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 #include <sys/syslimits.h>
23 
24 #include <stdint.h>
25 #include <errno.h>
26 #include <imsg.h>
27 #include <limits.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 <fcntl.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 
43 #include "got_lib_sha1.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_inflate.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_parse.h"
48 #include "got_lib_privsep.h"
49 
50 #ifndef nitems
51 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
52 #endif
53 
54 #define GOT_PKTMAX	65536
55 
56 struct got_object *indexed;
57 static int chattygit;
58 static char *fetchbranch;
59 static char *upstream = "origin";
60 static struct got_object_id zhash = {.sha1={0}};
61 
62 static char*
63 strip(char *ref)
64 {
65 	return ref;
66 }
67 
68 int
69 readn(int fd, void *buf, size_t n)
70 {
71 	ssize_t r, off;
72 
73 	off = 0;
74 	while (off != n) {
75 		r = read(fd, buf + off, n - off);
76 		if (r < 0)
77 			return -1;
78 		if (r == 0)
79 			return off;
80 		off += r;
81 	}
82 	return off;
83 }
84 
85 int
86 flushpkt(int fd)
87 {
88 	if(chattygit)
89 		fprintf(stderr, "writepkt: 0000\n");
90 	return write(fd, "0000", 4);
91 }
92 
93 
94 int
95 readpkt(int fd, char *buf, int nbuf)
96 {
97 	char len[5];
98 	char *e;
99 	int n, r;
100 
101 	if(readn(fd, len, 4) == -1){
102 		return -1;
103 	}
104 	len[4] = 0;
105 	n = strtol(len, &e, 16);
106 	if(n == 0){
107 		if(chattygit)
108 			fprintf(stderr, "readpkt: 0000\n");
109 		return 0;
110 	}
111 	if(e != len + 4 || n <= 4)
112 		err(1, "invalid packet line length");
113 	n  -= 4;
114 	if(n >= nbuf)
115 		err(1, "buffer too small");
116 	if((r = readn(fd, buf, n)) != n)
117 		return -1;
118 	buf[n] = 0;
119 	if(chattygit)
120 		fprintf(stderr, "readpkt: %s:\t%.*s\n", len, nbuf, buf);
121 	return n;
122 }
123 
124 int
125 writepkt(int fd, char *buf, int nbuf)
126 {
127 	char len[5];
128 	int i;
129 
130 	if (snprintf(len, sizeof(len), "%04x", nbuf + 4) >= sizeof(len))
131 		return -1;
132 	if(write(fd, len, 4) != 4)
133 		return -1;
134 	if(write(fd, buf, nbuf) != nbuf)
135 		return -1;
136 	if(chattygit){
137 		fprintf(stderr, "writepkt: %s:\t", len);
138 		fwrite(buf, 1, nbuf, stderr);
139 		for(i = 0; i < nbuf; i++){
140 			if(isprint(buf[i]))
141 				fputc(buf[i], stderr);
142 		}
143 	}
144 	return 0;
145 }
146 
147 
148 int
149 got_resolve_remote_ref(struct got_object_id *id, char *ref)
150 {
151 	char buf[128], *s;
152 	int r, f;
153 
154 	ref = strip(ref);
155 	if(!got_parse_sha1_digest(id->sha1, ref))
156 		return 0;
157 
158 	/* Slightly special handling: translate remote refs to local ones. */
159 	if (strcmp(ref, "HEAD") == 0) {
160 		if(snprintf(buf, sizeof(buf), ".git/HEAD") >= sizeof(buf))
161 			return -1;
162 	} else if(strstr(ref, "refs/heads") == ref) {
163 		ref += strlen("refs/heads");
164 		if(snprintf(buf, sizeof(buf),
165 		    ".git/refs/remotes/%s/%s", upstream, ref) >= sizeof(buf))
166 			return -1;
167 	} else if(strstr(ref, "refs/tags") == ref) {
168 		ref += strlen("refs/tags");
169 		if(snprintf(buf, sizeof(buf),
170 		    ".git/refs/tags/%s/%s", upstream, ref) >= sizeof(buf))
171 			return -1;
172 	} else {
173 		return -1;
174 	}
175 
176 	r = -1;
177 	s = strip(buf);
178 	if((f = open(s, O_RDONLY)) == -1)
