1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371
|
/* zlib.c
*
* zlib compression algorithm
*
*/
/* lsh, an implementation of the ssh protocol
*
* Copyright (C) 1998 Balzs Scheidler, Niels Mller
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation; either version 2 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#if HAVE_CONFIG_H
#include "config.h"
#endif
#include "compress.h"
#include "format.h"
#include "lsh_string.h"
#include "ssh.h"
#include "string_buffer.h"
#include "werror.h"
#include "xalloc.h"
#if WITH_ZLIB
#if HAVE_ZLIB_H
#include <zlib.h>
#endif
#include <assert.h>
static void do_free_zstream(z_stream *z);
#include "zlib.c.x"
/* GABA:
(class
(name zlib_instance)
(super compress_instance)
(vars
; Fail before producing larger packets than this
(max . uint32_t)
(rate . uint32_t)
(f pointer (function int "z_stream *" int))
(z indirect-special z_stream
#f do_free_zstream)))
*/
/* GABA:
(class
(name zlib_algorithm)
(super compress_algorithm)
(vars
(level . int)))
*/
/* Stored in the opaque pointer. */
struct zlib_type
{
int (*free_func)(z_stream *z);
const char *operation;
};
#define ZLIB_TYPE(z) ((struct zlib_type *)((z)->opaque))
static const struct zlib_type
zlib_inflate = { inflateEnd, "inflate" };
static const struct zlib_type
zlib_deflate = { deflateEnd, "deflate" };
/* zlib memory functions */
static void *
zlib_alloc(void *opaque UNUSED, unsigned int items, unsigned int size)
{
return lsh_space_alloc(items * size);
}
static void
zlib_free(void *opaque UNUSED, void *address)
{
lsh_space_free(address);
}
static void
do_free_zstream(z_stream *z)
{
/* We use the opaque pointer, as there's nothing else to help us
* figure if we should be calling inflateEnd or deflateEnd. */
const struct zlib_type *type = ZLIB_TYPE(z);
int res = type->free_func(z);
if (res != Z_OK)
debug("do_free_zstream (%z): Freeing failed: %z\n",
type->operation, z->msg ? z->msg : "No error");
}
/* Estimates of the resulting packet sizes. We use fixnum arithmetic,
* with one represented as 1<<10=1024. Only rates between 1/16 and 16
* are used. This may be a little too conservative; I have observed
* compression ratios of about 50. */
#define RATE_UNIT 1024
#define RATE_MAX (RATE_UNIT * 16)
#define RATE_MIN (RATE_UNIT / 16)
#define MARGIN 200
#define INSIGNIFICANT 100
static uint32_t estimate_size(uint32_t rate, uint32_t input, uint32_t max)
{
uint32_t guess = rate * input / RATE_UNIT + MARGIN;
return MIN(max, guess);
}
/* Assumes that input is nonzero */
static uint32_t estimate_update(uint32_t rate, uint32_t input, uint32_t output)
{
/* Decay old estimate */
rate = rate * 15 / 16;
/* NOTE: Following the envelope is suboptimal for small inputs. We
* do it only for input packets of reasonable size. This method
* could be improved.
*
* Perhaps a linear combination k * rate + (1-k) estimate, where k
* depends on the size of the sample (i.e. input) would make sense?
* Or use different rate estimates for different lengths? */
if (input > INSIGNIFICANT)
{
uint32_t estimate = output * RATE_UNIT / input;
if (estimate > RATE_MAX)
return RATE_MAX;
/* Follow the "envelope" */
rate = MAX(estimate, rate);
}
return MAX(rate, RATE_MIN);
}
/* Compress incoming data */
static struct lsh_string *
do_zlib(struct compress_instance *c,
struct lsh_string *packet,
int free)
{
CAST(zlib_instance, self, c);
struct string_buffer buffer;
/* LIMIT keeps track of the amount of storage we may still need to
* allocate. To detect that a packet grows unexpectedly large, we
* need a little extra buffer space beyond the maximum size. */
uint32_t limit = self->max + 1;
uint32_t input = lsh_string_length(packet);
uint32_t estimate;
debug("do_zlib (%z): length in: %i\n",
ZLIB_TYPE(&self->z)->operation, input);
if (!input)
{
werror("do_zlib (%z): Compressing empty packet.\n",
ZLIB_TYPE(&self->z)->operation);
return free ? packet : lsh_string_dup(packet);
}
estimate = estimate_size(self->rate, input, limit);
debug("do_zlib (%z): estimate: %i\n",
