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 <signal.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <ctype.h>
32 #include <sha1.h>
33 #include <fcntl.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <err.h>
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_path.h"
41 #include "got_version.h"
42 #include "got_fetch.h"
43 #include "got_reference.h"
44 
45 #include "got_lib_sha1.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_privsep.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_pkt.h"
52 #include "got_lib_gitproto.h"
53 #include "got_lib_ratelimit.h"
54 
55 #ifndef MIN
56 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 #endif
58 
59 #ifndef nitems
60 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
61 #endif
62 
63 struct got_object *indexed;
64 static int chattygot;
65 
66 static const struct got_capability got_capabilities[] = {
67 	{ GOT_CAPA_AGENT, "got/" GOT_VERSION_STR },
68 	{ GOT_CAPA_OFS_DELTA, NULL },
69 	{ GOT_CAPA_SIDE_BAND_64K, NULL },
70 };
71 
72 static void
73 match_remote_ref(struct got_pathlist_head *have_refs,
74     struct got_object_id *my_id, const char *refname)
75 {
76 	struct got_pathlist_entry *pe;
77 
78 	/* XXX zero-hash signifies we don't have this ref;
79 	 * we should use a flag instead */
80 	memset(my_id, 0, sizeof(*my_id));
81 
82 	TAILQ_FOREACH(pe, have_refs, entry) {
83 		struct got_object_id *id = pe->data;
84 		if (strcmp(pe->path, refname) == 0) {
85 			memcpy(my_id, id, sizeof(*my_id));
86 			break;
87 		}
88 	}
89 }
90 
91 static int
92 match_branch(const char *branch, const char *wanted_branch)
93 {
94 	if (strncmp(branch, "refs/heads/", 11) != 0)
95 		return 0;
96 
97 	if (strncmp(wanted_branch, "refs/heads/", 11) == 0)
98 		wanted_branch += 11;
99 
100 	return (strcmp(branch + 11, wanted_branch) == 0);
101 }
102 
103 static int
104 match_wanted_ref(const char *refname, const char *wanted_ref)
105 {
106 	if (strncmp(refname, "refs/", 5) != 0)
107 		return 0;
108 	refname += 5;
109 
110 	/*
111 	 * Prevent fetching of references that won't make any
112 	 * sense outside of the remote repository's context.
113 	 */
114 	if (strncmp(refname, "got/", 4) == 0)
115 		return 0;
116 	if (strncmp(refname, "remotes/", 8) == 0)
117 		return 0;
118 
119 	if (strncmp(wanted_ref, "refs/", 5) == 0)
120 		wanted_ref += 5;
121 
122 	/* Allow prefix match. */
123 	if (got_path_is_child(refname, wanted_ref, strlen(wanted_ref)))
124 		return 1;
125 
126 	/* Allow exact match. */
127 	return (strcmp(refname, wanted_ref) == 0);
128 }
129 
130 static const struct got_error *
131 send_fetch_server_progress(struct imsgbuf *ibuf, const char *msg, size_t msglen)
132 {
133 	if (msglen > MAX_IMSGSIZE - IMSG_HEADER_SIZE)
134 		return got_error(GOT_ERR_NO_SPACE);
135 
136 	if (msglen == 0)
137 		return NULL;
138 
139 	if (imsg_compose(ibuf, GOT_IMSG_FETCH_SERVER_PROGRESS, 0, 0, -1,
140 	    msg, msglen) == -1)
141 		return got_error_from_errno(
142 		    "imsg_compose FETCH_SERVER_PROGRESS");
143 
144 	return got_privsep_flush_imsg(ibuf);
145 }
146 
147 static const struct got_error *
148 send_fetch_download_progress(struct imsgbuf *ibuf, off_t bytes,
149     struct got_ratelimit *rl)
150 {
151 	const struct got_error *err;
152 	int elapsed = 0;
153 
154 	if (rl) {
155 		err = got_ratelimit_check(&elapsed, rl);
156 		if (err || !elapsed)
157 			return err;
158 	}
159 
160 	if (imsg_compose(ibuf, GOT_IMSG_FETCH_DOWNLOAD_PROGRESS, 0, 0, -1,
161 	    &bytes, sizeof(bytes)) == -1)
162 		return got_error_from_errno(
163 		    "imsg_compose FETCH_DOWNLOAD_PROGRESS");
164 
165 	return got_privsep_flush_imsg(ibuf);
166 }
167 
168 static const struct got_error *
169 send_fetch_done(struct imsgbuf *ibuf, uint8_t *pack_sha1)
170 {
171 	if (imsg_compose(ibuf, GOT_IMSG_FETCH_DONE, 0, 0, -1,
172 	    pack_sha1, SHA1_DIGEST_LENGTH) == -1)
173 		return got_error_from_errno("imsg_compose FETCH");
174 	return got_privsep_flush_imsg(ibuf);
175 }
176 
177 static const struct got_error *
178 fetch_progress(struct imsgbuf *ibuf, const char *buf, size_t len)
179 {
180 	size_t i;
181 
182 	if (len == 0)
183 		return NULL;
184 
185 	/*
186 	 * Truncate messages which exceed the maximum imsg payload size.
