lib/fetch.c

   1 /*
   2  * Copyright (c) 2018, 2019 Ori Bernstein <ori@openbsd.org>
   3  *
   4  * Permission to use, copy, modify, and distribute this software for any
   5  * purpose with or without fee is hereby granted, provided that the above
   6  * copyright notice and this permission notice appear in all copies.
   7  *
   8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
   9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  15  */
  16 #include "got_compat.h"
  17
  18 #include <sys/types.h>
  19 #include <sys/stat.h>
  20 #include <sys/queue.h>
  21 #include <sys/uio.h>
  22 #include <sys/socket.h>
  23 #include <sys/wait.h>
  24 #include <sys/resource.h>
  25 #include <sys/socket.h>
  26
  27 #include <errno.h>
  28 #include <err.h>
  29 #include <fcntl.h>
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <string.h>
  33 #include <stdint.h>
  34 #include <unistd.h>
  35 #include <zlib.h>
  36 #include <ctype.h>
  37 #include <limits.h>
  38 #include <time.h>
  39
  40 #include "got_error.h"
  41 #include "got_reference.h"
  42 #include "got_repository.h"
  43 #include "got_path.h"
  44 #include "got_cancel.h"
  45 #include "got_worktree.h"
  46 #include "got_object.h"
  47 #include "got_opentemp.h"
  48 #include "got_fetch.h"
  49
  50 #include "got_lib_hash.h"
  51 #include "got_lib_delta.h"
  52 #include "got_lib_inflate.h"
  53 #include "got_lib_object.h"
  54 #include "got_lib_object_parse.h"
  55 #include "got_lib_object_create.h"
  56 #include "got_lib_pack.h"
  57 #include "got_lib_privsep.h"
  58 #include "got_lib_object_cache.h"
  59 #include "got_lib_repository.h"
  60 #include "got_lib_dial.h"
  61 #include "got_lib_pkt.h"
  62
  63 #ifndef nitems
  64 #define nitems(_a)      (sizeof((_a)) / sizeof((_a)[0]))
  65 #endif
  66
  67 #ifndef ssizeof
  68 #define ssizeof(_x) ((ssize_t)(sizeof(_x)))
  69 #endif
  70
  71 #ifndef MIN
  72 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
  73 #endif
  74
  75 const struct got_error *
  76 got_fetch_connect(pid_t *fetchpid, int *fetchfd, const char *proto,
  77     const char *host, const char *port, const char *server_path,
  78     const char *jumphost, const char *identity_file, int verbosity)
  79 {
  80         const struct got_error *err = NULL;
  81
  82         *fetchpid = -1;
  83         *fetchfd = -1;
  84
  85         if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
  86                 err = got_dial_ssh(fetchpid, fetchfd, host, port,
  87                     server_path, jumphost, identity_file, GOT_DIAL_CMD_FETCH,
  88                     verbosity);
  89         else if (strcmp(proto, "git") == 0)
  90                 err = got_dial_git(fetchfd, host, port, server_path,
  91                     GOT_DIAL_CMD_FETCH);
  92         else if (strcmp(proto, "http") == 0 ||
  93             strcmp(proto, "git+http") == 0 ||
  94             strcmp(proto, "https") == 0 ||
  95             strcmp(proto, "git+https") == 0)
  96                 err = got_dial_http(fetchpid, fetchfd, host, port,
  97                     server_path, verbosity, strstr(proto, "https") != NULL);
  98         else
  99                 err = got_error_path(proto, GOT_ERR_BAD_PROTO);
 100         return err;
 101 }
 102
 103 const struct got_error *
 104 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
 105     struct got_pathlist_head *symrefs, const char *remote_name,
 106     int mirror_references, int fetch_all_branches,
 107     struct got_pathlist_head *wanted_branches,
 108     struct got_pathlist_head *wanted_refs, int list_refs_only, int verbosity,
 109     int fetchfd, struct got_repository *repo, const char *worktree_refname,
 110     const char *remote_head, int no_head, got_fetch_progress_cb progress_cb,
 111     void *progress_arg)
 112 {
 113         size_t i;
 114         int imsg_fetchfds[2], imsg_idxfds[2];
 115         int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
 116         int tmpfds[3];
 117         int fetchstatus, idxstatus, done = 0;
 118         const struct got_error *err;
 119         struct imsgbuf fetchibuf, idxibuf;
 120         pid_t fetchpid, idxpid;
 121         char *tmppackpath = NULL, *tmpidxpath = NULL;
 122         char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
 123         const char *repo_path = NULL;
 124         struct got_pathlist_head have_refs;
 125         struct got_pathlist_entry *pe, *new;
 126         struct got_reflist_head my_refs;
 127         struct got_reflist_entry *re;
 128         off_t packfile_size = 0;
 129         struct got_packfile_hdr pack_hdr;
 130         uint32_t nobj = 0;
 131         char *path;
 132         char *progress = NULL;
 133
 134         *pack_hash = NULL;
 135         memset(&fetchibuf, 0, sizeof(fetchibuf));
 136         memset(&idxibuf, 0, sizeof(idxibuf));
 137
 138         if (repo && got_repo_get_object_format(repo) != GOT_HASH_SHA1)
 139                 return got_error_fmt(GOT_ERR_NOT_IMPL,
 140                     "sha256 object IDs unsupported in network protocol");
 141
 142         /*
 143          * Prevent fetching of references that won't make any
 144          * sense outside of the remote repository's context.
