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
  17 #include <sys/types.h>
  18 #include <sys/stat.h>
  19 #include <sys/queue.h>
  20 #include <sys/uio.h>
  21 #include <sys/socket.h>
  22 #include <sys/wait.h>
  23 #include <sys/resource.h>
  24 #include <sys/socket.h>
  25
  26 #include <errno.h>
  27 #include <err.h>
  28 #include <fcntl.h>
  29 #include <stdio.h>
  30 #include <stdlib.h>
  31 #include <string.h>
  32 #include <stdint.h>
  33 #include <unistd.h>
  34 #include <zlib.h>
  35 #include <ctype.h>
  36 #include <limits.h>
  37 #include <time.h>
  38
  39 #include "got_error.h"
  40 #include "got_reference.h"
  41 #include "got_repository.h"
  42 #include "got_path.h"
  43 #include "got_cancel.h"
  44 #include "got_worktree.h"
  45 #include "got_object.h"
  46 #include "got_opentemp.h"
  47 #include "got_fetch.h"
  48
  49 #include "got_lib_delta.h"
  50 #include "got_lib_inflate.h"
  51 #include "got_lib_object.h"
  52 #include "got_lib_object_parse.h"
  53 #include "got_lib_object_create.h"
  54 #include "got_lib_pack.h"
  55 #include "got_lib_sha1.h"
  56 #include "got_lib_privsep.h"
  57 #include "got_lib_object_cache.h"
  58 #include "got_lib_repository.h"
  59 #include "got_lib_dial.h"
  60 #include "got_lib_pkt.h"
  61
  62 #ifndef nitems
  63 #define nitems(_a)      (sizeof((_a)) / sizeof((_a)[0]))
  64 #endif
  65
  66 #ifndef ssizeof
  67 #define ssizeof(_x) ((ssize_t)(sizeof(_x)))
  68 #endif
  69
  70 #ifndef MIN
  71 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
  72 #endif
  73
  74 const struct got_error *
  75 got_fetch_connect(pid_t *fetchpid, int *fetchfd, const char *proto,
  76     const char *host, const char *port, const char *server_path, int verbosity)
  77 {
  78         const struct got_error *err = NULL;
  79
  80         *fetchpid = -1;
  81         *fetchfd = -1;
  82
  83         if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
  84                 err = got_dial_ssh(fetchpid, fetchfd, host, port,
  85                     server_path, GOT_DIAL_DIRECTION_FETCH, verbosity);
  86         else if (strcmp(proto, "git") == 0)
  87                 err = got_dial_git(fetchfd, host, port, server_path,
  88                     GOT_DIAL_DIRECTION_FETCH);
  89         else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
  90                 err = got_error_path(proto, GOT_ERR_NOT_IMPL);
  91         else
  92                 err = got_error_path(proto, GOT_ERR_BAD_PROTO);
  93         return err;
  94 }
  95
  96 const struct got_error *
  97 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
  98     struct got_pathlist_head *symrefs, const char *remote_name,
  99     int mirror_references, int fetch_all_branches,
 100     struct got_pathlist_head *wanted_branches,
 101     struct got_pathlist_head *wanted_refs, int list_refs_only, int verbosity,
 102     int fetchfd, struct got_repository *repo, const char *worktree_refname,
 103     got_fetch_progress_cb progress_cb, void *progress_arg)
 104 {
 105         size_t i;
 106         int imsg_fetchfds[2], imsg_idxfds[2];
 107         int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
 108         int tmpfds[3];
 109         int fetchstatus, idxstatus, done = 0;
 110         const struct got_error *err;
 111         struct imsgbuf fetchibuf, idxibuf;
 112         pid_t fetchpid, idxpid;
 113         char *tmppackpath = NULL, *tmpidxpath = NULL;
 114         char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
 115         const char *repo_path = NULL;
 116         struct got_pathlist_head have_refs;
 117         struct got_pathlist_entry *pe;
 118         struct got_reflist_head my_refs;
 119         struct got_reflist_entry *re;
 120         off_t packfile_size = 0;
 121         struct got_packfile_hdr pack_hdr;
 122         uint32_t nobj = 0;
 123         char *path;
 124         char *progress = NULL;
 125
 126         *pack_hash = NULL;
 127
 128         /*
 129          * Prevent fetching of references that won't make any
 130          * sense outside of the remote repository's context.
