xref: /freebsd/sys/kern/subr_compressor.c (revision 05f0f0e9) 
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2014, 2017 Mark Johnston <markj@FreeBSD.org>
5  * Copyright (c) 2017 Conrad Meyer <cem@FreeBSD.org>
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions are
9  * met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in
14  *    the documentation and/or other materials provided with the
15  *    distribution.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 
30 /*
31  * Subroutines used for writing compressed user process and kernel core dumps.
32  */
33 
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
36 
37 #include "opt_gzio.h"
38 #include "opt_zstdio.h"
39 
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 
43 #include <sys/compressor.h>
44 #include <sys/kernel.h>
45 #include <sys/linker_set.h>
46 #include <sys/malloc.h>
47 
48 MALLOC_DEFINE(M_COMPRESS, "compressor", "kernel compression subroutines");
49 
50 struct compressor_methods {
51 	int format;
52 	void *(* const init)(size_t, int);
53 	void (* const reset)(void *);
54 	int (* const write)(void *, void *, size_t, compressor_cb_t, void *);
55 	void (* const fini)(void *);
56 };
57 
58 struct compressor {
59 	const struct compressor_methods *methods;
60 	compressor_cb_t cb;
61 	void *priv;
62 	void *arg;
63 };
64 
65 SET_DECLARE(compressors, struct compressor_methods);
66 
67 #ifdef GZIO
68 
69 #include <sys/zutil.h>
70 
71 struct gz_stream {
72 	uint8_t		*gz_buffer;	/* output buffer */
73 	size_t		gz_bufsz;	/* output buffer size */
74 	off_t		gz_off;		/* offset into the output stream */
75 	uint32_t	gz_crc;		/* stream CRC32 */
76 	z_stream	gz_stream;	/* zlib state */
77 };
78 
79 static void 	*gz_init(size_t maxiosize, int level);
80 static void	gz_reset(void *stream);
81 static int	gz_write(void *stream, void *data, size_t len, compressor_cb_t,
82 		    void *);
83 static void	gz_fini(void *stream);
84 
85 static void *
86 gz_alloc(void *arg __unused, u_int n, u_int sz)
87 {
88 
89 	/*
90 	 * Memory for zlib state is allocated using M_NODUMP since it may be
91 	 * used to compress a kernel dump, and we don't want zlib to attempt to
92 	 * compress its own state.
93 	 */
94 	return (malloc(n * sz, M_COMPRESS, M_WAITOK | M_ZERO | M_NODUMP));
95 }
96 
97 static void
98 gz_free(void *arg __unused, void *ptr)
99 {
100 
101 	free(ptr, M_COMPRESS);
102 }
103 
104 static void *
105 gz_init(size_t maxiosize, int level)
106 {
107 	struct gz_stream *s;
108 	int error;
109 
110 	s = gz_alloc(NULL, 1, roundup2(sizeof(*s), PAGE_SIZE));
111 	s->gz_buffer = gz_alloc(NULL, 1, maxiosize);
112 	s->gz_bufsz = maxiosize;
113 
114 	s->gz_stream.zalloc = gz_alloc;
115 	s->gz_stream.zfree = gz_free;
116 	s->gz_stream.opaque = NULL;
117 	s->gz_stream.next_in = Z_NULL;
118 	s->gz_stream.avail_in = 0;
119 
120 	error = deflateInit2(&s->gz_stream, level, Z_DEFLATED, -MAX_WBITS,
121 	    DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY);
122 	if (error != 0)
123 		goto fail;
124 
125 	gz_reset(s);
126 
127 	return (s);
128 
129 fail:
130 	gz_free(NULL, s);
131 	return (NULL);
132 }
133 
134 static void
135 gz_reset(void *stream)
136 {
137 	struct gz_stream *s;
138 	uint8_t *hdr;
139 	const size_t hdrlen = 10;
140 
141 	s = stream;
142 	s->gz_off = 0;
143 	s->gz_crc = ~0U;
144 
145 	(void)deflateReset(&s->gz_stream);
146 	s->gz_stream.avail_out = s->gz_bufsz;
147 	s->gz_stream.next_out = s->gz_buffer;
148 
149 	/* Write the gzip header to the output buffer. */
150 	hdr = s->gz_buffer;
151 	memset(hdr, 0, hdrlen);
152 	hdr[0] = 0x1f;
153 	hdr[1] = 0x8b;
154 	hdr[2] = Z_DEFLATED;
155 	hdr[9] = OS_CODE;
156 	s->gz_stream.next_out += hdrlen;
157 	s->gz_stream.avail_out -= hdrlen;
158 }
159 
160 static int
161 gz_write(void *stream, void *data, size_t len, compressor_cb_t cb,
162     void *arg)
163 {
164 	struct gz_stream *s;
165 	uint8_t trailer[8];
166 	size_t room;
167 	int error, zerror, zflag;
168 
169 	s = stream;
170 	zflag = data == NULL ? Z_FINISH : Z_NO_FLUSH;
171 
172 	if (len > 0) {
173 		s->gz_stream.avail_in = len;
174 		s->gz_stream.next_in = data;
175 		s->gz_crc = crc32_raw(data, len, s->gz_crc);
176 	} else
177 		s->gz_crc ^= ~0U;
178 
179 	error = 0;
180 	do {
181 		zerror = deflate(&s->gz_stream, zflag);
182 		if (zerror != Z_OK && zerror != Z_STREAM_END) {
183 			error = EIO;
184 			break;
185 		}
186 
187 		if (s->gz_stream.avail_out == 0 || zerror == Z_STREAM_END) {
188 			/*
189 			 * Our output buffer is full or there's nothing left
190 			 * to produce, so we're flushing the buffer.
