2 * Copyright (c) 2018, 2019 Ori Bernstein <ori@openbsd.org>
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.
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.
17 #include <sys/types.h>
19 #include <sys/queue.h>
21 #include <sys/socket.h>
23 #include <sys/resource.h>
24 #include <sys/socket.h>
43 #include <netinet/in.h>
45 #include "got_error.h"
46 #include "got_reference.h"
47 #include "got_repository.h"
49 #include "got_cancel.h"
50 #include "got_worktree.h"
51 #include "got_object.h"
52 #include "got_opentemp.h"
53 #include "got_fetch.h"
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_sha1.h"
62 #include "got_lib_privsep.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
67 #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
71 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
75 hassuffix(char *base, char *suf)
81 if (ns <= nb && strcmp(base + (nb - ns), suf) == 0)
86 static const struct got_error *
87 dial_ssh(pid_t *fetchpid, int *fetchfd, const char *host, const char *port,
88 const char *path, const char *direction, int verbosity)
90 const struct got_error *error = NULL;
99 argv[i++] = GOT_FETCH_PATH_SSH;
102 argv[i++] = (char *)port;
104 if (verbosity == -1) {
107 /* ssh(1) allows up to 3 "-v" options. */
108 for (j = 0; j < MIN(3, verbosity); j++)
112 argv[i++] = (char *)host;
113 argv[i++] = (char *)cmd;
114 argv[i++] = (char *)path;
118 return got_error_from_errno("pipe");
122 error = got_error_from_errno("fork");
126 } else if (pid == 0) {
131 n = snprintf(cmd, sizeof(cmd), "git-%s-pack", direction);
132 if (n < 0 || n >= sizeof(cmd))
134 if (execv(GOT_FETCH_PATH_SSH, argv) == -1)
136 abort(); /* not reached */
145 static const struct got_error *
146 dial_git(int *fetchfd, const char *host, const char *port, const char *path,
147 const char *direction)
149 const struct got_error *err = NULL;
150 struct addrinfo hints, *servinfo, *p;
151 char *cmd = NULL, *pkt = NULL;
152 int fd = -1, totlen, r, eaicode;
157 port = GOT_DEFAULT_GIT_PORT_STR;
159 memset(&hints, 0, sizeof hints);
160 hints.ai_family = AF_UNSPEC;
161 hints.ai_socktype = SOCK_STREAM;
162 eaicode = getaddrinfo(host, port, &hints, &servinfo);
165 snprintf(msg, sizeof(msg), "%s: %s", host,
166 gai_strerror(eaicode));
167 return got_error_msg(GOT_ERR_ADDRINFO, msg);
170 for (p = servinfo; p != NULL; p = p->ai_next) {
171 if ((fd = socket(p->ai_family, p->ai_socktype,
172 p->ai_protocol)) == -1)
174 if (connect(fd, p->ai_addr, p->ai_addrlen) == 0)
176 err = got_error_from_errno("connect");
182 if (asprintf(&cmd, "git-%s-pack %s", direction, path) == -1) {
183 err = got_error_from_errno("asprintf");
186 totlen = 4 + strlen(cmd) + 1 + strlen("host=") + strlen(host) + 1;
187 if (asprintf(&pkt, "%04x%s", totlen, cmd) == -1) {
188 err = got_error_from_errno("asprintf");
191 r = write(fd, pkt, strlen(pkt) + 1);
193 err = got_error_from_errno("write");
196 if (asprintf(&pkt, "host=%s", host) == -1) {
197 err = got_error_from_errno("asprintf");
200 r = write(fd, pkt, strlen(pkt) + 1);
202 err = got_error_from_errno("write");
216 const struct got_error *
217 got_fetch_connect(pid_t *fetchpid, int *fetchfd, const char *proto,
218 const char *host, const char *port, const char *server_path, int verbosity)
220 const struct got_error *err = NULL;
225 if (strcmp(proto, "ssh") == 0 || strcmp(proto, "git+ssh") == 0)
226 err = dial_ssh(fetchpid, fetchfd, host, port, server_path,
227 "upload", verbosity);
228 else if (strcmp(proto, "git") == 0)
229 err = dial_git(fetchfd, host, port, server_path, "upload");
