Root/lib/crc32.c

1/*
2 * Oct 15, 2000 Matt Domsch <Matt_Domsch@dell.com>
3 * Nicer crc32 functions/docs submitted by linux@horizon.com. Thanks!
4 * Code was from the public domain, copyright abandoned. Code was
5 * subsequently included in the kernel, thus was re-licensed under the
6 * GNU GPL v2.
7 *
8 * Oct 12, 2000 Matt Domsch <Matt_Domsch@dell.com>
9 * Same crc32 function was used in 5 other places in the kernel.
10 * I made one version, and deleted the others.
11 * There are various incantations of crc32(). Some use a seed of 0 or ~0.
12 * Some xor at the end with ~0. The generic crc32() function takes
13 * seed as an argument, and doesn't xor at the end. Then individual
14 * users can do whatever they need.
15 * drivers/net/smc9194.c uses seed ~0, doesn't xor with ~0.
16 * fs/jffs2 uses seed 0, doesn't xor with ~0.
17 * fs/partitions/efi.c uses seed ~0, xor's with ~0.
18 *
19 * This source code is licensed under the GNU General Public License,
20 * Version 2. See the file COPYING for more details.
21 */
22
23#include <linux/crc32.h>
24#include <linux/kernel.h>
25#include <linux/module.h>
26#include <linux/compiler.h>
27#include <linux/types.h>
28#include <linux/init.h>
29#include <asm/atomic.h>
30#include "crc32defs.h"
31#if CRC_LE_BITS == 8
32# define tole(x) __constant_cpu_to_le32(x)
33#else
34# define tole(x) (x)
35#endif
36
37#if CRC_BE_BITS == 8
38# define tobe(x) __constant_cpu_to_be32(x)
39#else
40# define tobe(x) (x)
41#endif
42#include "crc32table.h"
43
44MODULE_AUTHOR("Matt Domsch <Matt_Domsch@dell.com>");
45MODULE_DESCRIPTION("Ethernet CRC32 calculations");
46MODULE_LICENSE("GPL");
47
48#if CRC_LE_BITS == 8 || CRC_BE_BITS == 8
49
50static inline u32
51crc32_body(u32 crc, unsigned char const *buf, size_t len, const u32 (*tab)[256])
52{
53# ifdef __LITTLE_ENDIAN
54# define DO_CRC(x) crc = tab[0][(crc ^ (x)) & 255] ^ (crc >> 8)
55# define DO_CRC4 crc = tab[3][(crc) & 255] ^ \
56        tab[2][(crc >> 8) & 255] ^ \
57        tab[1][(crc >> 16) & 255] ^ \
58        tab[0][(crc >> 24) & 255]
59# else
60# define DO_CRC(x) crc = tab[0][((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
61# define DO_CRC4 crc = tab[0][(crc) & 255] ^ \
62        tab[1][(crc >> 8) & 255] ^ \
63        tab[2][(crc >> 16) & 255] ^ \
64        tab[3][(crc >> 24) & 255]
65# endif
66    const u32 *b;
67    size_t rem_len;
68
69    /* Align it */
70    if (unlikely((long)buf & 3 && len)) {
71        do {
72            DO_CRC(*buf++);
73        } while ((--len) && ((long)buf)&3);
74    }
75    rem_len = len & 3;
76    /* load data 32 bits wide, xor data 32 bits wide. */
77    len = len >> 2;
78    b = (const u32 *)buf;
79    for (--b; len; --len) {
80        crc ^= *++b; /* use pre increment for speed */
81        DO_CRC4;
82    }
83    len = rem_len;
84    /* And the last few bytes */
85    if (len) {
86        u8 *p = (u8 *)(b + 1) - 1;
87        do {
88            DO_CRC(*++p); /* use pre increment for speed */
89        } while (--len);
90    }
91    return crc;
92#undef DO_CRC
93#undef DO_CRC4
94}
95#endif
96/**
97 * crc32_le() - Calculate bitwise little-endian Ethernet AUTODIN II CRC32
98 * @crc: seed value for computation. ~0 for Ethernet, sometimes 0 for
99 * other uses, or the previous crc32 value if computing incrementally.
100 * @p: pointer to buffer over which CRC is run
101 * @len: length of buffer @p
102 */
103u32 __pure crc32_le(u32 crc, unsigned char const *p, size_t len);
104
105#if CRC_LE_BITS == 1
106/*
107 * In fact, the table-based code will work in this case, but it can be
108 * simplified by inlining the table in ?: form.
