Do not talk about "diff" in rev-list documentation.
[git] / upload-pack.c
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"
13
14 static const char upload_pack_usage[] = "git-upload-pack [--strict] [--timeout=nn] <dir>";
15
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)
22
23 #define SHALLOW         (1u << 16)
24 #define NOT_SHALLOW     (1u << 17)
25 #define CLIENT_SHALLOW  (1u << 18)
26
27 static unsigned long oldest_have;
28
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;
32 static struct object_array have_obj;
33 static struct object_array want_obj;
34 static unsigned int timeout;
35 /* 0 for no sideband,
36  * otherwise maximum packet size (up to 65520 bytes).
37  */
38 static int use_sideband;
39 static int debug_fd;
40
41 static void reset_timeout(void)
42 {
43         alarm(timeout);
44 }
45
46 static int strip(char *line, int len)
47 {
48         if (len && line[len-1] == '\n')
49                 line[--len] = 0;
50         return len;
51 }
52
53 static ssize_t send_client_data(int fd, const char *data, ssize_t sz)
54 {
55         if (use_sideband)
56                 return send_sideband(1, fd, data, sz, use_sideband);
57         if (fd == 3)
58                 /* emergency quit */
59                 fd = 2;
60         if (fd == 2) {
61                 /* XXX: are we happy to lose stuff here? */
62                 xwrite(fd, data, sz);
63                 return sz;
64         }
65         return safe_write(fd, data, sz);
66 }
67
68 static FILE *pack_pipe = NULL;
69 static void show_commit(struct commit *commit)
70 {
71         if (commit->object.flags & BOUNDARY)
72                 fputc('-', pack_pipe);
73         if (fputs(sha1_to_hex(commit->object.sha1), pack_pipe) < 0)
74                 die("broken output pipe");
75         fputc('\n', pack_pipe);
76         fflush(pack_pipe);
77         free(commit->buffer);
78         commit->buffer = NULL;
79 }
80
81 static void show_object(struct object_array_entry *p)
82 {
83         /* An object with name "foo\n0000000..." can be used to
84          * confuse downstream git-pack-objects very badly.
85          */
86         const char *ep = strchr(p->name, '\n');
87         if (ep) {
88                 fprintf(pack_pipe, "%s %.*s\n", sha1_to_hex(p->item->sha1),
89                        (int) (ep - p->name),
90                        p->name);
91         }
92         else
93                 fprintf(pack_pipe, "%s %s\n",
94                                 sha1_to_hex(p->item->sha1), p->name);
95 }
96
97 static void show_edge(struct commit *commit)
98 {
99         fprintf(pack_pipe, "-%s\n", sha1_to_hex(commit->object.sha1));
100 }
101
102 static int do_rev_list(int fd, void *create_full_pack)
103 {
104         int i;
105         struct rev_info revs;
106
107         pack_pipe = fdopen(fd, "w");
108         if (create_full_pack)
109                 use_thin_pack = 0; /* no point doing it */
110         init_revisions(&revs, NULL);
111         revs.tag_objects = 1;
112         revs.tree_objects = 1;
113         revs.blob_objects = 1;
114         if (use_thin_pack)
115                 revs.edge_hint = 1;
116
117         if (create_full_pack) {
118                 const char *args[] = {"rev-list", "--all", NULL};
119                 setup_revisions(2, args, &revs, NULL);
120         } else {
121                 for (i = 0; i < want_obj.nr; i++) {
122                         struct object *o = want_obj.objects[i].item;
123                         /* why??? */
124                         o->flags &= ~UNINTERESTING;
125                         add_pending_object(&revs, o, NULL);
126                 }
127                 for (i = 0; i < have_obj.nr; i++) {
128                         struct object *o = have_obj.objects[i].item;
129                         o->flags |= UNINTERESTING;
130                         add_pending_object(&revs, o, NULL);
131                 }
132                 setup_revisions(0, NULL, &revs, NULL);
133         }
134         if (prepare_revision_walk(&revs))
135                 die("revision walk setup failed");
136         mark_edges_uninteresting(revs.commits, &revs, show_edge);
137         traverse_commit_list(&revs, show_commit, show_object);
138         return 0;
139 }
140
141 static void create_pack_file(void)
142 {
143         struct async rev_list;
144         struct child_process pack_objects;
145         int create_full_pack = (nr_our_refs == want_obj.nr && !have_obj.nr);
146         char data[8193], progress[128];
147         char abort_msg[] = "aborting due to possible repository "
148                 "corruption on the remote side.";
149         int buffered = -1;
150         ssize_t sz;
