1 /*
2  * Copyright (c) 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  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/time.h>
21 #include <sys/stat.h>
22 
23 #include <stdint.h>
24 #include <errno.h>
25 #include <imsg.h>
26 #include <limits.h>
27 #include <poll.h>
28 #include <signal.h>
29 #include <stdio.h>
30 #include <stdlib.h>
31 #include <string.h>
32 #include <ctype.h>
33 #include <sha1.h>
34 #include <sha2.h>
35 #include <fcntl.h>
36 #include <unistd.h>
37 #include <zlib.h>
38 #include <err.h>
39 
40 #include "got_error.h"
41 #include "got_object.h"
42 #include "got_path.h"
43 #include "got_version.h"
44 #include "got_fetch.h"
45 #include "got_reference.h"
46 
47 #include "got_lib_hash.h"
48 #include "got_lib_delta.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_parse.h"
51 #include "got_lib_privsep.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_pkt.h"
54 #include "got_lib_gitproto.h"
55 #include "got_lib_ratelimit.h"
56 
57 #ifndef MIN
58 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
59 #endif
60 
61 #ifndef nitems
62 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
63 #endif
64 
65 struct got_object *indexed;
66 static int chattygot;
67 
68 static const struct got_capability got_capabilities[] = {
69 	{ GOT_CAPA_AGENT, "got/" GOT_VERSION_STR },
70 	{ GOT_CAPA_OFS_DELTA, NULL },
71 	{ GOT_CAPA_SIDE_BAND_64K, NULL },
72 };
73 
74 static void
75 match_remote_ref(struct got_pathlist_head *have_refs,
76     struct got_object_id *my_id, const char *refname)
77 {
78 	struct got_pathlist_entry *pe;
79 
80 	/* XXX zero-hash signifies we don't have this ref;
81 	 * we should use a flag instead */
82 	memset(my_id, 0, sizeof(*my_id));
83 
84 	TAILQ_FOREACH(pe, have_refs, entry) {
85 		struct got_object_id *id = pe->data;
86 		if (strcmp(pe->path, refname) == 0) {
87 			memcpy(my_id, id, sizeof(*my_id));
88 			break;
89 		}
90 	}
91 }
92 
93 static int
94 match_branch(const char *branch, const char *wanted_branch)
95 {
96 	if (strncmp(branch, "refs/heads/", 11) != 0)
97 		return 0;
98 
99 	if (strncmp(wanted_branch, "refs/heads/", 11) == 0)
100 		wanted_branch += 11;
101 
102 	return (strcmp(branch + 11, wanted_branch) == 0);
103 }
104 
105 static int
106 match_wanted_ref(const char *refname, const char *wanted_ref)
107 {
108 	if (strncmp(refname, "refs/", 5) != 0)
109 		return 0;
110 	refname += 5;
111 
112 	/*
113 	 * Prevent fetching of references that won't make any
114 	 * sense outside of the remote repository's context.
115 	 */
116 	if (strncmp(refname, "got/", 4) == 0)
117 		return 0;
118 	if (strncmp(refname, "remotes/", 8) == 0)
119 		return 0;
120 
121 	if (strncmp(wanted_ref, "refs/", 5) == 0)
122 		wanted_ref += 5;
123 
124 	/* Allow prefix match. */
125 	if (got_path_is_child(refname, wanted_ref, strlen(wanted_ref)))
126 		return 1;
127 
128 	/* Allow exact match. */
129 	return (strcmp(refname, wanted_ref) == 0);
130 }
131 
132 static const struct got_error *
133 send_fetch_server_progress(struct imsgbuf *ibuf, const char *msg, size_t msglen)
134 {
135 	if (msglen > MAX_IMSGSIZE - IMSG_HEADER_SIZE)
136 		return got_error(GOT_ERR_NO_SPACE);
137 
138 	if (msglen == 0)
139 		return NULL;
140 
141 	if (imsg_compose(ibuf, GOT_IMSG_FETCH_SERVER_PROGRESS, 0, 0, -1,
142 	    msg, msglen) == -1)
143 		return got_error_from_errno(
144 		    "imsg_compose FETCH_SERVER_PROGRESS");
145 
146 	return got_privsep_flush_imsg(ibuf);
147 }
148 
149 static const struct got_error *
150 send_fetch_download_progress(struct imsgbuf *ibuf, off_t bytes,
151     struct got_ratelimit *rl)
152 {
153 	const struct got_error *err;
154 	int elapsed = 0;
155 
156 	if (rl) {
157 		err = got_ratelimit_check(&elapsed, rl);
158 		if (err || !elapsed)
159 			return err;
160 	}
161 
162 	if (imsg_compose(ibuf, GOT_IMSG_FETCH_DOWNLOAD_PROGRESS, 0, 0, -1,
163 	    &bytes, sizeof(bytes)) == -1)
164 		return got_error_from_errno(
165 		    "imsg_compose FETCH_DOWNLOAD_PROGRESS");
166 
167 	return got_privsep_flush_imsg(ibuf);
168 }
169 
170 static const struct got_error *
171 send_fetch_done(struct imsgbuf *ibuf, uint8_t *pack_sha1)
172 {
173 	if (imsg_compose(ibuf, GOT_IMSG_FETCH_DONE, 0, 0, -1,
174 	    pack_sha1, SHA1_DIGEST_LENGTH) == -1)
175 		return got_error_from_errno("imsg_compose FETCH");
176 	return got_privsep_flush_imsg(ibuf);
177 }
178 
179 static const struct got_error *
180 fetch_progress(struct imsgbuf *ibuf, const char *buf, size_t len)
181 {
182 	size_t i;
183 
184 	if (len == 0)
185 		return NULL;
186 
187 	/*
188 	 * Truncate messages which exceed the maximum imsg payload size.