109 */
110
111u32 __pure crc32_le(u32 crc, unsigned char const *p, size_t len)
112{
113    int i;
114    while (len--) {
115        crc ^= *p++;
116        for (i = 0; i < 8; i++)
117            crc = (crc >> 1) ^ ((crc & 1) ? CRCPOLY_LE : 0);
118    }
119    return crc;
120}
121#else /* Table-based approach */
122
123u32 __pure crc32_le(u32 crc, unsigned char const *p, size_t len)
124{
125# if CRC_LE_BITS == 8
126    const u32 (*tab)[] = crc32table_le;
127
128    crc = __cpu_to_le32(crc);
129    crc = crc32_body(crc, p, len, tab);
130    return __le32_to_cpu(crc);
131# elif CRC_LE_BITS == 4
132    while (len--) {
133        crc ^= *p++;
134        crc = (crc >> 4) ^ crc32table_le[crc & 15];
135        crc = (crc >> 4) ^ crc32table_le[crc & 15];
136    }
137    return crc;
138# elif CRC_LE_BITS == 2
139    while (len--) {
140        crc ^= *p++;
141        crc = (crc >> 2) ^ crc32table_le[crc & 3];
142        crc = (crc >> 2) ^ crc32table_le[crc & 3];
143        crc = (crc >> 2) ^ crc32table_le[crc & 3];
144        crc = (crc >> 2) ^ crc32table_le[crc & 3];
145    }
146    return crc;
147# endif
148}
149#endif
150
151/**
152 * crc32_be() - Calculate bitwise big-endian Ethernet AUTODIN II CRC32
153 * @crc: seed value for computation. ~0 for Ethernet, sometimes 0 for
154 * other uses, or the previous crc32 value if computing incrementally.
155 * @p: pointer to buffer over which CRC is run
156 * @len: length of buffer @p
157 */
158u32 __pure crc32_be(u32 crc, unsigned char const *p, size_t len);
159
160#if CRC_BE_BITS == 1
161/*
162 * In fact, the table-based code will work in this case, but it can be
163 * simplified by inlining the table in ?: form.
164 */
165
166u32 __pure crc32_be(u32 crc, unsigned char const *p, size_t len)
167{
168    int i;
169    while (len--) {
170        crc ^= *p++ << 24;
171        for (i = 0; i < 8; i++)
172            crc =
173                (crc << 1) ^ ((crc & 0x80000000) ? CRCPOLY_BE :
174                      0);
175    }
176    return crc;
177}
178
179#else /* Table-based approach */
180u32 __pure crc32_be(u32 crc, unsigned char const *p, size_t len)
181{
182# if CRC_BE_BITS == 8
183    const u32 (*tab)[] = crc32table_be;
184
185    crc = __cpu_to_be32(crc);
186    crc = crc32_body(crc, p, len, tab);
187    return __be32_to_cpu(crc);
188# elif CRC_BE_BITS == 4
189    while (len--) {
190        crc ^= *p++ << 24;
191        crc = (crc << 4) ^ crc32table_be[crc >> 28];
192        crc = (crc << 4) ^ crc32table_be[crc >> 28];
193    }
194    return crc;
195# elif CRC_BE_BITS == 2
196    while (len--) {
197        crc ^= *p++ << 24;
198        crc = (crc << 2) ^ crc32table_be[crc >> 30];
199        crc = (crc << 2) ^ crc32table_be[crc >> 30];
200        crc = (crc << 2) ^ crc32table_be[crc >> 30];
201        crc = (crc << 2) ^ crc32table_be[crc >> 30];
202    }
203    return crc;
204# endif
205}
206#endif
207
208EXPORT_SYMBOL(crc32_le);
209EXPORT_SYMBOL(crc32_be);
210
211/*
212 * A brief CRC tutorial.
213 *
214 * A CRC is a long-division remainder. You add the CRC to the message,
215 * and the whole thing (message+CRC) is a multiple of the given
216 * CRC polynomial. To check the CRC, you can either check that the
217 * CRC matches the recomputed value, *or* you can check that the
218 * remainder computed on the message+CRC is 0. This latter approach
219 * is used by a lot of hardware implementations, and is why so many
220 * protocols put the end-of-frame flag after the CRC.
221 *
222 * It's actually the same long division you learned in school, except that
223 * - We're working in binary, so the digits are only 0 and 1, and
224 * - When dividing polynomials, there are no carries. Rather than add and
225 * subtract, we just xor. Thus, we tend to get a bit sloppy about
226 * the difference between adding and subtracting.
227 *
228 * A 32-bit CRC polynomial is actually 33 bits long. But since it's
229 * 33 bits long, bit 32 is always going to be set, so usually the CRC
230 * is written in hex with the most significant bit omitted. (If you're
231 * familiar with the IEEE 754 floating-point format, it's the same idea.)
232 *
233 * Note that a CRC is computed over a string of *bits*, so you have
234 * to decide on the endianness of the bits within each byte. To get
235 * the best error-detecting properties, this should correspond to the
236 * order they're actually sent. For example, standard RS-232 serial is
237 * little-endian; the most significant bit (sometimes used for parity)
238 * is sent last. And when appending a CRC word to a message, you should
239 * do it in the right order, matching the endianness.
