xref: /freebsd/contrib/xz/src/liblzma/simple/x86.c (revision 3b35e7ee)
1*3b35e7eeSXin LI // SPDX-License-Identifier: 0BSD
2*3b35e7eeSXin LI 
381ad8388SMartin Matuska ///////////////////////////////////////////////////////////////////////////////
481ad8388SMartin Matuska //
581ad8388SMartin Matuska /// \file       x86.c
681ad8388SMartin Matuska /// \brief      Filter for x86 binaries (BCJ filter)
781ad8388SMartin Matuska ///
881ad8388SMartin Matuska //  Authors:    Igor Pavlov
981ad8388SMartin Matuska //              Lasse Collin
1081ad8388SMartin Matuska //
1181ad8388SMartin Matuska ///////////////////////////////////////////////////////////////////////////////
1281ad8388SMartin Matuska 
1381ad8388SMartin Matuska #include "simple_private.h"
1481ad8388SMartin Matuska 
1581ad8388SMartin Matuska 
1681ad8388SMartin Matuska #define Test86MSByte(b) ((b) == 0 || (b) == 0xFF)
1781ad8388SMartin Matuska 
1881ad8388SMartin Matuska 
191456f0f9SXin LI typedef struct {
2081ad8388SMartin Matuska 	uint32_t prev_mask;
2181ad8388SMartin Matuska 	uint32_t prev_pos;
221456f0f9SXin LI } lzma_simple_x86;
2381ad8388SMartin Matuska 
2481ad8388SMartin Matuska 
2581ad8388SMartin Matuska static size_t
x86_code(void * simple_ptr,uint32_t now_pos,bool is_encoder,uint8_t * buffer,size_t size)261456f0f9SXin LI x86_code(void *simple_ptr, uint32_t now_pos, bool is_encoder,
2781ad8388SMartin Matuska 		uint8_t *buffer, size_t size)
2881ad8388SMartin Matuska {
29*3b35e7eeSXin LI 	static const uint32_t MASK_TO_BIT_NUMBER[5] = { 0, 1, 2, 2, 3 };
3081ad8388SMartin Matuska 
311456f0f9SXin LI 	lzma_simple_x86 *simple = simple_ptr;
3281ad8388SMartin Matuska 	uint32_t prev_mask = simple->prev_mask;
3381ad8388SMartin Matuska 	uint32_t prev_pos = simple->prev_pos;
3481ad8388SMartin Matuska 
3581ad8388SMartin Matuska 	if (size < 5)
3681ad8388SMartin Matuska 		return 0;
3781ad8388SMartin Matuska 
3881ad8388SMartin Matuska 	if (now_pos - prev_pos > 5)
3981ad8388SMartin Matuska 		prev_pos = now_pos - 5;
4081ad8388SMartin Matuska 
4181ad8388SMartin Matuska 	const size_t limit = size - 5;
4281ad8388SMartin Matuska 	size_t buffer_pos = 0;
4381ad8388SMartin Matuska 
4481ad8388SMartin Matuska 	while (buffer_pos <= limit) {
4581ad8388SMartin Matuska 		uint8_t b = buffer[buffer_pos];
4681ad8388SMartin Matuska 		if (b != 0xE8 && b != 0xE9) {
4781ad8388SMartin Matuska 			++buffer_pos;
4881ad8388SMartin Matuska 			continue;
4981ad8388SMartin Matuska 		}
5081ad8388SMartin Matuska 
5181ad8388SMartin Matuska 		const uint32_t offset = now_pos + (uint32_t)(buffer_pos)
5281ad8388SMartin Matuska 				- prev_pos;
5381ad8388SMartin Matuska 		prev_pos = now_pos + (uint32_t)(buffer_pos);
5481ad8388SMartin Matuska 
5581ad8388SMartin Matuska 		if (offset > 5) {
5681ad8388SMartin Matuska 			prev_mask = 0;
5781ad8388SMartin Matuska 		} else {
5881ad8388SMartin Matuska 			for (uint32_t i = 0; i < offset; ++i) {
5981ad8388SMartin Matuska 				prev_mask &= 0x77;
6081ad8388SMartin Matuska 				prev_mask <<= 1;
6181ad8388SMartin Matuska 			}
6281ad8388SMartin Matuska 		}
6381ad8388SMartin Matuska 
6481ad8388SMartin Matuska 		b = buffer[buffer_pos + 4];
6581ad8388SMartin Matuska 
66*3b35e7eeSXin LI 		if (Test86MSByte(b) && (prev_mask >> 1) <= 4
67*3b35e7eeSXin LI 			&& (prev_mask >> 1) != 3) {
6881ad8388SMartin Matuska 
6981ad8388SMartin Matuska 			uint32_t src = ((uint32_t)(b) << 24)
7081ad8388SMartin Matuska 				| ((uint32_t)(buffer[buffer_pos + 3]) << 16)
7181ad8388SMartin Matuska 				| ((uint32_t)(buffer[buffer_pos + 2]) << 8)
7281ad8388SMartin Matuska 				| (buffer[buffer_pos + 1]);
7381ad8388SMartin Matuska 
7481ad8388SMartin Matuska 			uint32_t dest;
7581ad8388SMartin Matuska 			while (true) {
7681ad8388SMartin Matuska 				if (is_encoder)
7781ad8388SMartin Matuska 					dest = src + (now_pos + (uint32_t)(
7881ad8388SMartin Matuska 							buffer_pos) + 5);
7981ad8388SMartin Matuska 				else
8081ad8388SMartin Matuska 					dest = src - (now_pos + (uint32_t)(
