2 2181e0c8 2019-03-19 stsp * Copyright (c) 2019 Stefan Sperling <stsp@openbsd.org>
4 2181e0c8 2019-03-19 stsp * Permission to use, copy, modify, and distribute this software for any
5 2181e0c8 2019-03-19 stsp * purpose with or without fee is hereby granted, provided that the above
6 2181e0c8 2019-03-19 stsp * copyright notice and this permission notice appear in all copies.
8 2181e0c8 2019-03-19 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 2181e0c8 2019-03-19 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 2181e0c8 2019-03-19 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 2181e0c8 2019-03-19 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 2181e0c8 2019-03-19 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 2181e0c8 2019-03-19 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 2181e0c8 2019-03-19 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
19 2181e0c8 2019-03-19 stsp #include <errno.h>
20 2181e0c8 2019-03-19 stsp #include <stdio.h>
21 2181e0c8 2019-03-19 stsp #include <stdlib.h>
22 2181e0c8 2019-03-19 stsp #include <string.h>
23 2181e0c8 2019-03-19 stsp #include <zlib.h>
24 2181e0c8 2019-03-19 stsp #include <time.h>
26 2181e0c8 2019-03-19 stsp #include "got_error.h"
27 2181e0c8 2019-03-19 stsp #include "got_object.h"
28 324d37e7 2019-05-11 stsp #include "got_path.h"
30 2181e0c8 2019-03-19 stsp #include "got_lib_deflate.h"
33 2181e0c8 2019-03-19 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
36 2181e0c8 2019-03-19 stsp const struct got_error *
37 3b9e6fcf 2021-06-05 stsp got_deflate_init(struct got_deflate_buf *zb, uint8_t *outbuf, size_t bufsize)
39 2181e0c8 2019-03-19 stsp const struct got_error *err = NULL;
42 2181e0c8 2019-03-19 stsp memset(&zb->z, 0, sizeof(zb->z));
44 2181e0c8 2019-03-19 stsp zb->z.zalloc = Z_NULL;
45 2181e0c8 2019-03-19 stsp zb->z.zfree = Z_NULL;
46 2181e0c8 2019-03-19 stsp zerr = deflateInit(&zb->z, Z_DEFAULT_COMPRESSION);
47 2181e0c8 2019-03-19 stsp if (zerr != Z_OK) {
48 2181e0c8 2019-03-19 stsp if (zerr == Z_ERRNO)
49 638f9024 2019-05-13 stsp return got_error_from_errno("deflateInit");
50 2181e0c8 2019-03-19 stsp if (zerr == Z_MEM_ERROR) {
51 2181e0c8 2019-03-19 stsp errno = ENOMEM;
52 638f9024 2019-05-13 stsp return got_error_from_errno("deflateInit");
54 2181e0c8 2019-03-19 stsp return got_error(GOT_ERR_COMPRESSION);
57 2181e0c8 2019-03-19 stsp zb->inlen = zb->outlen = bufsize;
59 2181e0c8 2019-03-19 stsp zb->inbuf = calloc(1, zb->inlen);
60 2181e0c8 2019-03-19 stsp if (zb->inbuf == NULL) {
61 638f9024 2019-05-13 stsp err = got_error_from_errno("calloc");
65 2181e0c8 2019-03-19 stsp zb->flags = 0;
66 2181e0c8 2019-03-19 stsp if (outbuf == NULL) {
67 2181e0c8 2019-03-19 stsp zb->outbuf = calloc(1, zb->outlen);
68 2181e0c8 2019-03-19 stsp if (zb->outbuf == NULL) {
69 638f9024 2019-05-13 stsp err = got_error_from_errno("calloc");
72 2181e0c8 2019-03-19 stsp zb->flags |= GOT_DEFLATE_F_OWN_OUTBUF;
74 2181e0c8 2019-03-19 stsp zb->outbuf = outbuf;
77 2181e0c8 2019-03-19 stsp got_deflate_end(zb);
82 dbaa2362 2021-09-28 thomas csum_output(struct got_deflate_checksum *csum, const uint8_t *buf, size_t len)
84 91b40e30 2021-05-21 stsp if (csum->output_crc)
85 91b40e30 2021-05-21 stsp *csum->output_crc = crc32(*csum->output_crc, buf, len);