151         const char *argv[10];
152         int arg = 0;
153
154         rev_list.proc = do_rev_list;
155         /* .data is just a boolean: any non-NULL value will do */
156         rev_list.data = create_full_pack ? &rev_list : NULL;
157         if (start_async(&rev_list))
158                 die("git-upload-pack: unable to fork git-rev-list");
159
160         argv[arg++] = "pack-objects";
161         argv[arg++] = "--stdout";
162         if (!no_progress)
163                 argv[arg++] = "--progress";
164         if (use_ofs_delta)
165                 argv[arg++] = "--delta-base-offset";
166         if (use_include_tag)
167                 argv[arg++] = "--include-tag";
168         argv[arg++] = NULL;
169
170         memset(&pack_objects, 0, sizeof(pack_objects));
171         pack_objects.in = rev_list.out; /* start_command closes it */
172         pack_objects.out = -1;
173         pack_objects.err = -1;
174         pack_objects.git_cmd = 1;
175         pack_objects.argv = argv;
176
177         if (start_command(&pack_objects))
178                 die("git-upload-pack: unable to fork git-pack-objects");
179
180         /* We read from pack_objects.err to capture stderr output for
181          * progress bar, and pack_objects.out to capture the pack data.
182          */
183
184         while (1) {
185                 struct pollfd pfd[2];
186                 int pe, pu, pollsize;
187
188                 reset_timeout();
189
190                 pollsize = 0;
191                 pe = pu = -1;
192
193                 if (0 <= pack_objects.out) {
194                         pfd[pollsize].fd = pack_objects.out;
195                         pfd[pollsize].events = POLLIN;
196                         pu = pollsize;
197                         pollsize++;
198                 }
199                 if (0 <= pack_objects.err) {
200                         pfd[pollsize].fd = pack_objects.err;
201                         pfd[pollsize].events = POLLIN;
202                         pe = pollsize;
203                         pollsize++;
204                 }
205
206                 if (!pollsize)
207                         break;
208
209                 if (poll(pfd, pollsize, -1) < 0) {
210                         if (errno != EINTR) {
211                                 error("poll failed, resuming: %s",
212                                       strerror(errno));
213                                 sleep(1);
214                         }
215                         continue;
216                 }
217                 if (0 <= pu && (pfd[pu].revents & (POLLIN|POLLHUP))) {
218                         /* Data ready; we keep the last byte to ourselves
219                          * in case we detect broken rev-list, so that we
220                          * can leave the stream corrupted.  This is
221                          * unfortunate -- unpack-objects would happily
222                          * accept a valid packdata with trailing garbage,
223                          * so appending garbage after we pass all the
224                          * pack data is not good enough to signal
225                          * breakage to downstream.
226                          */
227                         char *cp = data;
228                         ssize_t outsz = 0;
229                         if (0 <= buffered) {
230                                 *cp++ = buffered;
231                                 outsz++;
232                         }
233                         sz = xread(pack_objects.out, cp,
234                                   sizeof(data) - outsz);
235                         if (0 < sz)
236                                         ;
237                         else if (sz == 0) {
238                                 close(pack_objects.out);
239                                 pack_objects.out = -1;
240                         }
241                         else
242                                 goto fail;
243                         sz += outsz;
244                         if (1 < sz) {
245                                 buffered = data[sz-1] & 0xFF;
246                                 sz--;
247                         }
248                         else
249                                 buffered = -1;
250                         sz = send_client_data(1, data, sz);
251                         if (sz < 0)
252                                 goto fail;
253                 }
254                 if (0 <= pe && (pfd[pe].revents & (POLLIN|POLLHUP))) {
255                         /* Status ready; we ship that in the side-band
256                          * or dump to the standard error.
