From 9f63db991c3b39f960a45da0d318837f6c9bd5b7 Mon Sep 17 00:00:00 2001
From: Thomas Harte <thomas.harte@gmail.com>
Date: Fri, 22 Sep 2023 11:07:09 -0400
Subject: [PATCH] Capture default segments, fix base/index confusion.

---
 InstructionSets/x86/Decoder.cpp               | 19 ++++++-------
 InstructionSets/x86/Instruction.hpp           | 16 +++++++++++
 OSBindings/Mac/Clock SignalTests/8088Tests.mm | 27 +++++++++++++------
 3 files changed, 43 insertions(+), 19 deletions(-)

diff --git a/InstructionSets/x86/Decoder.cpp b/InstructionSets/x86/Decoder.cpp
index 46bd274a3..d1d41184a 100644
--- a/InstructionSets/x86/Decoder.cpp
+++ b/InstructionSets/x86/Decoder.cpp
@@ -663,19 +663,16 @@ std::pair<int, typename Decoder<model>::InstructionT> Decoder<model>::decode(con
 				// Classic 16-bit decoding: mode picks a displacement size,
 				// and a few fixed index+base pairs are defined.
 				constexpr ScaleIndexBase rm_table[8] = {
-					ScaleIndexBase(0, Source::eBX, Source::eSI),
-					ScaleIndexBase(0, Source::eBX, Source::eDI),
-					ScaleIndexBase(0, Source::eBP, Source::eSI),
-					ScaleIndexBase(0, Source::eBP, Source::eDI),
-					ScaleIndexBase(0, Source::None, Source::eSI),
-					ScaleIndexBase(0, Source::None, Source::eDI),
-					ScaleIndexBase(0, Source::None, Source::eBP),
-					ScaleIndexBase(0, Source::None, Source::eBX),
+					ScaleIndexBase(0, Source::eSI, Source::eBX),
+					ScaleIndexBase(0, Source::eDI, Source::eBX),
+					ScaleIndexBase(0, Source::eSI, Source::eBP),
+					ScaleIndexBase(0, Source::eDI, Source::eBP),
+					ScaleIndexBase(0, Source::eSI, Source::None),
+					ScaleIndexBase(0, Source::eDI, Source::None),
+					ScaleIndexBase(0, Source::eBP, Source::None),
+					ScaleIndexBase(0, Source::eBX, Source::None),
 				};
 
-				// TODO: unless overridden, BP and SP are always relative to ss; SI to ds; DI to es.
-				// Capture that.
-
 				sib_ = rm_table[rm];
 			}
 		}
diff --git a/InstructionSets/x86/Instruction.hpp b/InstructionSets/x86/Instruction.hpp
index 13fbf22b3..9808ded0e 100644
--- a/InstructionSets/x86/Instruction.hpp
+++ b/InstructionSets/x86/Instruction.hpp
@@ -623,6 +623,22 @@ class DataPointer {
 			return sib_.index();
 		}
 
+		/// @returns The default segment to use for this access.
+		constexpr Source default_segment() const {
+			switch(source_) {
+				default:	return Source::None;
+
+				case Source::Indirect:
+				case Source::IndirectNoBase:
+					switch(base()) {
+						default:			return Source::DS;
+						case Source::eBP:
+						case Source::eSP:	return Source::SS;
+						case Source::eDI:	return Source::ES;
+					}
+			}
+		}
+
 		template <bool obscure_indirectNoBase = false> constexpr Source base() const {
 			if constexpr (obscure_indirectNoBase) {
 				return (source_ <= Source::IndirectNoBase) ? Source::None : sib_.base();
diff --git a/OSBindings/Mac/Clock SignalTests/8088Tests.mm b/OSBindings/Mac/Clock SignalTests/8088Tests.mm
index 3678630d8..87d74b433 100644
--- a/OSBindings/Mac/Clock SignalTests/8088Tests.mm	
+++ b/OSBindings/Mac/Clock SignalTests/8088Tests.mm	
@@ -131,14 +131,25 @@ constexpr char TestSuiteHome[] = "/Users/tharte/Projects/ProcessorTests/8088/v1"
 					instruction.operation_size() == InstructionSet::x86::DataSize::Byte ? 2 : 4
 				);
 
-			case Source::Indirect:
-				return (std::stringstream() <<
-					'[' <<
-						InstructionSet::x86::to_string(pointer.index(), data_size(instruction.address_size())) << '+' <<
-						InstructionSet::x86::to_string(pointer.base(), data_size(instruction.address_size())) << '+' <<
-						std::setfill('0') << std::setw(4) << std::uppercase << std::hex << instruction.offset() << 'h' <<
-					']'
-				).str();
+			case Source::Indirect: {
+				std::stringstream stream;
+
+				Source segment = Source::None;
+				switch(instruction.data_segment()) {
+					default: 			segment = instruction.data_segment();	break;
+					case Source::None:	segment = pointer.default_segment();	break;
+				}
+				if(segment != Source::None && segment != Source::DS) {
+					stream << InstructionSet::x86::to_string(segment, InstructionSet::x86::DataSize::None) << ':';
+				}
+
+				stream << '[';
+				stream << InstructionSet::x86::to_string(pointer.base(), data_size(instruction.address_size())) << '+';
+				stream << InstructionSet::x86::to_string(pointer.index(), data_size(instruction.address_size())) << '+';
+				stream << std::setfill('0') << std::setw(4) << std::uppercase << std::hex << instruction.offset() << 'h';
+				stream << ']';
+				return stream.str();
+			}
 
 			case Source::DirectAddress:
 				return (std::stringstream() <<