230 else if (strcmp(proto, "http") == 0 || strcmp(proto, "git+http") == 0)
231 err = got_error_path(proto, GOT_ERR_NOT_IMPL);
233 err = got_error_path(proto, GOT_ERR_BAD_PROTO);
237 const struct got_error *
238 got_fetch_parse_uri(char **proto, char **host, char **port,
239 char **server_path, char **repo_name, const char *uri)
241 const struct got_error *err = NULL;
245 *proto = *host = *port = *server_path = *repo_name = NULL;
247 p = strstr(uri, "://");
249 /* Try parsing Git's "scp" style URL syntax. */
250 *proto = strdup("ssh");
252 err = got_error_from_errno("strdup");
258 err = got_error(GOT_ERR_PARSE_URI);
261 /* No slashes allowed before first colon. */
264 err = got_error(GOT_ERR_PARSE_URI);
267 *host = strndup(s, q - s);
269 err = got_error_from_errno("strndup");
274 *proto = strndup(uri, p - uri);
276 err = got_error_from_errno("strndup");
282 if (p == NULL || strlen(p) == 1) {
283 err = got_error(GOT_ERR_PARSE_URI);
287 q = memchr(s, ':', p - s);
289 *host = strndup(s, q - s);
291 err = got_error_from_errno("strndup");
294 *port = strndup(q + 1, p - (q + 1));
296 err = got_error_from_errno("strndup");
300 *host = strndup(s, p - s);
302 err = got_error_from_errno("strndup");
308 while (p[0] == '/' && p[1] == '/')
310 *server_path = strdup(p);
311 if (*server_path == NULL) {
312 err = got_error_from_errno("strdup");
317 if (!p || strlen(p) <= 1) {
318 err = got_error(GOT_ERR_PARSE_URI);
324 err = got_error(GOT_ERR_PARSE_URI);
327 if (hassuffix(p, ".git"))
329 *repo_name = strndup(p, (p + n) - p);
330 if (*repo_name == NULL) {
331 err = got_error_from_errno("strndup");
350 const struct got_error*
351 got_fetch_pack(struct got_object_id **pack_hash, struct got_pathlist_head *refs,
352 struct got_pathlist_head *symrefs, const char *remote_name,
353 int mirror_references, int fetch_all_branches,
354 struct got_pathlist_head *wanted_branches,
355 struct got_pathlist_head *wanted_refs, int list_refs_only, int verbosity,
356 int fetchfd, struct got_repository *repo,
357 got_fetch_progress_cb progress_cb, void *progress_arg)
359 int imsg_fetchfds[2], imsg_idxfds[2];
360 int packfd = -1, npackfd = -1, idxfd = -1, nidxfd = -1, nfetchfd = -1;
362 int fetchstatus, idxstatus, done = 0;
363 const struct got_error *err;
364 struct imsgbuf fetchibuf, idxibuf;
365 pid_t fetchpid, idxpid;
366 char *tmppackpath = NULL, *tmpidxpath = NULL;
367 char *packpath = NULL, *idxpath = NULL, *id_str = NULL;
368 const char *repo_path = NULL;
369 struct got_pathlist_head have_refs;
370 struct got_pathlist_entry *pe;
371 struct got_reflist_head my_refs;
372 struct got_reflist_entry *re;
373 off_t packfile_size = 0;
374 struct got_packfile_hdr pack_hdr;
376 char *ref_prefix = NULL;
377 size_t ref_prefixlen = 0;
379 char *progress = NULL;
384 * Prevent fetching of references that won't make any
385 * sense outside of the remote repository's context.
387 TAILQ_FOREACH(pe, wanted_refs, entry) {
388 const char *refname = pe->path;
389 if (strncmp(refname, "refs/got/", 9) == 0 ||
390 strncmp(refname, "got/", 4) == 0 ||
391 strncmp(refname, "refs/remotes/", 13) == 0 ||
392 strncmp(refname, "remotes/", 8) == 0)
393 return got_error_path(refname, GOT_ERR_FETCH_BAD_REF);
397 repo_path = got_repo_get_path_git_dir(repo);
399 for (i = 0; i < nitems(tmpfds); i++)
402 TAILQ_INIT(&have_refs);
403 SIMPLEQ_INIT(&my_refs);
405 if (!mirror_references) {
406 if (asprintf(&ref_prefix, "refs/remotes/%s/",
408 return got_error_from_errno("asprintf");
409 ref_prefixlen = strlen(ref_prefix);
