1 // SPDX-License-Identifier: 0BSD
2
3 ///////////////////////////////////////////////////////////////////////////////
4 //
5 /// \file alone_decoder.c
6 /// \brief Decoder for LZMA_Alone files
7 //
8 // Author: Lasse Collin
9 //
10 ///////////////////////////////////////////////////////////////////////////////
11
12 #include "alone_decoder.h"
13 #include "lzma_decoder.h"
14 #include "lz_decoder.h"
15
16
17 typedef struct {
18 lzma_next_coder next;
19
20 enum {
21 SEQ_PROPERTIES,
22 SEQ_DICTIONARY_SIZE,
23 SEQ_UNCOMPRESSED_SIZE,
24 SEQ_CODER_INIT,
25 SEQ_CODE,
26 } sequence;
27
28 /// If true, reject files that are unlikely to be .lzma files.
29 /// If false, more non-.lzma files get accepted and will give
30 /// LZMA_DATA_ERROR either immediately or after a few output bytes.
31 bool picky;
32
33 /// Position in the header fields
34 size_t pos;
35
36 /// Uncompressed size decoded from the header
37 lzma_vli uncompressed_size;
38
39 /// Memory usage limit
40 uint64_t memlimit;
41
42 /// Amount of memory actually needed (only an estimate)
43 uint64_t memusage;
44
45 /// Options decoded from the header needed to initialize
46 /// the LZMA decoder
47 lzma_options_lzma options;
48 } lzma_alone_coder;
49
50
51 static lzma_ret
alone_decode(void * coder_ptr,const lzma_allocator * allocator,const uint8_t * restrict in,size_t * restrict in_pos,size_t in_size,uint8_t * restrict out,size_t * restrict out_pos,size_t out_size,lzma_action action)52 alone_decode(void *coder_ptr, const lzma_allocator *allocator,
53 const uint8_t *restrict in, size_t *restrict in_pos,
54 size_t in_size, uint8_t *restrict out,
55 size_t *restrict out_pos, size_t out_size,
56 lzma_action action)
57 {
58 lzma_alone_coder *coder = coder_ptr;
59
60 while (*out_pos < out_size
61 && (coder->sequence == SEQ_CODE || *in_pos < in_size))
62 switch (coder->sequence) {
63 case SEQ_PROPERTIES:
64 if (lzma_lzma_lclppb_decode(&coder->options, in[*in_pos]))
65 return LZMA_FORMAT_ERROR;
66
67 coder->sequence = SEQ_DICTIONARY_SIZE;
68 ++*in_pos;
69 break;
70
71 case SEQ_DICTIONARY_SIZE:
72 coder->options.dict_size
73 |= (size_t)(in[*in_pos]) << (coder->pos * 8);
74
75 if (++coder->pos == 4) {
76 if (coder->picky && coder->options.dict_size
77 != UINT32_MAX) {
78 // A hack to ditch tons of false positives:
79 // We allow only dictionary sizes that are
80 // 2^n or 2^n + 2^(n-1). LZMA_Alone created
81 // only files with 2^n, but accepts any
82 // dictionary size.
83 uint32_t d = coder->options.dict_size - 1;
84 d |= d >> 2;
85 d |= d >> 3;
86 d |= d >> 4;
87 d |= d >> 8;
88 d |= d >> 16;
89 ++d;
90
91 if (d != coder->options.dict_size)
92 return LZMA_FORMAT_ERROR;
93 }
94
95 coder->pos = 0;
96 coder->sequence = SEQ_UNCOMPRESSED_SIZE;
97 }
98
99 ++*in_pos;
100 break;
101
102 case SEQ_UNCOMPRESSED_SIZE:
103 coder->uncompressed_size
104 |= (lzma_vli)(in[*in_pos]) << (coder->pos * 8);
105 ++*in_pos;
106 if (++coder->pos < 8)
107 break;
108
109 // Another hack to ditch false positives: Assume that
110 // if the uncompressed size is known, it must be less
111 // than 256 GiB.
112 //
113 // FIXME? Without picky we allow > LZMA_VLI_MAX which doesn't
114 // really matter in this specific situation (> LZMA_VLI_MAX is
115 // safe in the LZMA decoder) but it's somewhat weird still.
116 if (coder->picky
117 && coder->uncompressed_size != LZMA_VLI_UNKNOWN
118 && coder->uncompressed_size
119 >= (LZMA_VLI_C(1) << 38))
120 return LZMA_FORMAT_ERROR;
121
122 // Use LZMA_FILTER_LZMA1EXT features to specify the
123 // uncompressed size and that the end marker is allowed
124 // even when the uncompressed size is known. Both .lzma
125 // header and LZMA1EXT use UINT64_MAX indicate that size
126 // is unknown.
