fetch-pack: Use a strbuf to compose the want list
[git.git] / upload-pack.c
blob38ddac2e8607f4cc6a801eeb49ba755dd89ee81c
1 #include "cache.h"
2 #include "refs.h"
3 #include "pkt-line.h"
4 #include "sideband.h"
5 #include "tag.h"
6 #include "object.h"
7 #include "commit.h"
8 #include "exec_cmd.h"
9 #include "diff.h"
10 #include "revision.h"
11 #include "list-objects.h"
12 #include "run-command.h"
14 static const char upload_pack_usage[] = "git upload-pack [--strict] [--timeout=nn] <dir>";
16 /* bits #0..7 in revision.h, #8..10 in commit.c */
17 #define THEY_HAVE (1u << 11)
18 #define OUR_REF (1u << 12)
19 #define WANTED (1u << 13)
20 #define COMMON_KNOWN (1u << 14)
21 #define REACHABLE (1u << 15)
23 #define SHALLOW (1u << 16)
24 #define NOT_SHALLOW (1u << 17)
25 #define CLIENT_SHALLOW (1u << 18)
27 static unsigned long oldest_have;
29 static int multi_ack, nr_our_refs;
30 static int use_thin_pack, use_ofs_delta, use_include_tag;
31 static int no_progress, daemon_mode;
32 static int shallow_nr;
33 static struct object_array have_obj;
34 static struct object_array want_obj;
35 static struct object_array extra_edge_obj;
36 static unsigned int timeout;
37 /* 0 for no sideband,
38 * otherwise maximum packet size (up to 65520 bytes).
40 static int use_sideband;
41 static int debug_fd;
43 static void reset_timeout(void)
45 alarm(timeout);
48 static int strip(char *line, int len)
50 if (len && line[len-1] == '\n')
51 line[--len] = 0;
52 return len;
55 static ssize_t send_client_data(int fd, const char *data, ssize_t sz)
57 if (use_sideband)
58 return send_sideband(1, fd, data, sz, use_sideband);
59 if (fd == 3)
60 /* emergency quit */
61 fd = 2;
62 if (fd == 2) {
63 /* XXX: are we happy to lose stuff here? */
64 xwrite(fd, data, sz);
65 return sz;
67 return safe_write(fd, data, sz);
70 static FILE *pack_pipe = NULL;
71 static void show_commit(struct commit *commit, void *data)
73 if (commit->object.flags & BOUNDARY)
74 fputc('-', pack_pipe);
75 if (fputs(sha1_to_hex(commit->object.sha1), pack_pipe) < 0)
76 die("broken output pipe");
77 fputc('\n', pack_pipe);
78 fflush(pack_pipe);
79 free(commit->buffer);
80 commit->buffer = NULL;
83 static void show_object(struct object *obj, const struct name_path *path, const char *component)
85 /* An object with name "foo\n0000000..." can be used to
86 * confuse downstream git-pack-objects very badly.
88 const char *name = path_name(path, component);
89 const char *ep = strchr(name, '\n');
90 if (ep) {
91 fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(obj->sha1),
92 (int) (ep - name),
93 name);
95 else
96 fprintf(pack_pipe, "%s %s\n",
97 sha1_to_hex(obj->sha1), name);
98 free((char *)name);
101 static void show_edge(struct commit *commit)
103 fprintf(pack_pipe, "-%s\n", sha1_to_hex(commit->object.sha1));
106 static int do_rev_list(int fd, void *create_full_pack)
108 int i;
109 struct rev_info revs;
111 pack_pipe = xfdopen(fd, "w");
112 init_revisions(&revs, NULL);
113 revs.tag_objects = 1;
114 revs.tree_objects = 1;
115 revs.blob_objects = 1;
116 if (use_thin_pack)
117 revs.edge_hint = 1;
119 if (create_full_pack) {
120 const char *args[] = {"rev-list", "--all", NULL};
121 setup_revisions(2, args, &revs, NULL);
122 } else {
123 for (i = 0; i < want_obj.nr; i++) {
124 struct object *o = want_obj.objects[i].item;
125 /* why??? */
126 o->flags &= ~UNINTERESTING;
127 add_pending_object(&revs, o, NULL);
129 for (i = 0; i < have_obj.nr; i++) {
130 struct object *o = have_obj.objects[i].item;
131 o->flags |= UNINTERESTING;
132 add_pending_object(&revs, o, NULL);
134 setup_revisions(0, NULL, &revs, NULL);
136 if (prepare_revision_walk(&revs))
137 die("revision walk setup failed");
138 mark_edges_uninteresting(revs.commits, &revs, show_edge);
139 if (use_thin_pack)
140 for (i = 0; i < extra_edge_obj.nr; i++)
141 fprintf(pack_pipe, "-%s\n", sha1_to_hex(
142 extra_edge_obj.objects[i].item->sha1));
143 traverse_commit_list(&revs, show_commit, show_object, NULL);
144 fflush(pack_pipe);
145 fclose(pack_pipe);
146 return 0;
149 static int feed_msg_to_hook(int fd, const char *fmt, ...)
