Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
5098 | clevermous | 1 | loader_start: |
2 | ; start address; this code will be injected after the init code |
||
3 | ; (some commands below "B32" in the kernel) |
||
4 | mov edi, 0x280000 |
||
5 | lea ebx, [edi+loader_size1+16] |
||
6 | lea edx, [ebx+4] |
||
7 | loader_patch1: |
||
8 | mov esi, 0 ; will be patched: start address to copy |
||
9 | mov ecx, 0 ; will be patched: size of data to copy |
||
10 | push esi |
||
11 | rep movsb |
||
12 | jmp edx |
||
13 | loader_size1 = $ - loader_start |
||
14 | |||
15 | loader_patch2: |
||
16 | dd 0x280000 + loader_size |
||
17 | dd 0 ; will be patched: start value for code |
||
18 | ; (LZMA-specific) |
||
19 | dd -1 |
||
20 | dd _RangeDecoderBitDecode_edx - loader_start + 0x280000 |
||
21 | dd _RangeDecoderBitDecode - loader_start + 0x280000 |
||
22 | RangeDecoderBitDecode equ dword [ebx] |
||
23 | RangeDecoderBitDecode_edx equ dword [ebx-4] |
||
24 | code_ equ ebx-12 |
||
25 | range equ ebx-8 |
||
26 | |||
27 | rep1 equ ebx-28 |
||
28 | rep2 equ ebx-24 |
||
29 | rep3 equ ebx-20 |
||
30 | inptr_ldr equ ebx-16 |
||
31 | |||
32 | pb equ 0 ; pos state bits |
||
33 | lp equ 0 ; literal pos state bits |
||
34 | lc equ 3 ; literal context bits |
||
35 | posStateMask equ ((1 shl pb)-1) |
||
36 | literalPosMask equ ((1 shl lp)-1) |
||
37 | |||
38 | kNumPosBitsMax = 4 |
||
39 | kNumPosStatesMax = (1 shl kNumPosBitsMax) |
||
40 | |||
41 | kLenNumLowBits = 3 |
||
42 | kLenNumLowSymbols = (1 shl kLenNumLowBits) |
||
43 | kLenNumMidBits = 3 |
||
44 | kLenNumMidSymbols = (1 shl kLenNumMidBits) |
||
45 | kLenNumHighBits = 8 |
||
46 | kLenNumHighSymbols = (1 shl kLenNumHighBits) |
||
47 | |||
48 | LenChoice = 0 |
||
49 | LenChoice2 = 1 |
||
50 | LenLow = 2 |
||
51 | LenMid = (LenLow + (kNumPosStatesMax shl kLenNumLowBits)) |
||
52 | LenHigh = (LenMid + (kNumPosStatesMax shl kLenNumMidBits)) |
||
53 | kNumLenProbs = (LenHigh + kLenNumHighSymbols) |
||
54 | |||
55 | kNumStates = 12 |
||
56 | kNumLitStates = 7 |
||
57 | kStartPosModelIndex = 4 |
||
58 | kEndPosModelIndex = 14 |
||
59 | kNumFullDistances = (1 shl (kEndPosModelIndex/2)) |
||
60 | kNumPosSlotBits = 6 |
||
61 | kNumLenToPosStates = 4 |
||
62 | kNumAlignBits = 4 |
||
63 | kAlignTableSize = (1 shl kNumAlignBits) |
||
64 | kMatchMinLen = 2 |
||
65 | |||
66 | IsMatch = 0 |
||
67 | IsRep = 0xC0 ; (IsMatch + (kNumStates shl kNumPosBitsMax)) |
||
68 | IsRepG0 = 0xCC ; (IsRep + kNumStates) |
||
69 | IsRepG1 = 0xD8 ; (IsRepG0 + kNumStates) |
||
70 | IsRepG2 = 0xE4 ; (IsRepG1 + kNumStates) |
||
71 | IsRep0Long = 0xF0 ; (IsRepG2 + kNumStates) |
||
72 | PosSlot = 0x1B0 ; (IsRep0Long + (kNumStates shl kNumPosBitsMax)) |
||
73 | SpecPos = 0x2B0 ; (PosSlot + (kNumLenToPosStates shl kNumPosSlotBits)) |
||
74 | Align_ = 0x322 ; (SpecPos + kNumFullDistances - kEndPosModelIndex) |
||
75 | Lencoder = 0x332 ; (Align_ + kAlignTableSize) |
||
76 | RepLencoder = 0x534 ; (Lencoder + kNumLenProbs) |
||
