From 30c2c0f050ad083e70a7e75b521c112465c212a2 Mon Sep 17 00:00:00 2001
From: Thomas Harte <thomas.harte@gmail.com>
Date: Thu, 7 Jan 2021 21:59:00 -0500
Subject: [PATCH] Attempts to complete operand recognition.
---
Processors/Decoders/x86/x86.cpp | 55 ++++++++++++++++++++++++++-------
Processors/Decoders/x86/x86.hpp | 16 ++++++++--
2 files changed, 58 insertions(+), 13 deletions(-)
diff --git a/Processors/Decoders/x86/x86.cpp b/Processors/Decoders/x86/x86.cpp
index 2837cec08..236289bf9 100644
--- a/Processors/Decoders/x86/x86.cpp
+++ b/Processors/Decoders/x86/x86.cpp
@@ -285,9 +285,9 @@ Instruction Decoder::decode(const uint8_t *source, size_t length) {
case 0xe7: AddrReg(OUT, AX, 2, 1); break;
case 0xe8: RegData(CALLD, None, 2); break;
- case 0xe9: RegData(JMP, None, 2); break;
- case 0xea: Far(JMP); break;
- case 0xeb: Jump(JMP); break;
+ case 0xe9: RegData(JMPN, None, 2); break;
+ case 0xea: Far(JMPF); break;
+ case 0xeb: Jump(JMPN); break;
case 0xec: Complete(IN, DX, AL, 1); break;
case 0xed: Complete(IN, DX, AX, 1); break;
@@ -306,14 +306,8 @@ Instruction Decoder::decode(const uint8_t *source, size_t length) {
case 0xfc: Complete(CLD, None, None, 1); break;
case 0xfd: Complete(STD, None, None, 1); break;
- /*
- Unimplemented (but should be):
-
- 0xfe, 0xff
-
- [and consider which others are unused but seem to be
- known to consume a second byte?]
- */
+ case 0xfe: MemRegReg(Invalid, MemRegINC_DEC, 1); break;
+ case 0xff: MemRegReg(Invalid, MemRegINC_to_PUSH, 1); break;
// Other prefix bytes.
case 0xf0: lock_ = true; break;
@@ -444,6 +438,45 @@ Instruction Decoder::decode(const uint8_t *source, size_t length) {
}
break;
+ case ModRegRMFormat::MemRegINC_DEC:
+ source_ = destination_ = memreg;
+
+ switch(reg) {
+ default:
+ reset_parsing();
+ return Instruction(consumed_);
+
+ case 0: operation_ = Operation::INC; break;
+ case 1: operation_ = Operation::DEC; break;
+ }
+ break;
+
+ case ModRegRMFormat::MemRegINC_to_PUSH:
+ source_ = destination_ = memreg;
+
+ switch(reg) {
+ default:
+ reset_parsing();
+ return Instruction(consumed_);
+
+ case 0: operation_ = Operation::INC; break;
+ case 1: operation_ = Operation::DEC; break;
+ case 2: operation_ = Operation::CALLN; break;
+ case 3:
+ operation_ = Operation::CALLF;
+ operand_size_ = 4;
+ source_ = Source::Immediate;
+ break;
+ case 4: operation_ = Operation::JMPN; break;
+ case 5:
+ operation_ = Operation::JMPF;
+ operand_size_ = 4;
+ source_ = Source::Immediate;
+ break;
+ case 6: operation_ = Operation::PUSH; break;
+ }
+ break;
+
case ModRegRMFormat::MemRegPOP:
source_ = destination_ = memreg;
diff --git a/Processors/Decoders/x86/x86.hpp b/Processors/Decoders/x86/x86.hpp
index bb6a7b312..7f43b8ca9 100644
--- a/Processors/Decoders/x86/x86.hpp
+++ b/Processors/Decoders/x86/x86.hpp
@@ -41,7 +41,7 @@ enum class Operation: uint8_t {
CALLF,
/// Displacement call; followed by a 16-bit operand providing a call offset.
CALLD,
- /* TODO: other CALLs */
+ CALLN,
/// Convert byte into word; source will be AL, destination will be AH.
CBW,
/// Clear carry flag; no source or destination provided.
@@ -68,7 +68,9 @@ enum class Operation: uint8_t {
INC, INT, INT3, INTO, IRET,
JO, JNO, JB, JNB, JE, JNE, JBE, JNBE,
JS, JNS, JP, JNP, JL, JNL, JLE, JNLE,
- JMP, JCXZ,
+ JMPN,
+ JMPF,
+ JCXZ,
LAHF, LDS, LEA,
LODS, LOOPE, LOOPNE, MOV, MOVS, MUL, NEG, NOP, NOT, OR, OUT,
POP, POPF, PUSH, PUSHF, RCL, RCR, REP, ROL, ROR, SAHF,
@@ -210,6 +212,16 @@ struct Decoder {
// register based on the reg field.
SegReg,
+ // Parse for mode and register/memory fields, populating the
+ // source_ and destination_ fields with the result. Uses the
+ // 'register' field to pick INC or DEC.
+ MemRegINC_DEC,
+
+ // Parse for mode and register/memory fields, populating the
+ // source_ and destination_ fields with the result. Uses the
+ // 'register' field to pick from INC/DEC/CALL/JMP/PUSH, altering
+ // the source to ::Immediate and setting an operand size if necessary.
+ MemRegINC_to_PUSH,
} modregrm_format_ = ModRegRMFormat::MemReg_Reg;
// Ephemeral decoding state.