151 int cnt;
152 char buf[1024];
153 va_list params;
155 va_start(params, fmt);
156 cnt = vsprintf(buf, fmt, params);
157 va_end(params);
158 return write_in_full(fd, buf, cnt) != cnt;
161 static int feed_obj_to_hook(const char *label, struct object_array *oa, int i, int fd)
163 return feed_msg_to_hook(fd, "%s %s\n", label,
164 sha1_to_hex(oa->objects[i].item->sha1));
167 static int run_post_upload_pack_hook(size_t total, struct timeval *tv)
169 const char *argv[2];
170 struct child_process proc;
171 int err, i;
173 argv[0] = "hooks/post-upload-pack";
174 argv[1] = NULL;
176 if (access(argv[0], X_OK) < 0)
177 return 0;
179 memset(&proc, 0, sizeof(proc));
180 proc.argv = argv;
181 proc.in = -1;
182 proc.stdout_to_stderr = 1;
183 err = start_command(&proc);
184 if (err)
185 return err;
186 for (i = 0; !err && i < want_obj.nr; i++)
187 err |= feed_obj_to_hook("want", &want_obj, i, proc.in);
188 for (i = 0; !err && i < have_obj.nr; i++)
189 err |= feed_obj_to_hook("have", &have_obj, i, proc.in);
190 if (!err)
191 err |= feed_msg_to_hook(proc.in, "time %ld.%06ld\n",
192 (long)tv->tv_sec, (long)tv->tv_usec);
193 if (!err)
194 err |= feed_msg_to_hook(proc.in, "size %ld\n", (long)total);
195 if (!err)
196 err |= feed_msg_to_hook(proc.in, "kind %s\n",
197 (nr_our_refs == want_obj.nr && !have_obj.nr)
198 ? "clone" : "fetch");
199 if (close(proc.in))
200 err = 1;
201 if (finish_command(&proc))
202 err = 1;
203 return err;
206 static void create_pack_file(void)
208 struct timeval start_tv, tv;
209 struct async rev_list;
210 struct child_process pack_objects;
211 int create_full_pack = (nr_our_refs == want_obj.nr && !have_obj.nr);
212 char data[8193], progress[128];
213 char abort_msg[] = "aborting due to possible repository "
214 "corruption on the remote side.";
215 int buffered = -1;
216 ssize_t sz, total_sz;
217 const char *argv[10];
218 int arg = 0;
220 gettimeofday(&start_tv, NULL);
221 total_sz = 0;
222 if (shallow_nr) {
223 rev_list.proc = do_rev_list;
224 rev_list.data = 0;
225 if (start_async(&rev_list))
226 die("git upload-pack: unable to fork git-rev-list");
227 argv[arg++] = "pack-objects";
228 } else {
229 argv[arg++] = "pack-objects";
230 argv[arg++] = "--revs";
231 if (create_full_pack)
232 argv[arg++] = "--all";
233 else if (use_thin_pack)
234 argv[arg++] = "--thin";
237 argv[arg++] = "--stdout";
238 if (!no_progress)
239 argv[arg++] = "--progress";
240 if (use_ofs_delta)
241 argv[arg++] = "--delta-base-offset";
242 if (use_include_tag)
243 argv[arg++] = "--include-tag";
244 argv[arg++] = NULL;
246 memset(&pack_objects, 0, sizeof(pack_objects));