77 | Literal = 0x736 ; (RepLencoder + kNumLenProbs) |
||
78 | |||
79 | LZMA_BASE_SIZE = 1846 ; must be ==Literal |
||
80 | LZMA_LIT_SIZE = 768 |
||
81 | |||
82 | kNumTopBits = 24 |
||
83 | kTopValue = (1 shl kNumTopBits) |
||
84 | |||
85 | kNumBitModelTotalBits = 11 |
||
86 | kBitModelTotal = (1 shl kNumBitModelTotalBits) |
||
87 | kNumMoveBits = 5 |
||
88 | |||
89 | uninit_base = 2C0000h |
||
90 | |||
91 | p = uninit_base |
||
92 | |||
93 | unpacker: |
||
94 | xor ebp, ebp |
||
95 | xor eax, eax |
||
96 | dec eax |
||
97 | lea edi, [rep1] |
||
98 | stosd |
||
99 | stosd |
||
100 | stosd |
||
101 | xchg eax, esi |
||
102 | ; mov ecx, Literal + (LZMA_LIT_SIZE shl (lc+lp)) |
||
103 | mov ch, (Literal + (LZMA_LIT_SIZE shl (lc+lp)) + 0xFF) shr 8 |
||
104 | mov eax, kBitModelTotal/2 |
||
105 | mov edi, p |
||
106 | rep stosd |
||
107 | pop edi |
||
108 | push edi |
||
109 | .main_loop: |
||
110 | ..loader_patch3: |
||
111 | cmp edi, dword 0 ; will be patched: end of data to unpack |
||
112 | jae .main_loop_done |
||
113 | if posStateMask |
||
114 | mov edx, edi |
||
115 | and edx, posStateMask |
||
116 | else |
||
117 | xor edx, edx |
||
118 | end if |
||
119 | push eax ; al = previous byte |
||
120 | lea eax, [ebp + ((p+IsMatch*4) shr (kNumPosBitsMax+2))] |
||
121 | shl eax, kNumPosBitsMax+2 |
||
122 | if posStateMask |
||
123 | call RangeDecoderBitDecode_edx |
||
124 | else |
||
125 | call RangeDecoderBitDecode |
||
126 | end if |
||
127 | pop eax |
||
128 | jc .1 |
||
129 | movzx eax, al |
||
130 | if literalPosMask |
||
131 | mov ah, dl |
||
132 | and ah, literalPosMask |
||
133 | end if |
||
134 | if ((LZMA_LIT_SIZE*4) and ((1 shl (8-lc)) - 1)) <> 0 |
||
135 | shr eax, 8-lc |
||
136 | imul eax, LZMA_LIT_SIZE*4 |
||
137 | else |
||
138 | and al, not ((1 shl (8-lc)) - 1) |
||
139 | imul eax, (LZMA_LIT_SIZE*4) shr (8-lc) |
||
140 | end if |
||
141 | add eax, p+Literal*4 |
||
142 | mov dl, 1 |
||
143 | cmp ebp, kNumLitStates |
||
144 | jb .literal |
||
145 | mov cl, [edi + esi] |
||
146 | .lx0: |
||
147 | add cl, cl |
||
148 | adc dh, 1 |
||
149 | call RangeDecoderBitDecode_edx |
||
150 | adc dl, dl |
||
151 | jc .lx1 |
||
152 | xor dh, dl |
||
153 | test dh, 1 |
||
154 | mov dh, 0 |
||
155 | jnz .lx0 |
||
156 | .literal: |
||
157 | @@: |
||
158 | call RangeDecoderBitDecode_edx |
||
159 | adc dl, dl |
||
160 | jnc @b |
||
161 | .lx1: |
||
162 | mov eax, ebp |
||
163 | cmp al, 4 |
||
164 | jb @f |
||
165 | cmp al, 10 |
||
166 | mov al, 3 |
||
167 | jb @f |
||
168 | mov al, 6 |
||
169 | @@: sub ebp, eax |
||
170 | xchg eax, edx |
||
171 | .stosb_main_loop: |
||
172 | stosb |
||
173 | jmp .main_loop |
||
174 | .1: |
||
175 | lea eax, [p + IsRep*4 + ebp*4] |
||
176 | call RangeDecoderBitDecode |
||
177 | jnc .10 |
||
178 | add eax, (IsRepG0 - IsRep)*4 ;lea eax, [p + IsRepG0*4 + ebp*4] |
||
179 | call RangeDecoderBitDecode |
||
180 | jc .111 |
||
181 | mov eax, ebp |
||
182 | shl eax, kNumPosBitsMax+2 |
||
183 | add eax, p + IsRep0Long*4 |