 145          */
 146         RB_FOREACH(pe, got_pathlist_head, wanted_refs) {
 147                 const char *refname = pe->path;
 148                 if (strncmp(refname, "refs/got/", 9) == 0 ||
 149                     strncmp(refname, "got/", 4) == 0 ||
 150                     strncmp(refname, "refs/remotes/", 13) == 0 ||
 151                     strncmp(refname, "remotes/", 8) == 0)
 152                         return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
 153         }
 154
 155         if (!list_refs_only)
 156                 repo_path = got_repo_get_path_git_dir(repo);
 157
 158         for (i = 0; i < nitems(tmpfds); i++)
 159                 tmpfds[i] = -1;
 160
 161         RB_INIT(&have_refs);
 162         TAILQ_INIT(&my_refs);
 163
 164         if (!list_refs_only) {
 165                 err = got_ref_list(&my_refs, repo, NULL,
 166                     got_ref_cmp_by_name, NULL);
 167                 if (err)
 168                         goto done;
 169         }
 170
 171         TAILQ_FOREACH(re, &my_refs, entry) {
 172                 struct got_object_id *id;
 173                 char *refname;
 174
 175                 if (got_ref_is_symbolic(re->ref))
 176                         continue;
 177
 178                 err = got_ref_resolve(&id, repo, re->ref);
 179                 if (err)
 180                         goto done;
 181                 refname = strdup(got_ref_get_name(re->ref));
 182                 if (refname == NULL) {
 183                         err = got_error_from_errno("strdup");
 184                         goto done;
 185                 }
 186                 err = got_pathlist_insert(&new, &have_refs, refname, id);
 187                 if (err)
 188                         goto done;
 189                 if (new == NULL){
 190                         free(refname);
 191                         free(id);
 192                 }
 193
 194         }
 195
 196         if (list_refs_only) {
 197                 packfd = got_opentempfd();
 198                 if (packfd == -1) {
 199                         err = got_error_from_errno("got_opentempfd");
 200                         goto done;
 201                 }
 202         } else {
 203                 if (asprintf(&path, "%s/%s/fetching.pack",
 204                     repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
 205                         err = got_error_from_errno("asprintf");
 206                         goto done;
 207                 }
 208                 err = got_opentemp_named_fd(&tmppackpath, &packfd, path, "");
 209                 free(path);
 210                 if (err)
 211                         goto done;
 212                 if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
 213                         err = got_error_from_errno2("fchmod", tmppackpath);
 214                         goto done;
 215                 }
 216         }
 217         if (list_refs_only) {
 218                 idxfd = got_opentempfd();
 219                 if (idxfd == -1) {
 220                         err = got_error_from_errno("got_opentempfd");
 221                         goto done;
 222                 }
 223         } else {
 224                 if (asprintf(&path, "%s/%s/fetching.idx",
 225                     repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
 226                         err = got_error_from_errno("asprintf");
 227                         goto done;
 228                 }
 229                 err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
 230                 free(path);
 231                 if (err)
 232                         goto done;
 233                 if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
 234                         err = got_error_from_errno2("fchmod", tmpidxpath);
 235                         goto done;
 236                 }
 237         }
 238         nidxfd = dup(idxfd);
 239         if (nidxfd == -1) {
 240                 err = got_error_from_errno("dup");
 241                 goto done;
 242         }
 243