 131          */
 132         TAILQ_FOREACH(pe, wanted_refs, entry) {
 133                 const char *refname = pe->path;
 134                 if (strncmp(refname, "refs/got/", 9) == 0 ||
 135                     strncmp(refname, "got/", 4) == 0 ||
 136                     strncmp(refname, "refs/remotes/", 13) == 0 ||
 137                     strncmp(refname, "remotes/", 8) == 0)
 138                         return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
 139         }
 140
 141         if (!list_refs_only)
 142                 repo_path = got_repo_get_path_git_dir(repo);
 143
 144         for (i = 0; i < nitems(tmpfds); i++)
 145                 tmpfds[i] = -1;
 146
 147         TAILQ_INIT(&have_refs);
 148         TAILQ_INIT(&my_refs);
 149
 150         if (!list_refs_only) {
 151                 err = got_ref_list(&my_refs, repo, NULL,
 152                     got_ref_cmp_by_name, NULL);
 153                 if (err)
 154                         goto done;
 155         }
 156
 157         TAILQ_FOREACH(re, &my_refs, entry) {
 158                 struct got_object_id *id;
 159                 const char *refname;
 160
 161                 if (got_ref_is_symbolic(re->ref))
 162                         continue;
 163
 164                 err = got_ref_resolve(&id, repo, re->ref);
 165                 if (err)
 166                         goto done;
 167                 refname = strdup(got_ref_get_name(re->ref));
 168                 if (refname == NULL) {
 169                         err = got_error_from_errno("strdup");
 170                         goto done;
 171                 }
 172                 err = got_pathlist_append(&have_refs, refname, id);
 173                 if (err)
 174                         goto done;
 175         }
 176
 177         if (list_refs_only) {
 178                 packfd = got_opentempfd();
 179                 if (packfd == -1) {
 180                         err = got_error_from_errno("got_opentempfd");
 181                         goto done;
 182                 }
 183         } else {
 184                 if (asprintf(&path, "%s/%s/fetching.pack",
 185                     repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
 186                         err = got_error_from_errno("asprintf");
 187                         goto done;
 188                 }
 189                 err = got_opentemp_named_fd(&tmppackpath, &packfd, path, "");
 190                 free(path);
 191                 if (err)
 192                         goto done;
 193                 if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
 194                         err = got_error_from_errno2("fchmod", tmppackpath);
 195                         goto done;
 196                 }
 197         }
 198         if (list_refs_only) {
 199                 idxfd = got_opentempfd();
 200                 if (idxfd == -1) {
 201                         err = got_error_from_errno("got_opentempfd");
 202                         goto done;
 203                 }
 204         } else {
 205                 if (asprintf(&path, "%s/%s/fetching.idx",
 206                     repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
 207                         err = got_error_from_errno("asprintf");
 208                         goto done;
 209                 }
 210                 err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
 211                 free(path);
 212                 if (err)
 213                         goto done;
 214                 if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
 215                         err = got_error_from_errno2("fchmod", tmpidxpath);
 216                         goto done;
 217                 }
 218         }
 219         nidxfd = dup(idxfd);
 220         if (nidxfd == -1) {
 221                 err = got_error_from_errno("dup");
 222                 goto done;
 223         }
 224
 225         for (i = 0; i < nitems(tmpfds); i++) {
 226                 tmpfds[i] = got_opentempfd();
 227                 if (tmpfds[i] == -1) {
 228                         err = got_error_from_errno("got_opentempfd");
 229                         goto done;
 230                 }
 231         }
 232
 233         if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
 234                 err = got_error_from_errno("socketpair");
 235                 goto done;
 236         }
 237
 238         fetchpid = fork();
 239         if (fetchpid == -1) {
 240                 err = got_error_from_errno("fork");
 241                 goto done;
 242         } else if (fetchpid == 0){
 243                 got_privsep_exec_child(imsg_fetchfds,
 244                     GOT_PATH_PROG_FETCH_PACK, tmppackpath);
 245         }
 246
 247         if (close(imsg_fetchfds[1]) == -1) {
 248                 err = got_error_from_errno("close");
 249                 goto done;
 250         }
 251         imsg_init(&fetchibuf, imsg_fetchfds[0]);
 252         nfetchfd = dup(fetchfd);
 253         if (nfetchfd == -1) {
 254                 err = got_error_from_errno("dup");
 255                 goto done;
 256         }
 257         err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
 258             fetch_all_branches, wanted_branches, wanted_refs,
 259             list_refs_only, worktree_refname, verbosity);
 260         if (err != NULL)
 261                 goto done;
 262         nfetchfd = -1;
 263         npackfd = dup(packfd);
 264         if (npackfd == -1) {
 265                 err = got_error_from_errno("dup");
 266                 goto done;
 267         }
 268         err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
 269         if (err != NULL)
 270                 goto done;
 271         npackfd = -1;
 272
 273         packfile_size = 0;
 274         progress = calloc(GOT_PKT_MAX, 1);
 275         if (progress == NULL) {
 276                 err = got_error_from_errno("calloc");
 277                 goto done;
 278         }
 279
 280         *pack_hash = calloc(1, sizeof(**pack_hash));
 281         if (*pack_hash == NULL) {
 282                 err = got_error_from_errno("calloc");
 283                 goto done;
 284         }
 285
 286         while (!done) {
 287                 struct got_object_id *id = NULL;
 288                 char *refname = NULL;
 289                 char *server_progress = NULL;
 290                 off_t packfile_size_cur = 0;
 291
 292                 err = got_privsep_recv_fetch_progress(&done,
 293                     &id, &refname, symrefs, &server_progress,
 294                     &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
 295                 if (err != NULL)
 296                         goto done;
 297                 /* Don't report size progress for an empty pack file. */
 298                 if (packfile_size_cur <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
 299                         packfile_size_cur = 0;
 300                 if (!done && refname && id) {
 301                         err = got_pathlist_insert(NULL, refs, refname, id);
 302                         if (err)
 303                                 goto done;
 304                 } else if (!done && server_progress) {
 305                         char *p;
 306                         /*
 307                          * XXX git-daemon tends to send batched output with
 308                          * lines spanning separate packets. Buffer progress
 309                          * output until we see a CR or LF to avoid giving
 310                          * partial lines of progress output to the callback.
