WIP

3 files changed, 140 insertions, 33 deletions

diff --git a/src/disasm.cpp b/src/disasm.cpp
index 3d1ac4a..d88a27e 100644
--- a/src/disasm.cpp
+++ b/src/disasm.cpp
@@ -37,7 +37,6 @@ DisasmNode::~DisasmNode()
 
 DisasmNode &DisasmMap::insertNode(uint32_t address, NodeType type)
 {
-    ASSERT(address < _code_size);
     if (IsInstruction(type)) {
         address = AlignInstructionAddress(address);
     }
@@ -73,6 +72,65 @@ DisasmNode &DisasmMap::insertNode(uint32_t address, NodeType type)
     return *node;
 }
 
+DisasmNode &DisasmMap::insertNodeQuickPeek(uint32_t address, NodeType type)
+{
+    ASSERT(IsInstruction(type));
+    address = AlignInstructionAddress(address);
+    auto *node = findNodeByAddress(address);
+    if (node) {
+        // Instruction nodes take precedence over data nodes. If a node that
+        // was previously accessed only as data now turns out to be an
+        // instruction, then it must become an instruction node.
+        // XXX: Traced data must not be classified as instruction. But the
+        // traced data support is yet to come.
+        if (IsInstruction(type) && !IsInstruction(node->type)) {
+            if (0 == (node->size & 1) && 0 == (node->address & 1)) {
+                *const_cast<NodeType*>(&node->type) = type;
+                // Make sure it is OpCode::kNone so it will be properly disassembled
+                node->op = Op{};
+            }
+        }
+        return *node;
+    }
+    node = new DisasmNode(DisasmNode::Simple(type, address));
+    ASSERT(node);
+    // Spread across the size
+    for (size_t o = 0; o < node->size; o++) {
+        ASSERT(_map[address + o] == nullptr || _map[address + o] == node);
+        _map[address + o] = node;
+    }
+    return *node;
+}
+
+DisasmNode *DisasmMap::mergeNodes(DisasmNode *primary, DisasmNode *secondary)
+{
+    ASSERT(primary->address < secondary->address);
+    ASSERT(primary->address + primary->size >= secondary->address);
+    ASSERT(primary->address + primary->size >= secondary->address + secondary->size);
+    ReferenceNode *rnode{secondary->ref_by};
+    while (rnode) {
+        for (size_t i = 0; i < rnode->refs_count; i--) {
+            primary->AddReferencedBy(rnode->refs[i].address, rnode->refs[i].type);
+        }
+        ReferenceNode *prev = rnode;
+        rnode = rnode->next;
+        delete prev;
+    }
+    if (secondary.ref_kinds & kRef1Mask) {
+        DisasmNode *node = _map[secondary.ref1_addr];
+        ASSERT(node);
+        node->RemoveReferencedBy(secondary.ref1_addr);
+    }
+    if (secondary.ref_kinds & kRef2Mask) {
+        DisasmNode *node = _map[secondary.ref2_addr];
+        ASSERT(node);
+        node->RemoveReferencedBy(secondary.ref2_addr);
+    }
+    secondary->ref_by = secondary->last_ref_by = nullptr;
+    delete secondary;
+    return primary;
+}
+
 DisasmNode &DisasmMap::insertReferencedBy(
         const uint32_t by_addr,
         const uint32_t ref_addr,
@@ -195,49 +253,40 @@ static constexpr bool IsNextLikelyAnInstruction(const Op &op)
             op.opcode != OpCode::kSTOP);
 }
 
