1*01826a49SYabin Cui /*
2*01826a49SYabin Cui * Copyright (c) Meta Platforms, Inc. and affiliates.
3*01826a49SYabin Cui * All rights reserved.
4*01826a49SYabin Cui *
5*01826a49SYabin Cui * This source code is licensed under both the BSD-style license (found in the
6*01826a49SYabin Cui * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7*01826a49SYabin Cui * in the COPYING file in the root directory of this source tree).
8*01826a49SYabin Cui * You may select, at your option, one of the above-listed licenses.
9*01826a49SYabin Cui */
10*01826a49SYabin Cui
11*01826a49SYabin Cui /**
12*01826a49SYabin Cui * This fuzz target performs a zstd round-trip test (compress & decompress),
13*01826a49SYabin Cui * compares the result with the original, and calls abort() on corruption.
14*01826a49SYabin Cui */
15*01826a49SYabin Cui
16*01826a49SYabin Cui #include <stddef.h>
17*01826a49SYabin Cui #include <stdlib.h>
18*01826a49SYabin Cui #include <stdio.h>
19*01826a49SYabin Cui #include <string.h>
20*01826a49SYabin Cui #include "common/cpu.h"
21*01826a49SYabin Cui #include "compress/hist.h"
22*01826a49SYabin Cui #include "common/huf.h"
23*01826a49SYabin Cui #include "fuzz_helpers.h"
24*01826a49SYabin Cui #include "fuzz_data_producer.h"
25*01826a49SYabin Cui #include "common/bits.h"
26*01826a49SYabin Cui
adjustTableLog(size_t tableLog,size_t maxSymbol)27*01826a49SYabin Cui static size_t adjustTableLog(size_t tableLog, size_t maxSymbol)
28*01826a49SYabin Cui {
29*01826a49SYabin Cui size_t const alphabetSize = maxSymbol + 1;
30*01826a49SYabin Cui size_t minTableLog = ZSTD_highbit32(alphabetSize) + 1;
31*01826a49SYabin Cui if ((alphabetSize & (alphabetSize - 1)) != 0) {
32*01826a49SYabin Cui ++minTableLog;
33*01826a49SYabin Cui }
34*01826a49SYabin Cui assert(minTableLog <= 9);
35*01826a49SYabin Cui if (tableLog < minTableLog)
36*01826a49SYabin Cui return minTableLog;
37*01826a49SYabin Cui else
38*01826a49SYabin Cui return tableLog;
39*01826a49SYabin Cui }
40*01826a49SYabin Cui
LLVMFuzzerTestOneInput(const uint8_t * src,size_t size)41*01826a49SYabin Cui int LLVMFuzzerTestOneInput(const uint8_t *src, size_t size)
42*01826a49SYabin Cui {
43*01826a49SYabin Cui FUZZ_dataProducer_t *producer = FUZZ_dataProducer_create(src, size);
44*01826a49SYabin Cui /* Select random parameters: #streams, X1 or X2 decoding, bmi2 */
45*01826a49SYabin Cui int const streams = FUZZ_dataProducer_int32Range(producer, 0, 1);
46*01826a49SYabin Cui int const symbols = FUZZ_dataProducer_int32Range(producer, 0, 1);
47*01826a49SYabin Cui int const flags = 0
48*01826a49SYabin Cui | (ZSTD_cpuid_bmi2(ZSTD_cpuid()) && FUZZ_dataProducer_int32Range(producer, 0, 1) ? HUF_flags_bmi2 : 0)
49*01826a49SYabin Cui | (FUZZ_dataProducer_int32Range(producer, 0, 1) ? HUF_flags_optimalDepth : 0)
50*01826a49SYabin Cui | (FUZZ_dataProducer_int32Range(producer, 0, 1) ? HUF_flags_preferRepeat : 0)
51*01826a49SYabin Cui | (FUZZ_dataProducer_int32Range(producer, 0, 1) ? HUF_flags_suspectUncompressible : 0)
52*01826a49SYabin Cui | (FUZZ_dataProducer_int32Range(producer, 0, 1) ? HUF_flags_disableAsm : 0)
53*01826a49SYabin Cui | (FUZZ_dataProducer_int32Range(producer, 0, 1) ? HUF_flags_disableFast : 0);
54*01826a49SYabin Cui /* Select a random cBufSize - it may be too small */
55*01826a49SYabin Cui size_t const cBufSize = FUZZ_dataProducer_uint32Range(producer, 0, 4 * size);
56*01826a49SYabin Cui /* Select a random tableLog - we'll adjust it up later */
57*01826a49SYabin Cui size_t tableLog = FUZZ_dataProducer_uint32Range(producer, 1, 12);
58*01826a49SYabin Cui size_t const kMaxSize = 256 * 1024;
59*01826a49SYabin Cui size = FUZZ_dataProducer_remainingBytes(producer);
60*01826a49SYabin Cui if (size > kMaxSize)
61*01826a49SYabin Cui size = kMaxSize;
62*01826a49SYabin Cui
63*01826a49SYabin Cui if (size <= 1) {
64*01826a49SYabin Cui FUZZ_dataProducer_free(producer);
