1 | /* |
---|
2 | * Copyright (C) 1996-2002 Markus Franz Xaver Johannes Oberhumer |
---|
3 | * |
---|
4 | * This file is free software; you can redistribute it and/or |
---|
5 | * modify it under the terms of the GNU General Public License as |
---|
6 | * published by the Free Software Foundation; either version 2 of |
---|
7 | * the License, or (at your option) any later version. |
---|
8 | * |
---|
9 | * Originally this code was part of ucl the data compression library |
---|
10 | * for upx the ``Ultimate Packer of eXecutables''. |
---|
11 | * |
---|
12 | * - Converted to gas assembly, and refitted to work with etherboot. |
---|
13 | * Eric Biederman 20 Aug 2002 |
---|
14 | * |
---|
15 | * - Structure modified to be a subroutine call rather than an |
---|
16 | * executable prefix. |
---|
17 | * Michael Brown 30 Mar 2004 |
---|
18 | * |
---|
19 | * - Modified to be compilable as either 16-bit or 32-bit code. |
---|
20 | * Michael Brown 9 Mar 2005 |
---|
21 | */ |
---|
22 | |
---|
23 | FILE_LICENCE ( GPL2_OR_LATER ) |
---|
24 | |
---|
25 | /**************************************************************************** |
---|
26 | * This file provides the decompress() and decompress16() functions |
---|
27 | * which can be called in order to decompress an image compressed with |
---|
28 | * the nrv2b utility in src/util. |
---|
29 | * |
---|
30 | * These functions are designed to be called by the prefix. They are |
---|
31 | * position-independent code. |
---|
32 | * |
---|
33 | * The same basic assembly code is used to compile both |
---|
34 | * decompress() and decompress16(). |
---|
35 | **************************************************************************** |
---|
36 | */ |
---|
37 | |
---|
38 | .text |
---|
39 | .arch i386 |
---|
40 | .section ".prefix.lib", "ax", @progbits |
---|
41 | |
---|
42 | #ifdef CODE16 |
---|
43 | /**************************************************************************** |
---|
44 | * decompress16 (real-mode near call, position independent) |
---|
45 | * |
---|
46 | * Decompress data in 16-bit mode |
---|
47 | * |
---|
48 | * Parameters (passed via registers): |
---|
49 | * %ds:%esi - Start of compressed input data |
---|
50 | * %es:%edi - Start of output buffer |
---|
51 | * Returns: |
---|
52 | * %ds:%esi - End of compressed input data |
---|
53 | * %es:%edi - End of decompressed output data |
---|
54 | * All other registers are preserved |
---|
55 | * |
---|
56 | * NOTE: It would be possible to build a smaller version of the |
---|
57 | * decompression code for -DKEEP_IT_REAL by using |
---|
58 | * #define REG(x) x |
---|
59 | * to use 16-bit registers where possible. This would impose limits |
---|
60 | * that the compressed data size must be in the range [1,65533-%si] |
---|
61 | * and the uncompressed data size must be in the range [1,65536-%di] |
---|
62 | * (where %si and %di are the input values for those registers). Note |
---|
63 | * particularly that the lower limit is 1, not 0, and that the upper |
---|
64 | * limit on the input (compressed) data really is 65533, since the |
---|
65 | * algorithm may read up to three bytes beyond the end of the input |
---|
66 | * data, since it reads dwords. |
---|
67 | **************************************************************************** |
---|
68 | */ |
---|
69 | |
---|
70 | #define REG(x) e ## x |
---|
71 | #define ADDR32 addr32 |
---|
72 | |
---|
73 | .code16 |
---|
74 | .globl decompress16 |
---|
75 | decompress16: |
---|
76 | |
---|
77 | #else /* CODE16 */ |
---|
78 | |
---|
79 | /**************************************************************************** |