412 if (!list_refs_only) {
413 err = got_ref_list(&my_refs, repo, NULL,
414 got_ref_cmp_by_name, NULL);
419 SIMPLEQ_FOREACH(re, &my_refs, entry) {
420 struct got_object_id *id;
423 if (got_ref_is_symbolic(re->ref))
426 refname = got_ref_get_name(re->ref);
428 if (mirror_references) {
430 err = got_ref_resolve(&id, repo, re->ref);
433 name = strdup(refname);
435 err = got_error_from_errno("strdup");
438 err = got_pathlist_append(&have_refs, name, id);
444 if (strncmp("refs/tags/", refname, 10) == 0) {
447 err = got_ref_resolve(&id, repo, re->ref);
450 tagname = strdup(refname);
451 if (tagname == NULL) {
452 err = got_error_from_errno("strdup");
455 err = got_pathlist_append(&have_refs, tagname, id);
462 if (strncmp(ref_prefix, refname, ref_prefixlen) == 0) {
465 err = got_ref_resolve(&id, repo, re->ref);
469 if (asprintf(&branchname, "refs/heads/%s",
470 refname + ref_prefixlen) == -1) {
471 err = got_error_from_errno("asprintf");
474 err = got_pathlist_append(&have_refs, branchname, id);
482 if (list_refs_only) {
483 packfd = got_opentempfd();
485 err = got_error_from_errno("got_opentempfd");
489 if (asprintf(&path, "%s/%s/fetching.pack",
490 repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
491 err = got_error_from_errno("asprintf");
494 err = got_opentemp_named_fd(&tmppackpath, &packfd, path);
499 if (list_refs_only) {
500 idxfd = got_opentempfd();
502 err = got_error_from_errno("got_opentempfd");
506 if (asprintf(&path, "%s/%s/fetching.idx",
507 repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
508 err = got_error_from_errno("asprintf");
511 err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
518 err = got_error_from_errno("dup");
522 for (i = 0; i < nitems(tmpfds); i++) {
523 tmpfds[i] = got_opentempfd();
524 if (tmpfds[i] == -1) {
525 err = got_error_from_errno("got_opentempfd");
530 if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fetchfds) == -1) {
531 err = got_error_from_errno("socketpair");
536 if (fetchpid == -1) {
537 err = got_error_from_errno("fork");
539 } else if (fetchpid == 0){
540 got_privsep_exec_child(imsg_fetchfds,
541 GOT_PATH_PROG_FETCH_PACK, tmppackpath);
544 if (close(imsg_fetchfds[1]) != 0) {
545 err = got_error_from_errno("close");
548 imsg_init(&fetchibuf, imsg_fetchfds[0]);
549 nfetchfd = dup(fetchfd);
550 if (nfetchfd == -1) {
551 err = got_error_from_errno("dup");
554 err = got_privsep_send_fetch_req(&fetchibuf, nfetchfd, &have_refs,
555 fetch_all_branches, wanted_branches, wanted_refs,
556 list_refs_only, verbosity);
560 npackfd = dup(packfd);
562 err = got_error_from_errno("dup");
565 err = got_privsep_send_fetch_outfd(&fetchibuf, npackfd);
571 progress = calloc(GOT_FETCH_PKTMAX, 1);
572 if (progress == NULL) {
573 err = got_error_from_errno("calloc");
577 *pack_hash = calloc(1, sizeof(**pack_hash));
578 if (*pack_hash == NULL) {
579 err = got_error_from_errno("calloc");
584 struct got_object_id *id = NULL;
585 char *refname = NULL;
586 char *server_progress = NULL;
587 off_t packfile_size_cur = 0;
589 err = got_privsep_recv_fetch_progress(&done,
590 &id, &refname, symrefs, &server_progress,
591 &packfile_size_cur, (*pack_hash)->sha1, &fetchibuf);
594 if (!done && refname && id) {
595 err = got_pathlist_insert(NULL, refs, refname, id);
598 } else if (!done && server_progress) {
601 * XXX git-daemon tends to send batched output with
602 * lines spanning separate packets. Buffer progress
603 * output until we see a CR or LF to avoid giving
604 * partial lines of progress output to the callback.