179 		goto err;
180 	if(readn(f, buf, sizeof(buf)) < 40)
181 		goto err;
182 	if(!got_parse_sha1_digest(id->sha1, buf))
183 		goto err;
184 err:
185 	close(f);
186 	if(r == -1 && strstr(buf, "ref:") == buf)
187 		return got_resolve_remote_ref(id, buf + strlen("ref:"));
188 	return r;
189 }
190 
191 static int
192 got_check_pack_hash(int fd, size_t sz, uint8_t *hcomp)
193 {
194 	SHA1_CTX ctx;
195 	uint8_t hexpect[SHA1_DIGEST_LENGTH];
196 	char s1[SHA1_DIGEST_STRING_LENGTH + 1];
197 	char s2[SHA1_DIGEST_STRING_LENGTH + 1];
198 	uint8_t buf[32*1024];
199 	ssize_t n, r, nr;
200 
201 	if(sz < 28)
202 		return -1;
203 
204 	n = 0;
205 	SHA1Init(&ctx);
206 	while(n < sz - 20){
207 		nr = sizeof(buf);
208 		if(sz - n - 20 < sizeof(buf))
209 			nr = sz - n - 20;
210 		r = readn(fd, buf, nr);
211 		if(r != nr)
212 			return -1;
213 		SHA1Update(&ctx, buf, nr);
214 		n += r;
215 	}
216 	SHA1Final(hcomp, &ctx);
217 
218 	if(readn(fd, hexpect, sizeof(hexpect)) != sizeof(hexpect))
219 		errx(1, "truncated packfile");
220 	if(memcmp(hcomp, hexpect, SHA1_DIGEST_LENGTH) != 0){
221 		got_sha1_digest_to_str(hcomp, s1, sizeof(s1));
222 		got_sha1_digest_to_str(hexpect, s2, sizeof(s2));
223 		printf("hash mismatch %s != %s\n", s1, s2);
224 		return -1;
225 	}
226 	return 0;
227 }
228 
229 int
230 got_has_object(struct got_object_id *obj)
231 {
232 	return 0;
233 }
234 
235 /*static */int
236 got_make_pack_dir(char *path)
237 {
238 	char s[128];
239 	char *p;
240 
241 	if(snprintf(s, sizeof(s), "%s", path) >= sizeof(s))
242 		return -1;
243 	for(p=strchr(s+1, '/'); p; p=strchr(p+1, '/')){
244 		*p = 0;
245 		if (mkdir(s, 0755) == -1)
246 			if(errno != EEXIST)
247 				return -1;
248 		*p = '/';
249 	}
250 	return 0;
251 }
252 
253 static int
254 got_match_branch(char *br, char *pat)
255 {
256 	char name[128];
257 
258 	if(strstr(pat, "refs/heads") == pat) {
259 		if (snprintf(name, sizeof(name), "%s", pat) >= sizeof(name))
260 			return -1;
261 	} else if(strstr(pat, "heads")) {
262 		if (snprintf(name, sizeof(name), "refs/%s", pat) >= sizeof(name))
263 			return -1;
264 	} else {
265 		if (snprintf(name, sizeof(name), "refs/heads/%s", pat) >= sizeof(name))
266 			return -1;
267 	}
268 	return strcmp(br, name) == 0;
269 }
270 
271 static const struct got_error *
272 got_tokenize_refline(char **tokens, char *line, int len)
273 {
274 	const struct got_error *err = NULL;
275 	char *p;
276 	size_t i, n = 0;
277 	const int maxtokens = 3;
278 
279 	for (i = 0; i < maxtokens; i++)
280 		tokens[i] = NULL;
281 
282 	for (i = 0; n < len && i < maxtokens; i++) {
283 		while (isspace(*line)) {
284 			line++;
285 			n++;
286 		}
287 		p = line;
288 		while (*line != '\0' &&
289 		    (!isspace(*line) || i == maxtokens - 1)) {
290 			line++;
291 			n++;
292 		}
293 		tokens[i] = strndup(p, line - p);
294 		if (tokens[i] == NULL) {
295 			err = got_error_from_errno("strndup");
296 			goto done;
297 		}
298 		/* Skip \0 field-delimiter at end of token. */
299 		while (line[0] == '\0' && n < len) {
300 			line++;
301 			n++;
302 		}
303 	}
304 	if (i <= 2)
305 		err = got_error(GOT_ERR_NOT_REF);
306 done:
307 	if (err) {
308 		int j;
309 		for (j = 0; j < i; j++)
310 			free(tokens[j]);
311 			tokens[j] = NULL;
312 	}
313 	return err;
314 }
315 
316 struct got_capability {
317 	const char *key;