257                          */
258                         sz = xread(pack_objects.err, progress,
259                                   sizeof(progress));
260                         if (0 < sz)
261                                 send_client_data(2, progress, sz);
262                         else if (sz == 0) {
263                                 close(pack_objects.err);
264                                 pack_objects.err = -1;
265                         }
266                         else
267                                 goto fail;
268                 }
269         }
270
271         if (finish_command(&pack_objects)) {
272                 error("git-upload-pack: git-pack-objects died with error.");
273                 goto fail;
274         }
275         if (finish_async(&rev_list))
276                 goto fail;      /* error was already reported */
277
278         /* flush the data */
279         if (0 <= buffered) {
280                 data[0] = buffered;
281                 sz = send_client_data(1, data, 1);
282                 if (sz < 0)
283                         goto fail;
284                 fprintf(stderr, "flushed.\n");
285         }
286         if (use_sideband)
287                 packet_flush(1);
288         return;
289
290  fail:
291         send_client_data(3, abort_msg, sizeof(abort_msg));
292         die("git-upload-pack: %s", abort_msg);
293 }
294
295 static int got_sha1(char *hex, unsigned char *sha1)
296 {
297         struct object *o;
298         int we_knew_they_have = 0;
299
300         if (get_sha1_hex(hex, sha1))
301                 die("git-upload-pack: expected SHA1 object, got '%s'", hex);
302         if (!has_sha1_file(sha1))
303                 return -1;
304
305         o = lookup_object(sha1);
306         if (!(o && o->parsed))
307                 o = parse_object(sha1);
308         if (!o)
309                 die("oops (%s)", sha1_to_hex(sha1));
310         if (o->type == OBJ_COMMIT) {
311                 struct commit_list *parents;
312                 struct commit *commit = (struct commit *)o;
313                 if (o->flags & THEY_HAVE)
314                         we_knew_they_have = 1;
315                 else
316                         o->flags |= THEY_HAVE;
317                 if (!oldest_have || (commit->date < oldest_have))
318                         oldest_have = commit->date;
319                 for (parents = commit->parents;
320                      parents;
321                      parents = parents->next)
322                         parents->item->object.flags |= THEY_HAVE;
323         }
324         if (!we_knew_they_have) {
325                 add_object_array(o, NULL, &have_obj);
326                 return 1;
327         }
328         return 0;
329 }
330
331 static int reachable(struct commit *want)
332 {
333         struct commit_list *work = NULL;
334
335         insert_by_date(want, &work);
336         while (work) {
337                 struct commit_list *list = work->next;
338                 struct commit *commit = work->item;
339                 free(work);
340                 work = list;
341
342                 if (commit->object.flags & THEY_HAVE) {
343                         want->object.flags |= COMMON_KNOWN;
344                         break;
345                 }
346                 if (!commit->object.parsed)
347                         parse_object(commit->object.sha1);
348                 if (commit->object.flags & REACHABLE)
349                         continue;
350                 commit->object.flags |= REACHABLE;
351                 if (commit->date < oldest_have)
352                         continue;
353                 for (list = commit->parents; list; list = list->next) {
354                         struct commit *parent = list->item;
355                         if (!(parent->object.flags & REACHABLE))
356                                 insert_by_date(parent, &work);
357                 }
358         }
359         want->object.flags |= REACHABLE;
360         clear_commit_marks(want, REACHABLE);
361         free_commit_list(work);
362         return (want->object.flags & COMMON_KNOWN);
363 }
364
365 static int ok_to_give_up(void)
366 {
367         int i;
368
369         if (!have_obj.nr)
370                 return 0;
371
372         for (i = 0; i < want_obj.nr; i++) {
373                 struct object *want = want_obj.objects[i].item;
374
375                 if (want->flags & COMMON_KNOWN)
376                         continue;
377                 want = deref_tag(want, "a want line", 0);
378                 if (!want || want->type != OBJ_COMMIT) {
379                         /* no way to tell if this is reachable by
380                          * looking at the ancestry chain alone, so
381                          * leave a note to ourselves not to worry about
382                          * this object anymore.
