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