240 *
241 * Just like with ordinary division, the remainder is always smaller than
242 * the divisor (the CRC polynomial) you're dividing by. Each step of the
243 * division, you take one more digit (bit) of the dividend and append it
244 * to the current remainder. Then you figure out the appropriate multiple
245 * of the divisor to subtract to being the remainder back into range.
246 * In binary, it's easy - it has to be either 0 or 1, and to make the
247 * XOR cancel, it's just a copy of bit 32 of the remainder.
248 *
249 * When computing a CRC, we don't care about the quotient, so we can
250 * throw the quotient bit away, but subtract the appropriate multiple of
251 * the polynomial from the remainder and we're back to where we started,
252 * ready to process the next bit.
253 *
254 * A big-endian CRC written this way would be coded like:
255 * for (i = 0; i < input_bits; i++) {
256 * multiple = remainder & 0x80000000 ? CRCPOLY : 0;
257 * remainder = (remainder << 1 | next_input_bit()) ^ multiple;
258 * }
259 * Notice how, to get at bit 32 of the shifted remainder, we look
260 * at bit 31 of the remainder *before* shifting it.
261 *
262 * But also notice how the next_input_bit() bits we're shifting into
263 * the remainder don't actually affect any decision-making until
264 * 32 bits later. Thus, the first 32 cycles of this are pretty boring.
265 * Also, to add the CRC to a message, we need a 32-bit-long hole for it at
266 * the end, so we have to add 32 extra cycles shifting in zeros at the
267 * end of every message,
268 *
269 * So the standard trick is to rearrage merging in the next_input_bit()
270 * until the moment it's needed. Then the first 32 cycles can be precomputed,
271 * and merging in the final 32 zero bits to make room for the CRC can be
272 * skipped entirely.
273 * This changes the code to:
274 * for (i = 0; i < input_bits; i++) {
275 * remainder ^= next_input_bit() << 31;
276 * multiple = (remainder & 0x80000000) ? CRCPOLY : 0;
277 * remainder = (remainder << 1) ^ multiple;
278 * }
279 * With this optimization, the little-endian code is simpler:
280 * for (i = 0; i < input_bits; i++) {
281 * remainder ^= next_input_bit();
282 * multiple = (remainder & 1) ? CRCPOLY : 0;
283 * remainder = (remainder >> 1) ^ multiple;
284 * }
285 *
286 * Note that the other details of endianness have been hidden in CRCPOLY
287 * (which must be bit-reversed) and next_input_bit().
288 *
289 * However, as long as next_input_bit is returning the bits in a sensible
290 * order, we can actually do the merging 8 or more bits at a time rather
291 * than one bit at a time:
292 * for (i = 0; i < input_bytes; i++) {
293 * remainder ^= next_input_byte() << 24;
294 * for (j = 0; j < 8; j++) {
295 * multiple = (remainder & 0x80000000) ? CRCPOLY : 0;
296 * remainder = (remainder << 1) ^ multiple;
297 * }
298 * }
299 * Or in little-endian:
300 * for (i = 0; i < input_bytes; i++) {
301 * remainder ^= next_input_byte();
302 * for (j = 0; j < 8; j++) {
303 * multiple = (remainder & 1) ? CRCPOLY : 0;
304 * remainder = (remainder << 1) ^ multiple;
305 * }
306 * }
307 * If the input is a multiple of 32 bits, you can even XOR in a 32-bit
308 * word at a time and increase the inner loop count to 32.
309 *
310 * You can also mix and match the two loop styles, for example doing the
311 * bulk of a message byte-at-a-time and adding bit-at-a-time processing
312 * for any fractional bytes at the end.
313 *
314 * The only remaining optimization is to the byte-at-a-time table method.
315 * Here, rather than just shifting one bit of the remainder to decide
316 * in the correct multiple to subtract, we can shift a byte at a time.
317 * This produces a 40-bit (rather than a 33-bit) intermediate remainder,
318 * but again the multiple of the polynomial to subtract depends only on
319 * the high bits, the high 8 bits in this case.
320 *
321 * The multiple we need in that case is the low 32 bits of a 40-bit
322 * value whose high 8 bits are given, and which is a multiple of the
323 * generator polynomial. This is simply the CRC-32 of the given
324 * one-byte message.
325 *
326 * Two more details: normally, appending zero bits to a message which
327 * is already a multiple of a polynomial produces a larger multiple of that
328 * polynomial. To enable a CRC to detect this condition, it's common to
329 * invert the CRC before appending it. This makes the remainder of the
330 * message+crc come out not as zero, but some fixed non-zero value.
331 *
332 * The same problem applies to zero bits prepended to the message, and
333 * a similar solution is used. Instead of starting with a remainder of
334 * 0, an initial remainder of all ones is used. As long as you start
335 * the same way on decoding, it doesn't make a difference.
