Set up for better tree diff optimizations
[git/repo.git] / upload-pack.c
blob498bf50eb86f9e437dae89fc631499b73a8ec116
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"
13 static const char upload_pack_usage[] = "git-upload-pack [--strict] [--timeout=nn] <dir>";
15 /* bits #0..7 in revision.h, #8..10 in commit.c */
16 #define THEY_HAVE (1u << 11)
17 #define OUR_REF (1u << 12)
18 #define WANTED (1u << 13)
19 #define COMMON_KNOWN (1u << 14)
20 #define REACHABLE (1u << 15)
22 #define SHALLOW (1u << 16)
23 #define NOT_SHALLOW (1u << 17)
24 #define CLIENT_SHALLOW (1u << 18)
26 static unsigned long oldest_have;
28 static int multi_ack, nr_our_refs;
29 static int use_thin_pack, use_ofs_delta, no_progress;
30 static struct object_array have_obj;
31 static struct object_array want_obj;
32 static unsigned int timeout;
33 /* 0 for no sideband,
34 * otherwise maximum packet size (up to 65520 bytes).
36 static int use_sideband;
38 static void reset_timeout(void)
40 alarm(timeout);
43 static int strip(char *line, int len)
45 if (len && line[len-1] == '\n')
46 line[--len] = 0;
47 return len;
50 static ssize_t send_client_data(int fd, const char *data, ssize_t sz)
52 if (use_sideband)
53 return send_sideband(1, fd, data, sz, use_sideband);
54 if (fd == 3)
55 /* emergency quit */
56 fd = 2;
57 if (fd == 2) {
58 /* XXX: are we happy to lose stuff here? */
59 xwrite(fd, data, sz);
60 return sz;
62 return safe_write(fd, data, sz);
65 FILE *pack_pipe = NULL;
66 static void show_commit(struct commit *commit)
68 if (commit->object.flags & BOUNDARY)
69 fputc('-', pack_pipe);
70 if (fputs(sha1_to_hex(commit->object.sha1), pack_pipe) < 0)
71 die("broken output pipe");
72 fputc('\n', pack_pipe);
73 fflush(pack_pipe);
74 free(commit->buffer);
75 commit->buffer = NULL;
78 static void show_object(struct object_array_entry *p)
80 /* An object with name "foo\n0000000..." can be used to
81 * confuse downstream git-pack-objects very badly.
83 const char *ep = strchr(p->name, '\n');
84 if (ep) {
85 fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(p->item->sha1),
86 (int) (ep - p->name),
87 p->name);
89 else
90 fprintf(pack_pipe, "%s %s\n",
91 sha1_to_hex(p->item->sha1), p->name);
94 static void show_edge(struct commit *commit)
96 fprintf(pack_pipe, "-%s\n", sha1_to_hex(commit->object.sha1));
99 static void create_pack_file(void)
101 /* Pipes between rev-list to pack-objects, pack-objects to us
102 * and pack-objects error stream for progress bar.
104 int lp_pipe[2], pu_pipe[2], pe_pipe[2];
105 pid_t pid_rev_list, pid_pack_objects;
106 int create_full_pack = (nr_our_refs == want_obj.nr && !have_obj.nr);
107 char data[8193], progress[128];
108 char abort_msg[] = "aborting due to possible repository "
109 "corruption on the remote side.";
110 int buffered = -1;
112 if (pipe(lp_pipe) < 0)
113 die("git-upload-pack: unable to create pipe");
114 pid_rev_list = fork();
115 if (pid_rev_list < 0)
116 die("git-upload-pack: unable to fork git-rev-list");
118 if (!pid_rev_list) {
119 int i;
120 struct rev_info revs;
122 pack_pipe = fdopen(lp_pipe[1], "w");
124 if (create_full_pack)
125 use_thin_pack = 0; /* no point doing it */
126 init_revisions(&revs, NULL);
127 revs.tag_objects = 1;
128 revs.tree_objects = 1;
129 revs.blob_objects = 1;
130 if (use_thin_pack)
131 revs.edge_hint = 1;
133 if (create_full_pack) {
134 const char *args[] = {"rev-list", "--all", NULL};
135 setup_revisions(2, args, &revs, NULL);