-void DisasmMap::Disasm(
+void DisasmMap::disasmProper(
         const DataView &code, const Settings &s, size_t at, bool nested)
 {
-    at = AlignInstructionAddress(at);
-    _code_size = code.size;
-    ASSERT(_code_size <= kRomSizeBytes);
     // Some of logic of this function is covered by integration tests in
     // `test_walk_and_follow_jumps.bash`.
     bool inside_code_span = nested;
-    while (at < code.size) {
-        DisasmNode *node;
-        if (_type == DisasmMapType::kTraced) {
-            node = _map[at];
-            if (!node) {
-                if (inside_code_span) {
-                    node = &insertNode(at, NodeType::kTracedInstruction);
-                } else {
-                    at += kInstructionSizeStepBytes;
-                    continue;
-                }
+    for (at = AlignInstructionAddress(at); at < code.size;) {
+        DisasmNode *node = _map[at];
+        if (!node) {
+            if (!inside_code_span) {
+                at += kInstructionSizeStepBytes;
+                continue;
             }
-            ASSERT(node->address == at);
-        } else {
             node = &insertNode(at, NodeType::kTracedInstruction);
         }
+        ASSERT(node->address == at);
         const bool perform_disasm = node->IsYetToBeHandled(_type) || inside_code_span;
         if (perform_disasm) {
-            const auto size = node->Disasm(code, s);
-            if (canBeAllocated(*node)) {
-                // Spread across the size
-                const size_t address = node->address;
-                for (size_t o = 0; o < size; o++) {
-                    ASSERT(_map[address + o] == nullptr || _map[address + o] == node);
-                    _map[address + o] = node;
-                }
-            } else {
+            node->Disasm(code, s);
+            if (!canBeAllocated(*node)) {
                 node->DisasmAsRaw(code);
             }
+            // Spread across the size
+            const size_t size = node->size;
+            const size_t address = node->address;
+            for (size_t o = 0; o < size; o++) {
+                ASSERT(_map[address + o] == nullptr || _map[address + o] == node);
+                _map[address + o] = node;
+            }
         }
         inside_code_span = s.walk && IsNextLikelyAnInstruction(node->op);
         at += node->size;
         // NOTE: There is not much information about a reference passed further,
-        // so just don't add a reference of immediate if s.imm_labels is false
+        // so just don't add a reference of immediate if s.imm_labels is not
         // enabled.
         const bool has_ref1 = (node->ref_kinds & kRef1ImmMask)
             ? s.imm_labels
@@ -251,7 +300,7 @@ void DisasmMap::Disasm(
                     node->address, node->ref1_addr, type, ref_type);
             if (ref_node.IsYetToBeHandled(_type)) {
                 if (s.follow_jumps) {
-                    Disasm(code, s, ref_node.address, true);
+                    disasmProper(code, s, ref_node.address, true);
                 } else {
                     ref_node.DisasmAsRaw(code);
                 }
@@ -267,7 +316,7 @@ void DisasmMap::Disasm(
                     node->address, node->ref2_addr, type, ref_type);
             if (ref_node.IsYetToBeHandled(_type)) {
                 if (s.follow_jumps) {
-                    Disasm(code, s, ref_node.address, true);
+                    disasmProper(code, s, ref_node.address, true);
                 } else {
                     ref_node.DisasmAsRaw(code);
                 }
@@ -279,6 +328,54 @@ void DisasmMap::Disasm(
     }
 }
 