247 pack_objects.in = shallow_nr ? rev_list.out : -1;
248 pack_objects.out = -1;
249 pack_objects.err = -1;
250 pack_objects.git_cmd = 1;
251 pack_objects.argv = argv;
253 if (start_command(&pack_objects))
254 die("git upload-pack: unable to fork git-pack-objects");
256 /* pass on revisions we (don't) want */
257 if (!shallow_nr) {
258 FILE *pipe_fd = xfdopen(pack_objects.in, "w");
259 if (!create_full_pack) {
260 int i;
261 for (i = 0; i < want_obj.nr; i++)
262 fprintf(pipe_fd, "%s\n", sha1_to_hex(want_obj.objects[i].item->sha1));
263 fprintf(pipe_fd, "--not\n");
264 for (i = 0; i < have_obj.nr; i++)
265 fprintf(pipe_fd, "%s\n", sha1_to_hex(have_obj.objects[i].item->sha1));
268 fprintf(pipe_fd, "\n");
269 fflush(pipe_fd);
270 fclose(pipe_fd);
274 /* We read from pack_objects.err to capture stderr output for
275 * progress bar, and pack_objects.out to capture the pack data.
278 while (1) {
279 struct pollfd pfd[2];
280 int pe, pu, pollsize;
282 reset_timeout();
284 pollsize = 0;
285 pe = pu = -1;
287 if (0 <= pack_objects.out) {
288 pfd[pollsize].fd = pack_objects.out;
289 pfd[pollsize].events = POLLIN;
290 pu = pollsize;
291 pollsize++;
293 if (0 <= pack_objects.err) {
294 pfd[pollsize].fd = pack_objects.err;
295 pfd[pollsize].events = POLLIN;
296 pe = pollsize;
297 pollsize++;
300 if (!pollsize)
301 break;
303 if (poll(pfd, pollsize, -1) < 0) {
304 if (errno != EINTR) {
305 error("poll failed, resuming: %s",
306 strerror(errno));
307 sleep(1);
309 continue;
311 if (0 <= pu && (pfd[pu].revents & (POLLIN|POLLHUP))) {
312 /* Data ready; we keep the last byte to ourselves
313 * in case we detect broken rev-list, so that we
314 * can leave the stream corrupted. This is
315 * unfortunate -- unpack-objects would happily
316 * accept a valid packdata with trailing garbage,
317 * so appending garbage after we pass all the
318 * pack data is not good enough to signal
319 * breakage to downstream.
321 char *cp = data;
322 ssize_t outsz = 0;
323 if (0 <= buffered) {
324 *cp++ = buffered;
325 outsz++;
327 sz = xread(pack_objects.out, cp,
328 sizeof(data) - outsz);
329 if (0 < sz)
330 total_sz += sz;
331 else if (sz == 0) {
332 close(pack_objects.out);
333 pack_objects.out = -1;
335 else
336 goto fail;
337 sz += outsz;
338 if (1 < sz) {
339 buffered = data[sz-1] & 0xFF;
340 sz--;
342 else
343 buffered = -1;
344 sz = send_client_data(1, data, sz);
345 if (sz < 0)
346 goto fail;
348 if (0 <= pe && (pfd[pe].revents & (POLLIN|POLLHUP))) {
349 /* Status ready; we ship that in the side-band
350 * or dump to the standard error.