606 if (strlcat(progress, server_progress,
607 GOT_FETCH_PKTMAX) >= GOT_FETCH_PKTMAX) {
608 progress[0] = '\0'; /* discard */
611 while ((p = strchr(progress, '\r')) != NULL ||
612 (p = strchr(progress, '\n')) != NULL) {
617 if (asprintf(&s, "%s%s", progress,
618 c == '\n' ? "\n" : "") == -1) {
619 err = got_error_from_errno("asprintf");
622 err = progress_cb(progress_arg, s,
623 packfile_size_cur, 0, 0, 0, 0);
627 n = strlen(progress);
628 if (n < GOT_FETCH_PKTMAX - 1) {
629 memmove(progress, &progress[n + 1],
630 GOT_FETCH_PKTMAX - n - 1);
634 free(server_progress);
637 } else if (!done && packfile_size_cur != packfile_size) {
638 err = progress_cb(progress_arg, NULL,
639 packfile_size_cur, 0, 0, 0, 0);
642 packfile_size = packfile_size_cur;
645 if (waitpid(fetchpid, &fetchstatus, 0) == -1) {
646 err = got_error_from_errno("waitpid");
650 if (lseek(packfd, 0, SEEK_SET) == -1) {
651 err = got_error_from_errno("lseek");
655 /* If zero data was fetched without error we are already up-to-date. */
656 if (packfile_size == 0) {
660 } else if (packfile_size < sizeof(pack_hdr) + SHA1_DIGEST_LENGTH) {
661 err = got_error_msg(GOT_ERR_BAD_PACKFILE, "short pack file");
666 n = read(packfd, &pack_hdr, sizeof(pack_hdr));
668 err = got_error_from_errno("read");
671 if (n != sizeof(pack_hdr)) {
672 err = got_error(GOT_ERR_IO);
675 if (pack_hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)) {
676 err = got_error_msg(GOT_ERR_BAD_PACKFILE,
677 "bad pack file signature");
680 if (pack_hdr.version != htobe32(GOT_PACKFILE_VERSION)) {
681 err = got_error_msg(GOT_ERR_BAD_PACKFILE,
682 "bad pack file version");
685 nobj = be32toh(pack_hdr.nobjects);
687 packfile_size > sizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
688 return got_error_msg(GOT_ERR_BAD_PACKFILE,
689 "bad pack file with zero objects");
691 packfile_size <= sizeof(pack_hdr) + SHA1_DIGEST_LENGTH)
692 return got_error_msg(GOT_ERR_BAD_PACKFILE,
693 "empty pack file with non-zero object count");
697 * If the pack file contains no objects, we may only need to update
698 * references in our repository. The caller will take care of that.
703 if (lseek(packfd, 0, SEEK_SET) == -1) {
704 err = got_error_from_errno("lseek");
708 if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
709 err = got_error_from_errno("socketpair");
714 err= got_error_from_errno("fork");
716 } else if (idxpid == 0)
717 got_privsep_exec_child(imsg_idxfds,
718 GOT_PATH_PROG_INDEX_PACK, tmppackpath);
719 if (close(imsg_idxfds[1]) != 0) {
720 err = got_error_from_errno("close");
723 imsg_init(&idxibuf, imsg_idxfds[0]);
725 npackfd = dup(packfd);
727 err = got_error_from_errno("dup");
730 err = got_privsep_send_index_pack_req(&idxibuf, (*pack_hash)->sha1,
735 err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
739 for (i = 0; i < nitems(tmpfds); i++) {
740 err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
747 int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
749 err = got_privsep_recv_index_progress(&done, &nobj_total,
750 &nobj_indexed, &nobj_loose, &nobj_resolved,
754 if (nobj_indexed != 0) {
755 err = progress_cb(progress_arg, NULL,
756 packfile_size, nobj_total,
757 nobj_indexed, nobj_loose, nobj_resolved);
761 imsg_clear(&idxibuf);
763 if (close(imsg_idxfds[0]) == -1) {
764 err = got_error_from_errno("close");
767 if (waitpid(idxpid, &idxstatus, 0) == -1) {
768 err = got_error_from_errno("waitpid");
772 err = got_object_id_str(&id_str, *pack_hash);
775 if (asprintf(&packpath, "%s/%s/pack-%s.pack",
776 repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
777 err = got_error_from_errno("asprintf");
781 if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
782 repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
783 err = got_error_from_errno("asprintf");
787 if (rename(tmppackpath, packpath) == -1) {
788 err = got_error_from_errno3("rename", tmppackpath, packpath);
793 if (rename(tmpidxpath, idxpath) == -1) {
794 err = got_error_from_errno3("rename", tmpidxpath, idxpath);
801 if (tmppackpath && unlink(tmppackpath) == -1 && err == NULL)
802 err = got_error_from_errno2("unlink", tmppackpath);
803 if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
804 err = got_error_from_errno2("unlink", tmpidxpath);
805 if (nfetchfd != -1 && close(nfetchfd) == -1 && err == NULL)
806 err = got_error_from_errno("close");
807 if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
808 err = got_error_from_errno("close");
809 if (packfd != -1 && close(packfd) == -1 && err == NULL)
810 err = got_error_from_errno("close");
811 if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
812 err = got_error_from_errno("close");
813 for (i = 0; i < nitems(tmpfds); i++) {
814 if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
815 err = got_error_from_errno("close");
824 TAILQ_FOREACH(pe, &have_refs, entry) {
825 free((char *)pe->path);
828 got_pathlist_free(&have_refs);
829 got_ref_list_free(&my_refs);
833 TAILQ_FOREACH(pe, refs, entry) {
834 free((void *)pe->path);
837 got_pathlist_free(refs);
838 TAILQ_FOREACH(pe, symrefs, entry) {
839 free((void *)pe->path);
842 got_pathlist_free(symrefs);