191 			 */
192 			len = s->gz_bufsz - s->gz_stream.avail_out;
193 			if (zerror == Z_STREAM_END) {
194 				/*
195 				 * Try to pack as much of the trailer into the
196 				 * output buffer as we can.
197 				 */
198 				((uint32_t *)trailer)[0] = s->gz_crc;
199 				((uint32_t *)trailer)[1] =
200 				    s->gz_stream.total_in;
201 				room = MIN(sizeof(trailer),
202 				    s->gz_bufsz - len);
203 				memcpy(s->gz_buffer + len, trailer, room);
204 				len += room;
205 			}
206 
207 			error = cb(s->gz_buffer, len, s->gz_off, arg);
208 			if (error != 0)
209 				break;
210 
211 			s->gz_off += len;
212 			s->gz_stream.next_out = s->gz_buffer;
213 			s->gz_stream.avail_out = s->gz_bufsz;
214 
215 			/*
216 			 * If we couldn't pack the trailer into the output
217 			 * buffer, write it out now.
218 			 */
219 			if (zerror == Z_STREAM_END && room < sizeof(trailer))
220 				error = cb(trailer + room,
221 				    sizeof(trailer) - room, s->gz_off, arg);
222 		}
223 	} while (zerror != Z_STREAM_END &&
224 	    (zflag == Z_FINISH || s->gz_stream.avail_in > 0));
225 
226 	return (error);
227 }
228 
229 static void
230 gz_fini(void *stream)
231 {
232 	struct gz_stream *s;
233 
234 	s = stream;
235 	(void)deflateEnd(&s->gz_stream);
236 	gz_free(NULL, s->gz_buffer);
237 	gz_free(NULL, s);
238 }
239 
240 struct compressor_methods gzip_methods = {
241 	.format = COMPRESS_GZIP,
242 	.init = gz_init,
243 	.reset = gz_reset,
244 	.write = gz_write,
245 	.fini = gz_fini,
246 };
247 DATA_SET(compressors, gzip_methods);
248 
249 #endif /* GZIO */
250 
251 #ifdef ZSTDIO
252 
253 #define	ZSTD_STATIC_LINKING_ONLY
254 #include <contrib/zstd/lib/zstd.h>
255 
256 struct zstdio_stream {
257 	ZSTD_CCtx	*zst_stream;
258 	ZSTD_inBuffer	zst_inbuffer;
259 	ZSTD_outBuffer	zst_outbuffer;
260 	uint8_t *	zst_buffer;	/* output buffer */
261 	size_t		zst_maxiosz;	/* Max output IO size */
262 	off_t		zst_off;	/* offset into the output stream */
263 	void *		zst_static_wkspc;
264 };
265 
266 static void 	*zstdio_init(size_t maxiosize, int level);
267 static void	zstdio_reset(void *stream);
268 static int	zstdio_write(void *stream, void *data, size_t len,
269 		    compressor_cb_t, void *);
270 static void	zstdio_fini(void *stream);
271 
272 static void *
273 zstdio_init(size_t maxiosize, int level)
274 {
275 	ZSTD_CCtx *dump_compressor;
276 	struct zstdio_stream *s;
277 	void *wkspc, *owkspc, *buffer;
278 	size_t wkspc_size, buf_size;
279 
280 	wkspc_size = ZSTD_estimateCStreamSize(level);
281 	owkspc = wkspc = malloc(wkspc_size + 8, M_COMPRESS,
282 	    M_WAITOK | M_NODUMP);
283 	/* Zstd API requires 8-byte alignment. */
284 	if ((uintptr_t)wkspc % 8 != 0)
285 		wkspc = (void *)roundup2((uintptr_t)wkspc, 8);
286 
287 	dump_compressor = ZSTD_initStaticCCtx(wkspc, wkspc_size);
288 	if (dump_compressor == NULL) {
289 		free(owkspc, M_COMPRESS);
290 		printf("%s: workspace too small.\n", __func__);
291 		return (NULL);
292 	}
293 