 244         for (i = 0; i < nitems(tmpfds); i++) {
 245                 tmpfds[i] = got_opentempfd();
 246                 if (tmpfds[i] == -1) {
 247                         err = got_error_from_errno("got_opentempfd");
 248                         goto done;
 249                 }
 250         }
 251
 252         if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
 253                 err = got_error_from_errno("socketpair");
 254                 goto done;
 255         }
 256
 257         fetchpid = fork();
 258         if (fetchpid == -1) {
 259                 err = got_error_from_errno("fork");
 260                 goto done;
 261         } else if (fetchpid == 0){
 262                 got_privsep_exec_child(imsg_fetchfds,
 263                     GOT_PATH_PROG_FETCH_PACK, tmppackpath);
 264         }
 265
 266         if (close(imsg_fetchfds[1]) == -1) {
 267                 err = got_error_from_errno("close");
 268                 goto done;
 269         }
 270         if (imsgbuf_init(&fetchibuf, imsg_fetchfds[0]) == -1) {
 271                 err = got_error_from_errno("imsgbuf_init");
 272                 goto done;
 273         }
 274         imsgbuf_allow_fdpass(&fetchibuf);
 275         nfetchfd = dup(fetchfd);
 276         if (nfetchfd == -1) {
 277                 err = got_error_from_errno("dup");
 278                 goto done;
 279         }
 280         err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
 281             fetch_all_branches, wanted_branches, wanted_refs,
 282             list_refs_only, worktree_refname, remote_head, no_head, verbosity);
 283         if (err != NULL)
 284                 goto done;
 285         nfetchfd = -1;
 286         npackfd = dup(packfd);
 287         if (npackfd == -1) {
 288                 err = got_error_from_errno("dup");
 289                 goto done;
 290         }
 291         err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
 292         if (err != NULL)
 293                 goto done;
 294         npackfd = -1;
 295
 296         packfile_size = 0;
 297         progress = calloc(GOT_PKT_MAX, 1);
 298         if (progress == NULL) {
 299                 err = got_error_from_errno("calloc");
 300                 goto done;
 301         }
 302
 303         *pack_hash = calloc(1, sizeof(**pack_hash));
 304         if (*pack_hash == NULL) {
 305                 err = got_error_from_errno("calloc");
 306                 goto done;
 307         }
 308
 309         while (!done) {
 310                 struct got_object_id *id = NULL;
 311                 char *refname = NULL;
 312                 char *server_progress = NULL;
 313                 off_t packfile_size_cur = 0;
 314
 315                 err = got_privsep_recv_fetch_progress(&done,
 316                     &id, &refname, symrefs, &server_progress,
 317                     &packfile_size_cur, (*pack_hash)->hash, &fetchibuf);
 318                 if (err != NULL)
 319                         goto done;
 320                 /* Don't report size progress for an empty pack file. */
 321                 if (packfile_size_cur <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
 322                         packfile_size_cur = 0;
 323                 if (!done && refname && id) {
 324                         err = got_pathlist_insert(&pe, refs, refname, id);
 325                         if (err || pe == NULL) {
 326                                 free(id);
 327                                 free(refname);
 328                                 if (err != NULL)
 329                                         goto done;
 330                         }
 331                 } else if (!done && server_progress) {
 332                         char *p;
 333                         /*
 334                          * XXX git-daemon tends to send batched output with
 335                          * lines spanning separate packets. Buffer progress
 336                          * output until we see a CR or LF to avoid giving
 337                          * partial lines of progress output to the callback.