352 sz = xread(pack_objects.err, progress,
353 sizeof(progress));
354 if (0 < sz)
355 send_client_data(2, progress, sz);
356 else if (sz == 0) {
357 close(pack_objects.err);
358 pack_objects.err = -1;
360 else
361 goto fail;
365 if (finish_command(&pack_objects)) {
366 error("git upload-pack: git-pack-objects died with error.");
367 goto fail;
369 if (shallow_nr && finish_async(&rev_list))
370 goto fail; /* error was already reported */
372 /* flush the data */
373 if (0 <= buffered) {
374 data[0] = buffered;
375 sz = send_client_data(1, data, 1);
376 if (sz < 0)
377 goto fail;
378 fprintf(stderr, "flushed.\n");
380 if (use_sideband)
381 packet_flush(1);
383 gettimeofday(&tv, NULL);
384 tv.tv_sec -= start_tv.tv_sec;
385 if (tv.tv_usec < start_tv.tv_usec) {
386 tv.tv_sec--;
387 tv.tv_usec += 1000000;
389 tv.tv_usec -= start_tv.tv_usec;
390 if (run_post_upload_pack_hook(total_sz, &tv))
391 warning("post-upload-hook failed");
392 return;
394 fail:
395 send_client_data(3, abort_msg, sizeof(abort_msg));
396 die("git upload-pack: %s", abort_msg);
399 static int got_sha1(char *hex, unsigned char *sha1)
401 struct object *o;
402 int we_knew_they_have = 0;
404 if (get_sha1_hex(hex, sha1))
405 die("git upload-pack: expected SHA1 object, got '%s'", hex);
406 if (!has_sha1_file(sha1))
407 return -1;
409 o = lookup_object(sha1);
410 if (!(o && o->parsed))
411 o = parse_object(sha1);
412 if (!o)
413 die("oops (%s)", sha1_to_hex(sha1));
414 if (o->type == OBJ_COMMIT) {
415 struct commit_list *parents;
416 struct commit *commit = (struct commit *)o;
417 if (o->flags & THEY_HAVE)
418 we_knew_they_have = 1;
419 else
420 o->flags |= THEY_HAVE;
421 if (!oldest_have || (commit->date < oldest_have))
422 oldest_have = commit->date;
423 for (parents = commit->parents;
424 parents;
425 parents = parents->next)
426 parents->item->object.flags |= THEY_HAVE;
428 if (!we_knew_they_have) {
429 add_object_array(o, NULL, &have_obj);
430 return 1;
432 return 0;
435 static int reachable(struct commit *want)
437 struct commit_list *work = NULL;
439 insert_by_date(want, &work);
440 while (work) {
441 struct commit_list *list = work->next;
442 struct commit *commit = work->item;
443 free(work);
444 work = list;
446 if (commit->object.flags & THEY_HAVE) {
447 want->object.flags |= COMMON_KNOWN;
448 break;
450 if (!commit->object.parsed)
451 parse_object(commit->object.sha1);
452 if (commit->object.flags & REACHABLE)
453 continue;
454 commit->object.flags |= REACHABLE;
455 if (commit->date < oldest_have)
456 continue;
457 for (list = commit->parents; list; list = list->next) {
458 struct commit *parent = list->item;
459 if (!(parent->object.flags & REACHABLE))
460 insert_by_date(parent, &work);
463 want->object.flags |= REACHABLE;
464 clear_commit_marks(want, REACHABLE);
465 free_commit_list(work);
466 return (want->object.flags & COMMON_KNOWN);
469 static int ok_to_give_up(void)
471 int i;
473 if (!have_obj.nr)
474 return 0;
476 for (i = 0; i < want_obj.nr; i++) {
477 struct object *want = want_obj.objects[i].item;
479 if (want->flags & COMMON_KNOWN)
480 continue;
481 want = deref_tag(want, "a want line", 0);
482 if (!want || want->type != OBJ_COMMIT) {
483 /* no way to tell if this is reachable by
484 * looking at the ancestry chain alone, so
485 * leave a note to ourselves not to worry about
486 * this object anymore.