 311                          */
 312                         if (strlcat(progress, server_progress,
 313                             GOT_PKT_MAX) >= GOT_PKT_MAX) {
 314                                 progress[0] = '\0'; /* discard */
 315                                 continue;
 316                         }
 317                         while ((p = strchr(progress, '\r')) != NULL ||
 318                             (p = strchr(progress, '\n')) != NULL) {
 319                                 char *s;
 320                                 size_t n;
 321                                 char c = *p;
 322                                 *p = '\0';
 323                                 if (asprintf(&s, "%s%s", progress,
 324                                     c == '\n' ? "\n" : "") == -1) {
 325                                         err = got_error_from_errno("asprintf");
 326                                         goto done;
 327                                 }
 328                                 err = progress_cb(progress_arg, s,
 329                                     packfile_size_cur, 0, 0, 0, 0);
 330                                 free(s);
 331                                 if (err)
 332                                         break;
 333                                 n = strlen(progress);
 334                                 if (n < GOT_PKT_MAX - 1) {
 335                                         memmove(progress, &progress[n + 1],
 336                                             GOT_PKT_MAX - n - 1);
 337                                 } else
 338                                         progress[0] = '\0';
 339                         }
 340                         free(server_progress);
 341                         if (err)
 342                                 goto done;
 343                 } else if (!done && packfile_size_cur != packfile_size) {
 344                         err = progress_cb(progress_arg, NULL,
 345                             packfile_size_cur, 0, 0, 0, 0);
 346                         if (err)
 347                                 break;
 348                         packfile_size = packfile_size_cur;
 349                 }
 350         }
 351         if (close(imsg_fetchfds[0]) == -1) {
 352                 err = got_error_from_errno("close");
 353                 goto done;
 354         }
 355         if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
 356                 err = got_error_from_errno("waitpid");
 357                 goto done;
 358         }
 359
 360         if (lseek(packfd, 0, SEEK_SET) == -1) {
 361                 err = got_error_from_errno("lseek");
 362                 goto done;
 363         }
 364
 365         /* If zero data was fetched without error we are already up-to-date. */
 366         if (packfile_size == 0) {
 367                 free(*pack_hash);
 368                 *pack_hash = NULL;
 369                 goto done;
 370         } else if (packfile_size < ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
 371                 err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
 372                 goto done;
 373         } else {
 374                 ssize_t n;
 375
 376                 n = read(packfd, &pack_hdr, ssizeof(pack_hdr));
 377                 if (n == -1) {
 378                         err = got_error_from_errno("read");
 379                         goto done;
 380                 }
 381                 if (n != ssizeof(pack_hdr)) {
 382                         err = got_error(GOT_ERR_IO);
 383                         goto done;
 384                 }
 385                 if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
 386                         err = got_error_msg(GOT_ERR_BAD_PACKFILE,
 387                             "bad pack file signature");
 388                         goto done;
 389                 }
 390                 if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
 391                         err = got_error_msg(GOT_ERR_BAD_PACKFILE,
 392                             "bad pack file version");
 393                         goto done;
 394                 }
 395                 nobj = be32toh(pack_hdr.nobjects);
 396                 if (nobj == 0 &&
 397                     packfile_size > ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
 398                         return got_error_msg(GOT_ERR_BAD_PACKFILE,
 399                             "bad pack file with zero objects");
 400                 if (nobj != 0 &&
 401                     packfile_size <= ssizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
 402                         return got_error_msg(GOT_ERR_BAD_PACKFILE,
 403                             "empty pack file with non-zero object count");
 404         }
 405
 406         /*
 407          * If the pack file contains no objects, we may only need to update
 408          * references in our repository. The caller will take care of that.