127 coder->options.ext_flags = LZMA_LZMA1EXT_ALLOW_EOPM;
128 lzma_set_ext_size(coder->options, coder->uncompressed_size);
129
130 // Calculate the memory usage so that it is ready
131 // for SEQ_CODER_INIT. We know that lc/lp/pb are valid
132 // so we can use the _nocheck variant.
133 coder->memusage
134 = lzma_lzma_decoder_memusage_nocheck(&coder->options)
135 + LZMA_MEMUSAGE_BASE;
136
137 coder->pos = 0;
138 coder->sequence = SEQ_CODER_INIT;
139 FALLTHROUGH;
140
141 case SEQ_CODER_INIT: {
142 if (coder->memusage > coder->memlimit)
143 return LZMA_MEMLIMIT_ERROR;
144
145 lzma_filter_info filters[2] = {
146 {
147 .id = LZMA_FILTER_LZMA1EXT,
148 .init = &lzma_lzma_decoder_init,
149 .options = &coder->options,
150 }, {
151 .init = NULL,
152 }
153 };
154
155 return_if_error(lzma_next_filter_init(&coder->next,
156 allocator, filters));
157
158 coder->sequence = SEQ_CODE;
159 break;
160 }
161
162 case SEQ_CODE: {
163 return coder->next.code(coder->next.coder,
164 allocator, in, in_pos, in_size,
165 out, out_pos, out_size, action);
166 }
167
168 default:
169 return LZMA_PROG_ERROR;
170 }
171
172 return LZMA_OK;
173 }
174
175
176 static void
alone_decoder_end(void * coder_ptr,const lzma_allocator * allocator)177 alone_decoder_end(void *coder_ptr, const lzma_allocator *allocator)
178 {
179 lzma_alone_coder *coder = coder_ptr;
180 lzma_next_end(&coder->next, allocator);
181 lzma_free(coder, allocator);
182 return;
183 }
184
185
186 static lzma_ret
alone_decoder_memconfig(void * coder_ptr,uint64_t * memusage,uint64_t * old_memlimit,uint64_t new_memlimit)187 alone_decoder_memconfig(void *coder_ptr, uint64_t *memusage,
188 uint64_t *old_memlimit, uint64_t new_memlimit)
189 {
190 lzma_alone_coder *coder = coder_ptr;
191
192 *memusage = coder->memusage;
193 *old_memlimit = coder->memlimit;
194
195 if (new_memlimit != 0) {
196 if (new_memlimit < coder->memusage)
197 return LZMA_MEMLIMIT_ERROR;
198
199 coder->memlimit = new_memlimit;
200 }
201
202 return LZMA_OK;
203 }
204
205
206 extern lzma_ret
lzma_alone_decoder_init(lzma_next_coder * next,const lzma_allocator * allocator,uint64_t memlimit,bool picky)207 lzma_alone_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator,
208 uint64_t memlimit, bool picky)
209 {
210 lzma_next_coder_init(&lzma_alone_decoder_init, next, allocator);
211
212 lzma_alone_coder *coder = next->coder;
213
214 if (coder == NULL) {
215 coder = lzma_alloc(sizeof(lzma_alone_coder), allocator);
216 if (coder == NULL)
217 return LZMA_MEM_ERROR;
218
219 next->coder = coder;
220 next->code = &alone_decode;
221 next->end = &alone_decoder_end;
222 next->memconfig = &alone_decoder_memconfig;
223 coder->next = LZMA_NEXT_CODER_INIT;
224 }
225
226 coder->sequence = SEQ_PROPERTIES;
227 coder->picky = picky;
228 coder->pos = 0;
229 coder->options.dict_size = 0;
230 coder->options.preset_dict = NULL;
231 coder->options.preset_dict_size = 0;
232 coder->uncompressed_size = 0;
233 coder->memlimit = my_max(1, memlimit);
234 coder->memusage = LZMA_MEMUSAGE_BASE;
235
236 return LZMA_OK;
237 }
238
239
240 extern LZMA_API(lzma_ret)
lzma_alone_decoder(lzma_stream * strm,uint64_t memlimit)241 lzma_alone_decoder(lzma_stream *strm, uint64_t memlimit)
242 {
243 lzma_next_strm_init(lzma_alone_decoder_init, strm, memlimit, false);
244
245 strm->internal->supported_actions[LZMA_RUN] = true;
246 strm->internal->supported_actions[LZMA_FINISH] = true;
247
248 return LZMA_OK;
249 }
250