136 } else {
137 for (i = 0; i < want_obj.nr; i++) {
138 struct object *o = want_obj.objects[i].item;
139 /* why??? */
140 o->flags &= ~UNINTERESTING;
141 add_pending_object(&revs, o, NULL);
143 for (i = 0; i < have_obj.nr; i++) {
144 struct object *o = have_obj.objects[i].item;
145 o->flags |= UNINTERESTING;
146 add_pending_object(&revs, o, NULL);
148 setup_revisions(0, NULL, &revs, NULL);
150 prepare_revision_walk(&revs);
151 mark_edges_uninteresting(revs.commits, &revs, show_edge);
152 traverse_commit_list(&revs, show_commit, show_object);
153 exit(0);
156 if (pipe(pu_pipe) < 0)
157 die("git-upload-pack: unable to create pipe");
158 if (pipe(pe_pipe) < 0)
159 die("git-upload-pack: unable to create pipe");
160 pid_pack_objects = fork();
161 if (pid_pack_objects < 0) {
162 /* daemon sets things up to ignore TERM */
163 kill(pid_rev_list, SIGKILL);
164 die("git-upload-pack: unable to fork git-pack-objects");
166 if (!pid_pack_objects) {
167 const char *argv[10];
168 int i = 0;
170 dup2(lp_pipe[0], 0);
171 dup2(pu_pipe[1], 1);
172 dup2(pe_pipe[1], 2);
174 close(lp_pipe[0]);
175 close(lp_pipe[1]);
176 close(pu_pipe[0]);
177 close(pu_pipe[1]);
178 close(pe_pipe[0]);
179 close(pe_pipe[1]);
181 argv[i++] = "pack-objects";
182 argv[i++] = "--stdout";
183 if (!no_progress)
184 argv[i++] = "--progress";
185 if (use_ofs_delta)
186 argv[i++] = "--delta-base-offset";
187 argv[i++] = NULL;
189 execv_git_cmd(argv);
190 kill(pid_rev_list, SIGKILL);
191 die("git-upload-pack: unable to exec git-pack-objects");
194 close(lp_pipe[0]);
195 close(lp_pipe[1]);
197 /* We read from pe_pipe[0] to capture stderr output for
198 * progress bar, and pu_pipe[0] to capture the pack data.
200 close(pe_pipe[1]);
201 close(pu_pipe[1]);
203 while (1) {
204 const char *who;
205 struct pollfd pfd[2];
206 pid_t pid;
207 int status;
208 ssize_t sz;
209 int pe, pu, pollsize;
211 reset_timeout();
213 pollsize = 0;
214 pe = pu = -1;
216 if (0 <= pu_pipe[0]) {
217 pfd[pollsize].fd = pu_pipe[0];
218 pfd[pollsize].events = POLLIN;
219 pu = pollsize;
220 pollsize++;
222 if (0 <= pe_pipe[0]) {
223 pfd[pollsize].fd = pe_pipe[0];
224 pfd[pollsize].events = POLLIN;
225 pe = pollsize;
226 pollsize++;
229 if (pollsize) {
230 if (poll(pfd, pollsize, -1) < 0) {
231 if (errno != EINTR) {
232 error("poll failed, resuming: %s",
233 strerror(errno));
234 sleep(1);
236 continue;
238 if (0 <= pu && (pfd[pu].revents & (POLLIN|POLLHUP))) {
239 /* Data ready; we keep the last byte
240 * to ourselves in case we detect
241 * broken rev-list, so that we can
242 * leave the stream corrupted. This
243 * is unfortunate -- unpack-objects
244 * would happily accept a valid pack
245 * data with trailing garbage, so
246 * appending garbage after we pass all
247 * the pack data is not good enough to
248 * signal breakage to downstream.
250 char *cp = data;
251 ssize_t outsz = 0;
252 if (0 <= buffered) {
253 *cp++ = buffered;
254 outsz++;
256 sz = xread(pu_pipe[0], cp,
257 sizeof(data) - outsz);
258 if (0 < sz)
260 else if (sz == 0) {
261 close(pu_pipe[0]);
262 pu_pipe[0] = -1;
264 else
265 goto fail;
266 sz += outsz;
267 if (1 < sz) {
268 buffered = data[sz-1] & 0xFF;
269 sz--;
271 else
272 buffered = -1;
273 sz = send_client_data(1, data, sz);
274 if (sz < 0)
275 goto fail;
277 if (0 <= pe && (pfd[pe].revents & (POLLIN|POLLHUP))) {
278 /* Status ready; we ship that in the side-band
279 * or dump to the standard error.
