Blob


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 */
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/syslimits.h>
24 #include <sys/resource.h>
25 #include <sys/socket.h>
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 <sha1.h>
35 #include <zlib.h>
36 #include <ctype.h>
37 #include <limits.h>
38 #include <imsg.h>
39 #include <time.h>
40 #include <uuid.h>
41 #include <netdb.h>
42 #include <netinet/in.h>
44 #include "got_error.h"
45 #include "got_reference.h"
46 #include "got_repository.h"
47 #include "got_path.h"
48 #include "got_cancel.h"
49 #include "got_worktree.h"
50 #include "got_object.h"
51 #include "got_opentemp.h"
52 #include "got_fetch.h"
54 #include "got_lib_delta.h"
55 #include "got_lib_inflate.h"
56 #include "got_lib_object.h"
57 #include "got_lib_object_parse.h"
58 #include "got_lib_object_create.h"
59 #include "got_lib_pack.h"
60 #include "got_lib_sha1.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_object_cache.h"
63 #include "got_lib_repository.h"
65 #ifndef nitems
66 #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
67 #endif
69 #ifndef MIN
70 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
71 #endif
73 static int
74 hassuffix(char *base, char *suf)
75 {
76 int nb, ns;
78 nb = strlen(base);
79 ns = strlen(suf);
80 if (ns <= nb && strcmp(base + (nb - ns), suf) == 0)
81 return 1;
82 return 0;
83 }
85 static const struct got_error *
86 dial_ssh(int *fetchfd, const char *host, const char *port, const char *path,
87 const char *direction, int verbosity)
88 {
89 const struct got_error *error = NULL;
90 int pid, pfd[2];
91 char cmd[64];
92 char *argv[9];
93 int i = 0;
95 *fetchfd = -1;
97 argv[0] = GOT_FETCH_PATH_SSH;
98 if (verbosity == -1) {
99 argv[1 + i++] = "-q";
100 } else {
101 /* ssh(1) allows up to 3 "-v" options. */
102 for (i = 0; i < MIN(3, verbosity); i++)
103 argv[1 + i] = "-v";
105 argv[1 + i] = "--";
106 argv[2 + i] = (char *)host;
107 argv[3 + i] = (char *)cmd;
108 argv[4 + i] = (char *)path;
109 argv[5 + i] = NULL;
111 if (pipe(pfd) == -1)
112 return got_error_from_errno("pipe");
114 pid = fork();
115 if (pid == -1) {
116 error = got_error_from_errno("fork");
117 close(pfd[0]);
118 close(pfd[1]);
119 return error;
120 } else if (pid == 0) {
121 int n;
122 close(pfd[1]);
123 dup2(pfd[0], 0);
124 dup2(pfd[0], 1);
125 n = snprintf(cmd, sizeof(cmd), "git-%s-pack", direction);
126 if (n < 0 || n >= sizeof(cmd))
127 err(1, "snprintf");
128 if (execv(GOT_FETCH_PATH_SSH, argv) == -1)
129 err(1, "execl");
130 abort(); /* not reached */
131 } else {
132 close(pfd[0]);
133 *fetchfd = pfd[1];
134 return NULL;
138 static const struct got_error *
139 dial_git(int *fetchfd, const char *host, const char *port, const char *path,
140 const char *direction)
142 const struct got_error *err = NULL;
143 struct addrinfo hints, *servinfo, *p;
144 char *cmd = NULL, *pkt = NULL;
145 int fd = -1, totlen, r, eaicode;
147 *fetchfd = -1;
149 memset(&hints, 0, sizeof hints);
150 hints.ai_family = AF_UNSPEC;
151 hints.ai_socktype = SOCK_STREAM;
152 eaicode = getaddrinfo(host, port, &hints, &servinfo);
153 if (eaicode) {
154 char msg[512];
155 snprintf(msg, sizeof(msg), "%s: %s", host,
156 gai_strerror(eaicode));
157 return got_error_msg(GOT_ERR_ADDRINFO, msg);
160 for (p = servinfo; p != NULL; p = p->ai_next) {
161 if ((fd = socket(p->ai_family, p->ai_socktype,
162 p->ai_protocol)) == -1)
163 continue;
164 if (connect(fd, p->ai_addr, p->ai_addrlen) == 0)
165 break;
166 err = got_error_from_errno("connect");
167 close(fd);
169 if (p == NULL)
170 goto done;