383                          */
384                         want_obj.objects[i].item->flags |= COMMON_KNOWN;
385                         continue;
386                 }
387                 if (!reachable((struct commit *)want))
388                         return 0;
389         }
390         return 1;
391 }
392
393 static int get_common_commits(void)
394 {
395         static char line[1000];
396         unsigned char sha1[20];
397         char hex[41], last_hex[41];
398         int len;
399
400         save_commit_buffer = 0;
401
402         for(;;) {
403                 len = packet_read_line(0, line, sizeof(line));
404                 reset_timeout();
405
406                 if (!len) {
407                         if (have_obj.nr == 0 || multi_ack)
408                                 packet_write(1, "NAK\n");
409                         continue;
410                 }
411                 len = strip(line, len);
412                 if (!prefixcmp(line, "have ")) {
413                         switch (got_sha1(line+5, sha1)) {
414                         case -1: /* they have what we do not */
415                                 if (multi_ack && ok_to_give_up())
416                                         packet_write(1, "ACK %s continue\n",
417                                                      sha1_to_hex(sha1));
418                                 break;
419                         default:
420                                 memcpy(hex, sha1_to_hex(sha1), 41);
421                                 if (multi_ack) {
422                                         const char *msg = "ACK %s continue\n";
423                                         packet_write(1, msg, hex);
424                                         memcpy(last_hex, hex, 41);
425                                 }
426                                 else if (have_obj.nr == 1)
427                                         packet_write(1, "ACK %s\n", hex);
428                                 break;
429                         }
430                         continue;
431                 }
432                 if (!strcmp(line, "done")) {
433                         if (have_obj.nr > 0) {
434                                 if (multi_ack)
435                                         packet_write(1, "ACK %s\n", last_hex);
436                                 return 0;
437                         }
438                         packet_write(1, "NAK\n");
439                         return -1;
440                 }
441                 die("git-upload-pack: expected SHA1 list, got '%s'", line);
442         }
443 }
444
445 static void receive_needs(void)
446 {
447         struct object_array shallows = {0, 0, NULL};
448         static char line[1000];
449         int len, depth = 0;
450
451         if (debug_fd)
452                 write_in_full(debug_fd, "#S\n", 3);
453         for (;;) {
454                 struct object *o;
455                 unsigned char sha1_buf[20];
456                 len = packet_read_line(0, line, sizeof(line));
457                 reset_timeout();
458                 if (!len)
459                         break;
460                 if (debug_fd)
461                         write_in_full(debug_fd, line, len);
462
463                 if (!prefixcmp(line, "shallow ")) {
464                         unsigned char sha1[20];
465                         struct object *object;
466                         use_thin_pack = 0;
467                         if (get_sha1(line + 8, sha1))
468                                 die("invalid shallow line: %s", line);
469                         object = parse_object(sha1);
470                         if (!object)
471                                 die("did not find object for %s", line);
472                         object->flags |= CLIENT_SHALLOW;
473                         add_object_array(object, NULL, &shallows);
474                         continue;
475                 }
476                 if (!prefixcmp(line, "deepen ")) {
477                         char *end;
478                         use_thin_pack = 0;
479                         depth = strtol(line + 7, &end, 0);
480                         if (end == line + 7 || depth <= 0)
481                                 die("Invalid deepen: %s", line);
482                         continue;
483                 }
484                 if (prefixcmp(line, "want ") ||
485                     get_sha1_hex(line+5, sha1_buf))
486                         die("git-upload-pack: protocol error, "
487                             "expected to get sha, not '%s'", line);
488                 if (strstr(line+45, "multi_ack"))
489                         multi_ack = 1;
490                 if (strstr(line+45, "thin-pack"))
491                         use_thin_pack = 1;
492                 if (strstr(line+45, "ofs-delta"))
493                         use_ofs_delta = 1;
494                 if (strstr(line+45, "side-band-64k"))
495                         use_sideband = LARGE_PACKET_MAX;
496                 else if (strstr(line+45, "side-band"))
497                         use_sideband = DEFAULT_PACKET_MAX;
498                 if (strstr(line+45, "no-progress"))
499                         no_progress = 1;
500                 if (strstr(line+45, "include-tag"))
501                         use_include_tag = 1;
502
503                 /* We have sent all our refs already, and the other end
504                  * should have chosen out of them; otherwise they are
505                  * asking for nonsense.
506                  *
507                  * Hmph.  We may later want to allow "want" line that
508                  * asks for something like "master~10" (symbolic)...
509                  * would it make sense?  I don't know.
