1 /*
2  * Copyright (c) Facebook, Inc.
3  * All rights reserved.
4  *
5  * This source code is licensed under both the BSD-style license (found in the
6  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7  * in the COPYING file in the root directory of this source tree).
8  * You may select, at your option, one of the above-listed licenses.
9  */
10 
11 #define ZSTD_STATIC_LINKING_ONLY
12 #define ZDICT_STATIC_LINKING_ONLY
13 
14 #include <string.h>
15 
16 #include "zstd_helpers.h"
17 #include "fuzz_helpers.h"
18 #include "zstd.h"
19 #include "zdict.h"
20 
21 const int kMinClevel = -3;
22 const int kMaxClevel = 19;
23 
set(ZSTD_CCtx * cctx,ZSTD_cParameter param,int value)24 static void set(ZSTD_CCtx *cctx, ZSTD_cParameter param, int value)
25 {
26     FUZZ_ZASSERT(ZSTD_CCtx_setParameter(cctx, param, value));
27 }
28 
setRand(ZSTD_CCtx * cctx,ZSTD_cParameter param,unsigned min,unsigned max,FUZZ_dataProducer_t * producer)29 static void setRand(ZSTD_CCtx *cctx, ZSTD_cParameter param, unsigned min,
30                     unsigned max, FUZZ_dataProducer_t *producer) {
31     unsigned const value = FUZZ_dataProducer_uint32Range(producer, min, max);
32     set(cctx, param, value);
33 }
34 
FUZZ_randomCParams(size_t srcSize,FUZZ_dataProducer_t * producer)35 ZSTD_compressionParameters FUZZ_randomCParams(size_t srcSize, FUZZ_dataProducer_t *producer)
36 {
37     /* Select compression parameters */
38     ZSTD_compressionParameters cParams;
39     cParams.windowLog = FUZZ_dataProducer_uint32Range(producer, ZSTD_WINDOWLOG_MIN, 15);
40     cParams.hashLog = FUZZ_dataProducer_uint32Range(producer, ZSTD_HASHLOG_MIN, 15);
41     cParams.chainLog = FUZZ_dataProducer_uint32Range(producer, ZSTD_CHAINLOG_MIN, 16);
42     cParams.searchLog = FUZZ_dataProducer_uint32Range(producer, ZSTD_SEARCHLOG_MIN, 9);
43     cParams.minMatch = FUZZ_dataProducer_uint32Range(producer, ZSTD_MINMATCH_MIN,
44                                           ZSTD_MINMATCH_MAX);
45     cParams.targetLength = FUZZ_dataProducer_uint32Range(producer, 0, 512);
46     cParams.strategy = FUZZ_dataProducer_uint32Range(producer, ZSTD_STRATEGY_MIN, ZSTD_STRATEGY_MAX);
47     return ZSTD_adjustCParams(cParams, srcSize, 0);
48 }
49 
FUZZ_randomFParams(FUZZ_dataProducer_t * producer)50 ZSTD_frameParameters FUZZ_randomFParams(FUZZ_dataProducer_t *producer)
51 {
52     /* Select frame parameters */
53     ZSTD_frameParameters fParams;
54     fParams.contentSizeFlag = FUZZ_dataProducer_uint32Range(producer, 0, 1);
55     fParams.checksumFlag = FUZZ_dataProducer_uint32Range(producer, 0, 1);
56     fParams.noDictIDFlag = FUZZ_dataProducer_uint32Range(producer, 0, 1);
57     return fParams;
58 }
59 
FUZZ_randomParams(size_t srcSize,FUZZ_dataProducer_t * producer)60 ZSTD_parameters FUZZ_randomParams(size_t srcSize, FUZZ_dataProducer_t *producer)
61 {
62     ZSTD_parameters params;
63     params.cParams = FUZZ_randomCParams(srcSize, producer);
64     params.fParams = FUZZ_randomFParams(producer);
65     return params;
66 }
67 
FUZZ_setRandomParameters(ZSTD_CCtx * cctx,size_t srcSize,FUZZ_dataProducer_t * producer)68 void FUZZ_setRandomParameters(ZSTD_CCtx *cctx, size_t srcSize, FUZZ_dataProducer_t *producer)
69 {
70     ZSTD_compressionParameters cParams = FUZZ_randomCParams(srcSize, producer);
71     set(cctx, ZSTD_c_windowLog, cParams.windowLog);
72     set(cctx, ZSTD_c_hashLog, cParams.hashLog);
73     set(cctx, ZSTD_c_chainLog, cParams.chainLog);
74     set(cctx, ZSTD_c_searchLog, cParams.searchLog);
75     set(cctx, ZSTD_c_minMatch, cParams.minMatch);