65*01826a49SYabin Cui return 0;
66*01826a49SYabin Cui }
67*01826a49SYabin Cui
68*01826a49SYabin Cui uint32_t maxSymbol = 255;
69*01826a49SYabin Cui
70*01826a49SYabin Cui U32 count[256];
71*01826a49SYabin Cui size_t const mostFrequent = HIST_count(count, &maxSymbol, src, size);
72*01826a49SYabin Cui FUZZ_ZASSERT(mostFrequent);
73*01826a49SYabin Cui if (mostFrequent == size) {
74*01826a49SYabin Cui /* RLE */
75*01826a49SYabin Cui FUZZ_dataProducer_free(producer);
76*01826a49SYabin Cui return 0;
77*01826a49SYabin Cui
78*01826a49SYabin Cui }
79*01826a49SYabin Cui FUZZ_ASSERT(maxSymbol <= 255);
80*01826a49SYabin Cui tableLog = adjustTableLog(tableLog, maxSymbol);
81*01826a49SYabin Cui
82*01826a49SYabin Cui size_t const wkspSize = HUF_WORKSPACE_SIZE;
83*01826a49SYabin Cui void* wksp = FUZZ_malloc(wkspSize);
84*01826a49SYabin Cui void* rBuf = FUZZ_malloc(size);
85*01826a49SYabin Cui void* cBuf = FUZZ_malloc(cBufSize);
86*01826a49SYabin Cui HUF_CElt* ct = (HUF_CElt*)FUZZ_malloc(HUF_CTABLE_SIZE(maxSymbol));
87*01826a49SYabin Cui HUF_DTable* dt = (HUF_DTable*)FUZZ_malloc(HUF_DTABLE_SIZE(tableLog) * sizeof(HUF_DTable));
88*01826a49SYabin Cui dt[0] = tableLog * 0x01000001;
89*01826a49SYabin Cui
90*01826a49SYabin Cui tableLog = HUF_optimalTableLog(tableLog, size, maxSymbol, wksp, wkspSize, ct, count, flags);
91*01826a49SYabin Cui FUZZ_ASSERT(tableLog <= 12);
92*01826a49SYabin Cui tableLog = HUF_buildCTable_wksp(ct, count, maxSymbol, tableLog, wksp, wkspSize);
93*01826a49SYabin Cui FUZZ_ZASSERT(tableLog);
94*01826a49SYabin Cui size_t const tableSize = HUF_writeCTable_wksp(cBuf, cBufSize, ct, maxSymbol, tableLog, wksp, wkspSize);
95*01826a49SYabin Cui if (ERR_isError(tableSize)) {
96*01826a49SYabin Cui /* Errors on uncompressible data or cBufSize too small */
97*01826a49SYabin Cui goto _out;
98*01826a49SYabin Cui }
99*01826a49SYabin Cui FUZZ_ZASSERT(tableSize);
100*01826a49SYabin Cui if (symbols == 0) {
101*01826a49SYabin Cui FUZZ_ZASSERT(HUF_readDTableX1_wksp(dt, cBuf, tableSize, wksp, wkspSize, flags));
102*01826a49SYabin Cui } else {
103*01826a49SYabin Cui size_t const ret = HUF_readDTableX2_wksp(dt, cBuf, tableSize, wksp, wkspSize, flags);
104*01826a49SYabin Cui if (ERR_getErrorCode(ret) == ZSTD_error_tableLog_tooLarge) {
105*01826a49SYabin Cui FUZZ_ZASSERT(HUF_readDTableX1_wksp(dt, cBuf, tableSize, wksp, wkspSize, flags));
106*01826a49SYabin Cui } else {
107*01826a49SYabin Cui FUZZ_ZASSERT(ret);
108*01826a49SYabin Cui }
109*01826a49SYabin Cui }
110*01826a49SYabin Cui
111*01826a49SYabin Cui size_t cSize;
112*01826a49SYabin Cui size_t rSize;
113*01826a49SYabin Cui if (streams == 0) {
114*01826a49SYabin Cui cSize = HUF_compress1X_usingCTable(cBuf, cBufSize, src, size, ct, flags);
115*01826a49SYabin Cui FUZZ_ZASSERT(cSize);
116*01826a49SYabin Cui if (cSize != 0)
117*01826a49SYabin Cui rSize = HUF_decompress1X_usingDTable(rBuf, size, cBuf, cSize, dt, flags);
118*01826a49SYabin Cui } else {
119*01826a49SYabin Cui cSize = HUF_compress4X_usingCTable(cBuf, cBufSize, src, size, ct, flags);
120*01826a49SYabin Cui FUZZ_ZASSERT(cSize);
121*01826a49SYabin Cui if (cSize != 0)
122*01826a49SYabin Cui rSize = HUF_decompress4X_usingDTable(rBuf, size, cBuf, cSize, dt, flags);
123*01826a49SYabin Cui }
124*01826a49SYabin Cui if (cSize != 0) {
125*01826a49SYabin Cui FUZZ_ZASSERT(rSize);
126*01826a49SYabin Cui FUZZ_ASSERT_MSG(rSize == size, "Incorrect regenerated size");
127*01826a49SYabin Cui FUZZ_ASSERT_MSG(!FUZZ_memcmp(src, rBuf, size), "Corruption!");
128*01826a49SYabin Cui }
129*01826a49SYabin Cui _out:
130*01826a49SYabin Cui free(rBuf);
131*01826a49SYabin Cui free(cBuf);
132*01826a49SYabin Cui free(ct);
133*01826a49SYabin Cui free(dt);
134*01826a49SYabin Cui free(wksp);
135*01826a49SYabin Cui FUZZ_dataProducer_free(producer);
136*01826a49SYabin Cui return 0;
137*01826a49SYabin Cui }
138