187 	 * Server may send up to 64k.
188 	 */
189 	if (len > MAX_IMSGSIZE - IMSG_HEADER_SIZE)
190 		len = MAX_IMSGSIZE - IMSG_HEADER_SIZE;
191 
192 	/* Only allow printable ASCII. */
193 	for (i = 0; i < len; i++) {
194 		if (isprint((unsigned char)buf[i]) ||
195 		    isspace((unsigned char)buf[i]))
196 			continue;
197 		return got_error_msg(GOT_ERR_BAD_PACKET,
198 		    "non-printable progress message received from server");
199 	}
200 
201 	return send_fetch_server_progress(ibuf, buf, len);
202 }
203 
204 static const struct got_error *
205 fetch_error(const char *buf, size_t len)
206 {
207 	static char msg[1024];
208 	size_t i;
209 
210 	for (i = 0; i < len && i < sizeof(msg) - 1; i++) {
211 		if (!isprint((unsigned char)buf[i]))
212 			return got_error_msg(GOT_ERR_BAD_PACKET,
213 			    "non-printable error message received from server");
214 		msg[i] = buf[i];
215 	}
216 	msg[i] = '\0';
217 	return got_error_msg(GOT_ERR_FETCH_FAILED, msg);
218 }
219 
220 static const struct got_error *
221 send_fetch_symrefs(struct imsgbuf *ibuf, struct got_pathlist_head *symrefs)
222 {
223 	struct ibuf *wbuf;
224 	size_t len, nsymrefs = 0;
225 	struct got_pathlist_entry *pe;
226 
227 	len = sizeof(struct got_imsg_fetch_symrefs);
228 	TAILQ_FOREACH(pe, symrefs, entry) {
229 		const char *target = pe->data;
230 		len += sizeof(struct got_imsg_fetch_symref) +
231 		    pe->path_len + strlen(target);
232 		nsymrefs++;
233 	}
234 
235 	if (len >= MAX_IMSGSIZE - IMSG_HEADER_SIZE)
236 		return got_error(GOT_ERR_NO_SPACE);
237 
238 	wbuf = imsg_create(ibuf, GOT_IMSG_FETCH_SYMREFS, 0, 0, len);
239 	if (wbuf == NULL)
240 		return got_error_from_errno("imsg_create FETCH_SYMREFS");
241 
242 	/* Keep in sync with struct got_imsg_fetch_symrefs definition! */
243 	if (imsg_add(wbuf, &nsymrefs, sizeof(nsymrefs)) == -1)
244 		return got_error_from_errno("imsg_add FETCH_SYMREFS");
245 
246 	TAILQ_FOREACH(pe, symrefs, entry) {
247 		const char *name = pe->path;
248 		size_t name_len = pe->path_len;
249 		const char *target = pe->data;
250 		size_t target_len = strlen(target);
251 
252 		/* Keep in sync with struct got_imsg_fetch_symref definition! */
253 		if (imsg_add(wbuf, &name_len, sizeof(name_len)) == -1)
254 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
255 		if (imsg_add(wbuf, &target_len, sizeof(target_len)) == -1)
256 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
257 		if (imsg_add(wbuf, name, name_len) == -1)
258 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
259 		if (imsg_add(wbuf, target, target_len) == -1)
260 			return got_error_from_errno("imsg_add FETCH_SYMREFS");
261 	}
262 
263 	wbuf->fd = -1;
264 	imsg_close(ibuf, wbuf);
265 	return got_privsep_flush_imsg(ibuf);
266 }
267 
268 static const struct got_error *
269 send_fetch_ref(struct imsgbuf *ibuf, struct got_object_id *refid,
270     const char *refname)
271 {
272 	struct ibuf *wbuf;
273 	size_t len, reflen = strlen(refname);
274 
275 	len = sizeof(struct got_imsg_fetch_ref) + reflen;
276 	if (len >= MAX_IMSGSIZE - IMSG_HEADER_SIZE)
277 		return got_error(GOT_ERR_NO_SPACE);
278 
279 	wbuf = imsg_create(ibuf, GOT_IMSG_FETCH_REF, 0, 0, len);