||
184 | call RangeDecoderBitDecode_edx |
||
185 | jc .1101 |
||
186 | cmp ebp, 7 |
||
187 | sbb ebp, ebp |
||
188 | lea ebp, [ebp+ebp+11] |
||
189 | mov al, [edi + esi] |
||
190 | jmp .stosb_main_loop |
||
191 | .111: |
||
192 | add eax, (IsRepG1 - IsRepG0) * 4 ;lea eax, [p + IsRepG1*4 + ebp*4] |
||
193 | call RangeDecoderBitDecode |
||
194 | xchg esi, [rep1] |
||
195 | jnc @f |
||
196 | add eax, (IsRepG2 - IsRepG1) * 4 ;lea eax, [p + IsRepG2*4 + ebp*4] |
||
197 | call RangeDecoderBitDecode |
||
198 | xchg esi, [rep2] |
||
199 | jnc @f |
||
200 | xchg esi, [rep3] |
||
201 | @@: |
||
202 | .1101: |
||
203 | mov eax, p + RepLencoder*4 |
||
204 | call LzmaLenDecode |
||
205 | push 8 |
||
206 | jmp .rmu |
||
207 | .10: |
||
208 | xchg esi, [rep1] |
||
209 | xchg esi, [rep2] |
||
210 | mov [rep3], esi |
||
211 | mov eax, p + Lencoder*4 |
||
212 | call LzmaLenDecode |
||
213 | push kNumLenToPosStates-1 |
||
214 | pop edx |
||
215 | cmp edx, ecx |
||
216 | jb @f |
||
217 | mov edx, ecx |
||
218 | @@: |
||
219 | push ecx |
||
220 | push kNumPosSlotBits |
||
221 | pop ecx |
||
222 | mov eax, p+PosSlot*4 |
||
223 | shl edx, cl |
||
224 | call RangeDecoderBitTreeDecode |
||
225 | mov esi, ecx |
||
226 | cmp ecx, kStartPosModelIndex |
||
227 | jb .l6 |
||
228 | mov edx, ecx |
||
229 | xor eax, eax |
||
230 | shr ecx, 1 |
||
231 | adc al, 2 |
||
232 | dec ecx |
||
233 | shl eax, cl |
||
234 | mov esi, eax |
||
235 | sub eax, edx |
||
236 | lea eax, [p + (SpecPos - 1)*4 + eax*4] |
||
237 | cmp edx, kEndPosModelIndex |
||
238 | jb .l59 |
||
239 | ; call RangeDecoderDecodeDirectBits |
||
240 | ;RangeDecoderDecodeDirectBits: |
||
241 | xor eax, eax |
||
242 | .l: |
||
243 | shr dword [range], 1 |
||
244 | add eax, eax |
||
245 | mov edx, [code_] |
||
246 | sub edx, [range] |
||
247 | jb @f |
||
248 | mov [code_], edx |
||
249 | add al, 1 shl kNumAlignBits |
||
250 | @@: |
||
251 | call update_decoder |
||
252 | dec ecx |
||
253 | cmp ecx, kNumAlignBits |
||
254 | jnz .l |
||
255 | ; ret |
||
256 | add esi, eax |
||
257 | mov eax, p+Align_*4 |
||
258 | .l59: |
||
259 | ; call RangeDecoderReverseBitTreeDecode_addesi |
||
260 | ;_RangeDecoderReverseBitTreeDecode_addesi: |
||
261 | ; in: eax->probs,ecx=numLevels |
||
262 | ; out: esi+=length; destroys edx |
||
263 | push edi |
||
264 | xor edx, edx |
||
265 | inc edx |
||
266 | mov edi, edx |
||
267 | @@: |
||
268 | call RangeDecoderBitDecode_edx |
||
269 | jnc .591 |
||
270 | add esi, edi |
||
271 | stc |
||
272 | .591: |
||
273 | adc edx, edx |
||
274 | add edi, edi |
||
275 | loop @b |
||
276 | pop edi |
||
277 | ; ret |
||
278 | .l6: |
||
279 | pop ecx |
||
280 | not esi |
||
281 | push 7 |
||
282 | .rmu: |
||
283 | cmp ebp, 7 |
||
284 | pop ebp |
||
285 | jb @f |
||
286 | add ebp, 3 |
||
287 | @@: |
||
288 | .repmovsb: |
||
289 | inc ecx |
||
290 | push esi |
||
291 | add esi, edi |
||
292 | rep movsb |
||
293 | lodsb |
||
294 | pop esi |
||
295 | jmp .stosb_main_loop |
||