87 91b40e30 2021-05-21 stsp if (csum->output_sha1)
88 91b40e30 2021-05-21 stsp SHA1Update(csum->output_sha1, buf, len);
91 2181e0c8 2019-03-19 stsp const struct got_error *
92 e8f02263 2022-01-23 thomas got_deflate_read(struct got_deflate_buf *zb, FILE *f, off_t len,
93 e8f02263 2022-01-23 thomas size_t *outlenp, off_t *consumed)
95 2181e0c8 2019-03-19 stsp size_t last_total_out = zb->z.total_out;
96 2181e0c8 2019-03-19 stsp z_stream *z = &zb->z;
97 2181e0c8 2019-03-19 stsp int ret = Z_ERRNO;
99 2181e0c8 2019-03-19 stsp z->next_out = zb->outbuf;
100 2181e0c8 2019-03-19 stsp z->avail_out = zb->outlen;
102 2181e0c8 2019-03-19 stsp *outlenp = 0;
103 e8f02263 2022-01-23 thomas *consumed = 0;
105 e8f02263 2022-01-23 thomas size_t last_total_in = z->total_in;
106 2181e0c8 2019-03-19 stsp if (z->avail_in == 0) {
107 e8f02263 2022-01-23 thomas size_t n = 0;
108 e8f02263 2022-01-23 thomas if (*consumed < len) {
109 e8f02263 2022-01-23 thomas n = fread(zb->inbuf, 1,
110 e8f02263 2022-01-23 thomas MIN(zb->inlen, len - *consumed), f);
112 2181e0c8 2019-03-19 stsp if (n == 0) {
113 2181e0c8 2019-03-19 stsp if (ferror(f))
114 2181e0c8 2019-03-19 stsp return got_ferror(f, GOT_ERR_IO);
116 4e4a7005 2019-04-13 stsp ret = deflate(z, Z_FINISH);
119 2181e0c8 2019-03-19 stsp z->next_in = zb->inbuf;
120 2181e0c8 2019-03-19 stsp z->avail_in = n;
122 2181e0c8 2019-03-19 stsp ret = deflate(z, Z_NO_FLUSH);
123 e8f02263 2022-01-23 thomas *consumed += z->total_in - last_total_in;
124 2181e0c8 2019-03-19 stsp } while (ret == Z_OK && z->avail_out > 0);
126 2181e0c8 2019-03-19 stsp if (ret == Z_OK) {
127 2181e0c8 2019-03-19 stsp zb->flags |= GOT_DEFLATE_F_HAVE_MORE;
129 2181e0c8 2019-03-19 stsp if (ret != Z_STREAM_END)
130 2181e0c8 2019-03-19 stsp return got_error(GOT_ERR_COMPRESSION);
131 2181e0c8 2019-03-19 stsp zb->flags &= ~GOT_DEFLATE_F_HAVE_MORE;
134 2181e0c8 2019-03-19 stsp *outlenp = z->total_out - last_total_out;
135 2181e0c8 2019-03-19 stsp return NULL;
138 9249e7e3 2022-05-12 thomas static const struct got_error *
139 9249e7e3 2022-05-12 thomas deflate_read_mmap(struct got_deflate_buf *zb, uint8_t *map, size_t offset,
140 9249e7e3 2022-05-12 thomas size_t len, size_t *outlenp, size_t *consumed, int flush_on_eof)
142 2b0ae357 2022-01-10 thomas z_stream *z = &zb->z;
143 2b0ae357 2022-01-10 thomas size_t last_total_out = z->total_out;
144 2b0ae357 2022-01-10 thomas int ret = Z_ERRNO;
146 2b0ae357 2022-01-10 thomas z->next_out = zb->outbuf;
147 2b0ae357 2022-01-10 thomas z->avail_out = zb->outlen;
149 2b0ae357 2022-01-10 thomas *outlenp = 0;
150 2b0ae357 2022-01-10 thomas *consumed = 0;
152 2b0ae357 2022-01-10 thomas size_t last_total_in = z->total_in;
153 2b0ae357 2022-01-10 thomas if (z->avail_in == 0) {
154 2b0ae357 2022-01-10 thomas z->next_in = map + offset + *consumed;
155 f6a55b40 2022-02-12 thomas if (len - *consumed > UINT_MAX)
156 f6a55b40 2022-02-12 thomas z->avail_in = UINT_MAX;
158 f6a55b40 2022-02-12 thomas z->avail_in = len - *consumed;
159 2b0ae357 2022-01-10 thomas if (z->avail_in == 0) {
160 2b0ae357 2022-01-10 thomas /* EOF */
161 9249e7e3 2022-05-12 thomas if (flush_on_eof)
162 9249e7e3 2022-05-12 thomas ret = deflate(z, Z_FINISH);