280 	if (wbuf == NULL)
281 		return got_error_from_errno("imsg_create FETCH_REF");
282 
283 	/* Keep in sync with struct got_imsg_fetch_ref definition! */
284 	if (imsg_add(wbuf, refid, sizeof(*refid)) == -1)
285 		return got_error_from_errno("imsg_add FETCH_REF");
286 	if (imsg_add(wbuf, refname, reflen) == -1)
287 		return got_error_from_errno("imsg_add FETCH_REF");
288 
289 	wbuf->fd = -1;
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_sha1_digest(want->sha1, id_str)) {
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, struct imsgbuf *ibuf)
337 {
338 	const struct got_error *err = NULL;
339 	char buf[GOT_PKT_MAX];
340 	char hashstr[SHA1_DIGEST_STRING_LENGTH];
341 	struct got_object_id *have, *want;
342 	int is_firstpkt = 1, nref = 0, refsz = 16;
343 	int i, n, nwant = 0, nhave = 0, acked = 0;
344 	off_t packsz = 0, last_reported_packsz = 0;
345 	char *id_str = NULL, *default_id_str = NULL, *refname = NULL;
346 	char *server_capabilities = NULL, *my_capabilities = NULL;
347 	const char *default_branch = NULL;
348 	struct got_pathlist_head symrefs;
349 	struct got_pathlist_entry *pe;
350 	int sent_my_capabilites = 0, have_sidebands = 0;
351 	int found_branch = 0;
352 	SHA1_CTX sha1_ctx;
353 	uint8_t sha1_buf[SHA1_DIGEST_LENGTH];
354 	size_t sha1_buf_len = 0;
355 	ssize_t w;
356 	struct got_ratelimit rl;
357 
358 	TAILQ_INIT(&symrefs);
359 	SHA1Init(&sha1_ctx);
360 	got_ratelimit_init(&rl, 0, 500);
361 
362 	have = malloc(refsz * sizeof(have[0]));
363 	if (have == NULL)
364 		return got_error_from_errno("malloc");
365 	want = malloc(refsz * sizeof(want[0]));
366 	if (want == NULL) {
367 		err = got_error_from_errno("malloc");
368 		goto done;
369 	}
370 	while (1) {
371 		err = got_pkt_readpkt(&n, fd, buf, sizeof(buf), chattygot);
372 		if (err)
373 			goto done;
374 		if (n == 0)
375 			break;
376 		if (n >= 4 && strncmp(buf, "ERR ", 4) == 0) {
377 			err = fetch_error(&buf[4], n - 4);
378 			goto done;
379 		}
380 		free(id_str);
381 		free(refname);
382 		err = got_gitproto_parse_refline(&id_str, &refname,
383 		    &server_capabilities, buf, n);
384 		if (err)
385 			goto done;
386 
387 		if (refsz == nref + 1) {
388 			struct got_object_id *h, *w;
389 
390 			refsz *= 2;
391 			h = reallocarray(have, refsz, sizeof(have[0]));
392 			if (h == NULL) {
393 				err = got_error_from_errno("reallocarray");
394 				goto done;
395 			}
396 			have = h;
397 			w = reallocarray(want, refsz, sizeof(want[0]));
398 			if (w == NULL) {
399 				err = got_error_from_errno("reallocarray");
400 				goto done;
401 			}
402 			want = w;
403 		}
404 
405 		if (is_firstpkt) {
406 			if (chattygot && server_capabilities[0] != '\0')
407 				fprintf(stderr, "%s: server capabilities: %s\n",
408 				    getprogname(), server_capabilities);
409 			err = got_gitproto_match_capabilities(&my_capabilities,
410 			    &symrefs, server_capabilities,
411 			    got_capabilities, nitems(got_capabilities));
412 			if (err)
413 				goto done;
414 			if (chattygot)
415 				fprintf(stderr, "%s: my capabilities:%s\n",
416 				    getprogname(), my_capabilities != NULL ?
