diff options
author | Gabe Black <gabeblack@google.com> | 2014-12-04 15:53:54 -0800 |
---|---|---|
committer | Gabe Black <gabeblack@google.com> | 2014-12-04 15:53:54 -0800 |
commit | 22aaa5867f2449e2a73b7891fc34072c12c199b3 (patch) | |
tree | 6083ecfd34307076c8d9c55a450e6cc118025b95 /src/arch/x86/types.hh | |
parent | 3069c28a021d3f8c29221e537d48ee382c30646f (diff) | |
download | gem5-22aaa5867f2449e2a73b7891fc34072c12c199b3.tar.xz |
x86: Rework opcode parsing to support 3 byte opcodes properly.
Instead of counting the number of opcode bytes in an instruction and recording
each byte before the actual opcode, we can represent the path we took to get to
the actual opcode byte by using a type code. That has a couple of advantages.
First, we can disambiguate the properties of opcodes of the same length which
have different properties. Second, it reduces the amount of data stored in an
ExtMachInst, making them slightly easier/faster to create and process. This
also adds some flexibility as far as how different types of opcodes are
handled, which might come in handy if we decide to support VEX or XOP
instructions.
This change also adds tables to support properly decoding 3 byte opcodes.
Before we would fall off the end of some arrays, on top of the ambiguity
described above.
This change doesn't measureably affect performance on the twolf benchmark.
--HG--
rename : src/arch/x86/isa/decoder/three_byte_opcodes.isa => src/arch/x86/isa/decoder/three_byte_0f38_opcodes.isa
rename : src/arch/x86/isa/decoder/three_byte_opcodes.isa => src/arch/x86/isa/decoder/three_byte_0f3a_opcodes.isa
Diffstat (limited to 'src/arch/x86/types.hh')
-rw-r--r-- | src/arch/x86/types.hh | 62 |
1 files changed, 36 insertions, 26 deletions
diff --git a/src/arch/x86/types.hh b/src/arch/x86/types.hh index a604c3efc..dd60c0aec 100644 --- a/src/arch/x86/types.hh +++ b/src/arch/x86/types.hh @@ -104,6 +104,33 @@ namespace X86ISA Bitfield<0> b; EndBitUnion(Rex) + enum OpcodeType { + BadOpcode, + OneByteOpcode, + TwoByteOpcode, + ThreeByte0F38Opcode, + ThreeByte0F3AOpcode + }; + + static inline const char * + opcodeTypeToStr(OpcodeType type) + { + switch (type) { + case BadOpcode: + return "bad"; + case OneByteOpcode: + return "one byte"; + case TwoByteOpcode: + return "two byte"; + case ThreeByte0F38Opcode: + return "three byte 0f38"; + case ThreeByte0F3AOpcode: + return "three byte 0f3a"; + default: + return "unrecognized!"; + } + } + BitUnion8(Opcode) Bitfield<7,3> top5; Bitfield<2,0> bottom3; @@ -136,16 +163,7 @@ namespace X86ISA //This holds all of the bytes of the opcode struct { - //The number of bytes in this opcode. Right now, we ignore that - //this can be 3 in some cases - uint8_t num; - //The first byte detected in a 2+ byte opcode. Should be 0xF0. - uint8_t prefixA; - //The second byte detected in a 3+ byte opcode. Could be 0x38-0x3F - //for some SSE instructions. 3dNow! instructions are handled as - //two byte opcodes and then split out further by the immediate - //byte. - uint8_t prefixB; + OpcodeType type; //The main opcode byte. The highest addressed byte in the opcode. Opcode op; } opcode; @@ -173,14 +191,12 @@ namespace X86ISA operator << (std::ostream & os, const ExtMachInst & emi) { ccprintf(os, "\n{\n\tleg = %#x,\n\trex = %#x,\n\t" - "op = {\n\t\tnum = %d,\n\t\top = %#x,\n\t\t" - "prefixA = %#x,\n\t\tprefixB = %#x\n\t},\n\t" + "op = {\n\t\ttype = %s,\n\t\top = %#x,\n\t\t},\n\t" "modRM = %#x,\n\tsib = %#x,\n\t" "immediate = %#x,\n\tdisplacement = %#x\n\t" "dispSize = %d}\n", (uint8_t)emi.legacy, (uint8_t)emi.rex, - emi.opcode.num, (uint8_t)emi.opcode.op, - emi.opcode.prefixA, emi.opcode.prefixB, + opcodeTypeToStr(emi.opcode.type), (uint8_t)emi.opcode.op, (uint8_t)emi.modRM, (uint8_t)emi.sib, emi.immediate, emi.displacement, emi.dispSize); return os; @@ -193,14 +209,10 @@ namespace X86ISA return false; if(emi1.rex != emi2.rex) return false; - if(emi1.opcode.num != emi2.opcode.num) + if(emi1.opcode.type != emi2.opcode.type) return false; if(emi1.opcode.op != emi2.opcode.op) return false; - if(emi1.opcode.prefixA != emi2.opcode.prefixA) - return false; - if(emi1.opcode.prefixB != emi2.opcode.prefixB) - return false; if(emi1.modRM != emi2.modRM) return false; if(emi1.sib != emi2.sib) @@ -284,13 +296,11 @@ __hash_namespace_begin template<> struct hash<X86ISA::ExtMachInst> { size_t operator()(const X86ISA::ExtMachInst &emi) const { - return (((uint64_t)emi.legacy << 56) | - ((uint64_t)emi.rex << 48) | - ((uint64_t)emi.modRM << 40) | - ((uint64_t)emi.sib << 32) | - ((uint64_t)emi.opcode.num << 24) | - ((uint64_t)emi.opcode.prefixA << 16) | - ((uint64_t)emi.opcode.prefixB << 8) | + return (((uint64_t)emi.legacy << 40) | + ((uint64_t)emi.rex << 32) | + ((uint64_t)emi.modRM << 24) | + ((uint64_t)emi.sib << 16) | + ((uint64_t)emi.opcode.type << 8) | ((uint64_t)emi.opcode.op)) ^ emi.immediate ^ emi.displacement ^ emi.mode ^ |