Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
1713 | mario79 | 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 | |||
23 | RangeDecoderBitDecode equ dword [ebx] |
||
24 | RangeDecoderBitDecode_edx equ dword [ebx-4] |
||
25 | code_ equ ebx-12 |
||
26 | range equ ebx-8 |
||
27 | |||
28 | rep1 equ ebx-28 |
||
29 | rep2 equ ebx-24 |
||
30 | rep3 equ ebx-20 |
||
31 | inptr_ldr equ ebx-16 |
||
32 | |||
33 | pb equ 2 ; pos state bits |
||
34 | lp equ 0 ; literal pos state bits |
||
35 | lc equ 3 ; literal context bits |
||
36 | posStateMask equ ((1 shl pb)-1) |
||
37 | literalPosMask equ ((1 shl lp)-1) |
||
38 | |||
39 | kNumPosBitsMax = 4 |
||
40 | kNumPosStatesMax = (1 shl kNumPosBitsMax) |
||
41 | |||
42 | kLenNumLowBits = 3 |
||
43 | kLenNumLowSymbols = (1 shl kLenNumLowBits) |
||
44 | kLenNumMidBits = 3 |
||
45 | kLenNumMidSymbols = (1 shl kLenNumMidBits) |
||
46 | kLenNumHighBits = 8 |
||
47 | kLenNumHighSymbols = (1 shl kLenNumHighBits) |
||
48 | |||
49 | LenChoice = 0 |
||
50 | LenChoice2 = 1 |
||
51 | LenLow = 2 |
||
52 | LenMid = (LenLow + (kNumPosStatesMax shl kLenNumLowBits)) |
||
53 | LenHigh = (LenMid + (kNumPosStatesMax shl kLenNumMidBits)) |
||
54 | kNumLenProbs = (LenHigh + kLenNumHighSymbols) |
||
55 | |||
56 | kNumStates = 12 |
||
57 | kNumLitStates = 7 |
||
58 | kStartPosModelIndex = 4 |
||
59 | kEndPosModelIndex = 14 |
||
60 | kNumFullDistances = (1 shl (kEndPosModelIndex/2)) |
||
61 | kNumPosSlotBits = 6 |
||
62 | kNumLenToPosStates = 4 |
||
63 | kNumAlignBits = 4 |
||
64 | kAlignTableSize = (1 shl kNumAlignBits) |
||
65 | kMatchMinLen = 2 |
||
66 | |||
67 | IsMatch = 0 |
||
68 | IsRep = 0xC0 ; (IsMatch + (kNumStates shl kNumPosBitsMax)) |
||
69 | IsRepG0 = 0xCC ; (IsRep + kNumStates) |
||
70 | IsRepG1 = 0xD8 ; (IsRepG0 + kNumStates) |
||
71 | IsRepG2 = 0xE4 ; (IsRepG1 + kNumStates) |
||
72 | IsRep0Long = 0xF0 ; (IsRepG2 + kNumStates) |
||
73 | PosSlot = 0x1B0 ; (IsRep0Long + (kNumStates shl kNumPosBitsMax)) |
||
74 | SpecPos = 0x2B0 ; (PosSlot + (kNumLenToPosStates shl kNumPosSlotBits)) |
||
75 | Align_ = 0x322 ; (SpecPos + kNumFullDistances - kEndPosModelIndex) |
||
76 | Lencoder = 0x332 ; (Align_ + kAlignTableSize) |
||
77 | RepLencoder = 0x534 ; (Lencoder + kNumLenProbs) |
||
78 | Literal = 0x736 ; (RepLencoder + kNumLenProbs) |
||
79 | |||
80 | LZMA_BASE_SIZE = 1846 ; must be ==Literal |
||
81 | LZMA_LIT_SIZE = 768 |
||
82 | |||
83 | kNumTopBits = 24 |
||
84 | kTopValue = (1 shl kNumTopBits) |
||
85 | |||
86 | kNumBitModelTotalBits = 11 |
||
87 | kBitModelTotal = (1 shl kNumBitModelTotalBits) |
||
88 | kNumMoveBits = 5 |
||
89 | |||
90 | uninit_base = 2C0000h |
||
91 | |||
92 | p = uninit_base |
||
93 | |||
94 | unpacker: |
||
95 | xor ebp, ebp |
||
96 | xor eax, eax |
||
97 | dec eax |
||
98 | lea edi, [rep1] |
||
99 | stosd |
||
100 | stosd |
||
101 | stosd |
||
102 | |||
103 | xchg eax, esi |
||
104 | ; mov ecx, Literal + (LZMA_LIT_SIZE shl (lc+lp)) |
||
105 | mov ch, (Literal + (LZMA_LIT_SIZE shl (lc+lp)) + 0xFF) shr 8 |
||
106 | mov eax, kBitModelTotal/2 |
||
107 | mov edi, p |
||
108 | rep stosd |
||
109 | |||
110 | pop edi |
||
111 | push edi |
||
112 | loader_patch3: |
||
113 | .main_loop: |
||
114 | cmp edi, dword 0 ; will be patched: end of data to unpack |
||
115 | jae .main_loop_done |
||
116 | |||