172 if (asprintf(&cmd, "git-%s-pack %s", direction, path) == -1) {
173 err = got_error_from_errno("asprintf");
174 goto done;
176 totlen = 4 + strlen(cmd) + 1 + strlen("host=") + strlen(host) + 1;
177 if (asprintf(&pkt, "%04x%s", totlen, cmd) == -1) {
178 err = got_error_from_errno("asprintf");
179 goto done;
181 r = write(fd, pkt, strlen(pkt) + 1);
182 if (r == -1) {
183 err = got_error_from_errno("write");
184 goto done;
186 if (asprintf(&pkt, "host=%s", host) == -1) {
187 err = got_error_from_errno("asprintf");
188 goto done;
190 r = write(fd, pkt, strlen(pkt) + 1);
191 if (r == -1) {
192 err = got_error_from_errno("write");
193 goto done;
195 done:
196 free(cmd);
197 free(pkt);
198 if (err) {
199 if (fd != -1)
200 close(fd);
201 } else
202 *fetchfd = fd;
203 return err;
206 const struct got_error *
207 got_fetch_connect(int *fetchfd, const char *proto, const char *host,
208 const char *port, const char *server_path, int verbosity)
210 const struct got_error *err = NULL;
212 *fetchfd = -1;
214 if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
215 err = dial_ssh(fetchfd, host, port, server_path, "upload",
216 verbosity);
217 else if (strcmp(proto, "git") == 0)
218 err = dial_git(fetchfd, host, port, server_path, "upload");
219 else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
220 err = got_error_path(proto, GOT_ERR_NOT_IMPL);
221 else
222 err = got_error_path(proto, GOT_ERR_BAD_PROTO);
223 return err;
226 const struct got_error *
227 got_fetch_parse_uri(char **proto, char **host, char **port,
228 char **server_path, char **repo_name, const char *uri)
230 const struct got_error *err = NULL;
231 char *s, *p, *q;
232 int n;
234 *proto = *host = *port = *server_path = *repo_name = NULL;
236 p = strstr(uri, "://");
237 if (!p) {
238 /* Try parsing Git's "scp" style URL syntax. */
239 *proto = strdup("ssh");
240 if (proto == NULL) {
241 err = got_error_from_errno("strdup");
242 goto done;
244 *port = strdup("22");
245 if (*port == NULL) {
246 err = got_error_from_errno("strdup");
247 goto done;
249 s = (char *)uri;
250 q = strchr(s, ':');
251 if (q == NULL) {
252 err = got_error(GOT_ERR_PARSE_URI);
253 goto done;
255 /* No slashes allowed before first colon. */
256 p = strchr(s, '/');
257 if (p && q > p) {
258 err = got_error(GOT_ERR_PARSE_URI);
259 goto done;
261 *host = strndup(s, q - s);
262 if (*host == NULL) {
263 err = got_error_from_errno("strndup");
264 goto done;
266 p = q + 1;
267 } else {
268 *proto = strndup(uri, p - uri);
269 if (proto == NULL) {
270 err = got_error_from_errno("strndup");
271 goto done;
273 s = p + 3;
275 p = strstr(s, "/");
276 if (p == NULL || strlen(p) == 1) {
277 err = got_error(GOT_ERR_PARSE_URI);
278 goto done;
281 q = memchr(s, ':', p - s);
282 if (q) {
283 *host = strndup(s, q - s);
284 if (*host == NULL) {
285 err = got_error_from_errno("strndup");
286 goto done;
288 *port = strndup(q + 1, p - (q + 1));
289 if (*port == NULL) {
290 err = got_error_from_errno("strndup");
291 goto done;
293 } else {
294 *host = strndup(s, p - s);
295 if (*host == NULL) {
296 err = got_error_from_errno("strndup");
297 goto done;
299 if (asprintf(port, "%u", GOT_DEFAULT_GIT_PORT) == -1) {
300 err = got_error_from_errno("asprintf");
301 goto done;
306 *server_path = strdup(p);
307 if (*server_path == NULL) {
308 err = got_error_from_errno("strdup");
309 goto done;
312 p = strrchr(p, '/');
313 if (!p || strlen(p) <= 1) {
314 err = got_error(GOT_ERR_PARSE_URI);
315 goto done;
317 p++;
318 n = strlen(p);
319 if (n == 0) {
320 err = got_error(GOT_ERR_PARSE_URI);
321 goto done;
323 if (hassuffix(p, ".git"))
