LEB128即"Little-Endian Base 128",基于128的小印第安序编码格式,是对任意有符号或者无符号整型数的可变长度的编码。
也即,用LEB128编码的正数,会根据数字的大小改变所占字节数。在android的.dex文件中,他只用来编码32bits的整型数。
格式:
上图只是指示性的用两个字节表示。编码的每个字节有效部分只有低7bits,每个字节的最高bit用来指示是否是最后一个字节。
非最高字节的bit7为0
最高字节的bit7为1
将leb128编码的数字转换为可读数字的规则是:除去每个字节的bit7,将每个字节剩余的7个bits拼接在一起,即为数字。
比如:
LEB128编码的0x02b0 ---> 转换后的数字0x0130
转换过程:
0x02b0 ---> 0000 0010 1011 0000 -->去除最高位--> 000 0010 011 0000 -->按4bits重排 --> 00 0001 0011 0000 --> 0x130
转换函数为:
代码位于:android/dalvik/libdex/leb128.h
1 /* 2 * Reads an unsigned LEB128 value, updating the given pointer to point 3 * just past the end of the read value. This function tolerates 4 * non-zero high-order bits in the fifth encoded byte. 5 */ 6 DEX_INLINE int readUnsignedLeb128(const u1** pStream) { 7 const u1* ptr = *pStream; 8 int result = *(ptr++); 9 10 if (result > 0x7f) { 11 int cur = *(ptr++); 12 result = (result & 0x7f) | ((cur & 0x7f) << 7); 13 if (cur > 0x7f) { 14 cur = *(ptr++); 15 result |= (cur & 0x7f) << 14; 16 if (cur > 0x7f) { 17 cur = *(ptr++); 18 result |= (cur & 0x7f) << 21; 19 if (cur > 0x7f) { 20 /* 21 * Note: We don't check to see if cur is out of 22 * range here, meaning we tolerate garbage in the 23 * high four-order bits. 24 */ 25 cur = *(ptr++); 26 result |= cur << 28; 27 } 28 } 29 } 30 } 31 32 *pStream = ptr; 33 return result; 34 } 35 36 /* 37 * Reads a signed LEB128 value, updating the given pointer to point 38 * just past the end of the read value. This function tolerates 39 * non-zero high-order bits in the fifth encoded byte. 40 */ 41 DEX_INLINE int readSignedLeb128(const u1** pStream) { 42 const u1* ptr = *pStream; 43 int result = *(ptr++); 44 45 if (result <= 0x7f) { 46 result = (result << 25) >> 25; 47 } else { 48 int cur = *(ptr++); 49 result = (result & 0x7f) | ((cur & 0x7f) << 7); 50 if (cur <= 0x7f) { 51 result = (result << 18) >> 18; 52 } else { 53 cur = *(ptr++); 54 result |= (cur & 0x7f) << 14; 55 if (cur <= 0x7f) { 56 result = (result << 11) >> 11; 57 } else { 58 cur = *(ptr++); 59 result |= (cur & 0x7f) << 21; 60 if (cur <= 0x7f) { 61 result = (result << 4) >> 4; 62 } else { 63 /* 64 * Note: We don't check to see if cur is out of 65 * range here, meaning we tolerate garbage in the 66 * high four-order bits. 67 */ 68 cur = *(ptr++); 69 result |= cur << 28; 70 } 71 } 72 } 73 } 74 75 *pStream = ptr; 76 return result; 77 }