189 	 * Server may send up to 64k.
190 	 */
191 	if (len > MAX_IMSGSIZE - IMSG_HEADER_SIZE)
192 		len = MAX_IMSGSIZE - IMSG_HEADER_SIZE;
193 
194 	/* Only allow printable ASCII. */
195 	for (i = 0; i < len; i++) {
196 		if (isprint((unsigned char)buf[i]) ||
197 		    isspace((unsigned char)buf[i]))
198 			continue;
199 		return got_error_msg(GOT_ERR_BAD_PACKET,
200 		    "non-printable progress message received from server");
201 	}
202 
203 	return send_fetch_server_progress(ibuf, buf, len);
204 }
205 
206 static const struct got_error *
207 fetch_error(const char *buf, size_t len)
208 {
209 	static char msg[1024];
210 	size_t i;
211 
212 	for (i = 0; i < len && i < sizeof(msg) - 1; i++) {
213 		if (!isprint((unsigned char)buf[i]))
214 			return got_error_msg(GOT_ERR_BAD_PACKET,
215 			    "non-printable error message received from server");
216 		msg[i] = buf[i];
217 	}
218 	msg[i] = '\0';
219 	return got_error_msg(GOT_ERR_FETCH_FAILED, msg);
220 }
221 
222 static const struct got_error *
223 send_fetch_symrefs(struct imsgbuf *ibuf, struct got_pathlist_head *symrefs)
224 {
225 	struct ibuf *wbuf;
226 	size_t len, nsymrefs = 0;
227 	struct got_pathlist_entry *pe;
228 
229 	len = sizeof(struct got_imsg_fetch_symrefs);
230 	TAILQ_FOREACH(pe, symrefs, entry) {
231 		const char *target = pe->data;
232 		len += sizeof(struct got_imsg_fetch_symref) +
233 		    pe->path_len + strlen(target);
234 		nsymrefs++;
235 	}
236 
237 	if (len >= MAX_IMSGSIZE - IMSG_HEADER_SIZE)
238 		return got_error(GOT_ERR_NO_SPACE);
239 
240 	wbuf = imsg_create(ibuf, GOT_IMSG_FETCH_SYMREFS, 0, 0, len);
241 	if (wbuf == NULL)
242 		return got_error_from_errno("imsg_create FETCH_SYMREFS");
243 
244 	/* Keep in sync with struct got_imsg_fetch_symrefs definition! */
245 	if (imsg_add(wbuf, &nsymrefs, sizeof(nsymrefs)) == -1)
246 		return got_error_from_errno("imsg_add FETCH_SYMREFS");
247 
248 	TAILQ_FOREACH(pe, symrefs, entry) {
249 		const char *name = pe->path;
250 		size_t name_len = pe->path_len;
251 		const char *target = pe->data;
252 		size_t target_len = strlen(target);
253 
254 		/* Keep in sync with struct got_imsg_fetch_symref definition! */
255 		if (imsg_add(wbuf, &name_len, sizeof(name_len)) == -1)
256 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
257 		if (imsg_add(wbuf, &target_len, sizeof(target_len)) == -1)
258 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
259 		if (imsg_add(wbuf, name, name_len) == -1)
260 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
261 		if (imsg_add(wbuf, target, target_len) == -1)
262 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
263 	}
264 
265 	imsg_close(ibuf, wbuf);
266 	return got_privsep_flush_imsg(ibuf);
267 }
268 
269 static const struct got_error *
270 send_fetch_ref(struct imsgbuf *ibuf, struct got_object_id *refid,
271     const char *refname)
272 {
273 	struct ibuf *wbuf;
274 	size_t len, reflen = strlen(refname);
275 
276 	len = sizeof(struct got_imsg_fetch_ref) + reflen;
277 	if (len >= MAX_IMSGSIZE - IMSG_HEADER_SIZE)
278 		return got_error(GOT_ERR_NO_SPACE);
279 
280 	wbuf = imsg_create(ibuf, GOT_IMSG_FETCH_REF, 0, 0, len);
281 	if (wbuf == NULL)
282 		return got_error_from_errno("imsg_create FETCH_REF");
283 
284 	/* Keep in sync with struct got_imsg_fetch_ref definition! */
285 	if (imsg_add(wbuf, refid, sizeof(*refid)) == -1)
286 		return got_error_from_errno("imsg_add FETCH_REF");
287 	if (imsg_add(wbuf, refname, reflen) == -1)
288 		return got_error_from_errno("imsg_add FETCH_REF");
289 
290 	imsg_close(ibuf, wbuf);
291 	return got_privsep_flush_imsg(ibuf);
292 }
293 
294 static const struct got_error *
295 fetch_ref(struct imsgbuf *ibuf, struct got_pathlist_head *have_refs,
296     struct got_object_id *have, struct got_object_id *want,
297     const char *refname, const char *id_str)