+void DisasmMap::disasmQuickPeek(const DataView &code, const Settings &s)
+{
+    for (size_t at = 0; at < code.size;) {
+        // QuickPeek mode goes always aligned by 2.
+        ASSERT((at & 1u) == 0);
+        DisasmNode *node = &insertNodeQuickPeek(at, NodeType::kTracedInstruction);
+        node->Disasm(code, s);
+        if ((node->size & 1) != 0) {
+            fprintf(stderr, "at=%zx, size=%zu ", at, node->size);
+            fprintf(stderr, "type=%u ", static_cast<unsigned>(node->type));
+            fprintf(stderr, "opcode=%u\n", static_cast<unsigned>(node->op.opcode));
+            ASSERT((node->size & 1) == 0);
+        }
+        // Spread across the size and merge if an intersection encountered
+        const size_t address = node->address;
+        const auto size = node->size;
+        for (size_t i = 0; i < size; i++) {
+            auto *const ptr = _map[node->address + i];
+            if (ptr != nullptr && ptr != node) {
+                node = mergeNodes(node, ptr);
+            }
+            _map[address + i] = node;
+        }
+        at += node->size;
+        // NOTE: There is not much information about a reference passed further,
+        // so just don't add a reference of immediate if s.imm_labels is not
+        // enabled.
+        const bool has_ref1 = (node->ref_kinds & kRef1ImmMask)
+            ? s.imm_labels
+            : (node->ref_kinds & kRef1Mask);
+        const bool has_code_ref1 = node->ref1_addr < code.size && has_ref1;
+        if (has_code_ref1) {
+            const NodeType type = (node->ref_kinds & (kRef1ReadMask | kRef1WriteMask))
+                ? NodeType::kRefData : NodeType::kRefInstruction;
+            const auto ref_type = ReferenceTypeFromRefKindMask1(node->ref_kinds);
+            insertReferencedBy(node->address, node->ref1_addr, type, ref_type);
+        }
+        const bool has_ref2 = (node->ref_kinds & kRef2Mask);
+        const bool has_code_ref2 = (has_ref2 && node->ref2_addr < code.size);
+        if (has_code_ref2) {
+            const NodeType type = (node->ref_kinds & (kRef2ReadMask | kRef2WriteMask))
+                ? NodeType::kRefData : NodeType::kRefInstruction;
+            const auto ref_type = ReferenceTypeFromRefKindMask2(node->ref_kinds);
+            insertReferencedBy(node->address, node->ref2_addr, type, ref_type);
+        }
+    }
+}
+
 DisasmMap::~DisasmMap()
 {
     ASSERT(_map != nullptr);
diff --git a/src/disasm.h b/src/disasm.h
index 13109a5..a3c589b 100644
--- a/src/disasm.h
+++ b/src/disasm.h
@@ -168,7 +168,6 @@ struct Symbol {
 class DisasmMap {
     const DisasmMapType _type;
     DisasmNode **_map{static_cast<DisasmNode **>(calloc(kRomSizeBytes, sizeof(*_map)))};
-    size_t _code_size{};
     Symbol *_symtab{};
     size_t _symtab_size{};
     TraceTable _tt{};
@@ -176,6 +175,9 @@ class DisasmMap {
     constexpr size_t findFirstSymbolAtAddress(
             uint32_t address, bool return_last_considered=false) const;
     DisasmNode &insertNode(uint32_t address, NodeType);
+    DisasmNode &insertNodeQuickPeek(uint32_t address, NodeType);
+    /// Returns primary, secondary ceases to exist
+    DisasmNode *mergeNodes(DisasmNode *primary, DisasmNode *secondary);
     DisasmNode &insertReferencedBy(
             const uint32_t by_addr,
             const uint32_t ref_addr,
@@ -183,6 +185,8 @@ class DisasmMap {
             const ReferenceType ref_type);
     constexpr bool canBeAllocated(const DisasmNode& node) const;
     constexpr size_t symbolsCount() const { return _symtab_size / sizeof *_symtab; }
+    void disasmQuickPeek(const DataView &code, const Settings &);
+    void disasmProper(const DataView &code, const Settings &, size_t from=0, bool nested=false);
 public:
     constexpr const Symbol *Symtab() const { return _symtab; }
     constexpr size_t SymbolsCount() const { return symbolsCount(); }
@@ -195,7 +199,13 @@ public:
     void InsertNode(uint32_t address, NodeType type);
     bool ApplySymbolsFromElf(const ELF::Image &);
     void ConsumeTraceTable(TraceTable &&);
-    void Disasm(const DataView &code, const Settings &, size_t from=0, bool nested=false);
+    void Disasm(const DataView &code, const Settings &s)
+    {
+        if (_type == DisasmMapType::kTraced) {
+            return disasmProper(code, s, 0, false);
+        }
+        return disasmQuickPeek(code, s);
+    }
     DisasmMap(DisasmMapType type): _type(type) {}
     ~DisasmMap();
 };
diff --git a/src/m68k.h b/src/m68k.h
index 8fe10ee..079d26b 100644
--- a/src/m68k.h
+++ b/src/m68k.h
@@ -20,7 +20,7 @@ enum class OpSize: int {
 };
 
 enum class OpCode: uint8_t {
-    kNone,
+    kNone = 0,
     kRaw, ///< Emits ".short"
     kRaw8, ///< Emits ".byte"
     kORI,