281 sz = xread(pe_pipe[0], progress,
282 sizeof(progress));
283 if (0 < sz)
284 send_client_data(2, progress, sz);
285 else if (sz == 0) {
286 close(pe_pipe[0]);
287 pe_pipe[0] = -1;
289 else
290 goto fail;
294 /* See if the children are still there */
295 if (pid_rev_list || pid_pack_objects) {
296 pid = waitpid(-1, &status, WNOHANG);
297 if (!pid)
298 continue;
299 who = ((pid == pid_rev_list) ? "git-rev-list" :
300 (pid == pid_pack_objects) ? "git-pack-objects" :
301 NULL);
302 if (!who) {
303 if (pid < 0) {
304 error("git-upload-pack: %s",
305 strerror(errno));
306 goto fail;
308 error("git-upload-pack: we weren't "
309 "waiting for %d", pid);
310 continue;
312 if (!WIFEXITED(status) || WEXITSTATUS(status) > 0) {
313 error("git-upload-pack: %s died with error.",
314 who);
315 goto fail;
317 if (pid == pid_rev_list)
318 pid_rev_list = 0;
319 if (pid == pid_pack_objects)
320 pid_pack_objects = 0;
321 if (pid_rev_list || pid_pack_objects)
322 continue;
325 /* both died happily */
326 if (pollsize)
327 continue;
329 /* flush the data */
330 if (0 <= buffered) {
331 data[0] = buffered;
332 sz = send_client_data(1, data, 1);
333 if (sz < 0)
334 goto fail;
335 fprintf(stderr, "flushed.\n");
337 if (use_sideband)
338 packet_flush(1);
339 return;
341 fail:
342 if (pid_pack_objects)
343 kill(pid_pack_objects, SIGKILL);
344 if (pid_rev_list)
345 kill(pid_rev_list, SIGKILL);
346 send_client_data(3, abort_msg, sizeof(abort_msg));
347 die("git-upload-pack: %s", abort_msg);
350 static int got_sha1(char *hex, unsigned char *sha1)
352 struct object *o;
353 int we_knew_they_have = 0;
355 if (get_sha1_hex(hex, sha1))
356 die("git-upload-pack: expected SHA1 object, got '%s'", hex);
357 if (!has_sha1_file(sha1))
358 return -1;
360 o = lookup_object(sha1);
361 if (!(o && o->parsed))
362 o = parse_object(sha1);
363 if (!o)
364 die("oops (%s)", sha1_to_hex(sha1));
365 if (o->type == OBJ_COMMIT) {
366 struct commit_list *parents;
367 struct commit *commit = (struct commit *)o;
368 if (o->flags & THEY_HAVE)
369 we_knew_they_have = 1;
370 else
371 o->flags |= THEY_HAVE;
372 if (!oldest_have || (commit->date < oldest_have))
373 oldest_have = commit->date;
374 for (parents = commit->parents;
375 parents;
376 parents = parents->next)
377 parents->item->object.flags |= THEY_HAVE;
379 if (!we_knew_they_have) {
380 add_object_array(o, NULL, &have_obj);
381 return 1;
383 return 0;
386 static int reachable(struct commit *want)
388 struct commit_list *work = NULL;
390 insert_by_date(want, &work);
391 while (work) {
392 struct commit_list *list = work->next;
393 struct commit *commit = work->item;
394 free(work);
395 work = list;
397 if (commit->object.flags & THEY_HAVE) {
398 want->object.flags |= COMMON_KNOWN;
399 break;
401 if (!commit->object.parsed)
402 parse_object(commit->object.sha1);
403 if (commit->object.flags & REACHABLE)
404 continue;
405 commit->object.flags |= REACHABLE;
406 if (commit->date < oldest_have)
407 continue;
408 for (list = commit->parents; list; list = list->next) {
409 struct commit *parent = list->item;
410 if (!(parent->object.flags & REACHABLE))
411 insert_by_date(parent, &work);
414 want->object.flags |= REACHABLE;
415 clear_commit_marks(want, REACHABLE);
416 free_commit_list(work);
417 return (want->object.flags & COMMON_KNOWN);
420 static int ok_to_give_up(void)
422 int i;
424 if (!have_obj.nr)
425 return 0;
427 for (i = 0; i < want_obj.nr; i++) {
428 struct object *want = want_obj.objects[i].item;
430 if (want->flags & COMMON_KNOWN)
431 continue;
432 want = deref_tag(want, "a want line", 0);
433 if (!want || want->type != OBJ_COMMIT) {
434 /* no way to tell if this is reachable by
435 * looking at the ancestry chain alone, so
436 * leave a note to ourselves not to worry about
437 * this object anymore.