417 				    my_capabilities : "");
418 			err = send_fetch_symrefs(ibuf, &symrefs);
419 			if (err)
420 				goto done;
421 			is_firstpkt = 0;
422 			if (!fetch_all_branches) {
423 				TAILQ_FOREACH(pe, &symrefs, entry) {
424 					const char *name = pe->path;
425 					const char *symref_target = pe->data;
426 					if (strcmp(name, GOT_REF_HEAD) != 0)
427 						continue;
428 					default_branch = symref_target;
429 					break;
430 				}
431 			}
432 			if (default_branch)
433 				continue;
434 		}
435 		if (strstr(refname, "^{}")) {
436 			if (chattygot) {
437 				fprintf(stderr, "%s: ignoring %s\n",
438 				    getprogname(), refname);
439 			}
440 			continue;
441 		}
442 		if (default_branch && default_id_str == NULL &&
443 		    strcmp(refname, default_branch) == 0) {
444 			default_id_str = strdup(id_str);
445 			if (default_id_str == NULL) {
446 				err = got_error_from_errno("strdup");
447 				goto done;
448 			}
449 		}
450 
451 		if (list_refs_only || strncmp(refname, "refs/tags/", 10) == 0) {
452 			err = fetch_ref(ibuf, have_refs, &have[nref],
453 			    &want[nref], refname, id_str);
454 			if (err)
455 				goto done;
456 			nref++;
457 		} else if (strncmp(refname, "refs/heads/", 11) == 0) {
458 			if (fetch_all_branches) {
459 				err = fetch_ref(ibuf, have_refs, &have[nref],
460 				    &want[nref], refname, id_str);
461 				if (err)
462 					goto done;
463 				nref++;
464 				found_branch = 1;
465 				continue;
466 			}
467 			TAILQ_FOREACH(pe, wanted_branches, entry) {
468 				if (match_branch(refname, pe->path))
469 					break;
470 			}
471 			if (pe != NULL || (worktree_branch != NULL &&
472 			    match_branch(refname, worktree_branch))) {
473 				err = fetch_ref(ibuf, have_refs, &have[nref],
474 				    &want[nref], refname, id_str);
475 				if (err)
476 					goto done;
477 				nref++;
478 				found_branch = 1;
479 			} else if (chattygot) {
480 				fprintf(stderr, "%s: ignoring %s\n",
481 				    getprogname(), refname);
482 			}
483 		} else {
484 			TAILQ_FOREACH(pe, wanted_refs, entry) {
485 				if (match_wanted_ref(refname, pe->path))
486 					break;
487 			}
488 			if (pe != NULL) {
489 				err = fetch_ref(ibuf, have_refs, &have[nref],
490 				    &want[nref], refname, id_str);
491 				if (err)
492 					goto done;
493 				nref++;
494 			} else if (chattygot) {
495 				fprintf(stderr, "%s: ignoring %s\n",
496 				    getprogname(), refname);
497 			}
498 		}
499 	}
500 
501 	if (list_refs_only)
502 		goto done;
503 
504 	if (!found_branch && default_branch && default_id_str &&
505 	    strncmp(default_branch, "refs/heads/", 11) == 0) {
506 		err = fetch_ref(ibuf, have_refs, &have[nref],
507 		    &want[nref], default_branch, default_id_str);
508 		if (err)
509 			goto done;
510 		nref++;
511 		found_branch = 1;
512 	}
513 
514 	/* Abort if we haven't found anything to fetch. */
515 	if (nref == 0) {
516 		err = got_error(GOT_ERR_FETCH_NO_BRANCH);
517 		goto done;
518 	}
519 
520 	for (i = 0; i < nref; i++) {
521 		if (got_object_id_cmp(&have[i], &want[i]) == 0)
522 			continue;
523 		got_sha1_digest_to_str(want[i].sha1, hashstr, sizeof(hashstr));
524 		n = snprintf(buf, sizeof(buf), "want %s%s\n", hashstr,
525 		    sent_my_capabilites || my_capabilities == NULL ?