117 | if posStateMask |
||
118 | mov edx, edi |
||
119 | and edx, posStateMask |
||
120 | else |
||
121 | xor edx, edx |
||
122 | end if |
||
123 | push eax ; al = previous byte |
||
124 | lea eax, [ebp + ((p+IsMatch*4) shr (kNumPosBitsMax+2))] |
||
125 | shl eax, kNumPosBitsMax+2 |
||
126 | if posStateMask |
||
127 | call RangeDecoderBitDecode_edx |
||
128 | else |
||
129 | call RangeDecoderBitDecode |
||
130 | end if |
||
131 | pop eax |
||
132 | jc .1 |
||
133 | |||
134 | movzx eax, al |
||
135 | if literalPosMask |
||
136 | mov ah, dl |
||
137 | and ah, literalPosMask |
||
138 | end if |
||
139 | if ((LZMA_LIT_SIZE*4) and ((1 shl (8-lc)) - 1)) <> 0 |
||
140 | shr eax, 8-lc |
||
141 | imul eax, LZMA_LIT_SIZE*4 |
||
142 | else |
||
143 | and al, not ((1 shl (8-lc)) - 1) |
||
144 | imul eax, (LZMA_LIT_SIZE*4) shr (8-lc) |
||
145 | end if |
||
146 | add eax, p+Literal*4 |
||
147 | mov dl, 1 |
||
148 | cmp ebp, kNumLitStates |
||
149 | jb .literal |
||
150 | |||
151 | mov cl, [edi + esi] |
||
152 | ;-------------------------------------- |
||
153 | .lx0: |
||
154 | add cl, cl |
||
155 | adc dh, 1 |
||
156 | call RangeDecoderBitDecode_edx |
||
157 | |||
158 | adc dl, dl |
||
159 | jc .lx1 |
||
160 | |||
161 | xor dh, dl |
||
162 | test dh, 1 |
||
163 | mov dh, 0 |
||
164 | jnz .lx0 |
||
165 | ;-------------------------------------- |
||
166 | .literal: |
||
167 | @@: |
||
168 | call RangeDecoderBitDecode_edx |
||
169 | adc dl, dl |
||
170 | jnc @b |
||
171 | ;-------------------------------------- |
||
172 | .lx1: |
||
173 | mov eax, ebp |
||
174 | cmp al, 4 |
||
175 | jb @f |
||
176 | |||
177 | cmp al, 10 |
||
178 | mov al, 3 |
||
179 | jb @f |
||
180 | |||
181 | mov al, 6 |
||
182 | @@: |
||
183 | sub ebp, eax |
||
184 | xchg eax, edx |
||
185 | ;-------------------------------------- |
||
186 | .stosb_main_loop: |
||
187 | stosb |
||
188 | jmp .main_loop |
||
189 | ;********************************************************************* |
||
190 | .1: |
||
191 | lea eax, [p + IsRep*4 + ebp*4] |
||
192 | call RangeDecoderBitDecode |
||
193 | jnc .10 |
||
194 | |||
195 | add eax, (IsRepG0 - IsRep)*4 ;lea eax, [p + IsRepG0*4 + ebp*4] |
||
196 | call RangeDecoderBitDecode |
||
197 | jc .111 |
||
198 | |||
199 | mov eax, ebp |
||
200 | shl eax, kNumPosBitsMax+2 |
||
201 | add eax, p + IsRep0Long*4 |
||
202 | call RangeDecoderBitDecode_edx |
||
203 | jc .1101 |
||
204 | |||
205 | cmp ebp, 7 |
||
206 | sbb ebp, ebp |
||
207 | lea ebp, [ebp+ebp+11] |
||
208 | mov al, [edi + esi] |
||
209 | jmp .stosb_main_loop |
||
210 | ;********************************************************************* |
||
211 | .111: |
||
212 | add eax, (IsRepG1 - IsRepG0) * 4 ;lea eax, [p + IsRepG1*4 + ebp*4] |
||
213 | call RangeDecoderBitDecode |
||
214 | xchg esi, [rep1] |
||
215 | jnc @f |
||
216 | |||
217 | add eax, (IsRepG2 - IsRepG1) * 4 ;lea eax, [p + IsRepG2*4 + ebp*4] |
||
218 | call RangeDecoderBitDecode |
||
219 | xchg esi, [rep2] |
||
220 | jnc @f |
||
221 | |||
222 | xchg esi, [rep3] |
||
223 | ;-------------------------------------- |
||
224 | @@: |
||
225 | .1101: |
||
226 | mov eax, p + RepLencoder*4 |
||
227 | call LzmaLenDecode1 |
||
228 | |||
229 | push 8 |
||
230 | jmp .rmu |
||
231 | ;********************************************************************* |
||
232 | .10: |
||