324 n -= 4;
325 *repo_name = strndup(p, (p + n) - p);
326 if (*repo_name == NULL) {
327 err = got_error_from_errno("strndup");
328 goto done;
330 done:
331 if (err) {
332 free(*proto);
333 *proto = NULL;
334 free(*host);
335 *host = NULL;
336 free(*port);
337 *port = NULL;
338 free(*server_path);
339 *server_path = NULL;
340 free(*repo_name);
341 *repo_name = NULL;
343 return err;
346 static const struct got_error *
347 check_pack_hash(int fd, size_t sz, uint8_t *hcomp)
349 SHA1_CTX ctx;
350 uint8_t hexpect[SHA1_DIGEST_LENGTH];
351 uint8_t buf[32 * 1024];
352 ssize_t n, r, nr;
354 if (sz < sizeof(struct got_packfile_hdr) + SHA1_DIGEST_LENGTH)
355 return got_error_msg(GOT_ERR_BAD_PACKFILE, "short packfile");
357 n = 0;
358 SHA1Init(&ctx);
359 while (n < sz - 20) {
360 nr = sizeof(buf);
361 if (sz - n - 20 < sizeof(buf))
362 nr = sz - n - 20;
363 r = read(fd, buf, nr);
364 if (r == -1)
365 return got_error_from_errno("read");
366 if (r != nr)
367 return got_error_msg(GOT_ERR_BAD_PACKFILE,
368 "short pack file");
369 SHA1Update(&ctx, buf, nr);
370 n += r;
372 SHA1Final(hcomp, &ctx);
374 r = read(fd, hexpect, sizeof(hexpect));
375 if (r == -1)
376 return got_error_from_errno("read");
377 if (r != sizeof(hexpect))
378 return got_error_msg(GOT_ERR_BAD_PACKFILE,
379 "short pack file");
381 if (memcmp(hcomp, hexpect, SHA1_DIGEST_LENGTH) != 0)
382 return got_error_msg(GOT_ERR_BAD_PACKFILE,
383 "packfile checksum mismatch");
385 return NULL;
388 const struct got_error*
389 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
390 struct got_pathlist_head *symrefs, const char *remote_name, int fetchfd,
391 struct got_repository *repo, got_fetch_progress_cb progress_cb,
392 void *progress_arg)
394 int imsg_fetchfds[2], imsg_idxfds[2];
395 int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
396 int tmpfds[3], i;
397 int fetchstatus, idxstatus, done = 0;
398 const struct got_error *err;
399 struct imsgbuf fetchibuf, idxibuf;
400 pid_t fetchpid, idxpid;
401 char *tmppackpath = NULL, *tmpidxpath = NULL;
402 char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
403 const char *repo_path = got_repo_get_path(repo);
404 struct got_pathlist_head have_refs;
405 struct got_pathlist_entry *pe;
406 struct got_reflist_head my_refs;
407 struct got_reflist_entry *re;
408 off_t packfile_size = 0;
409 char *ref_prefix = NULL;
410 size_t ref_prefixlen = 0;
411 char *path;
413 *pack_hash = NULL;
414 for (i = 0; i < nitems(tmpfds); i++)
415 tmpfds[i] = -1;
417 TAILQ_INIT(&have_refs);
418 SIMPLEQ_INIT(&my_refs);
420 if (asprintf(&ref_prefix, "refs/remotes/%s/", remote_name) == -1)
421 return got_error_from_errno("asprintf");
422 ref_prefixlen = strlen(ref_prefix);
424 err = got_ref_list(&my_refs, repo, NULL, got_ref_cmp_by_name, NULL);
425 if (err)
426 goto done;
428 SIMPLEQ_FOREACH(re, &my_refs, entry) {
429 struct got_object_id *id;
430 const char *refname;
432 if (got_ref_is_symbolic(re->ref))
433 continue;
435 refname = got_ref_get_name(re->ref);
436 if (strncmp("refs/tags/", refname, 10) == 0) {
437 char *tagname;
439 err = got_ref_resolve(&id, repo, re->ref);
440 if (err)
441 goto done;
442 tagname = strdup(refname);
443 if (tagname == NULL) {
444 err = got_error_from_errno("strdup");
445 goto done;
447 err = got_pathlist_append(&have_refs, tagname, id);
448 if (err) {
449 free(tagname);
450 goto done;
454 if (strncmp(ref_prefix, refname, ref_prefixlen) == 0) {
455 char *branchname;
457 err = got_ref_resolve(&id, repo, re->ref);
458 if (err)
459 goto done;
461 if (asprintf(&branchname, "refs/heads/%s",
462 refname + ref_prefixlen) == -1) {