298 {
299 	const struct got_error *err;
300 	char *theirs = NULL, *mine = NULL;
301 
302 	if (!got_parse_object_id(want, id_str, GOT_HASH_SHA1)) {
303 		err = got_error(GOT_ERR_BAD_OBJ_ID_STR);
304 		goto done;
305 	}
306 
307 	match_remote_ref(have_refs, have, refname);
308 	err = send_fetch_ref(ibuf, want, refname);
309 	if (err)
310 		goto done;
311 
312 	if (chattygot)
313 		fprintf(stderr, "%s: %s will be fetched\n",
314 		    getprogname(), refname);
315 	if (chattygot > 1) {
316 		err = got_object_id_str(&theirs, want);
317 		if (err)
318 			goto done;
319 		err = got_object_id_str(&mine, have);
320 		if (err)
321 			goto done;
322 		fprintf(stderr, "%s: remote: %s\n%s: local:  %s\n",
323 		    getprogname(), theirs, getprogname(), mine);
324 	}
325 done:
326 	free(theirs);
327 	free(mine);
328 	return err;
329 }
330 
331 static const struct got_error *
332 fetch_pack(int fd, int packfd, uint8_t *pack_sha1,
333     struct got_pathlist_head *have_refs, int fetch_all_branches,
334     struct got_pathlist_head *wanted_branches,
335     struct got_pathlist_head *wanted_refs, int list_refs_only,
336     const char *worktree_branch, const char *remote_head,
337     int no_head, struct imsgbuf *ibuf)
338 {
339 	const struct got_error *err = NULL;
340 	char buf[GOT_PKT_MAX];
341 	char hashstr[SHA1_DIGEST_STRING_LENGTH];
342 	struct got_object_id *have, *want;
343 	int is_firstpkt = 1, nref = 0, refsz = 16;
344 	int i, n, nwant = 0, nhave = 0, acked = 0;
345 	off_t packsz = 0, last_reported_packsz = 0;
346 	char *id_str = NULL, *default_id_str = NULL, *refname = NULL;
347 	char *server_capabilities = NULL, *my_capabilities = NULL;
348 	const char *default_branch = NULL;
349 	struct got_pathlist_head symrefs;
350 	struct got_pathlist_entry *pe;
351 	int sent_my_capabilites = 0, have_sidebands = 0;
352 	int found_branch = 0;
353 	struct got_hash ctx;
354 	uint8_t sha1_buf[SHA1_DIGEST_LENGTH];
355 	size_t sha1_buf_len = 0;
356 	ssize_t w;
357 	struct got_ratelimit rl;
358 
359 	TAILQ_INIT(&symrefs);
360 	got_hash_init(&ctx, GOT_HASH_SHA1);
361 	got_ratelimit_init(&rl, 0, 500);
362 
363 	have = malloc(refsz * sizeof(have[0]));
364 	if (have == NULL)
365 		return got_error_from_errno("malloc");
366 	want = malloc(refsz * sizeof(want[0]));
367 	if (want == NULL) {
368 		err = got_error_from_errno("malloc");
369 		goto done;
370 	}
371 	while (1) {
372 		err = got_pkt_readpkt(&n, fd, buf, sizeof(buf), chattygot,
373 		    INFTIM);
374 		if (err)
375 			goto done;
376 		if (n == 0)
377 			break;
378 		if (n >= 4 && strncmp(buf, "ERR ", 4) == 0) {
379 			err = fetch_error(&buf[4], n - 4);
380 			goto done;
381 		}
382 		free(id_str);
383 		free(refname);
384 		err = got_gitproto_parse_refline(&id_str, &refname,
385 		    &server_capabilities, buf, n);
386 		if (err)
387 			goto done;
388 
389 		if (refsz == nref + 1) {
390 			struct got_object_id *h, *w;
391 
392 			refsz *= 2;
393 			h = reallocarray(have, refsz, sizeof(have[0]));
394 			if (h == NULL) {
395 				err = got_error_from_errno("reallocarray");
396 				goto done;
397 			}
398 			have = h;
399 			w = reallocarray(want, refsz, sizeof(want[0]));
400 			if (w == NULL) {
401 				err = got_error_from_errno("reallocarray");
402 				goto done;
403 			}
404 			want = w;
405 		}
406 
407 		if (is_firstpkt) {
408 			if (chattygot && server_capabilities[0] != '\0')
409 				fprintf(stderr, "%s: server capabilities: %s\n",
410 				    getprogname(), server_capabilities);
411 			err = got_gitproto_match_capabilities(&my_capabilities,
412 			    &symrefs, server_capabilities,
413 			    got_capabilities, nitems(got_capabilities));
414 			if (err)
415 				goto done;
416 			if (chattygot)
417 				fprintf(stderr, "%s: my capabilities:%s\n",
418 				    getprogname(), my_capabilities != NULL ?