233 | xchg esi, [rep1] |
||
234 | xchg esi, [rep2] |
||
235 | mov [rep3], esi |
||
236 | mov eax, p + Lencoder*4 |
||
237 | call LzmaLenDecode1 |
||
238 | |||
239 | push kNumLenToPosStates-1 |
||
240 | pop edx |
||
241 | cmp edx, ecx |
||
242 | jb @f |
||
243 | |||
244 | mov edx, ecx |
||
245 | ;-------------------------------------- |
||
246 | @@: |
||
247 | push ecx |
||
248 | push kNumPosSlotBits |
||
249 | pop ecx |
||
250 | mov eax, p+PosSlot*4 |
||
251 | shl edx, cl |
||
252 | call RangeDecoderBitTreeDecode1 |
||
253 | |||
254 | mov esi, ecx |
||
255 | cmp ecx, kStartPosModelIndex |
||
256 | jb .l6 |
||
257 | |||
258 | mov edx, ecx |
||
259 | xor eax, eax |
||
260 | shr ecx, 1 |
||
261 | adc al, 2 |
||
262 | dec ecx |
||
263 | shl eax, cl |
||
264 | mov esi, eax |
||
265 | sub eax, edx |
||
266 | lea eax, [p + (SpecPos - 1)*4 + eax*4] |
||
267 | cmp edx, kEndPosModelIndex |
||
268 | jb .l59 |
||
269 | ; call RangeDecoderDecodeDirectBits |
||
270 | ;RangeDecoderDecodeDirectBits: |
||
271 | xor eax, eax |
||
272 | ;-------------------------------------- |
||
273 | .l: |
||
274 | shr dword [range], 1 |
||
275 | add eax, eax |
||
276 | mov edx, [code_] |
||
277 | sub edx, [range] |
||
278 | jb @f |
||
279 | |||
280 | mov [code_], edx |
||
281 | add al, 1 shl kNumAlignBits |
||
282 | ;-------------------------------------- |
||
283 | @@: |
||
284 | call update_decoder |
||
285 | dec ecx |
||
286 | cmp ecx, kNumAlignBits |
||
287 | jnz .l |
||
288 | ; ret |
||
289 | add esi, eax |
||
290 | mov eax, p+Align_*4 |
||
291 | ;-------------------------------------- |
||
292 | .l59: |
||
293 | ; call RangeDecoderReverseBitTreeDecode_addesi |
||
294 | ;_RangeDecoderReverseBitTreeDecode_addesi: |
||
295 | ; in: eax->probs,ecx=numLevels |
||
296 | ; out: esi+=length; destroys edx |
||
297 | push edi |
||
298 | xor edx, edx |
||
299 | inc edx |
||
300 | mov edi, edx |
||
301 | ;-------------------------------------- |
||
302 | @@: |
||
303 | call RangeDecoderBitDecode_edx |
||
304 | jnc .591 |
||
305 | |||
306 | add esi, edi |
||
307 | stc |
||
308 | ;-------------------------------------- |
||
309 | .591: |
||
310 | adc edx, edx |
||
311 | add edi, edi |
||
312 | loop @b |
||
313 | |||
314 | pop edi |
||
315 | ; ret |
||
316 | ;-------------------------------------- |
||
317 | .l6: |
||
318 | pop ecx |
||
319 | not esi |
||
320 | push 7 |
||
321 | ;-------------------------------------- |
||
322 | .rmu: |
||
323 | cmp ebp, 7 |
||
324 | pop ebp |
||
325 | jb @f |
||
326 | add ebp, 3 |
||
327 | ;-------------------------------------- |
||
328 | @@: |
||
329 | .repmovsb: |
||
330 | inc ecx |
||
331 | push esi |
||
332 | add esi, edi |
||
333 | rep movsb |
||
334 | |||
335 | lodsb |
||
336 | pop esi |
||
337 | jmp .stosb_main_loop |
||
338 | ;********************************************************************* |
||
339 | .main_loop_done: |
||
340 | include 'calltrick2.inc' |
||
341 | ret |
||
342 | ;********************************************************************* |
||
343 | _RangeDecoderBitDecode: |
||
344 | ; in: eax->prob |
||
345 | ; out: CF=bit |
||
346 | push edx |
||
347 | mov edx, [range] |
||
348 | shr edx, kNumBitModelTotalBits |
||
349 | imul edx, [eax] |
||
350 | cmp [code_], edx |
||
351 | jae .ae |
||
352 | |||
353 | mov [range], edx |
||