166 2b0ae357 2022-01-10 thomas ret = deflate(z, Z_NO_FLUSH);
167 2b0ae357 2022-01-10 thomas *consumed += z->total_in - last_total_in;
168 2b0ae357 2022-01-10 thomas } while (ret == Z_OK && z->avail_out > 0);
170 2b0ae357 2022-01-10 thomas if (ret == Z_OK) {
171 2b0ae357 2022-01-10 thomas zb->flags |= GOT_DEFLATE_F_HAVE_MORE;
173 2b0ae357 2022-01-10 thomas if (ret != Z_STREAM_END)
174 2b0ae357 2022-01-10 thomas return got_error(GOT_ERR_COMPRESSION);
175 2b0ae357 2022-01-10 thomas zb->flags &= ~GOT_DEFLATE_F_HAVE_MORE;
178 2b0ae357 2022-01-10 thomas *outlenp = z->total_out - last_total_out;
179 9249e7e3 2022-05-12 thomas return NULL;
182 9249e7e3 2022-05-12 thomas const struct got_error *
183 9249e7e3 2022-05-12 thomas got_deflate_read_mmap(struct got_deflate_buf *zb, uint8_t *map, size_t offset,
184 9249e7e3 2022-05-12 thomas size_t len, size_t *outlenp, size_t *consumed)
186 9249e7e3 2022-05-12 thomas return deflate_read_mmap(zb, map, offset, len, outlenp, consumed, 1);
189 9249e7e3 2022-05-12 thomas const struct got_error *
190 9249e7e3 2022-05-12 thomas got_deflate_flush(struct got_deflate_buf *zb, FILE *outfile,
191 9249e7e3 2022-05-12 thomas struct got_deflate_checksum *csum, off_t *outlenp)
194 9249e7e3 2022-05-12 thomas size_t n;
195 9249e7e3 2022-05-12 thomas z_stream *z = &zb->z;
197 9249e7e3 2022-05-12 thomas if (z->avail_in != 0)
198 9249e7e3 2022-05-12 thomas return got_error_msg(GOT_ERR_COMPRESSION,
199 9249e7e3 2022-05-12 thomas "cannot flush zb with pending input data");
202 9249e7e3 2022-05-12 thomas size_t avail, last_total_out = zb->z.total_out;
204 9249e7e3 2022-05-12 thomas z->next_out = zb->outbuf;
205 9249e7e3 2022-05-12 thomas z->avail_out = zb->outlen;
207 9249e7e3 2022-05-12 thomas ret = deflate(z, Z_FINISH);
208 9249e7e3 2022-05-12 thomas if (ret != Z_STREAM_END && ret != Z_OK)
209 9249e7e3 2022-05-12 thomas return got_error(GOT_ERR_COMPRESSION);
211 9249e7e3 2022-05-12 thomas avail = z->total_out - last_total_out;
212 9249e7e3 2022-05-12 thomas if (avail > 0) {
213 9249e7e3 2022-05-12 thomas n = fwrite(zb->outbuf, avail, 1, outfile);
214 9249e7e3 2022-05-12 thomas if (n != 1)
215 9249e7e3 2022-05-12 thomas return got_ferror(outfile, GOT_ERR_IO);
216 9249e7e3 2022-05-12 thomas if (csum)
217 9249e7e3 2022-05-12 thomas csum_output(csum, zb->outbuf, avail);
218 9249e7e3 2022-05-12 thomas if (outlenp)
219 9249e7e3 2022-05-12 thomas *outlenp += avail;
221 9249e7e3 2022-05-12 thomas } while (ret != Z_STREAM_END);
223 9249e7e3 2022-05-12 thomas zb->flags &= ~GOT_DEFLATE_F_HAVE_MORE;
224 2b0ae357 2022-01-10 thomas return NULL;
228 2181e0c8 2019-03-19 stsp got_deflate_end(struct got_deflate_buf *zb)
230 2181e0c8 2019-03-19 stsp free(zb->inbuf);
231 2181e0c8 2019-03-19 stsp if (zb->flags & GOT_DEFLATE_F_OWN_OUTBUF)
232 2181e0c8 2019-03-19 stsp free(zb->outbuf);
233 2181e0c8 2019-03-19 stsp deflateEnd(&zb->z);
236 2181e0c8 2019-03-19 stsp const struct got_error *
237 e8f02263 2022-01-23 thomas got_deflate_to_file(off_t *outlen, FILE *infile, off_t len,
238 e8f02263 2022-01-23 thomas FILE *outfile, struct got_deflate_checksum *csum)
240 2181e0c8 2019-03-19 stsp const struct got_error *err;
241 2181e0c8 2019-03-19 stsp size_t avail;