318 	const char *value;
319 };
320 static const struct got_capability got_capabilities[] = {
321 #if 0 /* got-index-pack is not ready for this */
322 	{ "ofs-delta", NULL },
323 #endif
324 	{ "agent", "got/" GOT_VERSION_STR },
325 };
326 
327 static const struct got_error *
328 match_capability(char **my_capabilities, const char *capa,
329     const struct got_capability *mycapa)
330 {
331 	char *equalsign;
332 	char *s;
333 
334 	equalsign = strchr(capa, '=');
335 	if (equalsign) {
336 		if (strncmp(capa, mycapa->key, equalsign - capa) != 0)
337 			return NULL;
338 	} else {
339 		if (strcmp(capa, mycapa->key) != 0)
340 			return NULL;
341 	}
342 
343 	if (asprintf(&s, "%s%s%s%s%s",
344 	    *my_capabilities != NULL ? *my_capabilities : "",
345 	    *my_capabilities != NULL ? " " : "",
346 	    mycapa->key,
347 	    mycapa->value != NULL ? "=" : "",
348 	    mycapa->value != NULL? mycapa->value : "") == -1)
349 		return got_error_from_errno("asprintf");
350 
351 	free(*my_capabilities);
352 	*my_capabilities = s;
353 	return NULL;
354 }
355 
356 static const struct got_error *
357 add_symref(struct got_pathlist_head *symrefs, const char *capa)
358 {
359 	const struct got_error *err = NULL;
360 	char *colon, *name = NULL, *target = NULL;
361 
362 	/* Need at least "A:B" */
363 	if (strlen(capa) < 3)
364 		return NULL;
365 
366 	colon = strchr(capa, ':');
367 	if (colon == NULL)
368 		return NULL;
369 
370 	*colon = '\0';
371 	name = strdup(capa);
372 	if (name == NULL)
373 		return got_error_from_errno("strdup");
374 
375 	target = strdup(colon + 1);
376 	if (target == NULL) {
377 		err = got_error_from_errno("strdup");
378 		goto done;
379 	}
380 
381 	/* We can't validate the ref itself here. The main process will. */
382 	err = got_pathlist_append(symrefs, name, target);
383 done:
384 	if (err) {
385 		free(name);
386 		free(target);
387 	}
388 	return err;
389 }
390 
391 static const struct got_error *
392 match_capabilities(char **my_capabilities, struct got_pathlist_head *symrefs,
393     char *server_capabilities)
394 {
395 	const struct got_error *err = NULL;
396 	char *capa, *equalsign;
397 	int i;
398 
399 	*my_capabilities = NULL;
400 	do {
401 		capa = strsep(&server_capabilities, " ");
402 		if (capa == NULL)
403 			return NULL;
404 
405 		equalsign = strchr(capa, '=');
406 		if (equalsign != NULL &&
407 		    strncmp(capa, "symref", equalsign - capa) == 0) {
408 			err = add_symref(symrefs, equalsign + 1);
409 			if (err)
410 				break;
411 			continue;
412 		}
413 
414 		for (i = 0; i < nitems(got_capabilities); i++) {
415 			err = match_capability(my_capabilities,
416 			    capa, &got_capabilities[i]);
417 			if (err)
418 				break;
419 		}
420 	} while (capa);
421 
422 	return err;
423 }
424 
425 static const struct got_error *
426 fetch_pack(int fd, int packfd, struct got_object_id *packid,
427     struct imsgbuf *ibuf)
428 {
429 	const struct got_error *err = NULL;
430 	char buf[GOT_PKTMAX], *sp[3];
431 	char hashstr[SHA1_DIGEST_STRING_LENGTH];
432 	struct got_object_id *have, *want;
433 	int is_firstpkt = 1, nref = 0, refsz = 16;
434 	int i, n, req;
435 	off_t packsz;
436 	char *my_capabilities = NULL;
437 	struct got_pathlist_head symrefs;
438 	struct got_pathlist_entry *pe;
439 
440 	TAILQ_INIT(&symrefs);
441 
442 	have = malloc(refsz * sizeof(have[0]));