 338                          */
 339                         if (strlcat(progress, server_progress,
 340                             GOT_PKT_MAX) >= GOT_PKT_MAX) {
 341                                 progress[0] = '\0'; /* discard */
 342                                 continue;
 343                         }
 344                         while ((p = strchr(progress, '\r')) != NULL ||
 345                             (p = strchr(progress, '\n')) != NULL) {
 346                                 char *s;
 347                                 size_t n;
 348                                 char c = *p;
 349                                 *p = '\0';
 350                                 if (asprintf(&s, "%s%s", progress,
 351                                     c == '\n' ? "\n" : "") == -1) {
 352                                         err = got_error_from_errno("asprintf");
 353                                         goto done;
 354                                 }
 355                                 err = progress_cb(progress_arg, s,
 356                                     packfile_size_cur, 0, 0, 0, 0);
 357                                 free(s);
 358                                 if (err)
 359                                         break;
 360                                 n = strlen(progress);
 361                                 if (n < GOT_PKT_MAX - 1) {
 362                                         memmove(progress, &progress[n + 1],
 363                                             GOT_PKT_MAX - n - 1);
 364                                 } else
 365                                         progress[0] = '\0';
 366                         }
 367                         free(server_progress);
 368                         if (err)
 369                                 goto done;
 370                 } else if (!done && packfile_size_cur != packfile_size) {
 371                         err = progress_cb(progress_arg, NULL,
 372                             packfile_size_cur, 0, 0, 0, 0);
 373                         if (err)
 374                                 break;
 375                         packfile_size = packfile_size_cur;
 376                 }
 377         }
 378         if (close(imsg_fetchfds[0]) == -1) {
 379                 err = got_error_from_errno("close");
 380                 goto done;
 381         }
 382         if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
 383                 err = got_error_from_errno("waitpid");
 384                 goto done;
 385         }
 386
 387         if (lseek(packfd, 0, SEEK_SET) == -1) {
 388                 err = got_error_from_errno("lseek");
 389                 goto done;
 390         }
 391
 392         /* If zero data was fetched without error we are already up-to-date. */
 393         if (packfile_size == 0) {
 394                 free(*pack_hash);
 395                 *pack_hash = NULL;
 396                 goto done;
 397         } else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
 398                 err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
 399                 goto done;
 400         } else {
 401                 ssize_t n;
 402
 403                 n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
 404                 if (n == -1) {
 405                         err = got_error_from_errno("read");
 406                         goto done;
 407                 }
 408                 if (n != ssizeof(pack_hdr)) {
 409                         err = got_error(GOT_ERR_IO);
 410                         goto done;
 411                 }
 412                 if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
 413                         err = got_error_msg(GOT_ERR_BAD_PACKFILE,
 414                             "bad pack file signature");
 415                         goto done;
 416                 }
 417                 if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
 418                         err = got_error_msg(GOT_ERR_BAD_PACKFILE,
 419                             "bad pack file version");
 420                         goto done;
 421                 }
 422                 nobj = be32toh(pack_hdr.nobjects);
 423                 if (nobj == 0 &&
 424                     packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
 425                         err = got_error_msg(GOT_ERR_BAD_PACKFILE,
 426                             "bad pack file with zero objects");
 427                         goto done;
 428                 }
 429                 if (nobj != 0 &&
 430                     packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
 431                         err = got_error_msg(GOT_ERR_BAD_PACKFILE,
 432                             "empty pack file with non-zero object count");
 433                         goto done;
 434                 }
 435         }
 436
 437         /*
 438          * If the pack file contains no objects, we may only need to update
 439          * references in our repository. The caller will take care of that.
