {
 557                 if (!http_fetch_pack(walker, altbase, hash))
 558                         return 0;
 559                 fetch_alternates(walker, data->alt->base);
 560                 altbase = altbase->next;
 561         }
 562         return error("Unable to find %s under %s", hash_to_hex(hash),
 563                      data->alt->base);
 564 }
 565
 566 static int fetch_ref(struct walker *walker, struct ref *ref)
 567 {
 568         struct walker_data *data = walker->data;
 569         return http_fetch_ref(data->alt->base, ref);
 570 }
 571
 572 static void cleanup(struct walker *walker)
 573 {
 574         struct walker_data *data = walker->data;
 575         struct alt_base *alt, *alt_next;
 576
 577         if (data) {
 578                 alt = data->alt;
 579                 while (alt) {
 580                         alt_next = alt->next;
 581
 582                         free(alt->base);
 583                         free(alt);
 584
 585                         alt = alt_next;
 586                 }
 587                 free(data);
 588                 walker->data = NULL;
 589         }
 590 }
 591
 592 struct walker *get_http_walker(const char *url)
 593 {
 594         char *s;
 595         struct walker_data *data = xmalloc(sizeof(struct walker_data));
 596         struct walker *walker = xmalloc(sizeof(struct walker));
 597
 598         data->alt = xmalloc(sizeof(*data->alt));
 599         data->alt->base = xstrdup(url);
 600         for (s = data->alt->base + strlen(data->alt->base) - 1; *s == '/'; --s)
 601                 *s = 0;
 602
 603         data->alt->got_indices = 0;
 604         data->alt->packs = NULL;
 605         data->alt->next = NULL;
 606         data->got_alternates = -1;
 607
 608         walker->corrupt_object_found = 0;
 609         walker->fetch = fetch;
 610         walker->fetch_ref = fetch_ref;
 611         walker->prefetch = prefetch;
 612         walker->cleanup = cleanup;
 613         walker->data = data;
 614
 615         add_fill_function(NULL, fill_active_slot);
 616
 617         return walker;
 618 }