Loading...
mach_o/NListSymbolTable.cpp dyld-1340 dyld-1122.1
--- dyld/dyld-1340/mach_o/NListSymbolTable.cpp
+++ dyld/dyld-1122.1/mach_o/NListSymbolTable.cpp
@@ -23,16 +23,18 @@
 
 #include "MemoryBuffer.h"
 #include <algorithm>
-#include <mach-o/loader.h>
 #include <sys/types.h>
 #include <assert.h>
 #include <limits.h>
 #include <stdlib.h>
 #include <string.h>
 
-#if !TARGET_OS_EXCLAVEKIT
+#if BUILDING_MACHO_WRITER
+#include <mach-o/loader.h>
+#include <mach-o/nlist.h>
 #include <mach-o/stab.h>
-#endif
+#include "Algorithm.h"
+#endif // BUILDING_MACHO_WRITER
 
 #include "Symbol.h"
 #include "NListSymbolTable.h"
@@ -94,7 +96,6 @@
 {
     bool dontDeadStrip = (n_desc & N_NO_DEAD_STRIP);
     bool cold          = (n_desc & N_COLD_FUNC);
-    bool isThumb       = (n_desc & N_ARM_THUMB_DEF);
     switch ( n_type & N_TYPE ) {
         case N_UNDF:
             if ( n_value == 0 )
@@ -103,22 +104,11 @@
                 return Symbol::makeHiddenTentativeDef(symbolName, n_value, GET_COMM_ALIGN(n_desc), dontDeadStrip, cold);
             else
                 return Symbol::makeTentativeDef(symbolName, n_value, GET_COMM_ALIGN(n_desc), dontDeadStrip, cold);
-        case N_ABS: {
-            Symbol::Scope scope = Symbol::Scope::global;
-
-            if ( (n_type & N_EXT) == 0 ) {
-                if ( n_type & N_PEXT )
-                    scope = Symbol::Scope::wasLinkageUnit;
-                else
-                    scope = Symbol::Scope::translationUnit;
-            }
-            else if ( n_type & N_PEXT )
-                scope = Symbol::Scope::linkageUnit;
+        case N_ABS:
+            if ( n_type & N_EXT )
+                return Symbol::makeAbsoluteExport(symbolName, n_value, dontDeadStrip);
             else
-                scope = Symbol::Scope::global;
-
-            return Symbol::makeAbsolute(symbolName, n_value, dontDeadStrip, scope);
-        }
+                return Symbol::makeAbsoluteLocal(symbolName, n_value, dontDeadStrip);
         case N_INDR: {
             const char* importName = symbolName;
             if ( n_value < _stringPoolSize )
@@ -136,45 +126,41 @@
         }
         case N_SECT: {
             if ( (n_type & N_EXT) == 0 ) {
-                if ( n_desc & N_ALT_ENTRY ) {
-                    if ( n_type & N_PEXT )
-                        return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::wasLinkageUnit, dontDeadStrip, cold, (n_desc & N_WEAK_DEF) != 0);
-                    else
-                        return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::translationUnit, dontDeadStrip, cold, (n_desc & N_WEAK_DEF) != 0);
-                }
+                if ( n_desc & N_ALT_ENTRY )
+                    return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::translationUnit, dontDeadStrip, cold);
                 else if ( n_type & N_PEXT ) {
                     if ( n_desc & N_WEAK_DEF )
-                        return Symbol::makeWeakDefWasPrivateExtern(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
+                        return Symbol::makeWeakDefWasPrivateExtern(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
                     else
-                        return Symbol::makeRegularWasPrivateExtern(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
+                        return Symbol::makeRegularWasPrivateExtern(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
                 } else
-                    return Symbol::makeRegularLocal(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
+                    return Symbol::makeRegularLocal(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
             }
             else if ( n_type & N_PEXT ) {
-                if ( n_desc & N_ALT_ENTRY )
-                    return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::linkageUnit, dontDeadStrip, cold, (n_desc & N_WEAK_DEF) != 0);
-                else if ( n_desc & N_WEAK_DEF )
-                    return Symbol::makeWeakDefHidden(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
-                else if ( n_desc & N_SYMBOL_RESOLVER ) // rdar://123349256 (ld-prime needs to handle internal resolvers)
-                    return Symbol::makeDynamicResolver(symbolName, n_sect, 0, n_value - _preferredLoadAddress, Symbol::Scope::linkageUnit);
+                if ( n_desc & N_WEAK_DEF )
+                    return Symbol::makeWeakDefHidden(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
+                else if ( n_desc & N_ALT_ENTRY )
+                    return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::linkageUnit, dontDeadStrip, cold);
                 else
-                    return Symbol::makeRegularHidden(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
-            }
-            else if ( n_desc & N_ALT_ENTRY ) {
-                return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::global, dontDeadStrip, cold, (n_desc & N_WEAK_DEF) != 0);
+                    return Symbol::makeRegularHidden(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
             }
             else if ( (n_desc & (N_WEAK_DEF|N_WEAK_REF)) == (N_WEAK_DEF|N_WEAK_REF) ) {
-                return Symbol::makeWeakDefAutoHide(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
+                return Symbol::makeWeakDefAutoHide(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
             }
             else if ( n_desc & N_WEAK_DEF ) {
-                return Symbol::makeWeakDefExport(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, isThumb);
-            }
-            else if ( n_desc & N_SYMBOL_RESOLVER ) {
-                return Symbol::makeDynamicResolver(symbolName, n_sect, 0, n_value - _preferredLoadAddress);
+                return Symbol::makeWeakDefExport(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold);
             }
             else {
-                bool neverStrip = (n_desc & REFERENCED_DYNAMICALLY);
-                return Symbol::makeRegularExport(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, neverStrip, isThumb);
+                if ( n_desc & N_ALT_ENTRY ) {
+                    return Symbol::makeAltEntry(symbolName, n_value - _preferredLoadAddress, n_sect, Symbol::Scope::global, dontDeadStrip, cold);
+                }
+                else if ( n_desc & N_SYMBOL_RESOLVER ) {
+                    return Symbol::makeDynamicResolver(symbolName, n_sect, 0, n_value - _preferredLoadAddress);
+                }
+                else {
+                    bool neverStrip = (n_desc & REFERENCED_DYNAMICALLY);
+                    return Symbol::makeRegularExport(symbolName, n_value - _preferredLoadAddress, n_sect, dontDeadStrip, cold, neverStrip);
+                }
             }
         }
     }
@@ -183,21 +169,10 @@
 