 440          */
 441         if (nobj == 0) {
 442                 free(*pack_hash);
 443                 *pack_hash = NULL;
 444                 goto done;
 445         }
 446
 447         if (lseek(packfd, 0, SEEK_SET) == -1) {
 448                 err = got_error_from_errno("lseek");
 449                 goto done;
 450         }
 451
 452         if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
 453                 err = got_error_from_errno("socketpair");
 454                 goto done;
 455         }
 456         idxpid = fork();
 457         if (idxpid == -1) {
 458                 err= got_error_from_errno("fork");
 459                 goto done;
 460         } else if (idxpid == 0)
 461                 got_privsep_exec_child(imsg_idxfds,
 462                     GOT_PATH_PROG_INDEX_PACK, tmppackpath);
 463         if (close(imsg_idxfds[1]) == -1) {
 464                 err = got_error_from_errno("close");
 465                 goto done;
 466         }
 467         if (imsgbuf_init(&idxibuf, imsg_idxfds[0]) == -1) {
 468                 err = got_error_from_errno("imsgbuf_init");
 469                 goto done;
 470         }
 471         imsgbuf_allow_fdpass(&idxibuf);
 472
 473         npackfd = dup(packfd);
 474         if (npackfd == -1) {
 475                 err = got_error_from_errno("dup");
 476                 goto done;
 477         }
 478         err = got_privsep_send_index_pack_req(&idxibuf, *pack_hash, npackfd);
 479         if (err != NULL)
 480                 goto done;
 481         npackfd = -1;
 482         err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
 483         if (err != NULL)
 484                 goto done;
 485         nidxfd = -1;
 486         for (i = 0; i < nitems(tmpfds); i++) {
 487                 err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
 488                 if (err != NULL)
 489                         goto done;
 490                 tmpfds[i] = -1;
 491         }
 492         done = 0;
 493         while (!done) {
 494                 int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
 495
 496                 err = got_privsep_recv_index_progress(&done, &nobj_total,
 497                     &nobj_indexed, &nobj_loose, &nobj_resolved,
 498                     &idxibuf);
 499                 if (err != NULL)
 500                         goto done;
 501                 if (nobj_indexed != 0) {
 502                         err = progress_cb(progress_arg, NULL,
 503                             packfile_size, nobj_total,
 504                             nobj_indexed, nobj_loose, nobj_resolved);
 505                         if (err)
 506                                 break;
 507                 }
 508         }
 509         if (close(imsg_idxfds[0]) == -1) {
 510                 err = got_error_from_errno("close");
 511                 goto done;
 512         }
 513         if (waitpid(idxpid, &idxstatus, 0) == -1) {
 514                 err = got_error_from_errno("waitpid");
 515                 goto done;
 516         }
 517
 518         err = got_object_id_str(&id_str, *pack_hash);
 519         if (err)
 520                 goto done;
 521         if (asprintf(&packpath, "%s/%s/pack-%s.pack",
 522             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 523                 err = got_error_from_errno("asprintf");
 524                 goto done;
 525         }
 526
 527         if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
 528             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 529                 err = got_error_from_errno("asprintf");
 530                 goto done;
 531         }
 532         free(id_str);
 533         id_str = NULL;
 534
 535         if (rename(tmppackpath, packpath) == -1) {
 536                 err = got_error_from_errno3("rename", tmppackpath, packpath);
 537                 goto done;
 538         }
 539         free(tmppackpath);
 540         tmppackpath = NULL;
 541         if (rename(tmpidxpath, idxpath) == -1) {
 542                 err = got_error_from_errno3("rename", tmpidxpath, idxpath);
 543                 goto done;
 544         }
 545         free(tmpidxpath);
 546         tmpidxpath = NULL;
 547
 548 done:
 549         if (fetchibuf.w)
 550                 imsgbuf_clear(&fetchibuf);
 551         if (idxibuf.w)
 552                 imsgbuf_clear(&idxibuf);
 553         if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
 554                 err = got_error_from_errno2("unlink", tmppackpath);
 555         if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
 556                 err = got_error_from_errno2("unlink", tmpidxpath);
 557         if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
 558                 err = got_error_from_errno("close");
 559         if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
 560                 err = got_error_from_errno("close");
 561         if (packfd != -1 && close(packfd) == -1 && err == NULL)
 562                 err = got_error_from_errno("close");
 563         if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
 564                 err = got_error_from_errno("close");
 565         for (i = 0; i < nitems(tmpfds); i++) {
 566                 if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
 567                         err = got_error_from_errno("close");
 568         }
 569         free(tmppackpath);
 570         free(tmpidxpath);
 571         free(idxpath);
 572         free(id_str);
 573         free(packpath);
 574         free(progress);
 575
 576         got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
 577         got_ref_list_free(&my_refs);
 578         if (err) {
 579                 free(*pack_hash);
 580                 *pack_hash = NULL;
 581                 got_pathlist_free(refs, GOT_PATHLIST_FREE_ALL);
 582                 got_pathlist_free(symrefs, GOT_PATHLIST_FREE_ALL);
 583         }
 584         return err;
 585 }