439 want_obj.objects[i].item->flags |= COMMON_KNOWN;
440 continue;
442 if (!reachable((struct commit *)want))
443 return 0;
445 return 1;
448 static int get_common_commits(void)
450 static char line[1000];
451 unsigned char sha1[20];
452 char hex[41], last_hex[41];
453 int len;
455 track_object_refs = 0;
456 save_commit_buffer = 0;
458 for(;;) {
459 len = packet_read_line(0, line, sizeof(line));
460 reset_timeout();
462 if (!len) {
463 if (have_obj.nr == 0 || multi_ack)
464 packet_write(1, "NAK\n");
465 continue;
467 len = strip(line, len);
468 if (!prefixcmp(line, "have ")) {
469 switch (got_sha1(line+5, sha1)) {
470 case -1: /* they have what we do not */
471 if (multi_ack && ok_to_give_up())
472 packet_write(1, "ACK %s continue\n",
473 sha1_to_hex(sha1));
474 break;
475 default:
476 memcpy(hex, sha1_to_hex(sha1), 41);
477 if (multi_ack) {
478 const char *msg = "ACK %s continue\n";
479 packet_write(1, msg, hex);
480 memcpy(last_hex, hex, 41);
482 else if (have_obj.nr == 1)
483 packet_write(1, "ACK %s\n", hex);
484 break;
486 continue;
488 if (!strcmp(line, "done")) {
489 if (have_obj.nr > 0) {
490 if (multi_ack)
491 packet_write(1, "ACK %s\n", last_hex);
492 return 0;
494 packet_write(1, "NAK\n");
495 return -1;
497 die("git-upload-pack: expected SHA1 list, got '%s'", line);
501 static void receive_needs(void)
503 struct object_array shallows = {0, 0, NULL};
504 static char line[1000];
505 int len, depth = 0;
507 for (;;) {
508 struct object *o;
509 unsigned char sha1_buf[20];
510 len = packet_read_line(0, line, sizeof(line));
511 reset_timeout();
512 if (!len)
513 break;
515 if (!prefixcmp(line, "shallow ")) {
516 unsigned char sha1[20];
517 struct object *object;
518 use_thin_pack = 0;
519 if (get_sha1(line + 8, sha1))
520 die("invalid shallow line: %s", line);
521 object = parse_object(sha1);
522 if (!object)
523 die("did not find object for %s", line);
524 object->flags |= CLIENT_SHALLOW;
525 add_object_array(object, NULL, &shallows);
526 continue;
528 if (!prefixcmp(line, "deepen ")) {
529 char *end;
530 use_thin_pack = 0;
531 depth = strtol(line + 7, &end, 0);
532 if (end == line + 7 || depth <= 0)
533 die("Invalid deepen: %s", line);
534 continue;
536 if (prefixcmp(line, "want ") ||
537 get_sha1_hex(line+5, sha1_buf))
538 die("git-upload-pack: protocol error, "
539 "expected to get sha, not '%s'", line);
540 if (strstr(line+45, "multi_ack"))
541 multi_ack = 1;
542 if (strstr(line+45, "thin-pack"))
543 use_thin_pack = 1;
544 if (strstr(line+45, "ofs-delta"))
545 use_ofs_delta = 1;
546 if (strstr(line+45, "side-band-64k"))
547 use_sideband = LARGE_PACKET_MAX;
548 else if (strstr(line+45, "side-band"))
549 use_sideband = DEFAULT_PACKET_MAX;
550 if (strstr(line+45, "no-progress"))
551 no_progress = 1;
553 /* We have sent all our refs already, and the other end
554 * should have chosen out of them; otherwise they are
555 * asking for nonsense.