526 		    "" : my_capabilities);
527 		if (n < 0 || (size_t)n >= sizeof(buf)) {
528 			err = got_error(GOT_ERR_NO_SPACE);
529 			goto done;
530 		}
531 		err = got_pkt_writepkt(fd, buf, n, chattygot);
532 		if (err)
533 			goto done;
534 		sent_my_capabilites = 1;
535 		nwant++;
536 	}
537 	err = got_pkt_flushpkt(fd, chattygot);
538 	if (err)
539 		goto done;
540 
541 	if (nwant == 0)
542 		goto done;
543 
544 	TAILQ_FOREACH(pe, have_refs, entry) {
545 		struct got_object_id *id = pe->data;
546 		got_sha1_digest_to_str(id->sha1, hashstr, sizeof(hashstr));
547 		n = snprintf(buf, sizeof(buf), "have %s\n", hashstr);
548 		if (n < 0 || (size_t)n >= sizeof(buf)) {
549 			err = got_error(GOT_ERR_NO_SPACE);
550 			goto done;
551 		}
552 		err = got_pkt_writepkt(fd, buf, n, chattygot);
553 		if (err)
554 			goto done;
555 		nhave++;
556 	}
557 
558 	while (nhave > 0 && !acked) {
559 		struct got_object_id common_id;
560 
561 		/* The server should ACK the object IDs we need. */
562 		err = got_pkt_readpkt(&n, fd, buf, sizeof(buf), chattygot);
563 		if (err)
564 			goto done;
565 		if (n >= 4 && strncmp(buf, "ERR ", 4) == 0) {
566 			err = fetch_error(&buf[4], n - 4);
567 			goto done;
568 		}
569 		if (n >= 4 && strncmp(buf, "NAK\n", 4) == 0) {
570 			/* Server has not located our objects yet. */
571 			continue;
572 		}
573 		if (n < 4 + SHA1_DIGEST_STRING_LENGTH ||
574 		    strncmp(buf, "ACK ", 4) != 0) {
575 			err = got_error_msg(GOT_ERR_BAD_PACKET,
576 			    "unexpected message from server");
577 			goto done;
578 		}
579 		if (!got_parse_sha1_digest(common_id.sha1, buf + 4)) {
580 			err = got_error_msg(GOT_ERR_BAD_PACKET,
581 			    "bad object ID in ACK packet from server");
582 			goto done;
583 		}
584 		acked++;
585 	}
586 
587 	n = strlcpy(buf, "done\n", sizeof(buf));
588 	err = got_pkt_writepkt(fd, buf, n, chattygot);
589 	if (err)
590 		goto done;
591 
592 	if (nhave == 0) {
593 		err = got_pkt_readpkt(&n, fd, buf, sizeof(buf), chattygot);
594 		if (err)
595 			goto done;
596 		if (n != 4 || strncmp(buf, "NAK\n", n) != 0) {
597 			err = got_error_msg(GOT_ERR_BAD_PACKET,
598 			    "unexpected message from server");
599 			goto done;
600 		}
601 	}
602 
603 	if (chattygot)
604 		fprintf(stderr, "%s: fetching...\n", getprogname());
605 
606 	if (my_capabilities != NULL &&
607 	    strstr(my_capabilities, GOT_CAPA_SIDE_BAND_64K) != NULL)
608 		have_sidebands = 1;
609 
610 	while (1) {
611 		ssize_t r = 0;
612 		int datalen = -1;
613 
614 		if (have_sidebands) {
615 			err = got_pkt_readhdr(&datalen, fd, chattygot);
616 			if (err)
617 				goto done;
618 			if (datalen <= 0)
619 				break;
620 
621 			/* Read sideband channel ID (one byte). */
622 			r = read(fd, buf, 1);
623 			if (r == -1) {
624 				err = got_error_from_errno("read");
625 				goto done;
626 			}
627 			if (r != 1) {
628 				err = got_error_msg(GOT_ERR_BAD_PACKET,
629 				    "short packet");
630 				goto done;
631 			}
632 			if (datalen > sizeof(buf) - 5) {
633 				err = got_error_msg(GOT_ERR_BAD_PACKET,
634 				    "bad packet length");
635 				goto done;
636 			}
637 			datalen--; /* sideband ID has been read */
638 			if (buf[0] == GOT_SIDEBAND_PACKFILE_DATA) {
639 				/* Read packfile data. */
640 				err = got_pkt_readn(&r, fd, buf, datalen);
641 				if (err)
642 					goto done;
643 				if (r != datalen) {
644 					err = got_error_msg(GOT_ERR_BAD_PACKET,
645 					    "packet too short");
646 					goto done;
647 				}
648 			} else if (buf[0] == GOT_SIDEBAND_PROGRESS_INFO) {
649 				err = got_pkt_readn(&r, fd, buf, datalen);
650 				if (err)
651 					goto done;
652 				if (r != datalen) {
653 					err = got_error_msg(GOT_ERR_BAD_PACKET,
654 					    "packet too short");
655 					goto done;
656 				}
657 				err = fetch_progress(ibuf, buf, r);
658 				if (err)
659 					goto done;
660 				continue;
661 			} else if (buf[0] == GOT_SIDEBAND_ERROR_INFO) {
662 				err = got_pkt_readn(&r, fd, buf, datalen);
663 				if (err)
664 					goto done;
665 				if (r != datalen) {
666 					err = got_error_msg(GOT_ERR_BAD_PACKET,
667 					    "packet too short");
668 					goto done;
669 				}
670 				err = fetch_error(buf, r);
671 				goto done;
672 			} else if (buf[0] == 'A') {
673 				err = got_pkt_readn(&r, fd, buf, datalen);
674 				if (err)
675 					goto done;
676 				if (r != datalen) {
677 					err = got_error_msg(GOT_ERR_BAD_PACKET,
678 					    "packet too short");
679 					goto done;
680 				}
681 				/*
682 				 * Git server responds with ACK after 'done'
683 				 * even though multi_ack is disabled?!?