296 | .main_loop_done: |
||
297 | include 'calltrick2.asm' |
||
298 | ret |
||
299 | |||
300 | _RangeDecoderBitDecode: |
||
301 | ; in: eax->prob |
||
302 | ; out: CF=bit |
||
303 | push edx |
||
304 | mov edx, [range] |
||
305 | shr edx, kNumBitModelTotalBits |
||
306 | imul edx, [eax] |
||
307 | cmp [code_], edx |
||
308 | jae .ae |
||
309 | mov [range], edx |
||
310 | mov edx, kBitModelTotal |
||
311 | sub edx, [eax] |
||
312 | shr edx, kNumMoveBits |
||
313 | add [eax], edx |
||
314 | .n: |
||
315 | pushfd |
||
316 | call update_decoder |
||
317 | popfd |
||
318 | pop edx |
||
319 | ret |
||
320 | .ae: |
||
321 | sub [range], edx |
||
322 | sub [code_], edx |
||
323 | mov edx, [eax] |
||
324 | shr edx, kNumMoveBits |
||
325 | sub [eax], edx |
||
326 | stc |
||
327 | jmp .n |
||
328 | |||
329 | update_decoder: |
||
330 | cmp byte [range+3], 0 ;cmp dword [range], kTopValue |
||
331 | jnz @f ;jae @f |
||
332 | shl dword [range], 8 |
||
333 | shl dword [code_], 8 |
||
334 | push eax |
||
335 | mov eax, [inptr_ldr] |
||
336 | mov al, [eax] |
||
337 | inc dword [inptr_ldr] |
||
338 | mov byte [code_], al |
||
339 | pop eax |
||
340 | @@: ret |
||
341 | |||
342 | _RangeDecoderBitDecode_edx: |
||
343 | push eax |
||
344 | lea eax, [eax+edx*4] |
||
345 | call RangeDecoderBitDecode |
||
346 | pop eax |
||
347 | ret |
||
348 | |||
349 | LzmaLenDecode: |
||
350 | ; in: eax->prob, edx=posState |
||
351 | ; out: ecx=len |
||
352 | |||
353 | ; LenChoice==0 |
||
354 | ; add eax, LenChoice*4 |
||
355 | if kLenNumMidBits <> kLenNumLowBits |
||
356 | error in optimization |
||
357 | end if |
||
358 | mov cl, kLenNumMidBits |
||
359 | call RangeDecoderBitDecode |
||
360 | jnc .0 |
||
361 | add eax, (LenChoice2-LenChoice)*4 |
||
362 | call RangeDecoderBitDecode |
||
363 | jc @f |
||
364 | if (kLenNumMidBits <> 3) | (LenMid-LenChoice2 > 0x7F + kLenNumMidBits) |
||
365 | shl edx, cl |
||
366 | add edx, LenMid-LenChoice2 |
||
367 | else |
||
368 | lea edx, [ecx + edx*8 - kLenNumMidBits + LenMid-LenChoice2] |
||
369 | end if |
||
370 | push kLenNumLowSymbols |
||
371 | jmp RangeDecoderBitTreeDecode.1 |
||
372 | @@: |
||
373 | mov edx, LenHigh-LenChoice2 |
||
374 | mov cl, kLenNumHighBits |
||
375 | push kLenNumLowSymbols + kLenNumMidSymbols |
||
376 | jmp RangeDecoderBitTreeDecode.1 |
||
377 | .0: |
||
378 | shl edx, cl |
||
379 | if LenLow = 2 |
||
380 | inc edx |
||
381 | inc edx |
||
382 | else |
||
383 | add edx, LenLow |
||
384 | end if |
||
385 | RangeDecoderBitTreeDecode: |
||
386 | ; in: eax+edx*4->probs,ecx=numLevels |
||
387 | ; out: ecx=length; destroys edx |
||
388 | push 0 |
||
389 | .1: |
||
390 | lea eax, [eax+edx*4] |
||
391 | xor edx, edx |
||
392 | inc edx |
||
393 | push ecx |
||
394 | @@: |
||
395 | call RangeDecoderBitDecode_edx |
||
396 | adc edx, edx |
||
397 | loop @b |
||
398 | pop ecx |
||
399 | btc edx, ecx |
||
400 | pop ecx |
||
401 | add ecx, edx |
||
402 | ret |
||
403 | |||
404 | loader_size = $ - loader_start>>> |