 409          */
 410         if (nobj == 0) {
 411                 free(*pack_hash);
 412                 *pack_hash = NULL;
 413                 goto done;
 414         }
 415
 416         if (lseek(packfd, 0, SEEK_SET) == -1) {
 417                 err = got_error_from_errno("lseek");
 418                 goto done;
 419         }
 420
 421         if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
 422                 err = got_error_from_errno("socketpair");
 423                 goto done;
 424         }
 425         idxpid = fork();
 426         if (idxpid == -1) {
 427                 err= got_error_from_errno("fork");
 428                 goto done;
 429         } else if (idxpid == 0)
 430                 got_privsep_exec_child(imsg_idxfds,
 431                     GOT_PATH_PROG_INDEX_PACK, tmppackpath);
 432         if (close(imsg_idxfds[1]) == -1) {
 433                 err = got_error_from_errno("close");
 434                 goto done;
 435         }
 436         imsg_init(&idxibuf, imsg_idxfds[0]);
 437
 438         npackfd = dup(packfd);
 439         if (npackfd == -1) {
 440                 err = got_error_from_errno("dup");
 441                 goto done;
 442         }
 443         err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
 444             npackfd);
 445         if (err != NULL)
 446                 goto done;
 447         npackfd = -1;
 448         err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
 449         if (err != NULL)
 450                 goto done;
 451         nidxfd = -1;
 452         for (i = 0; i < nitems(tmpfds); i++) {
 453                 err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
 454                 if (err != NULL)
 455                         goto done;
 456                 tmpfds[i] = -1;
 457         }
 458         done = 0;
 459         while (!done) {
 460                 int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
 461
 462                 err = got_privsep_recv_index_progress(&done, &nobj_total,
 463                     &nobj_indexed, &nobj_loose, &nobj_resolved,
 464                     &idxibuf);
 465                 if (err != NULL)
 466                         goto done;
 467                 if (nobj_indexed != 0) {
 468                         err = progress_cb(progress_arg, NULL,
 469                             packfile_size, nobj_total,
 470                             nobj_indexed, nobj_loose, nobj_resolved);
 471                         if (err)
 472                                 break;
 473                 }
 474         }
 475         if (close(imsg_idxfds[0]) == -1) {
 476                 err = got_error_from_errno("close");
 477                 goto done;
 478         }
 479         if (waitpid(idxpid, &idxstatus, 0) == -1) {
 480                 err = got_error_from_errno("waitpid");
 481                 goto done;
 482         }
 483
 484         err = got_object_id_str(&id_str, *pack_hash);
 485         if (err)
 486                 goto done;
 487         if (asprintf(&packpath, "%s/%s/pack-%s.pack",
 488             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 489                 err = got_error_from_errno("asprintf");
 490                 goto done;
 491         }
 492
 493         if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
 494             repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
 495                 err = got_error_from_errno("asprintf");
 496                 goto done;
 497         }
 498         free(id_str);
 499         id_str = NULL;
 500
 501         if (rename(tmppackpath, packpath) == -1) {
 502                 err = got_error_from_errno3("rename", tmppackpath, packpath);
 503                 goto done;
 504         }
 505         free(tmppackpath);
 506         tmppackpath = NULL;
 507         if (rename(tmpidxpath, idxpath) == -1) {
 508                 err = got_error_from_errno3("rename", tmpidxpath, idxpath);
 509                 goto done;
 510         }
 511         free(tmpidxpath);
 512         tmpidxpath = NULL;
 513
 514 done:
 515         if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
 516                 err = got_error_from_errno2("unlink", tmppackpath);
 517         if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
 518                 err = got_error_from_errno2("unlink", tmpidxpath);
 519         if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
 520                 err = got_error_from_errno("close");
 521         if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
 522                 err = got_error_from_errno("close");
 523         if (packfd != -1 && close(packfd) == -1 && err == NULL)
 524                 err = got_error_from_errno("close");
 525         if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
 526                 err = got_error_from_errno("close");
 527         for (i = 0; i < nitems(tmpfds); i++) {
 528                 if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
 529                         err = got_error_from_errno("close");
 530         }
 531         free(tmppackpath);
 532         free(tmpidxpath);
 533         free(idxpath);
 534         free(id_str);
 535         free(packpath);
 536         free(progress);
 537
 538         got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
 539         got_ref_list_free(&my_refs);
 540         if (err) {
 541                 free(*pack_hash);
 542                 *pack_hash = NULL;
 543                 got_pathlist_free(refs, GOT_PATHLIST_FREE_ALL);
 544                 got_pathlist_free(symrefs, GOT_PATHLIST_FREE_ALL);
 545         }
 546         return err;
 547 }