336 */
337
338#ifdef UNITTEST
339
340#include <stdlib.h>
341#include <stdio.h>
342
343#if 0 /*Not used at present */
344static void
345buf_dump(char const *prefix, unsigned char const *buf, size_t len)
346{
347    fputs(prefix, stdout);
348    while (len--)
349        printf(" %02x", *buf++);
350    putchar('\n');
351
352}
353#endif
354
355static void bytereverse(unsigned char *buf, size_t len)
356{
357    while (len--) {
358        unsigned char x = bitrev8(*buf);
359        *buf++ = x;
360    }
361}
362
363static void random_garbage(unsigned char *buf, size_t len)
364{
365    while (len--)
366        *buf++ = (unsigned char) random();
367}
368
369#if 0 /* Not used at present */
370static void store_le(u32 x, unsigned char *buf)
371{
372    buf[0] = (unsigned char) x;
373    buf[1] = (unsigned char) (x >> 8);
374    buf[2] = (unsigned char) (x >> 16);
375    buf[3] = (unsigned char) (x >> 24);
376}
377#endif
378
379static void store_be(u32 x, unsigned char *buf)
380{
381    buf[0] = (unsigned char) (x >> 24);
382    buf[1] = (unsigned char) (x >> 16);
383    buf[2] = (unsigned char) (x >> 8);
384    buf[3] = (unsigned char) x;
385}
386
387/*
388 * This checks that CRC(buf + CRC(buf)) = 0, and that
389 * CRC commutes with bit-reversal. This has the side effect
390 * of bytewise bit-reversing the input buffer, and returns
391 * the CRC of the reversed buffer.
392 */
393static u32 test_step(u32 init, unsigned char *buf, size_t len)
394{
395    u32 crc1, crc2;
396    size_t i;
397
398    crc1 = crc32_be(init, buf, len);
399    store_be(crc1, buf + len);
400    crc2 = crc32_be(init, buf, len + 4);
401    if (crc2)
402        printf("\nCRC cancellation fail: 0x%08x should be 0\n",
403               crc2);
404
405    for (i = 0; i <= len + 4; i++) {
406        crc2 = crc32_be(init, buf, i);
407        crc2 = crc32_be(crc2, buf + i, len + 4 - i);
408        if (crc2)
409            printf("\nCRC split fail: 0x%08x\n", crc2);
410    }
411
412    /* Now swap it around for the other test */
413
414    bytereverse(buf, len + 4);
415    init = bitrev32(init);
416    crc2 = bitrev32(crc1);
417    if (crc1 != bitrev32(crc2))
418        printf("\nBit reversal fail: 0x%08x -> 0x%08x -> 0x%08x\n",
419               crc1, crc2, bitrev32(crc2));
420    crc1 = crc32_le(init, buf, len);
421    if (crc1 != crc2)
422        printf("\nCRC endianness fail: 0x%08x != 0x%08x\n", crc1,
423               crc2);
424    crc2 = crc32_le(init, buf, len + 4);
425    if (crc2)
426        printf("\nCRC cancellation fail: 0x%08x should be 0\n",
427               crc2);
428
429    for (i = 0; i <= len + 4; i++) {
430        crc2 = crc32_le(init, buf, i);
431        crc2 = crc32_le(crc2, buf + i, len + 4 - i);
432        if (crc2)
433            printf("\nCRC split fail: 0x%08x\n", crc2);
434    }
435
436    return crc1;
437}
438
439#define SIZE 64
440#define INIT1 0
441#define INIT2 0
442
443int main(void)
444{
445    unsigned char buf1[SIZE + 4];
446    unsigned char buf2[SIZE + 4];
447    unsigned char buf3[SIZE + 4];
448    int i, j;
449    u32 crc1, crc2, crc3;
450
451    for (i = 0; i <= SIZE; i++) {
452        printf("\rTesting length %d...", i);
453        fflush(stdout);
454        random_garbage(buf1, i);
455        random_garbage(buf2, i);
456        for (j = 0; j < i; j++)
457            buf3[j] = buf1[j] ^ buf2[j];
458
459        crc1 = test_step(INIT1, buf1, i);
460        crc2 = test_step(INIT2, buf2, i);
461        /* Now check that CRC(buf1 ^ buf2) = CRC(buf1) ^ CRC(buf2) */
462        crc3 = test_step(INIT1 ^ INIT2, buf3, i);
463        if (crc3 != (crc1 ^ crc2))
464            printf("CRC XOR fail: 0x%08x != 0x%08x ^ 0x%08x\n",
465                   crc3, crc1, crc2);
466    }
467    printf("\nAll test complete. No failures expected.\n");
468    return 0;
469}
470
471#endif /* UNITTEST */
472

Archive Download this file



interactive