294 	(void)ZSTD_CCtx_setParameter(dump_compressor, ZSTD_p_checksumFlag, 1);
295 
296 	buf_size = ZSTD_CStreamOutSize() * 2;
297 	buffer = malloc(buf_size, M_COMPRESS, M_WAITOK | M_NODUMP);
298 
299 	s = malloc(sizeof(*s), M_COMPRESS, M_NODUMP | M_WAITOK);
300 	s->zst_buffer = buffer;
301 	s->zst_outbuffer.dst = buffer;
302 	s->zst_outbuffer.size = buf_size;
303 	s->zst_maxiosz = maxiosize;
304 	s->zst_stream = dump_compressor;
305 	s->zst_static_wkspc = owkspc;
306 
307 	zstdio_reset(s);
308 
309 	return (s);
310 }
311 
312 static void
313 zstdio_reset(void *stream)
314 {
315 	struct zstdio_stream *s;
316 	size_t res;
317 
318 	s = stream;
319 	res = ZSTD_resetCStream(s->zst_stream, 0);
320 	if (ZSTD_isError(res))
321 		panic("%s: could not reset stream %p: %s\n", __func__, s,
322 		    ZSTD_getErrorName(res));
323 
324 	s->zst_off = 0;
325 	s->zst_inbuffer.src = NULL;
326 	s->zst_inbuffer.size = 0;
327 	s->zst_inbuffer.pos = 0;
328 	s->zst_outbuffer.pos = 0;
329 }
330 
331 static int
332 zst_flush_intermediate(struct zstdio_stream *s, compressor_cb_t cb, void *arg)
333 {
334 	size_t bytes_to_dump;
335 	int error;
336 
337 	/* Flush as many full output blocks as possible. */
338 	/* XXX: 4096 is arbitrary safe HDD block size for kernel dumps */
339 	while (s->zst_outbuffer.pos >= 4096) {
340 		bytes_to_dump = rounddown(s->zst_outbuffer.pos, 4096);
341 
342 		if (bytes_to_dump > s->zst_maxiosz)
343 			bytes_to_dump = s->zst_maxiosz;
344 
345 		error = cb(s->zst_buffer, bytes_to_dump, s->zst_off, arg);
346 		if (error != 0)
347 			return (error);
348 
349 		/*
350 		 * Shift any non-full blocks up to the front of the output
351 		 * buffer.
352 		 */
353 		s->zst_outbuffer.pos -= bytes_to_dump;
354 		memmove(s->zst_outbuffer.dst,
355 		    (char *)s->zst_outbuffer.dst + bytes_to_dump,
356 		    s->zst_outbuffer.pos);
357 		s->zst_off += bytes_to_dump;
358 	}
359 	return (0);
360 }
361 
362 static int
363 zstdio_flush(struct zstdio_stream *s, compressor_cb_t cb, void *arg)
364 {
365 	size_t rc, lastpos;
366 	int error;
367 
368 	/*
369 	 * Positive return indicates unflushed data remaining; need to call
370 	 * endStream again after clearing out room in output buffer.
371 	 */
372 	rc = 1;
373 	lastpos = s->zst_outbuffer.pos;
374 	while (rc > 0) {
375 		rc = ZSTD_endStream(s->zst_stream, &s->zst_outbuffer);
376 		if (ZSTD_isError(rc)) {
377 			printf("%s: ZSTD_endStream failed (%s)\n", __func__,
378 			    ZSTD_getErrorName(rc));
379 			return (EIO);
380 		}
381 		if (lastpos == s->zst_outbuffer.pos) {
382 			printf("%s: did not make forward progress endStream %zu\n",
383 			    __func__, lastpos);
384 			return (EIO);
385 		}
386 
387 		error = zst_flush_intermediate(s, cb, arg);
388 		if (error != 0)
389 			return (error);
390 
391 		lastpos = s->zst_outbuffer.pos;
392 	}
393 
394 	/*
395 	 * We've already done an intermediate flush, so all full blocks have
396 	 * been written.  Only a partial block remains.  Padding happens in a
397 	 * higher layer.