76     set(cctx, ZSTD_c_targetLength, cParams.targetLength);
77     set(cctx, ZSTD_c_strategy, cParams.strategy);
78     /* Select frame parameters */
79     setRand(cctx, ZSTD_c_contentSizeFlag, 0, 1, producer);
80     setRand(cctx, ZSTD_c_checksumFlag, 0, 1, producer);
81     setRand(cctx, ZSTD_c_dictIDFlag, 0, 1, producer);
82     /* Select long distance matching parameters */
83     setRand(cctx, ZSTD_c_enableLongDistanceMatching, 0, 1, producer);
84     setRand(cctx, ZSTD_c_ldmHashLog, ZSTD_HASHLOG_MIN, 16, producer);
85     setRand(cctx, ZSTD_c_ldmMinMatch, ZSTD_LDM_MINMATCH_MIN,
86             ZSTD_LDM_MINMATCH_MAX, producer);
87     setRand(cctx, ZSTD_c_ldmBucketSizeLog, 0, ZSTD_LDM_BUCKETSIZELOG_MAX,
88             producer);
89     setRand(cctx, ZSTD_c_ldmHashRateLog, ZSTD_LDM_HASHRATELOG_MIN,
90             ZSTD_LDM_HASHRATELOG_MAX, producer);
91     /* Set misc parameters */
92     setRand(cctx, ZSTD_c_nbWorkers, 0, 2, producer);
93     setRand(cctx, ZSTD_c_rsyncable, 0, 1, producer);
94     setRand(cctx, ZSTD_c_useRowMatchFinder, 0, 2, producer);
95     setRand(cctx, ZSTD_c_enableDedicatedDictSearch, 0, 1, producer);
96     setRand(cctx, ZSTD_c_forceMaxWindow, 0, 1, producer);
97     setRand(cctx, ZSTD_c_literalCompressionMode, 0, 2, producer);
98     setRand(cctx, ZSTD_c_forceAttachDict, 0, 2, producer);
99     setRand(cctx, ZSTD_c_splitBlocks, 0, 1, producer);
100     setRand(cctx, ZSTD_c_deterministicRefPrefix, 0, 1, producer);
101     if (FUZZ_dataProducer_uint32Range(producer, 0, 1) == 0) {
102       setRand(cctx, ZSTD_c_srcSizeHint, ZSTD_SRCSIZEHINT_MIN, 2 * srcSize, producer);
103     }
104     if (FUZZ_dataProducer_uint32Range(producer, 0, 1) == 0) {
105       setRand(cctx, ZSTD_c_targetCBlockSize, ZSTD_TARGETCBLOCKSIZE_MIN, ZSTD_TARGETCBLOCKSIZE_MAX, producer);
106     }
107 }
108 
FUZZ_train(void const * src,size_t srcSize,FUZZ_dataProducer_t * producer)109 FUZZ_dict_t FUZZ_train(void const* src, size_t srcSize, FUZZ_dataProducer_t *producer)
110 {
111     size_t const dictSize = MAX(srcSize / 8, 1024);
112     size_t const totalSampleSize = dictSize * 11;
113     FUZZ_dict_t dict = { FUZZ_malloc(dictSize), dictSize };
114     char* const samples = (char*)FUZZ_malloc(totalSampleSize);
115     unsigned nbSamples = 100;
116     size_t* const samplesSizes = (size_t*)FUZZ_malloc(sizeof(size_t) * nbSamples);
117     size_t pos = 0;
118     size_t sample = 0;
119     ZDICT_fastCover_params_t params;
120 
121     for (sample = 0; sample < nbSamples; ++sample) {
122       size_t const remaining = totalSampleSize - pos;
123       size_t const offset = FUZZ_dataProducer_uint32Range(producer, 0, MAX(srcSize, 1) - 1);
124       size_t const limit = MIN(srcSize - offset, remaining);
125       size_t const toCopy = MIN(limit, remaining / (nbSamples - sample));
126       memcpy(samples + pos, src + offset, toCopy);
127       pos += toCopy;
128       samplesSizes[sample] = toCopy;
129     }
130     memset(samples + pos, 0, totalSampleSize - pos);
131 
132     memset(&params, 0, sizeof(params));
133     params.accel = 5;
134     params.k = 40;
135     params.d = 8;
136     params.f = 14;
137     params.zParams.compressionLevel = 1;
138     dict.size = ZDICT_trainFromBuffer_fastCover(dict.buff, dictSize,
139         samples, samplesSizes, nbSamples, params);
140     if (ZSTD_isError(dict.size)) {
141         free(dict.buff);
142         memset(&dict, 0, sizeof(dict));
143     }
144 
145     free(samplesSizes);
146     free(samples);
147 
148     return dict;
149 }
150