510                  */
511                 o = lookup_object(sha1_buf);
512                 if (!o || !(o->flags & OUR_REF))
513                         die("git-upload-pack: not our ref %s", line+5);
514                 if (!(o->flags & WANTED)) {
515                         o->flags |= WANTED;
516                         add_object_array(o, NULL, &want_obj);
517                 }
518         }
519         if (debug_fd)
520                 write_in_full(debug_fd, "#E\n", 3);
521         if (depth == 0 && shallows.nr == 0)
522                 return;
523         if (depth > 0) {
524                 struct commit_list *result, *backup;
525                 int i;
526                 backup = result = get_shallow_commits(&want_obj, depth,
527                         SHALLOW, NOT_SHALLOW);
528                 while (result) {
529                         struct object *object = &result->item->object;
530                         if (!(object->flags & (CLIENT_SHALLOW|NOT_SHALLOW))) {
531                                 packet_write(1, "shallow %s",
532                                                 sha1_to_hex(object->sha1));
533                                 register_shallow(object->sha1);
534                         }
535                         result = result->next;
536                 }
537                 free_commit_list(backup);
538                 for (i = 0; i < shallows.nr; i++) {
539                         struct object *object = shallows.objects[i].item;
540                         if (object->flags & NOT_SHALLOW) {
541                                 struct commit_list *parents;
542                                 packet_write(1, "unshallow %s",
543                                         sha1_to_hex(object->sha1));
544                                 object->flags &= ~CLIENT_SHALLOW;
545                                 /* make sure the real parents are parsed */
546                                 unregister_shallow(object->sha1);
547                                 object->parsed = 0;
548                                 if (parse_commit((struct commit *)object))
549                                         die("invalid commit");
550                                 parents = ((struct commit *)object)->parents;
551                                 while (parents) {
552                                         add_object_array(&parents->item->object,
553                                                         NULL, &want_obj);
554                                         parents = parents->next;
555                                 }
556                         }
557                         /* make sure commit traversal conforms to client */
558                         register_shallow(object->sha1);
559                 }
560                 packet_flush(1);
561         } else
562                 if (shallows.nr > 0) {
563                         int i;
564                         for (i = 0; i < shallows.nr; i++)
565                                 register_shallow(shallows.objects[i].item->sha1);
566                 }
567         free(shallows.objects);
568 }
569
570 static int send_ref(const char *refname, const unsigned char *sha1, int flag, void *cb_data)
571 {
572         static const char *capabilities = "multi_ack thin-pack side-band"
573                 " side-band-64k ofs-delta shallow no-progress"
574                 " include-tag";
575         struct object *o = parse_object(sha1);
576
577         if (!o)
578                 die("git-upload-pack: cannot find object %s:", sha1_to_hex(sha1));
579
580         if (capabilities)
581                 packet_write(1, "%s %s%c%s\n", sha1_to_hex(sha1), refname,
582                         0, capabilities);
583         else
584                 packet_write(1, "%s %s\n", sha1_to_hex(sha1), refname);
585         capabilities = NULL;
586         if (!(o->flags & OUR_REF)) {
587                 o->flags |= OUR_REF;
588                 nr_our_refs++;
589         }
590         if (o->type == OBJ_TAG) {
591                 o = deref_tag(o, refname, 0);
592                 if (o)
593                         packet_write(1, "%s %s^{}\n", sha1_to_hex(o->sha1), refname);
594         }
595         return 0;
596 }
597
598 static void upload_pack(void)
599 {
600         reset_timeout();
601         head_ref(send_ref, NULL);
602         for_each_ref(send_ref, NULL);
603         packet_flush(1);
604         receive_needs();
605         if (want_obj.nr) {
606                 get_common_commits();
607                 create_pack_file();
608         }
609 }
610
611 int main(int argc, char **argv)
612 {
613         char *dir;
614         int i;
615         int strict = 0;
616
617         for (i = 1; i < argc; i++) {
618                 char *arg = argv[i];
619
620                 if (arg[0] != '-')
621                         break;
622                 if (!strcmp(arg, "--strict")) {
623                         strict = 1;
624                         continue;
625                 }
626                 if (!prefixcmp(arg, "--timeout=")) {
627                         timeout = atoi(arg+10);
628                         continue;
629                 }
630                 if (!strcmp(arg, "--")) {
631                         i++;
632                         break;
633                 }
634         }
635
636         if (i != argc-1)
637                 usage(upload_pack_usage);
638
639         setup_path(NULL);
640
641         dir = argv[i];
642
643         if (!enter_repo(dir, strict))
644                 die("'%s': unable to chdir or not a git archive", dir);
645         if (is_repository_shallow())
646                 die("attempt to fetch/clone from a shallow repository");
647         if (getenv("GIT_DEBUG_SEND_PACK"))
648                 debug_fd = atoi(getenv("GIT_DEBUG_SEND_PACK"));
649         upload_pack();
650         return 0;
651 }