684 				 */
685 				buf[r] = '\0';
686 				if (strncmp(buf, "CK ", 3) == 0)
687 					continue; /* ignore */
688 				err = got_error_msg(GOT_ERR_BAD_PACKET,
689 				    "unexpected message from server");
690 				goto done;
691 			} else {
692 				err = got_error_msg(GOT_ERR_BAD_PACKET,
693 				    "unknown side-band received from server");
694 				goto done;
695 			}
696 		} else {
697 			/* No sideband channel. Every byte is packfile data. */
698 			err = got_pkt_readn(&r, fd, buf, sizeof buf);
699 			if (err)
700 				goto done;
701 			if (r <= 0)
702 				break;
703 		}
704 
705 		/*
706 		 * An expected SHA1 checksum sits at the end of the pack file.
707 		 * Since we don't know the file size ahead of time we have to
708 		 * keep SHA1_DIGEST_LENGTH bytes buffered and avoid mixing
709 		 * those bytes into our SHA1 checksum computation until we
710 		 * know for sure that additional pack file data bytes follow.
711 		 *
712 		 * We can assume r > 0 since otherwise the loop would exit.
713 		 */
714 		if (r < SHA1_DIGEST_LENGTH) {
715 			if (sha1_buf_len < SHA1_DIGEST_LENGTH) {
716 				/*
717 				 * If there's enough buffered + read data to
718 				 * fill up the buffer then shift a sufficient
719 				 * amount of bytes out at the front to make
720 				 * room, mixing those bytes into the checksum.
721 				 */
722 				if (sha1_buf_len > 0 &&
723 				    sha1_buf_len + r > SHA1_DIGEST_LENGTH) {
724 					size_t nshift = MIN(sha1_buf_len + r -
725 					    SHA1_DIGEST_LENGTH, sha1_buf_len);
726 					SHA1Update(&sha1_ctx, sha1_buf, nshift);
727 					memmove(sha1_buf, sha1_buf + nshift,
728 					    sha1_buf_len - nshift);
729 					sha1_buf_len -= nshift;
730 				}
731 
732 				/* Buffer potential checksum bytes. */
733 				memcpy(sha1_buf + sha1_buf_len, buf, r);
734 				sha1_buf_len += r;
735 			} else {
736 				/*
737 				 * Mix in previously buffered bytes which
738 				 * are not part of the checksum after all.
739 				 */
740 				SHA1Update(&sha1_ctx, sha1_buf, r);
741 
742 				/* Update potential checksum buffer. */
743 				memmove(sha1_buf, sha1_buf + r,
744 				    sha1_buf_len - r);
745 				memcpy(sha1_buf + sha1_buf_len - r, buf, r);
746 			}
747 		} else {
748 			/* Mix in any previously buffered bytes. */
749 			SHA1Update(&sha1_ctx, sha1_buf, sha1_buf_len);
750 
751 			/* Mix in bytes read minus potential checksum bytes. */
752 			SHA1Update(&sha1_ctx, buf, r - SHA1_DIGEST_LENGTH);
753 
754 			/* Buffer potential checksum bytes. */
755 			memcpy(sha1_buf, buf + r - SHA1_DIGEST_LENGTH,
756 			    SHA1_DIGEST_LENGTH);
757 			sha1_buf_len = SHA1_DIGEST_LENGTH;
758 		}
759 
760 		/* Write packfile data to temporary pack file. */
761 		w = write(packfd, buf, r);
762 		if (w == -1) {
763 			err = got_error_from_errno("write");
764 			goto done;
765 		}
766 		if (w != r) {
767 			err = got_error(GOT_ERR_IO);
768 			goto done;
769 		}
770 		packsz += w;
771 
772 		/* Don't send too many progress privsep messages. */
773 		if (packsz > last_reported_packsz + 1024) {
774 			err = send_fetch_download_progress(ibuf, packsz, &rl);
775 			if (err)
776 				goto done;
777 			last_reported_packsz = packsz;
778 		}
779 	}
780 	err = send_fetch_download_progress(ibuf, packsz, NULL);
781 	if (err)
782 		goto done;
783 
784 	SHA1Final(pack_sha1, &sha1_ctx);
785 	if (sha1_buf_len != SHA1_DIGEST_LENGTH ||
786 	    memcmp(pack_sha1, sha1_buf, sha1_buf_len) != 0) {
787 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
788 		    "pack file checksum mismatch");
789 	}
790 done:
791 	got_pathlist_free(&symrefs, GOT_PATHLIST_FREE_ALL);
792 	free(have);
793 	free(want);
794 	free(id_str);