557 * Hmph. We may later want to allow "want" line that
558 * asks for something like "master~10" (symbolic)...
559 * would it make sense? I don't know.
561 o = lookup_object(sha1_buf);
562 if (!o || !(o->flags & OUR_REF))
563 die("git-upload-pack: not our ref %s", line+5);
564 if (!(o->flags & WANTED)) {
565 o->flags |= WANTED;
566 add_object_array(o, NULL, &want_obj);
569 if (depth == 0 && shallows.nr == 0)
570 return;
571 if (depth > 0) {
572 struct commit_list *result, *backup;
573 int i;
574 backup = result = get_shallow_commits(&want_obj, depth,
575 SHALLOW, NOT_SHALLOW);
576 while (result) {
577 struct object *object = &result->item->object;
578 if (!(object->flags & (CLIENT_SHALLOW|NOT_SHALLOW))) {
579 packet_write(1, "shallow %s",
580 sha1_to_hex(object->sha1));
581 register_shallow(object->sha1);
583 result = result->next;
585 free_commit_list(backup);
586 for (i = 0; i < shallows.nr; i++) {
587 struct object *object = shallows.objects[i].item;
588 if (object->flags & NOT_SHALLOW) {
589 struct commit_list *parents;
590 packet_write(1, "unshallow %s",
591 sha1_to_hex(object->sha1));
592 object->flags &= ~CLIENT_SHALLOW;
593 /* make sure the real parents are parsed */
594 unregister_shallow(object->sha1);
595 object->parsed = 0;
596 parse_commit((struct commit *)object);
597 parents = ((struct commit *)object)->parents;
598 while (parents) {
599 add_object_array(&parents->item->object,
600 NULL, &want_obj);
601 parents = parents->next;
604 /* make sure commit traversal conforms to client */
605 register_shallow(object->sha1);
607 packet_flush(1);
608 } else
609 if (shallows.nr > 0) {
610 int i;
611 for (i = 0; i < shallows.nr; i++)
612 register_shallow(shallows.objects[i].item->sha1);
614 free(shallows.objects);
617 static int send_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
619 static const char *capabilities = "multi_ack thin-pack side-band"
620 " side-band-64k ofs-delta shallow no-progress";
621 struct object *o = parse_object(sha1);
623 if (!o)
624 die("git-upload-pack: cannot find object %s:", sha1_to_hex(sha1));
626 if (capabilities)
627 packet_write(1, "%s %s%c%s\n", sha1_to_hex(sha1), refname,
628 0, capabilities);
629 else
630 packet_write(1, "%s %s\n", sha1_to_hex(sha1), refname);
631 capabilities = NULL;
632 if (!(o->flags & OUR_REF)) {
633 o->flags |= OUR_REF;
634 nr_our_refs++;
636 if (o->type == OBJ_TAG) {
637 o = deref_tag(o, refname, 0);
638 packet_write(1, "%s %s^{}\n", sha1_to_hex(o->sha1), refname);
640 return 0;
643 static void upload_pack(void)
645 reset_timeout();
646 head_ref(send_ref, NULL);
647 for_each_ref(send_ref, NULL);
648 packet_flush(1);
649 receive_needs();
650 if (want_obj.nr) {
651 get_common_commits();
652 create_pack_file();
656 int main(int argc, char **argv)
658 char *dir;
659 int i;
660 int strict = 0;
662 for (i = 1; i < argc; i++) {
663 char *arg = argv[i];
665 if (arg[0] != '-')
666 break;
667 if (!strcmp(arg, "--strict")) {
668 strict = 1;
669 continue;
671 if (!prefixcmp(arg, "--timeout=")) {
672 timeout = atoi(arg+10);
673 continue;
675 if (!strcmp(arg, "--")) {
676 i++;
677 break;
681 if (i != argc-1)
682 usage(upload_pack_usage);
683 dir = argv[i];
685 if (!enter_repo(dir, strict))
686 die("'%s': unable to chdir or not a git archive", dir);
687 if (is_repository_shallow())
688 die("attempt to fetch/clone from a shallow repository");
689 upload_pack();
690 return 0;