242 e8f02263 2022-01-23 thomas off_t consumed;
243 2181e0c8 2019-03-19 stsp struct got_deflate_buf zb;
245 3b9e6fcf 2021-06-05 stsp err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
249 2181e0c8 2019-03-19 stsp *outlen = 0;
252 e8f02263 2022-01-23 thomas err = got_deflate_read(&zb, infile, len, &avail, &consumed);
255 e8f02263 2022-01-23 thomas len -= consumed;
256 2181e0c8 2019-03-19 stsp if (avail > 0) {
258 2181e0c8 2019-03-19 stsp n = fwrite(zb.outbuf, avail, 1, outfile);
259 2181e0c8 2019-03-19 stsp if (n != 1) {
260 2181e0c8 2019-03-19 stsp err = got_ferror(outfile, GOT_ERR_IO);
264 3b9e6fcf 2021-06-05 stsp csum_output(csum, zb.outbuf, avail);
265 2181e0c8 2019-03-19 stsp *outlen += avail;
267 2181e0c8 2019-03-19 stsp } while (zb.flags & GOT_DEFLATE_F_HAVE_MORE);
270 2181e0c8 2019-03-19 stsp got_deflate_end(&zb);
271 2181e0c8 2019-03-19 stsp return err;
274 2b0ae357 2022-01-10 thomas const struct got_error *
275 e8f02263 2022-01-23 thomas got_deflate_to_file_mmap(off_t *outlen, uint8_t *map, size_t offset,
276 2b0ae357 2022-01-10 thomas size_t len, FILE *outfile, struct got_deflate_checksum *csum)
278 2b0ae357 2022-01-10 thomas const struct got_error *err;
279 2b0ae357 2022-01-10 thomas size_t avail, consumed;
280 2b0ae357 2022-01-10 thomas struct got_deflate_buf zb;
282 2b0ae357 2022-01-10 thomas err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
284 2b0ae357 2022-01-10 thomas goto done;
286 2b0ae357 2022-01-10 thomas *outlen = 0;
288 2b0ae357 2022-01-10 thomas err = got_deflate_read_mmap(&zb, map, offset, len, &avail,
289 2b0ae357 2022-01-10 thomas &consumed);
291 2b0ae357 2022-01-10 thomas goto done;
292 2b0ae357 2022-01-10 thomas offset += consumed;
293 2b0ae357 2022-01-10 thomas len -= consumed;
294 2b0ae357 2022-01-10 thomas if (avail > 0) {
295 2b0ae357 2022-01-10 thomas size_t n;
296 2b0ae357 2022-01-10 thomas n = fwrite(zb.outbuf, avail, 1, outfile);
297 2b0ae357 2022-01-10 thomas if (n != 1) {
298 2b0ae357 2022-01-10 thomas err = got_ferror(outfile, GOT_ERR_IO);
299 2b0ae357 2022-01-10 thomas goto done;
301 2b0ae357 2022-01-10 thomas if (csum)
302 2b0ae357 2022-01-10 thomas csum_output(csum, zb.outbuf, avail);
303 2b0ae357 2022-01-10 thomas *outlen += avail;
305 2b0ae357 2022-01-10 thomas } while (zb.flags & GOT_DEFLATE_F_HAVE_MORE);
308 9249e7e3 2022-05-12 thomas got_deflate_end(&zb);
309 9249e7e3 2022-05-12 thomas return err;
312 9249e7e3 2022-05-12 thomas const struct got_error *
313 9249e7e3 2022-05-12 thomas got_deflate_append_to_file_mmap(struct got_deflate_buf *zb, off_t *outlen,
314 9249e7e3 2022-05-12 thomas uint8_t *map, size_t offset, size_t len, FILE *outfile,
315 9249e7e3 2022-05-12 thomas struct got_deflate_checksum *csum)
317 9249e7e3 2022-05-12 thomas const struct got_error *err;
318 9249e7e3 2022-05-12 thomas size_t avail, consumed;
321 9249e7e3 2022-05-12 thomas err = deflate_read_mmap(zb, map, offset, len, &avail,
322 9249e7e3 2022-05-12 thomas &consumed, 0);
325 9249e7e3 2022-05-12 thomas offset += consumed;
326 9249e7e3 2022-05-12 thomas len -= consumed;
327 9249e7e3 2022-05-12 thomas if (avail > 0) {
328 9249e7e3 2022-05-12 thomas size_t n;
329 9249e7e3 2022-05-12 thomas n = fwrite(zb->outbuf, avail, 1, outfile);
330 9249e7e3 2022-05-12 thomas if (n != 1) {