795 	free(default_id_str);
796 	free(refname);
797 	free(server_capabilities);
798 	return err;
799 }
800 
801 
802 int
803 main(int argc, char **argv)
804 {
805 	const struct got_error *err = NULL;
806 	int fetchfd = -1, packfd = -1;
807 	uint8_t pack_sha1[SHA1_DIGEST_LENGTH];
808 	struct imsgbuf ibuf;
809 	struct imsg imsg;
810 	struct got_pathlist_head have_refs;
811 	struct got_pathlist_head wanted_branches;
812 	struct got_pathlist_head wanted_refs;
813 	struct got_imsg_fetch_request fetch_req;
814 	struct got_imsg_fetch_have_ref href;
815 	struct got_imsg_fetch_wanted_branch wbranch;
816 	struct got_imsg_fetch_wanted_ref wref;
817 	size_t datalen, i;
818 	char *worktree_branch = NULL;
819 #if 0
820 	static int attached;
821 	while (!attached)
822 		sleep (1);
823 #endif
824 
825 	TAILQ_INIT(&have_refs);
826 	TAILQ_INIT(&wanted_branches);
827 	TAILQ_INIT(&wanted_refs);
828 
829 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
830 #ifndef PROFILE
831 	/* revoke access to most system calls */
832 	if (pledge("stdio recvfd", NULL) == -1) {
833 		err = got_error_from_errno("pledge");
834 		got_privsep_send_error(&ibuf, err);
835 		return 1;
836 	}
837 #endif
838 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
839 	if (err) {
840 		if (err->code == GOT_ERR_PRIVSEP_PIPE)
841 			err = NULL;
842 		goto done;
843 	}
844 	if (imsg.hdr.type == GOT_IMSG_STOP)
845 		goto done;
846 	if (imsg.hdr.type != GOT_IMSG_FETCH_REQUEST) {
847 		err = got_error(GOT_ERR_PRIVSEP_MSG);
848 		goto done;
849 	}
850 	datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
851 	if (datalen < sizeof(fetch_req)) {
852 		err = got_error(GOT_ERR_PRIVSEP_LEN);
853 		goto done;
854 	}
855 	memcpy(&fetch_req, imsg.data, sizeof(fetch_req));
856 	fetchfd = imsg.fd;
857 
858 	if (datalen != sizeof(fetch_req) +
859 	    fetch_req.worktree_branch_len) {
860 		err = got_error(GOT_ERR_PRIVSEP_LEN);
861 		goto done;
862 	}
863 
864 	if (fetch_req.worktree_branch_len != 0) {
865 		worktree_branch = strndup(imsg.data +
866 		    sizeof(fetch_req), fetch_req.worktree_branch_len);
867 		if (worktree_branch == NULL) {
868 			err = got_error_from_errno("strndup");
869 			goto done;
870 		}
871 	}
872 
873 	imsg_free(&imsg);
874 
875 	if (fetch_req.verbosity > 0)
876 		chattygot += fetch_req.verbosity;
877 
878 	for (i = 0; i < fetch_req.n_have_refs; i++) {
879 		struct got_object_id *id;
880 		char *refname;
881 
882 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
883 		if (err) {
884 			if (err->code == GOT_ERR_PRIVSEP_PIPE)
885 				err = NULL;
886 			goto done;
887 		}
888 		if (imsg.hdr.type == GOT_IMSG_STOP)
889 			goto done;
890 		if (imsg.hdr.type != GOT_IMSG_FETCH_HAVE_REF) {
891 			err = got_error(GOT_ERR_PRIVSEP_MSG);
892 			goto done;
893 		}
894 		datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
895 		if (datalen < sizeof(href)) {
896 			err = got_error(GOT_ERR_PRIVSEP_LEN);
897 			goto done;
898 		}
899 		memcpy(&href, imsg.data, sizeof(href));
900 		if (datalen - sizeof(href) < href.name_len) {
901 			err = got_error(GOT_ERR_PRIVSEP_LEN);
902 			goto done;
903 		}
904 		refname = strndup(imsg.data + sizeof(href), href.name_len);
905 		if (refname == NULL) {
906 			err = got_error_from_errno("strndup");
907 			goto done;
908 		}
909 
910 		id = malloc(sizeof(*id));
911 		if (id == NULL) {
912 			free(refname);
913 			err = got_error_from_errno("malloc");
914 			goto done;
915 		}
916 		memcpy(id, &href.id, sizeof(*id));
917 		err = got_pathlist_append(&have_refs, refname, id);
918 		if (err) {
919 			free(refname);
920 			free(id);
921 			goto done;
922 		}
923 
924 		imsg_free(&imsg);
925 	}
926 