419 				    my_capabilities : "");
420 			err = send_fetch_symrefs(ibuf, &symrefs);
421 			if (err)
422 				goto done;
423 			is_firstpkt = 0;
424 			if (!fetch_all_branches) {
425 				TAILQ_FOREACH(pe, &symrefs, entry) {
426 					const char *name = pe->path;
427 					const char *symref_target = pe->data;
428 					if (strcmp(name, GOT_REF_HEAD) != 0)
429 						continue;
430 					default_branch = symref_target;
431 					break;
432 				}
433 			}
434 			if (default_branch)
435 				continue;
436 		}
437 		if (strstr(refname, "^{}")) {
438 			if (chattygot) {
439 				fprintf(stderr, "%s: ignoring %s\n",
440 				    getprogname(), refname);
441 			}
442 			continue;
443 		}
444 		if (default_branch && default_id_str == NULL &&
445 		    strcmp(refname, default_branch) == 0) {
446 			default_id_str = strdup(id_str);
447 			if (default_id_str == NULL) {
448 				err = got_error_from_errno("strdup");
449 				goto done;
450 			}
451 		}
452 
453 		if (list_refs_only || strncmp(refname, "refs/tags/", 10) == 0) {
454 			err = fetch_ref(ibuf, have_refs, &have[nref],
455 			    &want[nref], refname, id_str);
456 			if (err)
457 				goto done;
458 			nref++;
459 		} else if (strncmp(refname, "refs/heads/", 11) == 0) {
460 			if (fetch_all_branches) {
461 				err = fetch_ref(ibuf, have_refs, &have[nref],
462 				    &want[nref], refname, id_str);
463 				if (err)
464 					goto done;
465 				nref++;
466 				found_branch = 1;
467 				continue;
468 			}
469 			TAILQ_FOREACH(pe, wanted_branches, entry) {
470 				if (match_branch(refname, pe->path))
471 					break;
472 			}
473 			if (pe != NULL || (worktree_branch != NULL &&
474 			    match_branch(refname, worktree_branch))) {
475 				err = fetch_ref(ibuf, have_refs, &have[nref],
476 				    &want[nref], refname, id_str);
477 				if (err)
478 					goto done;
479 				nref++;
480 				found_branch = 1;
481 			} else if (chattygot) {
482 				fprintf(stderr, "%s: ignoring %s\n",
483 				    getprogname(), refname);
484 			}
485 		} else {
486 			TAILQ_FOREACH(pe, wanted_refs, entry) {
487 				if (match_wanted_ref(refname, pe->path))
488 					break;
489 			}
490 			if (pe != NULL) {
491 				err = fetch_ref(ibuf, have_refs, &have[nref],
492 				    &want[nref], refname, id_str);
493 				if (err)
494 					goto done;
495 				nref++;
496 			} else if (chattygot) {
497 				fprintf(stderr, "%s: ignoring %s\n",
498 				    getprogname(), refname);
499 			}
500 		}
501 	}
502 
503 	if (list_refs_only)
504 		goto done;
505 
506 	/*
507 	 * If -b was not used and either none of the requested branches
508 	 * (got.conf, worktree) were found or the client already has the
509 	 * remote HEAD symref but its target changed, fetch remote's HEAD.
510 	 */
511 	if (!no_head && default_branch && default_id_str &&
512 	    strncmp(default_branch, "refs/heads/", 11) == 0) {
513 		int remote_head_changed = 0;
514 
515 		if (remote_head) {
516 			if (strcmp(remote_head, default_branch + 11) != 0)
517 				remote_head_changed = 1;
518 		}
519 
520 		if (!found_branch || remote_head_changed) {
521 			err = fetch_ref(ibuf, have_refs, &have[nref],
522 			    &want[nref], default_branch, default_id_str);
523 			if (err)
524 				goto done;
525 			nref++;
526 		}
527 	}
528 
529 	/* Abort if we haven't found anything to fetch. */
530 	if (nref == 0) {
531 		struct got_pathlist_entry *pe;
532 		static char msg[PATH_MAX + 33];
533 
534 		pe = TAILQ_FIRST(wanted_branches);
535 		if (pe) {
536 			snprintf(msg, sizeof(msg),
537 			    "branch \"%s\" not found on server", pe->path);
538 			err = got_error_msg(GOT_ERR_FETCH_NO_BRANCH, msg);
539 			goto done;
540 		}
541 
542 		pe = TAILQ_FIRST(wanted_refs);
543 		if (pe) {
544 			snprintf(msg, sizeof(msg),
545 			    "reference \"%s\" not found on server", pe->path);
546 			err = got_error_msg(GOT_ERR_FETCH_NO_BRANCH, msg);
547 			goto done;
548 		}
549 
550 		err = got_error(GOT_ERR_FETCH_NO_BRANCH);
551 		goto done;
552 	}
553 
554 	for (i = 0; i < nref; i++) {
555 		if (got_object_id_cmp(&have[i], &want[i]) == 0)
556 			continue;
557 		got_sha1_digest_to_str(want[i].sha1, hashstr, sizeof(hashstr));
558 		n = snprintf(buf, sizeof(buf), "want %s%s\n", hashstr,
559 		    sent_my_capabilites || my_capabilities == NULL ?