---|
80 | * decompress (32-bit protected-mode near call, position independent) |
---|
81 | * |
---|
82 | * Parameters (passed via registers): |
---|
83 | * %ds:%esi - Start of compressed input data |
---|
84 | * %es:%edi - Start of output buffer |
---|
85 | * Returns: |
---|
86 | * %ds:%esi - End of compressed input data |
---|
87 | * %es:%edi - End of decompressed output data |
---|
88 | * All other registers are preserved |
---|
89 | **************************************************************************** |
---|
90 | */ |
---|
91 | |
---|
92 | #define REG(x) e ## x |
---|
93 | #define ADDR32 |
---|
94 | |
---|
95 | .code32 |
---|
96 | .globl decompress |
---|
97 | decompress: |
---|
98 | |
---|
99 | #endif /* CODE16 */ |
---|
100 | |
---|
101 | #define xAX REG(ax) |
---|
102 | #define xCX REG(cx) |
---|
103 | #define xBP REG(bp) |
---|
104 | #define xSI REG(si) |
---|
105 | #define xDI REG(di) |
---|
106 | |
---|
107 | /* Save registers */ |
---|
108 | push %xAX |
---|
109 | pushl %ebx |
---|
110 | push %xCX |
---|
111 | push %xBP |
---|
112 | /* Do the decompression */ |
---|
113 | cld |
---|
114 | xor %xBP, %xBP |
---|
115 | dec %xBP /* last_m_off = -1 */ |
---|
116 | jmp dcl1_n2b |
---|
117 | |
---|
118 | decompr_literals_n2b: |
---|
119 | ADDR32 movsb |
---|
120 | decompr_loop_n2b: |
---|
121 | addl %ebx, %ebx |
---|
122 | jnz dcl2_n2b |
---|
123 | dcl1_n2b: |
---|
124 | call getbit32 |
---|
125 | dcl2_n2b: |
---|
126 | jc decompr_literals_n2b |
---|
127 | xor %xAX, %xAX |
---|
128 | inc %xAX /* m_off = 1 */ |
---|
129 | loop1_n2b: |
---|
130 | call getbit1 |
---|
131 | adc %xAX, %xAX /* m_off = m_off*2 + getbit() */ |
---|
132 | call getbit1 |
---|
133 | jnc loop1_n2b /* while(!getbit()) */ |
---|
134 | sub $3, %xAX |
---|
135 | jb decompr_ebpeax_n2b /* if (m_off == 2) goto decompr_ebpeax_n2b ? */ |
---|
136 | shl $8, %xAX |
---|
137 | ADDR32 movb (%xSI), %al /* m_off = (m_off - 3)*256 + src[ilen++] */ |
---|
138 | inc %xSI |
---|
139 | xor $-1, %xAX |
---|
140 | jz decompr_end_n2b /* if (m_off == 0xffffffff) goto decomp_end_n2b */ |
---|
141 | mov %xAX, %xBP /* last_m_off = m_off ?*/ |
---|
142 | decompr_ebpeax_n2b: |
---|
143 | xor %xCX, %xCX |
---|
144 | call getbit1 |
---|
145 | adc %xCX, %xCX /* m_len = getbit() */ |
---|
146 | call getbit1 |
---|
147 | adc %xCX, %xCX /* m_len = m_len*2 + getbit()) */ |
---|
148 | jnz decompr_got_mlen_n2b /* if (m_len == 0) goto decompr_got_mlen_n2b */ |
---|
149 | inc %xCX /* m_len++ */ |
---|
150 | loop2_n2b: |
---|
151 | call getbit1 |
---|
152 | adc %xCX, %xCX /* m_len = m_len*2 + getbit() */ |
---|
153 | call getbit1 |
---|
154 | jnc loop2_n2b /* while(!getbit()) */ |
---|
155 | inc %xCX |
---|
156 | inc %xCX /* m_len += 2 */ |
---|
157 | decompr_got_mlen_n2b: |
---|
158 | cmp $-0xd00, %xBP |
---|
159 | adc $1, %xCX /* m_len = m_len + 1 + (last_m_off > 0xd00) */ |
---|
160 | push %xSI |
---|
161 | ADDR32 lea (%xBP,%xDI), %xSI /* m_pos = dst + olen + -m_off */ |
---|
162 | rep |
---|
163 | es ADDR32 movsb /* dst[olen++] = *m_pos++ while(m_len > 0) */ |
---|
164 | pop %xSI |
---|
165 | jmp decompr_loop_n2b |
---|
166 | |
---|
167 | |
---|
168 | getbit1: |
---|
169 | addl %ebx, %ebx |
---|
170 | jnz 1f |
---|
171 | getbit32: |
---|
172 | ADDR32 movl (%xSI), %ebx |
---|
173 | sub $-4, %xSI /* sets carry flag */ |
---|
174 | adcl %ebx, %ebx |
---|
175 | 1: |
---|
176 | ret |
---|
177 | |
---|
178 | decompr_end_n2b: |
---|
179 | /* Restore registers and return */ |
---|
180 | pop %xBP |
---|
181 | pop %xCX |
---|
182 | popl %ebx |
---|
183 | pop %xAX |
---|
184 | ret |
---|