463 err = got_error_from_errno("asprintf");
464 goto done;
466 err = got_pathlist_append(&have_refs, branchname, id);
467 if (err) {
468 free(branchname);
469 goto done;
474 if (asprintf(&path, "%s/%s/fetching.pack",
475 repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
476 err = got_error_from_errno("asprintf");
477 goto done;
479 err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
480 free(path);
481 if (err)
482 goto done;
483 npackfd = dup(packfd);
484 if (npackfd == -1) {
485 err = got_error_from_errno("dup");
486 goto done;
488 if (asprintf(&path, "%s/%s/fetching.idx",
489 repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
490 err = got_error_from_errno("asprintf");
491 goto done;
493 err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
494 free(path);
495 if (err)
496 goto done;
497 nidxfd = dup(idxfd);
498 if (nidxfd == -1) {
499 err = got_error_from_errno("dup");
500 goto done;
503 for (i = 0; i < nitems(tmpfds); i++) {
504 tmpfds[i] = got_opentempfd();
505 if (tmpfds[i] == -1) {
506 err = got_error_from_errno("got_opentempfd");
507 goto done;
511 if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
512 err = got_error_from_errno("socketpair");
513 goto done;
516 fetchpid = fork();
517 if (fetchpid == -1) {
518 err = got_error_from_errno("fork");
519 goto done;
520 } else if (fetchpid == 0){
521 got_privsep_exec_child(imsg_fetchfds,
522 GOT_PATH_PROG_FETCH_PACK, tmppackpath);
525 if (close(imsg_fetchfds[1]) != 0) {
526 err = got_error_from_errno("close");
527 goto done;
529 imsg_init(&fetchibuf, imsg_fetchfds[0]);
530 nfetchfd = dup(fetchfd);
531 if (nfetchfd == -1) {
532 err = got_error_from_errno("dup");
533 goto done;
535 err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs);
536 if (err != NULL)
537 goto done;
538 nfetchfd = -1;
539 err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
540 if (err != NULL)
541 goto done;
542 npackfd = dup(packfd);
543 if (npackfd == -1) {
544 err = got_error_from_errno("dup");
545 goto done;
548 packfile_size = 0;
549 while (!done) {
550 struct got_object_id *id = NULL;
551 char *refname = NULL;
552 char *server_progress = NULL;
553 off_t packfile_size_cur = 0;
555 err = got_privsep_recv_fetch_progress(&done,
556 &id, &refname, symrefs, &server_progress,
557 &packfile_size_cur, &fetchibuf);
558 if (err != NULL)
559 goto done;
560 if (done) {
561 if (packfile_size > 0)
562 *pack_hash = id;
563 else
564 free(id);
566 else if (refname && id) {
567 err = got_pathlist_append(refs, refname, id);
568 if (err)
569 goto done;
570 } else if (server_progress) {
571 char *s, *s0 = server_progress;
572 while ((s = strsep(&s0, "\r")) != NULL) {
573 if (*s == '\0')
574 continue;
575 err = progress_cb(progress_arg, s,
576 packfile_size_cur, 0, 0, 0, 0);
577 if (err)
578 break;
580 free(server_progress);
581 if (err)
582 goto done;
583 } else if (packfile_size_cur != packfile_size) {
584 err = progress_cb(progress_arg, NULL,
585 packfile_size_cur, 0, 0, 0, 0);
586 if (err)
587 break;
588 packfile_size = packfile_size_cur;
591 if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
592 err = got_error_from_errno("waitpid");
593 goto done;
596 /* If zero data was fetched without error we are already up-to-date. */
597 if (packfile_size == 0)
598 goto done;
600 if (lseek(packfd, 0, SEEK_SET) == -1) {
601 err = got_error_from_errno("lseek");
602 goto done;
604 err = check_pack_hash(packfd, packfile_size, (*pack_hash)->sha1);
605 if (err)
606 goto done;
608 if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
609 err = got_error_from_errno("socketpair");