560 		    "" : my_capabilities);
561 		if (n < 0 || (size_t)n >= sizeof(buf)) {
562 			err = got_error(GOT_ERR_NO_SPACE);
563 			goto done;
564 		}
565 		err = got_pkt_writepkt(fd, buf, n, chattygot);
566 		if (err)
567 			goto done;
568 		sent_my_capabilites = 1;
569 		nwant++;
570 	}
571 	err = got_pkt_flushpkt(fd, chattygot);
572 	if (err)
573 		goto done;
574 
575 	if (nwant == 0)
576 		goto done;
577 
578 	TAILQ_FOREACH(pe, have_refs, entry) {
579 		struct got_object_id *id = pe->data;
580 		got_sha1_digest_to_str(id->sha1, hashstr, sizeof(hashstr));
581 		n = snprintf(buf, sizeof(buf), "have %s\n", hashstr);
582 		if (n < 0 || (size_t)n >= sizeof(buf)) {
583 			err = got_error(GOT_ERR_NO_SPACE);
584 			goto done;
585 		}
586 		err = got_pkt_writepkt(fd, buf, n, chattygot);
587 		if (err)
588 			goto done;
589 		nhave++;
590 	}
591 
592 	n = strlcpy(buf, "done\n", sizeof(buf));
593 	err = got_pkt_writepkt(fd, buf, n, chattygot);
594 	if (err)
595 		goto done;
596 
597 	while (nhave > 0 && !acked) {
598 		struct got_object_id common_id;
599 
600 		/* The server should ACK the object IDs we need. */
601 		err = got_pkt_readpkt(&n, fd, buf, sizeof(buf), chattygot,
602 		    INFTIM);
603 		if (err)
604 			goto done;
605 		if (n >= 4 && strncmp(buf, "ERR ", 4) == 0) {
606 			err = fetch_error(&buf[4], n - 4);
607 			goto done;
608 		}
609 		if (n >= 4 && strncmp(buf, "NAK\n", 4) == 0) {
610 			/*
611 			 * Server could not find a common ancestor.
612 			 * Perhaps it is an out-of-date mirror, or there
613 			 * is a repository with unrelated history.
614 			 */
615 			break;
616 		}
617 		if (n < 4 + SHA1_DIGEST_STRING_LENGTH ||
618 		    strncmp(buf, "ACK ", 4) != 0) {
619 			err = got_error_msg(GOT_ERR_BAD_PACKET,
620 			    "unexpected message from server");
621 			goto done;
622 		}
623 		if (!got_parse_object_id(&common_id, buf + 4,
624 		    GOT_HASH_SHA1)) {
625 			err = got_error_msg(GOT_ERR_BAD_PACKET,
626 			    "bad object ID in ACK packet from server");
627 			goto done;
628 		}
629 		acked++;
630 	}
631 
632 	if (nhave == 0) {
633 		err = got_pkt_readpkt(&n, fd, buf, sizeof(buf), chattygot,
634 		    INFTIM);
635 		if (err)
636 			goto done;
637 		if (n != 4 || strncmp(buf, "NAK\n", n) != 0) {
638 			err = got_error_msg(GOT_ERR_BAD_PACKET,
639 			    "unexpected message from server");
640 			goto done;
641 		}
642 	}
643 
644 	if (chattygot)
645 		fprintf(stderr, "%s: fetching...\n", getprogname());
646 
647 	if (my_capabilities != NULL &&
648 	    strstr(my_capabilities, GOT_CAPA_SIDE_BAND_64K) != NULL)
649 		have_sidebands = 1;
650 
651 	while (1) {
652 		ssize_t r = 0;
653 		int datalen = -1;
654 
655 		if (have_sidebands) {
656 			err = got_pkt_readhdr(&datalen, fd, chattygot, INFTIM);
657 			if (err)
658 				goto done;
659 			if (datalen <= 0)
660 				break;
661 
662 			/* Read sideband channel ID (one byte). */
663 			r = read(fd, buf, 1);
664 			if (r == -1) {
665 				err = got_error_from_errno("read");
666 				goto done;
667 			}
668 			if (r != 1) {
669 				err = got_error_msg(GOT_ERR_BAD_PACKET,
670 				    "short packet");
671 				goto done;
672 			}
673 			if (datalen > sizeof(buf) - 5) {
674 				err = got_error_msg(GOT_ERR_BAD_PACKET,
675 				    "bad packet length");
676 				goto done;
677 			}
678 			datalen--; /* sideband ID has been read */
679 			if (buf[0] == GOT_SIDEBAND_PACKFILE_DATA) {
680 				/* Read packfile data. */
681 				err = got_pkt_readn(&r, fd, buf, datalen,
682 				    INFTIM);
683 				if (err)
684 					goto done;
685 				if (r != datalen) {
686 					err = got_error_msg(GOT_ERR_BAD_PACKET,
687 					    "packet too short");
688 					goto done;
689 				}
690 			} else if (buf[0] == GOT_SIDEBAND_PROGRESS_INFO) {
691 				err = got_pkt_readn(&r, fd, buf, datalen,
692 				    INFTIM);
693 				if (err)
694 					goto done;
695 				if (r != datalen) {
696 					err = got_error_msg(GOT_ERR_BAD_PACKET,
697 					    "packet too short");
698 					goto done;
699 				}
700 				err = fetch_progress(ibuf, buf, r);
701 				if (err)
702 					goto done;
703 				continue;
704 			} else if (buf[0] == GOT_SIDEBAND_ERROR_INFO) {
705 				err = got_pkt_readn(&r, fd, buf, datalen,
706 				    INFTIM);
707 				if (err)
708 					goto done;
709 				if (r != datalen) {
710 					err = got_error_msg(GOT_ERR_BAD_PACKET,
711 					    "packet too short");
712 					goto done;
713 				}
714 				err = fetch_error(buf, r);
715 				goto done;
716 			} else if (buf[0] == 'A') {
717 				err = got_pkt_readn(&r, fd, buf, datalen,
718 				    INFTIM);
719 				if (err)
720 					goto done;
721 				if (r != datalen) {
722 					err = got_error_msg(GOT_ERR_BAD_PACKET,
723 					    "packet too short");
724 					goto done;
725 				}
726 				/*
727 				 * Git server responds with ACK after 'done'
728 				 * even though multi_ack is disabled?!?