443 	if (have == NULL)
444 		return got_error_from_errno("malloc");
445 	want = malloc(refsz * sizeof(want[0]));
446 	if (want == NULL) {
447 		err = got_error_from_errno("malloc");
448 		goto done;
449 	}
450 	if (chattygit)
451 		fprintf(stderr, "starting fetch\n");
452 	while (1) {
453 		n = readpkt(fd, buf, sizeof(buf));
454 		if (n == -1){
455 			err = got_error_from_errno("readpkt");
456 			goto done;
457 		}
458 		if (n == 0)
459 			break;
460 		if (n >= 4 && strncmp(buf, "ERR ", 4) == 0) {
461 			static char msg[1024];
462 			for (i = 0; i < n && i < sizeof(msg) - 1; i++) {
463 				if (!isprint(buf[i])) {
464 					err = got_error(GOT_ERR_FETCH_FAILED);
465 					goto done;
466 				}
467 				msg[i] = buf[i];
468 			}
469 			msg[i] = '\0';
470 			err = got_error_msg(GOT_ERR_FETCH_FAILED, msg);
471 			goto done;
472 		}
473 		err = got_tokenize_refline(sp, buf, n);
474 		if (err)
475 			goto done;
476 		if (chattygit && sp[2][0] != '\0')
477 			fprintf(stderr, "server capabilities: %s\n", sp[2]);
478 		if (is_firstpkt) {
479 			err = match_capabilities(&my_capabilities, &symrefs,
480 			    sp[2]);
481 			if (err)
482 				goto done;
483 			if (chattygit && my_capabilities)
484 				fprintf(stderr, "my matched capabilities: %s\n",
485 				    my_capabilities);
486 			err = got_privsep_send_fetch_symrefs(ibuf, &symrefs);
487 			if (err)
488 				goto done;
489 		}
490 		is_firstpkt = 0;
491 		if (strstr(sp[1], "^{}"))
492 			continue;
493 		if (fetchbranch && !got_match_branch(sp[1], fetchbranch))
494 			continue;
495 		if (refsz == nref + 1){
496 			refsz *= 2;
497 			have = realloc(have, refsz * sizeof(have[0]));
498 			if (have == NULL) {
499 				err = got_error_from_errno("realloc");
500 				goto done;
501 			}
502 			want = realloc(want, refsz * sizeof(want[0]));
503 			if (want == NULL) {
504 				err = got_error_from_errno("realloc");
505 				goto done;
506 			}
507 		}
508 		if (!got_parse_sha1_digest(want[nref].sha1, sp[0])) {
509 			err = got_error(GOT_ERR_BAD_OBJ_ID_STR);
510 			goto done;
511 		}
512 
513 		if (got_resolve_remote_ref(&have[nref], sp[1]) == -1)
514 			memset(&have[nref], 0, sizeof(have[nref]));
515 		err = got_privsep_send_fetch_progress(ibuf, &want[nref], sp[1]);
516 		if (err)
517 			goto done;
518 		if (chattygit)
519 			fprintf(stderr, "remote %s\n", sp[1]);
520 		nref++;
521 	}
522 
523 	req = 0;
524 	for (i = 0; i < nref; i++){
525 		if (got_object_id_cmp(&have[i], &want[i]) == 0)
526 			continue;
527 		if (got_has_object(&want[i]))
528 			continue;
529 		got_sha1_digest_to_str(want[i].sha1, hashstr, sizeof(hashstr));
530 		n = snprintf(buf, sizeof(buf), "want %s%s%s\n", hashstr,
531 		   i == 0 && my_capabilities ? " " : "",
532 		   i == 0 && my_capabilities ? my_capabilities : "");
533 		if (n >= sizeof(buf)) {
534 			err = got_error(GOT_ERR_NO_SPACE);
535 			goto done;
536 		}
537 		if (writepkt(fd, buf, n) == -1) {
538 			err = got_error_from_errno("writepkt");
539 			goto done;
540 		}
541 		req = 1;
542 	}
543 	flushpkt(fd);
544 	for (i = 0; i < nref; i++){
545 		if (got_object_id_cmp(&have[i], &zhash) == 0)
546 			continue;
547 		got_sha1_digest_to_str(want[i].sha1, hashstr, sizeof(hashstr));
548 		n = snprintf(buf, sizeof(buf), "have %s\n", hashstr);
549 		if (n >= sizeof(buf)) {
550 			err = got_error(GOT_ERR_NO_SPACE);