488 want_obj.objects[i].item->flags |= COMMON_KNOWN;
489 continue;
491 if (!reachable((struct commit *)want))
492 return 0;
494 return 1;
497 static int get_common_commits(void)
499 static char line[1000];
500 unsigned char sha1[20];
501 char hex[41], last_hex[41];
503 save_commit_buffer = 0;
505 for (;;) {
506 int len = packet_read_line(0, line, sizeof(line));
507 reset_timeout();
509 if (!len) {
510 if (have_obj.nr == 0 || multi_ack)
511 packet_write(1, "NAK\n");
512 continue;
514 strip(line, len);
515 if (!prefixcmp(line, "have ")) {
516 switch (got_sha1(line+5, sha1)) {
517 case -1: /* they have what we do not */
518 if (multi_ack && ok_to_give_up())
519 packet_write(1, "ACK %s continue\n",
520 sha1_to_hex(sha1));
521 break;
522 default:
523 memcpy(hex, sha1_to_hex(sha1), 41);
524 if (multi_ack) {
525 const char *msg = "ACK %s continue\n";
526 packet_write(1, msg, hex);
527 memcpy(last_hex, hex, 41);
529 else if (have_obj.nr == 1)
530 packet_write(1, "ACK %s\n", hex);
531 break;
533 continue;
535 if (!strcmp(line, "done")) {
536 if (have_obj.nr > 0) {
537 if (multi_ack)
538 packet_write(1, "ACK %s\n", last_hex);
539 return 0;
541 packet_write(1, "NAK\n");
542 return -1;
544 die("git upload-pack: expected SHA1 list, got '%s'", line);
548 static void receive_needs(void)
550 struct object_array shallows = {0, 0, NULL};
551 static char line[1000];
552 int len, depth = 0;
554 shallow_nr = 0;
555 if (debug_fd)
556 write_str_in_full(debug_fd, "#S\n");
557 for (;;) {
558 struct object *o;
559 unsigned char sha1_buf[20];
560 len = packet_read_line(0, line, sizeof(line));
561 reset_timeout();
562 if (!len)
563 break;
564 if (debug_fd)
565 write_in_full(debug_fd, line, len);
567 if (!prefixcmp(line, "shallow ")) {
568 unsigned char sha1[20];
569 struct object *object;
570 if (get_sha1(line + 8, sha1))
571 die("invalid shallow line: %s", line);
572 object = parse_object(sha1);
573 if (!object)
574 die("did not find object for %s", line);
575 object->flags |= CLIENT_SHALLOW;
576 add_object_array(object, NULL, &shallows);
577 continue;
579 if (!prefixcmp(line, "deepen ")) {
580 char *end;
581 depth = strtol(line + 7, &end, 0);
582 if (end == line + 7 || depth <= 0)
583 die("Invalid deepen: %s", line);
584 continue;
586 if (prefixcmp(line, "want ") ||
587 get_sha1_hex(line+5, sha1_buf))
588 die("git upload-pack: protocol error, "
589 "expected to get sha, not '%s'", line);
590 if (strstr(line+45, "multi_ack"))
591 multi_ack = 1;
592 if (strstr(line+45, "thin-pack"))
593 use_thin_pack = 1;
594 if (strstr(line+45, "ofs-delta"))
595 use_ofs_delta = 1;
596 if (strstr(line+45, "side-band-64k"))
597 use_sideband = LARGE_PACKET_MAX;
598 else if (strstr(line+45, "side-band"))
599 use_sideband = DEFAULT_PACKET_MAX;
600 if (strstr(line+45, "no-progress"))
601 no_progress = 1;
602 if (strstr(line+45, "include-tag"))
603 use_include_tag = 1;
605 /* We have sent all our refs already, and the other end
606 * should have chosen out of them; otherwise they are
607 * asking for nonsense.
609 * Hmph. We may later want to allow "want" line that
610 * asks for something like "master~10" (symbolic)...
611 * would it make sense? I don't know.