398 	 */
399 	if (s->zst_outbuffer.pos != 0) {
400 		error = cb(s->zst_buffer, s->zst_outbuffer.pos, s->zst_off,
401 		    arg);
402 		if (error != 0)
403 			return (error);
404 	}
405 
406 	return (0);
407 }
408 
409 static int
410 zstdio_write(void *stream, void *data, size_t len, compressor_cb_t cb,
411     void *arg)
412 {
413 	struct zstdio_stream *s;
414 	size_t lastpos, rc;
415 	int error;
416 
417 	s = stream;
418 	if (data == NULL)
419 		return (zstdio_flush(s, cb, arg));
420 
421 	s->zst_inbuffer.src = data;
422 	s->zst_inbuffer.size = len;
423 	s->zst_inbuffer.pos = 0;
424 	lastpos = 0;
425 
426 	while (s->zst_inbuffer.pos < s->zst_inbuffer.size) {
427 		rc = ZSTD_compressStream(s->zst_stream, &s->zst_outbuffer,
428 		    &s->zst_inbuffer);
429 		if (ZSTD_isError(rc)) {
430 			printf("%s: Compress failed on %p! (%s)\n",
431 			    __func__, data, ZSTD_getErrorName(rc));
432 			return (EIO);
433 		}
434 
435 		if (lastpos == s->zst_inbuffer.pos) {
436 			/*
437 			 * XXX: May need flushStream to make forward progress
438 			 */
439 			printf("ZSTD: did not make forward progress @pos %zu\n",
440 			    lastpos);
441 			return (EIO);
442 		}
443 		lastpos = s->zst_inbuffer.pos;
444 
445 		error = zst_flush_intermediate(s, cb, arg);
446 		if (error != 0)
447 			return (error);
448 	}
449 	return (0);
450 }
451 
452 static void
453 zstdio_fini(void *stream)
454 {
455 	struct zstdio_stream *s;
456 
457 	s = stream;
458 	if (s->zst_static_wkspc != NULL)
459 		free(s->zst_static_wkspc, M_COMPRESS);
460 	else
461 		ZSTD_freeCCtx(s->zst_stream);
462 	free(s->zst_buffer, M_COMPRESS);
463 	free(s, M_COMPRESS);
464 }
465 
466 static struct compressor_methods zstd_methods = {
467 	.format = COMPRESS_ZSTD,
468 	.init = zstdio_init,
469 	.reset = zstdio_reset,
470 	.write = zstdio_write,
471 	.fini = zstdio_fini,
472 };
473 DATA_SET(compressors, zstd_methods);
474 
475 #endif /* ZSTDIO */
476 
477 bool
478 compressor_avail(int format)
479 {
480 	struct compressor_methods **iter;
481 
482 	SET_FOREACH(iter, compressors) {
483 		if ((*iter)->format == format)
484 			return (true);
485 	}
486 	return (false);
487 }
488 
489 struct compressor *
490 compressor_init(compressor_cb_t cb, int format, size_t maxiosize, int level,
491     void *arg)
492 {
493 	struct compressor_methods **iter;
494 	struct compressor *s;
495 	void *priv;
496 
497 	SET_FOREACH(iter, compressors) {
498 		if ((*iter)->format == format)
499 			break;
500 	}
501 	if (iter == SET_LIMIT(compressors))
502 		return (NULL);
503 
504 	priv = (*iter)->init(maxiosize, level);
505 	if (priv == NULL)
506 		return (NULL);
507 
508 	s = malloc(sizeof(*s), M_COMPRESS, M_WAITOK | M_ZERO);
509 	s->methods = (*iter);
510 	s->priv = priv;
511 	s->cb = cb;
512 	s->arg = arg;
513 	return (s);
514 }
515 
516 void
517 compressor_reset(struct compressor *stream)
518 {
519 
520 	stream->methods->reset(stream->priv);
521 }
522 
523 int
524 compressor_write(struct compressor *stream, void *data, size_t len)
525 {
526 
527 	return (stream->methods->write(stream->priv, data, len, stream->cb,
528 	    stream->arg));
529 }
530 
531 int
532 compressor_flush(struct compressor *stream)
533 {
534 
535 	return (stream->methods->write(stream->priv, NULL, 0, stream->cb,
536 	    stream->arg));
537 }
538 
539 void
540 compressor_fini(struct compressor *stream)
541 {
542 
543 	stream->methods->fini(stream->priv);
544 }
545