331 9249e7e3 2022-05-12 thomas err = got_ferror(outfile, GOT_ERR_IO);
334 9249e7e3 2022-05-12 thomas if (csum)
335 9249e7e3 2022-05-12 thomas csum_output(csum, zb->outbuf, avail);
336 9249e7e3 2022-05-12 thomas if (outlen)
337 9249e7e3 2022-05-12 thomas *outlen += avail;
339 9249e7e3 2022-05-12 thomas } while ((zb->flags & GOT_DEFLATE_F_HAVE_MORE) && len > 0);
341 9249e7e3 2022-05-12 thomas return err;
344 9249e7e3 2022-05-12 thomas const struct got_error *
345 9249e7e3 2022-05-12 thomas got_deflate_to_mem_mmap(uint8_t **outbuf, size_t *outlen,
346 9249e7e3 2022-05-12 thomas size_t *consumed_total, struct got_deflate_checksum *csum, uint8_t *map,
347 9249e7e3 2022-05-12 thomas size_t offset, size_t len)
349 9249e7e3 2022-05-12 thomas const struct got_error *err;
350 9249e7e3 2022-05-12 thomas size_t avail, consumed;
351 9249e7e3 2022-05-12 thomas struct got_deflate_buf zb;
352 9249e7e3 2022-05-12 thomas void *newbuf;
353 9249e7e3 2022-05-12 thomas size_t nbuf = 1;
355 9249e7e3 2022-05-12 thomas if (outbuf) {
356 9249e7e3 2022-05-12 thomas *outbuf = malloc(GOT_DEFLATE_BUFSIZE);
357 9249e7e3 2022-05-12 thomas if (*outbuf == NULL)
358 9249e7e3 2022-05-12 thomas return got_error_from_errno("malloc");
359 9249e7e3 2022-05-12 thomas err = got_deflate_init(&zb, *outbuf, GOT_DEFLATE_BUFSIZE);
360 9249e7e3 2022-05-12 thomas if (err) {
361 9249e7e3 2022-05-12 thomas free(*outbuf);
362 9249e7e3 2022-05-12 thomas *outbuf = NULL;
363 9249e7e3 2022-05-12 thomas return err;
366 9249e7e3 2022-05-12 thomas err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
368 9249e7e3 2022-05-12 thomas return err;
371 9249e7e3 2022-05-12 thomas *outlen = 0;
372 9249e7e3 2022-05-12 thomas if (consumed_total)
373 9249e7e3 2022-05-12 thomas *consumed_total = 0;
375 9249e7e3 2022-05-12 thomas err = got_deflate_read_mmap(&zb, map, offset, len, &avail,
376 9249e7e3 2022-05-12 thomas &consumed);
378 9249e7e3 2022-05-12 thomas goto done;
379 9249e7e3 2022-05-12 thomas offset += consumed;
380 9249e7e3 2022-05-12 thomas if (consumed_total)
381 9249e7e3 2022-05-12 thomas *consumed_total += consumed;
382 9249e7e3 2022-05-12 thomas len -= consumed;
383 9249e7e3 2022-05-12 thomas if (avail > 0 && csum)
384 9249e7e3 2022-05-12 thomas csum_output(csum, zb.outbuf, avail);
385 9249e7e3 2022-05-12 thomas *outlen += avail;
386 9249e7e3 2022-05-12 thomas if ((zb.flags & GOT_DEFLATE_F_HAVE_MORE) && outbuf != NULL) {
387 9249e7e3 2022-05-12 thomas newbuf = reallocarray(*outbuf, ++nbuf,
388 9249e7e3 2022-05-12 thomas GOT_DEFLATE_BUFSIZE);
389 9249e7e3 2022-05-12 thomas if (newbuf == NULL) {
390 9249e7e3 2022-05-12 thomas err = got_error_from_errno("reallocarray");
391 9249e7e3 2022-05-12 thomas free(*outbuf);
392 9249e7e3 2022-05-12 thomas *outbuf = NULL;
393 9249e7e3 2022-05-12 thomas *outlen = 0;
394 9249e7e3 2022-05-12 thomas goto done;
396 9249e7e3 2022-05-12 thomas *outbuf = newbuf;
397 9249e7e3 2022-05-12 thomas zb.outbuf = newbuf + *outlen;
398 9249e7e3 2022-05-12 thomas zb.outlen = (nbuf * GOT_DEFLATE_BUFSIZE) - *outlen;
400 9249e7e3 2022-05-12 thomas } while (zb.flags & GOT_DEFLATE_F_HAVE_MORE);
402 2b0ae357 2022-01-10 thomas got_deflate_end(&zb);
403 2b0ae357 2022-01-10 thomas return err;