729 				 */
730 				buf[r] = '\0';
731 				if (strncmp(buf, "CK ", 3) == 0)
732 					continue; /* ignore */
733 				err = got_error_msg(GOT_ERR_BAD_PACKET,
734 				    "unexpected message from server");
735 				goto done;
736 			} else {
737 				err = got_error_msg(GOT_ERR_BAD_PACKET,
738 				    "unknown side-band received from server");
739 				goto done;
740 			}
741 		} else {
742 			/* No sideband channel. Every byte is packfile data. */
743 			err = got_pkt_readn(&r, fd, buf, sizeof buf, INFTIM);
744 			if (err)
745 				goto done;
746 			if (r <= 0)
747 				break;
748 		}
749 
750 		/*
751 		 * An expected SHA1 checksum sits at the end of the pack file.
752 		 * Since we don't know the file size ahead of time we have to
753 		 * keep SHA1_DIGEST_LENGTH bytes buffered and avoid mixing
754 		 * those bytes into our SHA1 checksum computation until we
755 		 * know for sure that additional pack file data bytes follow.
756 		 *
757 		 * We can assume r > 0 since otherwise the loop would exit.
758 		 */
759 		if (r < SHA1_DIGEST_LENGTH) {
760 			if (sha1_buf_len < SHA1_DIGEST_LENGTH) {
761 				/*
762 				 * If there's enough buffered + read data to
763 				 * fill up the buffer then shift a sufficient
764 				 * amount of bytes out at the front to make
765 				 * room, mixing those bytes into the checksum.
766 				 */
767 				if (sha1_buf_len > 0 &&
768 				    sha1_buf_len + r > SHA1_DIGEST_LENGTH) {
769 					size_t nshift = MIN(sha1_buf_len + r -
770 					    SHA1_DIGEST_LENGTH, sha1_buf_len);
771 					got_hash_update(&ctx, sha1_buf,
772 					    nshift);
773 					memmove(sha1_buf, sha1_buf + nshift,
774 					    sha1_buf_len - nshift);
775 					sha1_buf_len -= nshift;
776 				}
777 
778 				/* Buffer potential checksum bytes. */
779 				memcpy(sha1_buf + sha1_buf_len, buf, r);
780 				sha1_buf_len += r;
781 			} else {
782 				/*
783 				 * Mix in previously buffered bytes which
784 				 * are not part of the checksum after all.
785 				 */
786 				got_hash_update(&ctx, sha1_buf, r);
787 
788 				/* Update potential checksum buffer. */
789 				memmove(sha1_buf, sha1_buf + r,
790 				    sha1_buf_len - r);
791 				memcpy(sha1_buf + sha1_buf_len - r, buf, r);
792 			}
793 		} else {
794 			/* Mix in any previously buffered bytes. */
795 			got_hash_update(&ctx, sha1_buf, sha1_buf_len);
796 
797 			/* Mix in bytes read minus potential checksum bytes. */
798 			got_hash_update(&ctx, buf, r - SHA1_DIGEST_LENGTH);
799 
800 			/* Buffer potential checksum bytes. */
801 			memcpy(sha1_buf, buf + r - SHA1_DIGEST_LENGTH,
802 			    SHA1_DIGEST_LENGTH);
803 			sha1_buf_len = SHA1_DIGEST_LENGTH;
804 		}
805 
806 		/* Write packfile data to temporary pack file. */
807 		w = write(packfd, buf, r);
808 		if (w == -1) {
809 			err = got_error_from_errno("write");
810 			goto done;
811 		}
812 		if (w != r) {
813 			err = got_error(GOT_ERR_IO);
814 			goto done;
815 		}
816 		packsz += w;
817 
818 		/* Don't send too many progress privsep messages. */
819 		if (packsz > last_reported_packsz + 1024) {
820 			err = send_fetch_download_progress(ibuf, packsz, &rl);
821 			if (err)
822 				goto done;
823 			last_reported_packsz = packsz;
824 		}
825 	}
826 	err = send_fetch_download_progress(ibuf, packsz, NULL);
827 	if (err)
828 		goto done;
829 
830 	got_hash_final(&ctx, pack_sha1);
831 	if (sha1_buf_len != SHA1_DIGEST_LENGTH ||
832 	    memcmp(pack_sha1, sha1_buf, sha1_buf_len) != 0) {
833 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
834 		    "pack file checksum mismatch");
835 	}
836 done:
837 	got_pathlist_free(&symrefs, GOT_PATHLIST_FREE_ALL);
838 	free(have);
839 	free(want);
840 	free(id_str);
841 	free(default_id_str);
842 	free(refname);
843 	free(server_capabilities);
844 	return err;
845 }
846 
847 
848 int
849 main(int argc, char **argv)
850 {