354 | mov edx, kBitModelTotal |
||
355 | sub edx, [eax] |
||
356 | shr edx, kNumMoveBits |
||
357 | add [eax], edx |
||
358 | ;-------------------------------------- |
||
359 | .n: |
||
360 | pushfd |
||
361 | call update_decoder |
||
362 | |||
363 | popfd |
||
364 | pop edx |
||
365 | ret |
||
366 | ;********************************************************************* |
||
367 | .ae: |
||
368 | sub [range], edx |
||
369 | sub [code_], edx |
||
370 | mov edx, [eax] |
||
371 | shr edx, kNumMoveBits |
||
372 | sub [eax], edx |
||
373 | stc |
||
374 | jmp .n |
||
375 | ;********************************************************************* |
||
376 | update_decoder: |
||
377 | cmp byte [range+3], 0 ;cmp dword [range], kTopValue |
||
378 | jnz @f ;jae @f |
||
379 | |||
380 | shl dword [range], 8 |
||
381 | shl dword [code_], 8 |
||
382 | push eax |
||
383 | mov eax, [inptr_ldr] |
||
384 | mov al, [eax] |
||
385 | inc dword [inptr_ldr] |
||
386 | mov byte [code_], al |
||
387 | pop eax |
||
388 | @@: |
||
389 | ret |
||
390 | ;********************************************************************* |
||
391 | _RangeDecoderBitDecode_edx: |
||
392 | push eax |
||
393 | lea eax, [eax+edx*4] |
||
394 | call RangeDecoderBitDecode |
||
395 | |||
396 | pop eax |
||
397 | ret |
||
398 | ;********************************************************************* |
||
399 | LzmaLenDecode1: |
||
400 | ; in: eax->prob, edx=posState |
||
401 | ; out: ecx=len |
||
402 | |||
403 | ; LenChoice==0 |
||
404 | ; add eax, LenChoice*4 |
||
405 | if kLenNumMidBits <> kLenNumLowBits |
||
406 | error in optimization |
||
407 | end if |
||
408 | mov cl, kLenNumMidBits |
||
409 | call RangeDecoderBitDecode |
||
410 | jnc .0 |
||
411 | |||
412 | add eax, (LenChoice2-LenChoice)*4 |
||
413 | call RangeDecoderBitDecode |
||
414 | jc @f |
||
415 | if (kLenNumMidBits <> 3) | (LenMid-LenChoice2 > 0x7F + kLenNumMidBits) |
||
416 | shl edx, cl |
||
417 | add edx, LenMid-LenChoice2 |
||
418 | else |
||
419 | lea edx, [ecx + edx*8 - kLenNumMidBits + LenMid-LenChoice2] |
||
420 | end if |
||
421 | push kLenNumLowSymbols |
||
422 | jmp RangeDecoderBitTreeDecode1.1 |
||
423 | ;********************************************************************* |
||
424 | @@: |
||
425 | mov edx, LenHigh-LenChoice2 |
||
426 | mov cl, kLenNumHighBits |
||
427 | push kLenNumLowSymbols + kLenNumMidSymbols |
||
428 | jmp RangeDecoderBitTreeDecode1.1 |
||
429 | ;********************************************************************* |
||
430 | .0: |
||
431 | shl edx, cl |
||
432 | if LenLow = 2 |
||
433 | inc edx |
||
434 | inc edx |
||
435 | else |
||
436 | add edx, LenLow |
||
437 | end if |
||
438 | ;-------------------------------------- |
||
439 | RangeDecoderBitTreeDecode1: |
||
440 | ; in: eax+edx*4->probs,ecx=numLevels |
||
441 | ; out: ecx=length; destroys edx |
||
442 | push 0 |
||
443 | .1: |
||
444 | lea eax, [eax+edx*4] |
||
445 | xor edx, edx |
||
446 | inc edx |
||
447 | push ecx |
||
448 | @@: |
||
449 | call RangeDecoderBitDecode_edx |
||
450 | |||
451 | adc edx, edx |
||
452 | loop @b |
||
453 | |||
454 | pop ecx |
||
455 | btc edx, ecx |
||
456 | pop ecx |
||
457 | add ecx, edx |
||
458 | ret |
||
459 | loader_size = $ - loader_start |
||
460 | ;*********************************************************************>>> |