8181ad8388SMartin Matuska 							buffer_pos) + 5);
8281ad8388SMartin Matuska 
8381ad8388SMartin Matuska 				if (prev_mask == 0)
8481ad8388SMartin Matuska 					break;
8581ad8388SMartin Matuska 
8681ad8388SMartin Matuska 				const uint32_t i = MASK_TO_BIT_NUMBER[
8781ad8388SMartin Matuska 						prev_mask >> 1];
8881ad8388SMartin Matuska 
8981ad8388SMartin Matuska 				b = (uint8_t)(dest >> (24 - i * 8));
9081ad8388SMartin Matuska 
9181ad8388SMartin Matuska 				if (!Test86MSByte(b))
9281ad8388SMartin Matuska 					break;
9381ad8388SMartin Matuska 
94a8675d92SXin LI 				src = dest ^ ((1U << (32 - i * 8)) - 1);
9581ad8388SMartin Matuska 			}
9681ad8388SMartin Matuska 
9781ad8388SMartin Matuska 			buffer[buffer_pos + 4]
9881ad8388SMartin Matuska 					= (uint8_t)(~(((dest >> 24) & 1) - 1));
9981ad8388SMartin Matuska 			buffer[buffer_pos + 3] = (uint8_t)(dest >> 16);
10081ad8388SMartin Matuska 			buffer[buffer_pos + 2] = (uint8_t)(dest >> 8);
10181ad8388SMartin Matuska 			buffer[buffer_pos + 1] = (uint8_t)(dest);
10281ad8388SMartin Matuska 			buffer_pos += 5;
10381ad8388SMartin Matuska 			prev_mask = 0;
10481ad8388SMartin Matuska 
10581ad8388SMartin Matuska 		} else {
10681ad8388SMartin Matuska 			++buffer_pos;
10781ad8388SMartin Matuska 			prev_mask |= 1;
10881ad8388SMartin Matuska 			if (Test86MSByte(b))
10981ad8388SMartin Matuska 				prev_mask |= 0x10;
11081ad8388SMartin Matuska 		}
11181ad8388SMartin Matuska 	}
11281ad8388SMartin Matuska 
11381ad8388SMartin Matuska 	simple->prev_mask = prev_mask;
11481ad8388SMartin Matuska 	simple->prev_pos = prev_pos;
11581ad8388SMartin Matuska 
11681ad8388SMartin Matuska 	return buffer_pos;
11781ad8388SMartin Matuska }
11881ad8388SMartin Matuska 
11981ad8388SMartin Matuska 
12081ad8388SMartin Matuska static lzma_ret
x86_coder_init(lzma_next_coder * next,const lzma_allocator * allocator,const lzma_filter_info * filters,bool is_encoder)12153200025SRui Paulo x86_coder_init(lzma_next_coder *next, const lzma_allocator *allocator,
12281ad8388SMartin Matuska 		const lzma_filter_info *filters, bool is_encoder)
12381ad8388SMartin Matuska {
12481ad8388SMartin Matuska 	const lzma_ret ret = lzma_simple_coder_init(next, allocator, filters,
1251456f0f9SXin LI 			&x86_code, sizeof(lzma_simple_x86), 5, 1, is_encoder);
12681ad8388SMartin Matuska 
12781ad8388SMartin Matuska 	if (ret == LZMA_OK) {
1281456f0f9SXin LI 		lzma_simple_coder *coder = next->coder;
1291456f0f9SXin LI 		lzma_simple_x86 *simple = coder->simple;
1301456f0f9SXin LI 		simple->prev_mask = 0;
1311456f0f9SXin LI 		simple->prev_pos = (uint32_t)(-5);
13281ad8388SMartin Matuska 	}
13381ad8388SMartin Matuska 
13481ad8388SMartin Matuska 	return ret;
13581ad8388SMartin Matuska }
13681ad8388SMartin Matuska 
13781ad8388SMartin Matuska 
13873ed8e77SXin LI #ifdef HAVE_ENCODER_X86
13981ad8388SMartin Matuska extern lzma_ret
lzma_simple_x86_encoder_init(lzma_next_coder * next,const lzma_allocator * allocator,const lzma_filter_info * filters)14053200025SRui Paulo lzma_simple_x86_encoder_init(lzma_next_coder *next,
14153200025SRui Paulo 		const lzma_allocator *allocator,
14281ad8388SMartin Matuska 		const lzma_filter_info *filters)
14381ad8388SMartin Matuska {
14481ad8388SMartin Matuska 	return x86_coder_init(next, allocator, filters, true);
14581ad8388SMartin Matuska }
14673ed8e77SXin LI #endif
14781ad8388SMartin Matuska 
14881ad8388SMartin Matuska 
14973ed8e77SXin LI #ifdef HAVE_DECODER_X86
15081ad8388SMartin Matuska extern lzma_ret
lzma_simple_x86_decoder_init(lzma_next_coder * next,const lzma_allocator * allocator,const lzma_filter_info * filters)15153200025SRui Paulo lzma_simple_x86_decoder_init(lzma_next_coder *next,
15253200025SRui Paulo 		const lzma_allocator *allocator,
15381ad8388SMartin Matuska 		const lzma_filter_info *filters)
15481ad8388SMartin Matuska {
15581ad8388SMartin Matuska 	return x86_coder_init(next, allocator, filters, false);
15681ad8388SMartin Matuska }
15773ed8e77SXin LI #endif
158