ZLIB_TYPE(&self->z)->operation,
estimate);
string_buffer_init(&buffer, estimate);
limit -= lsh_string_length(buffer.partial);
/* The cast is needed because zlib, at least version 1.1.4, doesn't
use const */
self->z.next_in = (uint8_t *) lsh_string_data(packet);
self->z.avail_in = input;
for (;;)
{
int rc;
assert(buffer.left);
rc = lsh_string_zlib(buffer.partial, buffer.pos,
self->f, &self->z, Z_SYNC_FLUSH,
buffer.left);
switch (rc)
{
case Z_BUF_ERROR:
/* If avail_in is zero, this just means that all data have
* been flushed. */
if (self->z.avail_in)
werror("do_zlib (%z): Z_BUF_ERROR (probably harmless),\n"
" avail_in = %i, avail_out = %i\n",
ZLIB_TYPE(&self->z)->operation,
self->z.avail_in, self->z.avail_out);
/* Fall through */
case Z_OK:
break;
default:
werror("do_zlib: %z failed: %z\n",
ZLIB_TYPE(&self->z)->operation,
self->z.msg ? self->z.msg : "No error(?)");
if (free)
lsh_string_free(packet);
return NULL;
}
/* NOTE: It's not enough to check that avail_in is zero to
* determine that all data have been flushed. avail_in == 0 and
* avail_out > 0 implies that all data has been flushed, but if
* avail_in == avail_out == 0, we have to allocate more output
* space. */
if (!self->z.avail_in && !self->z.avail_out)
debug("do_zlib (%z): Both avail_in and avail_out are zero.\n",
ZLIB_TYPE(&self->z)->operation);
if (!self->z.avail_out)
{ /* All output space consumed */
uint32_t plength = lsh_string_length(buffer.partial);
if (!limit)
{
werror("do_zlib (%z): Packet grew too large!\n",
ZLIB_TYPE(&self->z)->operation);
if (free)
lsh_string_free(packet);
string_buffer_clear(&buffer);
return NULL;
}
/* Grow to about double size. */
string_buffer_grow(&buffer,
MIN(limit, plength + buffer.total + 100));
limit -= plength;
}
else if (!self->z.avail_in)
{ /* Compressed entire packet */
uint32_t output;
if (free)
lsh_string_free(packet);
packet =
string_buffer_final(&buffer, self->z.avail_out);
output = lsh_string_length(packet);
assert(output <= self->max);
debug("do_zlib (%z): length out: %i\n",
ZLIB_TYPE(&self->z)->operation,
output);
if (output > estimate)
debug("do_zlib (%z): Estimated size exceeded: input = %i, estimate = %i, output = %i\n",
ZLIB_TYPE(&self->z)->operation,
input, estimate, output);
self->rate = estimate_update(self->rate, input, output);
return packet;
}
}
}
static struct compress_instance *
make_zlib_instance(struct compress_algorithm *c, int mode)
{
CAST(zlib_algorithm, closure, c);
NEW(zlib_instance, res);
res->z.zalloc = zlib_alloc;
res->z.zfree = zlib_free;
switch (mode)
{
case COMPRESS_DEFLATE:
res->z.opaque = (void *) &zlib_deflate;
res->f = deflate;
res->super.codec = do_zlib;
/* Maximum expansion is 0.1% + 12 bytes. We use 1% + 12, to be
* conservative.
*
* FIXME: These figures are documented for the entire stream,
* does they really apply to all segments, separated by
* Z_SYNC_FLUSH ? */
res->max = SSH_MAX_PACKET + SSH_MAX_PACKET / 100 + 12;
res->rate = RATE_UNIT;
if (deflateInit(&res->z, closure->level) != Z_OK)
{
werror("deflateInit failed: %z\n",
res->z.msg ? res->z.msg : "No error(?)");
KILL(res);
return NULL;
}
break;
case COMPRESS_INFLATE:
res->z.opaque = (void *) &zlib_inflate;
res->f = inflate;
res->super.codec = do_zlib;
/* FIXME: Perhaps we ought to use the connection's
* rec_max_packet size? */
res->max = SSH_MAX_PACKET;
res->rate = 2 * RATE_UNIT;
if (inflateInit(&res->z) != Z_OK)
{
werror("inflateInit failed: %z\n",
res->z.msg ? res->z.msg : "No error(?)");
KILL(res);
return NULL;
}
break;
}
return &res->super;
}
struct compress_algorithm *make_zlib_algorithm(int level)
{
if ( (level != Z_DEFAULT_COMPRESSION)
&& ( (level < Z_NO_COMPRESSION)
|| (level > Z_BEST_COMPRESSION) ))
return NULL;
else
{
NEW(zlib_algorithm, closure);
closure->super.make_compress = make_zlib_instance;
closure->level = level;
return &closure->super;
}
}
struct compress_algorithm *make_zlib(void)
{
return make_zlib_algorithm(Z_DEFAULT_COMPRESSION);
}
#endif /* WITH_ZLIB */
|