mirrors/Bochs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

fixed comments for SHLD/SHRD instructrions and make code a little more clear

Browse Source
This commit is contained in:
Stanislav Shwartsman 2012-09-09 17:44:42 +00:00
parent d2aaa00336
commit 0386f49e03
1 changed files with 26 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
bochs/cpu/shift16.cc
View File

@ -40,19 +40,22 @@ BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::SHLD_EwGwM(bxInstruction_c *i)
bx_address eaddr = BX_CPU_CALL_METHODR(i->ResolveModrm, (i));
Bit16u op1_16 = read_RMW_virtual_word(i->seg(), eaddr);
Bit32u op1_16 = (Bit32u) read_RMW_virtual_word(i->seg(), eaddr);
if (count) {
Bit16u op2_16 = BX_READ_16BIT_REG(i->src());
Bit32u op2_16 = (Bit32u) BX_READ_16BIT_REG(i->src());
/* count < 32, since only lower 5 bits used */
temp_32 = ((Bit32u)(op1_16) << 16) | (op2_16); // double formed by op1:op2
temp_32 = (op1_16 << 16) | (op2_16); // double formed by op1:op2
result_32 = temp_32 << count;
// hack to act like x86 SHLD when count > 16
if (count > 16) {
// when count > 16 actually shifting op1:op2:op2 << count,
// for Pentium processor, when count > 16, actually shifting op1:op2:op2 << count,
// it is the same as shifting op2:op2 by count-16
// For P6 and later (CPU_LEVEL >= 6), when count > 16, actually shifting op1:op2:op1 << count,
// which is the same as shifting op2:op1 by count-16
// The behavior is undefined so both ways are correct, we prefer P6 way of implementation
result_32 |= (op1_16 << (count - 16));
}
@ -85,17 +88,20 @@ BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::SHLD_EwGwR(bxInstruction_c *i)
count &= 0x1f; // use only 5 LSB's
if (count) {
Bit16u op1_16 = BX_READ_16BIT_REG(i->dst());
Bit16u op2_16 = BX_READ_16BIT_REG(i->src());
Bit32u op1_16 = (Bit32u) BX_READ_16BIT_REG(i->dst());
Bit32u op2_16 = (Bit32u) BX_READ_16BIT_REG(i->src());
/* count < 32, since only lower 5 bits used */
temp_32 = ((Bit32u)(op1_16) << 16) | (op2_16); // double formed by op1:op2
temp_32 = (op1_16 << 16) | (op2_16); // double formed by op1:op2
result_32 = temp_32 << count;
// hack to act like x86 SHLD when count > 16
if (count > 16) {
// when count > 16 actually shifting op1:op2:op2 << count,
// for Pentium processor, when count > 16, actually shifting op1:op2:op2 << count,
// it is the same as shifting op2:op2 by count-16
// For P6 and later (CPU_LEVEL >= 6), when count > 16, actually shifting op1:op2:op1 << count,
// which is the same as shifting op2:op1 by count-16
// The behavior is undefined so both ways are correct, we prefer P6 way of implementation
result_32 |= (op1_16 << (count - 16));
}
@ -128,10 +134,10 @@ BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::SHRD_EwGwM(bxInstruction_c *i)
bx_address eaddr = BX_CPU_CALL_METHODR(i->ResolveModrm, (i));
Bit16u op1_16 = read_RMW_virtual_word(i->seg(), eaddr);
Bit32u op1_16 = (Bit32u) read_RMW_virtual_word(i->seg(), eaddr);
if (count) {
Bit16u op2_16 = BX_READ_16BIT_REG(i->src());
Bit32u op2_16 = (Bit32u) BX_READ_16BIT_REG(i->src());
/* count < 32, since only lower 5 bits used */
temp_32 = (op2_16 << 16) | op1_16; // double formed by op2:op1
@ -139,8 +145,11 @@ BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::SHRD_EwGwM(bxInstruction_c *i)
// hack to act like x86 SHRD when count > 16
if (count > 16) {
// when count > 16 actually shifting op2:op2:op1 >> count,
// for Pentium processor, when count > 16, actually shifting op2:op2:op1 >> count,
// it is the same as shifting op2:op2 by count-16
// For P6 and later (CPU_LEVEL >= 6), when count > 16, actually shifting op1:op2:op1 >> count,
// which is the same as shifting op1:op2 by count-16
// The behavior is undefined so both ways are correct, we prefer P6 way of implementation
result_32 |= (op1_16 << (32 - count));
}
@ -172,8 +181,8 @@ BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::SHRD_EwGwR(bxInstruction_c *i)
count &= 0x1f; /* use only 5 LSB's */
if (count) {
Bit16u op1_16 = BX_READ_16BIT_REG(i->dst());
Bit16u op2_16 = BX_READ_16BIT_REG(i->src());
Bit32u op1_16 = (Bit32u) BX_READ_16BIT_REG(i->dst());
Bit32u op2_16 = (Bit32u) BX_READ_16BIT_REG(i->src());
/* count < 32, since only lower 5 bits used */
temp_32 = (op2_16 << 16) | op1_16; // double formed by op2:op1
@ -181,8 +190,11 @@ BX_INSF_TYPE BX_CPP_AttrRegparmN(1) BX_CPU_C::SHRD_EwGwR(bxInstruction_c *i)
// hack to act like x86 SHRD when count > 16
if (count > 16) {
// when count > 16 actually shifting op2:op2:op1 >> count,
// for Pentium processor, when count > 16, actually shifting op2:op2:op1 >> count,
// it is the same as shifting op2:op2 by count-16
// For P6 and later (CPU_LEVEL >= 6), when count > 16, actually shifting op1:op2:op1 >> count,
// which is the same as shifting op1:op2 by count-16
// The behavior is undefined so both ways are correct, we prefer P6 way of implementation
result_32 |= (op1_16 << (32 - count));
}