927 	for (i = 0; i < fetch_req.n_wanted_branches; i++) {
928 		char *refname;
929 
930 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
931 		if (err) {
932 			if (err->code == GOT_ERR_PRIVSEP_PIPE)
933 				err = NULL;
934 			goto done;
935 		}
936 		if (imsg.hdr.type == GOT_IMSG_STOP)
937 			goto done;
938 		if (imsg.hdr.type != GOT_IMSG_FETCH_WANTED_BRANCH) {
939 			err = got_error(GOT_ERR_PRIVSEP_MSG);
940 			goto done;
941 		}
942 		datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
943 		if (datalen < sizeof(wbranch)) {
944 			err = got_error(GOT_ERR_PRIVSEP_LEN);
945 			goto done;
946 		}
947 		memcpy(&wbranch, imsg.data, sizeof(wbranch));
948 		if (datalen - sizeof(wbranch) < wbranch.name_len) {
949 			err = got_error(GOT_ERR_PRIVSEP_LEN);
950 			goto done;
951 		}
952 		refname = strndup(imsg.data + sizeof(wbranch),
953 		    wbranch.name_len);
954 		if (refname == NULL) {
955 			err = got_error_from_errno("strndup");
956 			goto done;
957 		}
958 
959 		err = got_pathlist_append(&wanted_branches, refname, NULL);
960 		if (err) {
961 			free(refname);
962 			goto done;
963 		}
964 
965 		imsg_free(&imsg);
966 	}
967 
968 	for (i = 0; i < fetch_req.n_wanted_refs; i++) {
969 		char *refname;
970 
971 		err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
972 		if (err) {
973 			if (err->code == GOT_ERR_PRIVSEP_PIPE)
974 				err = NULL;
975 			goto done;
976 		}
977 		if (imsg.hdr.type == GOT_IMSG_STOP)
978 			goto done;
979 		if (imsg.hdr.type != GOT_IMSG_FETCH_WANTED_REF) {
980 			err = got_error(GOT_ERR_PRIVSEP_MSG);
981 			goto done;
982 		}
983 		datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
984 		if (datalen < sizeof(wref)) {
985 			err = got_error(GOT_ERR_PRIVSEP_LEN);
986 			goto done;
987 		}
988 		memcpy(&wref, imsg.data, sizeof(wref));
989 		if (datalen - sizeof(wref) < wref.name_len) {
990 			err = got_error(GOT_ERR_PRIVSEP_LEN);
991 			goto done;
992 		}
993 		refname = strndup(imsg.data + sizeof(wref), wref.name_len);
994 		if (refname == NULL) {
995 			err = got_error_from_errno("strndup");
996 			goto done;
997 		}
998 
999 		err = got_pathlist_append(&wanted_refs, refname, NULL);
1000 		if (err) {
1001 			free(refname);
1002 			goto done;
1003 		}
1004 
1005 		imsg_free(&imsg);
1006 	}
1007 
1008 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
1009 	if (err) {
1010 		if (err->code == GOT_ERR_PRIVSEP_PIPE)
1011 			err = NULL;
1012 		goto done;
1013 	}
1014 	if (imsg.hdr.type == GOT_IMSG_STOP)
1015 		goto done;
1016 	if (imsg.hdr.type != GOT_IMSG_FETCH_OUTFD) {
1017 		err = got_error(GOT_ERR_PRIVSEP_MSG);
1018 		goto done;
1019 	}
1020 	if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
1021 		err = got_error(GOT_ERR_PRIVSEP_LEN);
1022 		goto done;
1023 	}
1024 	packfd = imsg.fd;
1025 
1026 	err = fetch_pack(fetchfd, packfd, pack_sha1, &have_refs,
1027 	    fetch_req.fetch_all_branches, &wanted_branches,
1028 	    &wanted_refs, fetch_req.list_refs_only,
1029 	    worktree_branch, &ibuf);
1030 done:
1031 	free(worktree_branch);
1032 	got_pathlist_free(&have_refs, GOT_PATHLIST_FREE_ALL);
1033 	got_pathlist_free(&wanted_branches, GOT_PATHLIST_FREE_PATH);
1034 	if (fetchfd != -1 && close(fetchfd) == -1 && err == NULL)
1035 		err = got_error_from_errno("close");
1036 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
1037 		err = got_error_from_errno("close");
1038 	if (err != NULL)
1039 		got_privsep_send_error(&ibuf, err);
1040 	else
1041 		err = send_fetch_done(&ibuf, pack_sha1);
1042 	if (err != NULL) {
1043 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
1044 		got_privsep_send_error(&ibuf, err);
1045 	}
1046 
1047 	exit(0);
1048 }