613 o = lookup_object(sha1_buf);
614 if (!o || !(o->flags & OUR_REF))
615 die("git upload-pack: not our ref %s", line+5);
616 if (!(o->flags & WANTED)) {
617 o->flags |= WANTED;
618 add_object_array(o, NULL, &want_obj);
621 if (debug_fd)
622 write_str_in_full(debug_fd, "#E\n");
624 if (!use_sideband && daemon_mode)
625 no_progress = 1;
627 if (depth == 0 && shallows.nr == 0)
628 return;
629 if (depth > 0) {
630 struct commit_list *result, *backup;
631 int i;
632 backup = result = get_shallow_commits(&want_obj, depth,
633 SHALLOW, NOT_SHALLOW);
634 while (result) {
635 struct object *object = &result->item->object;
636 if (!(object->flags & (CLIENT_SHALLOW|NOT_SHALLOW))) {
637 packet_write(1, "shallow %s",
638 sha1_to_hex(object->sha1));
639 register_shallow(object->sha1);
640 shallow_nr++;
642 result = result->next;
644 free_commit_list(backup);
645 for (i = 0; i < shallows.nr; i++) {
646 struct object *object = shallows.objects[i].item;
647 if (object->flags & NOT_SHALLOW) {
648 struct commit_list *parents;
649 packet_write(1, "unshallow %s",
650 sha1_to_hex(object->sha1));
651 object->flags &= ~CLIENT_SHALLOW;
652 /* make sure the real parents are parsed */
653 unregister_shallow(object->sha1);
654 object->parsed = 0;
655 if (parse_commit((struct commit *)object))
656 die("invalid commit");
657 parents = ((struct commit *)object)->parents;
658 while (parents) {
659 add_object_array(&parents->item->object,
660 NULL, &want_obj);
661 parents = parents->next;
663 add_object_array(object, NULL, &extra_edge_obj);
665 /* make sure commit traversal conforms to client */
666 register_shallow(object->sha1);
668 packet_flush(1);
669 } else
670 if (shallows.nr > 0) {
671 int i;
672 for (i = 0; i < shallows.nr; i++)
673 register_shallow(shallows.objects[i].item->sha1);
676 shallow_nr += shallows.nr;
677 free(shallows.objects);
680 static int send_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
682 static const char *capabilities = "multi_ack thin-pack side-band"
683 " side-band-64k ofs-delta shallow no-progress"
684 " include-tag";
685 struct object *o = parse_object(sha1);
687 if (!o)
688 die("git upload-pack: cannot find object %s:", sha1_to_hex(sha1));
690 if (capabilities)
691 packet_write(1, "%s %s%c%s\n", sha1_to_hex(sha1), refname,
692 0, capabilities);
693 else
694 packet_write(1, "%s %s\n", sha1_to_hex(sha1), refname);
695 capabilities = NULL;
696 if (!(o->flags & OUR_REF)) {
697 o->flags |= OUR_REF;
698 nr_our_refs++;
700 if (o->type == OBJ_TAG) {
701 o = deref_tag(o, refname, 0);
702 if (o)
703 packet_write(1, "%s %s^{}\n", sha1_to_hex(o->sha1), refname);
705 return 0;
708 static void upload_pack(void)
710 reset_timeout();
711 head_ref(send_ref, NULL);
712 for_each_ref(send_ref, NULL);
713 packet_flush(1);
714 receive_needs();
715 if (want_obj.nr) {
716 get_common_commits();
717 create_pack_file();
721 int main(int argc, char **argv)
723 char *dir;
724 int i;
725 int strict = 0;
727 git_extract_argv0_path(argv[0]);
728 read_replace_refs = 0;
730 for (i = 1; i < argc; i++) {
731 char *arg = argv[i];
733 if (arg[0] != '-')
734 break;
735 if (!strcmp(arg, "--strict")) {
736 strict = 1;
737 continue;
739 if (!prefixcmp(arg, "--timeout=")) {
740 timeout = atoi(arg+10);
741 daemon_mode = 1;
742 continue;
744 if (!strcmp(arg, "--")) {
745 i++;
746 break;
750 if (i != argc-1)
751 usage(upload_pack_usage);
753 setup_path();
755 dir = argv[i];
757 if (!enter_repo(dir, strict))
758 die("'%s' does not appear to be a git repository", dir);
759 if (is_repository_shallow())
760 die("attempt to fetch/clone from a shallow repository");
761 if (getenv("GIT_DEBUG_SEND_PACK"))
762 debug_fd = atoi(getenv("GIT_DEBUG_SEND_PACK"));
763 upload_pack();
764 return 0;