851 	const struct got_error *err = NULL;
852 	int fetchfd = -1, packfd = -1;
853 	uint8_t pack_sha1[SHA1_DIGEST_LENGTH];
854 	struct imsgbuf ibuf;
855 	struct imsg imsg;
856 	struct got_pathlist_head have_refs;
857 	struct got_pathlist_head wanted_branches;
858 	struct got_pathlist_head wanted_refs;
859 	struct got_imsg_fetch_request fetch_req;
860 	struct got_imsg_fetch_have_ref href;
861 	struct got_imsg_fetch_wanted_branch wbranch;
862 	struct got_imsg_fetch_wanted_ref wref;
863 	size_t datalen, i;
864 	char *remote_head = NULL, *worktree_branch = NULL;
865 #if 0
866 	static int attached;
867 	while (!attached)
868 		sleep (1);
869 #endif
870 
871 	TAILQ_INIT(&have_refs);
872 	TAILQ_INIT(&wanted_branches);
873 	TAILQ_INIT(&wanted_refs);
874 
875 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
876 #ifndef PROFILE
877 	/* revoke access to most system calls */
878 	if (pledge("stdio recvfd", NULL) == -1) {
879 		err = got_error_from_errno("pledge");
880 		got_privsep_send_error(&ibuf, err);
881 		return 1;
882 	}
883 #endif
884 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
885 	if (err) {
886 		if (err->code == GOT_ERR_PRIVSEP_PIPE)
887 			err = NULL;
888 		goto done;
889 	}
890 	if (imsg.hdr.type == GOT_IMSG_STOP)
891 		goto done;
892 	if (imsg.hdr.type != GOT_IMSG_FETCH_REQUEST) {
893 		err = got_error(GOT_ERR_PRIVSEP_MSG);
894 		goto done;
895 	}
896 	datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
897 	if (datalen < sizeof(fetch_req)) {
898 		err = got_error(GOT_ERR_PRIVSEP_LEN);
899 		goto done;
900 	}
901 	memcpy(&fetch_req, imsg.data, sizeof(fetch_req));
902 	fetchfd = imsg_get_fd(&imsg);
903 
904 	if (datalen != sizeof(fetch_req) +
905 	    fetch_req.worktree_branch_len + fetch_req.remote_head_len) {
906 		err = got_error(GOT_ERR_PRIVSEP_LEN);
907 		goto done;
908 	}
909 
910 	if (fetch_req.worktree_branch_len != 0) {
911 		worktree_branch = strndup(imsg.data +
912 		    sizeof(fetch_req), fetch_req.worktree_branch_len);
913 		if (worktree_branch == NULL) {
914 			err = got_error_from_errno("strndup");
915 			goto done;
916 		}
917 	}
918 
919 	if (fetch_req.remote_head_len != 0) {
920 		remote_head = strndup(imsg.data + sizeof(fetch_req) +
921 		    fetch_req.worktree_branch_len, fetch_req.remote_head_len);
922 		if (remote_head == NULL) {
923 			err = got_error_from_errno("strndup");
924 			goto done;
925 		}
926 	}
927 
928 	imsg_free(&imsg);
929 
930 	if (fetch_req.verbosity > 0)
931 		chattygot += fetch_req.verbosity;
932 
933 	for (i = 0; i < fetch_req.n_have_refs; i++) {
934 		struct got_object_id *id;
935 		char *refname;
936 
937 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
938 		if (err) {
939 			if (err->code == GOT_ERR_PRIVSEP_PIPE)
940 				err = NULL;
941 			goto done;
942 		}
943 		if (imsg.hdr.type == GOT_IMSG_STOP)
944 			goto done;
945 		if (imsg.hdr.type != GOT_IMSG_FETCH_HAVE_REF) {
946 			err = got_error(GOT_ERR_PRIVSEP_MSG);
947 			goto done;
948 		}
949 		datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
950 		if (datalen < sizeof(href)) {
951 			err = got_error(GOT_ERR_PRIVSEP_LEN);
952 			goto done;
953 		}
954 		memcpy(&href, imsg.data, sizeof(href));
955 		if (datalen - sizeof(href) < href.name_len) {
956 			err = got_error(GOT_ERR_PRIVSEP_LEN);
957 			goto done;
958 		}
959 		refname = strndup(imsg.data + sizeof(href), href.name_len);
960 		if (refname == NULL) {
961 			err = got_error_from_errno("strndup");
962 			goto done;
963 		}
964 
965 		id = malloc(sizeof(*id));
966 		if (id == NULL) {
967 			free(refname);
968 			err = got_error_from_errno("malloc");
969 			goto done;
970 		}
971 		memcpy(id, &href.id, sizeof(*id));
972 		err = got_pathlist_append(&have_refs, refname, id);
973 		if (err) {
974 			free(refname);
975 			free(id);
976 			goto done;
977 		}
978 