551 			goto done;
552 		}
553 		if (writepkt(fd, buf, n + 1) == -1) {
554 			err = got_error_from_errno("writepkt");
555 			goto done;
556 		}
557 	}
558 	if (!req){
559 		fprintf(stderr, "up to date\n");
560 		flushpkt(fd);
561 	}
562 	n = snprintf(buf, sizeof(buf), "done\n");
563 	if (writepkt(fd, buf, n) == -1) {
564 		err = got_error_from_errno("writepkt");
565 		goto done;
566 	}
567 	if (!req)
568 		return 0;
569 
570 	if ((n = readpkt(fd, buf, sizeof(buf))) == -1) {
571 		err = got_error_from_errno("readpkt");
572 		goto done;
573 	}
574 	buf[n] = 0;
575 
576 	if (chattygit)
577 		fprintf(stderr, "fetching...\n");
578 	packsz = 0;
579 	while (1) {
580 		ssize_t w;
581 		n = readn(fd, buf, sizeof buf);
582 		if (n == 0)
583 			break;
584 		if (n == -1) {
585 			err = got_error_from_errno("readn");
586 			goto done;
587 		}
588 		w = write(packfd, buf, n);
589 		if (w == -1) {
590 			err = got_error_from_errno("write");
591 			goto done;
592 		}
593 		if (w != n) {
594 			err = got_error(GOT_ERR_IO);
595 			goto done;
596 		}
597 		packsz += n;
598 	}
599 	if (lseek(packfd, 0, SEEK_SET) == -1) {
600 		err = got_error_from_errno("lseek");
601 		goto done;
602 	}
603 	if (got_check_pack_hash(packfd, packsz, packid->sha1) == -1)
604 		err = got_error(GOT_ERR_BAD_PACKFILE);
605 done:
606 	TAILQ_FOREACH(pe, &symrefs, entry) {
607 		free((void *)pe->path);
608 		free(pe->data);
609 	}
610 	got_pathlist_free(&symrefs);
611 	free(have);
612 	free(want);
613 	return err;
614 }
615 
616 
617 int
618 main(int argc, char **argv)
619 {
620 	const struct got_error *err = NULL;
621 	int fetchfd, packfd = -1;
622 	struct got_object_id packid;
623 	struct imsgbuf ibuf;
624 	struct imsg imsg;
625 
626 	if(getenv("GOT_DEBUG") != NULL){
627 		fprintf(stderr, "fetch-pack being chatty!\n");
628 		chattygit = 1;
629 	}
630 
631 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
632 	if((err = got_privsep_recv_imsg(&imsg, &ibuf, 0)) != 0) {
633 		if (err->code == GOT_ERR_PRIVSEP_PIPE)
634 			err = NULL;
635 		goto done;
636 	}
637 	if (imsg.hdr.type == GOT_IMSG_STOP)
638 		goto done;
639 	if (imsg.hdr.type != GOT_IMSG_FETCH_REQUEST) {
640 		err = got_error(GOT_ERR_PRIVSEP_MSG);
641 		goto done;
642 	}
643 	if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
644 		err = got_error(GOT_ERR_PRIVSEP_LEN);
645 		goto done;
646 	}
647 	fetchfd = imsg.fd;
648 
649 	if((err = got_privsep_recv_imsg(&imsg, &ibuf, 0)) != 0) {
650 		if (err->code == GOT_ERR_PRIVSEP_PIPE)
651 			err = NULL;
652 		goto done;
653 	}
654 	if (imsg.hdr.type == GOT_IMSG_STOP)
655 		goto done;
656 	if (imsg.hdr.type != GOT_IMSG_TMPFD) {
657 		err = got_error(GOT_ERR_PRIVSEP_MSG);
658 		goto done;
659 	}
660 	if (imsg.hdr.len - IMSG_HEADER_SIZE != 0) {
661 		err = got_error(GOT_ERR_PRIVSEP_LEN);
662 		goto done;
663 	}
664 	packfd = imsg.fd;
665 
666 	err = fetch_pack(fetchfd, packfd, &packid, &ibuf);
667 	if (err)
668 		goto done;
669 done:
670 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
671 		err = got_error_from_errno("close");
672 	if (err != NULL)
673 		got_privsep_send_error(&ibuf, err);
674 	else
675 		err = got_privsep_send_fetch_done(&ibuf, packid);
676 	if(err != NULL) {
677 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
678 		got_privsep_send_error(&ibuf, err);
679 	}
680 
681 	exit(0);
682 }