610 goto done;
612 idxpid = fork();
613 if (idxpid == -1) {
614 err= got_error_from_errno("fork");
615 goto done;
616 } else if (idxpid == 0)
617 got_privsep_exec_child(imsg_idxfds,
618 GOT_PATH_PROG_INDEX_PACK, tmppackpath);
619 if (close(imsg_idxfds[1]) != 0) {
620 err = got_error_from_errno("close");
621 goto done;
623 imsg_init(&idxibuf, imsg_idxfds[0]);
625 err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
626 npackfd);
627 if (err != NULL)
628 goto done;
629 npackfd = -1;
630 err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
631 if (err != NULL)
632 goto done;
633 nidxfd = -1;
634 for (i = 0; i < nitems(tmpfds); i++) {
635 err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
636 if (err != NULL)
637 goto done;
638 tmpfds[i] = -1;
640 done = 0;
641 while (!done) {
642 int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
644 err = got_privsep_recv_index_progress(&done, &nobj_total,
645 &nobj_indexed, &nobj_loose, &nobj_resolved,
646 &idxibuf);
647 if (err != NULL)
648 goto done;
649 if (nobj_indexed != 0) {
650 err = progress_cb(progress_arg, NULL,
651 packfile_size, nobj_total,
652 nobj_indexed, nobj_loose, nobj_resolved);
653 if (err)
654 break;
656 imsg_clear(&idxibuf);
658 if (close(imsg_idxfds[0]) == -1) {
659 err = got_error_from_errno("close");
660 goto done;
662 if (waitpid(idxpid, &idxstatus, 0) == -1) {
663 err = got_error_from_errno("waitpid");
664 goto done;
667 err = got_object_id_str(&id_str, *pack_hash);
668 if (err)
669 goto done;
670 if (asprintf(&packpath, "%s/%s/pack-%s.pack",
671 repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
672 err = got_error_from_errno("asprintf");
673 goto done;
676 if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
677 repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
678 err = got_error_from_errno("asprintf");
679 goto done;
682 if (rename(tmppackpath, packpath) == -1) {
683 err = got_error_from_errno3("rename", tmppackpath, packpath);
684 goto done;
686 free(tmppackpath);
687 tmppackpath = NULL;
688 if (rename(tmpidxpath, idxpath) == -1) {
689 err = got_error_from_errno3("rename", tmpidxpath, idxpath);
690 goto done;
692 free(tmpidxpath);
693 tmpidxpath = NULL;
695 done:
696 if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
697 err = got_error_from_errno2("unlink", tmppackpath);
698 if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
699 err = got_error_from_errno2("unlink", tmpidxpath);
700 if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
701 err = got_error_from_errno("close");
702 if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
703 err = got_error_from_errno("close");
704 if (packfd != -1 && close(packfd) == -1 && err == NULL)
705 err = got_error_from_errno("close");
706 if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
707 err = got_error_from_errno("close");
708 for (i = 0; i < nitems(tmpfds); i++) {
709 if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
710 err = got_error_from_errno("close");
712 free(tmppackpath);
713 free(tmpidxpath);
714 free(idxpath);
715 free(packpath);
716 free(ref_prefix);
718 TAILQ_FOREACH(pe, &have_refs, entry) {
719 free((char *)pe->path);
720 free(pe->data);
722 got_pathlist_free(&have_refs);
723 got_ref_list_free(&my_refs);
724 if (err) {
725 free(*pack_hash);
726 *pack_hash = NULL;
727 TAILQ_FOREACH(pe, refs, entry) {
728 free((void *)pe->path);
729 free(pe->data);
731 got_pathlist_free(refs);
732 TAILQ_FOREACH(pe, symrefs, entry) {
733 free((void *)pe->path);
734 free(pe->data);
736 got_pathlist_free(symrefs);
738 return err;