979 		imsg_free(&imsg);
980 	}
981 
982 	for (i = 0; i < fetch_req.n_wanted_branches; i++) {
983 		char *refname;
984 
985 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
986 		if (err) {
987 			if (err->code == GOT_ERR_PRIVSEP_PIPE)
988 				err = NULL;
989 			goto done;
990 		}
991 		if (imsg.hdr.type == GOT_IMSG_STOP)
992 			goto done;
993 		if (imsg.hdr.type != GOT_IMSG_FETCH_WANTED_BRANCH) {
994 			err = got_error(GOT_ERR_PRIVSEP_MSG);
995 			goto done;
996 		}
997 		datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
998 		if (datalen < sizeof(wbranch)) {
999 			err = got_error(GOT_ERR_PRIVSEP_LEN);
1000 			goto done;
1001 		}
1002 		memcpy(&wbranch, imsg.data, sizeof(wbranch));
1003 		if (datalen - sizeof(wbranch) < wbranch.name_len) {
1004 			err = got_error(GOT_ERR_PRIVSEP_LEN);
1005 			goto done;
1006 		}
1007 		refname = strndup(imsg.data + sizeof(wbranch),
1008 		    wbranch.name_len);
1009 		if (refname == NULL) {
1010 			err = got_error_from_errno("strndup");
1011 			goto done;
1012 		}
1013 
1014 		err = got_pathlist_append(&wanted_branches, refname, NULL);
1015 		if (err) {
1016 			free(refname);
1017 			goto done;
1018 		}
1019 
1020 		imsg_free(&imsg);
1021 	}
1022 
1023 	for (i = 0; i < fetch_req.n_wanted_refs; i++) {
1024 		char *refname;
1025 
1026 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
1027 		if (err) {
1028 			if (err->code == GOT_ERR_PRIVSEP_PIPE)
1029 				err = NULL;
1030 			goto done;
1031 		}
1032 		if (imsg.hdr.type == GOT_IMSG_STOP)
1033 			goto done;
1034 		if (imsg.hdr.type != GOT_IMSG_FETCH_WANTED_REF) {
1035 			err = got_error(GOT_ERR_PRIVSEP_MSG);
1036 			goto done;
1037 		}
1038 		datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
1039 		if (datalen < sizeof(wref)) {
1040 			err = got_error(GOT_ERR_PRIVSEP_LEN);
1041 			goto done;
1042 		}
1043 		memcpy(&wref, imsg.data, sizeof(wref));
1044 		if (datalen - sizeof(wref) < wref.name_len) {
1045 			err = got_error(GOT_ERR_PRIVSEP_LEN);
1046 			goto done;
1047 		}
1048 		refname = strndup(imsg.data + sizeof(wref), wref.name_len);
1049 		if (refname == NULL) {
1050 			err = got_error_from_errno("strndup");
1051 			goto done;
1052 		}
1053 
1054 		err = got_pathlist_append(&wanted_refs, refname, NULL);
1055 		if (err) {
1056 			free(refname);
1057 			goto done;
1058 		}
1059 
1060 		imsg_free(&imsg);
1061 	}
1062 
1063 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
1064 	if (err) {
1065 		if (err->code == GOT_ERR_PRIVSEP_PIPE)
1066 			err = NULL;
1067 		goto done;
1068 	}
1069 	if (imsg.hdr.type == GOT_IMSG_STOP)
1070 		goto done;
1071 	if (imsg.hdr.type != GOT_IMSG_FETCH_OUTFD) {
1072 		err = got_error(GOT_ERR_PRIVSEP_MSG);
1073 		goto done;
1074 	}
1075 	if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
1076 		err = got_error(GOT_ERR_PRIVSEP_LEN);
1077 		goto done;
1078 	}
1079 	packfd = imsg_get_fd(&imsg);
1080 
1081 	err = fetch_pack(fetchfd, packfd, pack_sha1, &have_refs,
1082 	    fetch_req.fetch_all_branches, &wanted_branches,
1083 	    &wanted_refs, fetch_req.list_refs_only,
1084 	    worktree_branch, remote_head, fetch_req.no_head, &ibuf);
1085 done:
1086 	free(worktree_branch);
1087 	free(remote_head);
1088 	got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
1089 	got_pathlist_free(&wanted_branches, GOT_PATHLIST_FREE_PATH);
1090 	if (fetchfd != -1 && close(fetchfd) == -1 && err == NULL)
1091 		err = got_error_from_errno("close");
1092 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
1093 		err = got_error_from_errno("close");
1094 	if (err != NULL)
1095 		got_privsep_send_error(&ibuf, err);
1096 	else
1097 		err = send_fetch_done(&ibuf, pack_sha1);
1098 	if (err != NULL) {
1099 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
1100 		got_privsep_send_error(&ibuf, err);
1101 	}
1102 
1103 	exit(0);
1104 }