From 0bd63cf00f4e3c6cb5fa8d843f89032a5b9bfd8b Mon Sep 17 00:00:00 2001
From: Thomas Harte <thomas.harte@gmail.com>
Date: Thu, 10 Feb 2022 09:35:05 -0500
Subject: [PATCH] Introduces the easy F page instructions.

---
 InstructionSets/x86/Decoder.cpp | 45 ++++++++++++++++++++++++++++-----
 InstructionSets/x86/Decoder.hpp |  4 ++-
 2 files changed, 42 insertions(+), 7 deletions(-)

diff --git a/InstructionSets/x86/Decoder.cpp b/InstructionSets/x86/Decoder.cpp
index 380bd8c5a..bf03357bf 100644
--- a/InstructionSets/x86/Decoder.cpp
+++ b/InstructionSets/x86/Decoder.cpp
@@ -81,6 +81,12 @@ std::pair<int, InstructionSet::x86::Instruction> Decoder::decode(const uint8_t *
 	displacement_size_ = 2;							\
 	operand_size_ = 1;								\
 
+#define undefined()	{												\
+	const auto result = std::make_pair(consumed_, Instruction());	\
+	reset_parsing();												\
+	return result;													\
+}
+
 	while(phase_ == Phase::Instruction && source != end) {
 		// Retain the instruction byte, in case additional decoding is deferred
 		// to the ModRegRM byte.
@@ -88,12 +94,6 @@ std::pair<int, InstructionSet::x86::Instruction> Decoder::decode(const uint8_t *
 		++source;
 		++consumed_;
 
-#define undefined()	{	\
-	const auto result = std::make_pair(consumed_, Instruction());	\
-	reset_parsing();	\
-	return result;	\
-}
-
 		switch(instr_) {
 			default: undefined();
 
@@ -112,6 +112,13 @@ std::pair<int, InstructionSet::x86::Instruction> Decoder::decode(const uint8_t *
 			PartialBlock(0x08, OR);						break;
 			case 0x0e: Complete(PUSH, CS, None, 2);		break;
 
+			// The 286 onwards have a further set of instructions
+			// prefixed with $0f.
+			case 0x0f:
+				if(model_ < Model::i80286) undefined();
+				phase_ = Phase::InstructionPageF;
+			break;
+
 			PartialBlock(0x10, ADC);					break;
 			case 0x16: Complete(PUSH, SS, None, 2);		break;
 			case 0x17: Complete(POP, None, SS, 2);		break;
@@ -167,6 +174,10 @@ std::pair<int, InstructionSet::x86::Instruction> Decoder::decode(const uint8_t *
 				if(model_ < Model::i80186) undefined();
 				MemRegReg(BOUND, Reg_MemReg, 2);
 			break;
+			case 0x63:
+				if(model_ < Model::i80286) undefined();
+				MemRegReg(ARPL, MemReg_Reg, 2);
+			break;
 			case 0x6c:	// INSB
 				if(model_ < Model::i80186) undefined();
 				Complete(INS, None, None, 1);
@@ -365,6 +376,28 @@ std::pair<int, InstructionSet::x86::Instruction> Decoder::decode(const uint8_t *
 		}
 	}
 
+	// MARK: - Additional F page of instructions.
+	if(phase_ == Phase::InstructionPageF && source != end) {
+		// Update the instruction acquired.
+		instr_ = 0x0f00 | *source;
+		++source;
+		++consumed_;
+
+		// NB: to reach here, the instruction set must be at least
+		// that of an 80286.
+		switch(instr_) {
+			default: undefined();
+
+			case 0x02:	MemRegReg(LAR, Reg_MemReg, 2);	break;
+			case 0x03:	MemRegReg(LSL, Reg_MemReg, 2);	break;
+			case 0x06:	Complete(CLTS, None, None, 1);	break;
+		}
+			// TODO: 0x0f 0x00 -> LLDT/SLDT/VERR/VERW/LTR/STR
+			// TODO: 0x0f 0x01 -> LGDT/LIDT/SGDT/SIDT/LMSW/SMSW
+			// TODO: 0x0f 0x05 -> LOADALL
+
+	}
+
 #undef Far
 #undef Jump
 #undef MemRegReg
diff --git a/InstructionSets/x86/Decoder.hpp b/InstructionSets/x86/Decoder.hpp
index 4f0cdcc7b..dd0182d36 100644
--- a/InstructionSets/x86/Decoder.hpp
+++ b/InstructionSets/x86/Decoder.hpp
@@ -48,6 +48,8 @@ class Decoder {
 		enum class Phase {
 			/// Captures all prefixes and continues until an instruction byte is encountered.
 			Instruction,
+			/// Having encountered a 0x0f first instruction byte, waits for the next byte fully to determine the instruction.
+			InstructionPageF,
 			/// Receives a ModRegRM byte and either populates the source_ and dest_ fields appropriately
 			/// or completes decoding of the instruction, as per the instruction format.
 			ModRegRM,
@@ -119,7 +121,7 @@ class Decoder {
 
 		// Ephemeral decoding state.
 		Operation operation_ = Operation::Invalid;
-		uint8_t instr_ = 0x00;	// TODO: is this desired, versus loading more context into ModRegRMFormat?
+		uint16_t instr_ = 0x0000;	// TODO: is this desired, versus loading more context into ModRegRMFormat?
 		int consumed_ = 0, operand_bytes_ = 0;
 
 		// Source and destination locations.