 void NListSymbolTable::forEachExportedSymbol(void (^callback)(const Symbol& symbol, uint32_t symbolIndex, bool& stop)) const
 {
-    if ( (_localsCount == 0) && (_globalsCount == 0) && (_undefsCount == 0) && (_nlistCount != 0) ) {
-        // if no LC_DYSYMTAB, need to scan whole table and selectively find global symbols
-        forEachSymbol(^(const char* symbolName, uint64_t n_value, uint8_t n_type, uint8_t n_sect, uint16_t n_desc, uint32_t symbolIndex, bool& stop) {
-            uint8_t type = n_type & N_TYPE;
-            if ( (n_type & N_EXT) && ((type == N_SECT) || (type == N_ABS) || (type == N_INDR)) && ((n_type & N_STAB) == 0))
-                callback(symbolFromNList(symbolName, n_value, n_type, n_sect, n_desc), symbolIndex, stop);
-        });
-    } else {
-        uint32_t globalsStartIndex = _localsCount;
-        forEachSymbol(globalsStartIndex, _globalsCount, ^(const char* symbolName, uint64_t n_value, uint8_t n_type, uint8_t n_sect, uint16_t n_desc, uint32_t symbolIndex, bool& stop) {
-            uint8_t type = n_type & N_TYPE;
-            if ( (n_type & N_EXT) && ((type == N_SECT) || (type == N_ABS) || (type == N_INDR)) && ((n_type & N_STAB) == 0))
-                callback(symbolFromNList(symbolName, n_value, n_type, n_sect, n_desc), symbolIndex, stop);
-        });
-    }
+    uint32_t globalsStartIndex = _localsCount;
+    forEachSymbol(globalsStartIndex, _globalsCount, ^(const char* symbolName, uint64_t n_value, uint8_t n_type, uint8_t n_sect, uint16_t n_desc, uint32_t symbolIndex, bool& stop) {
+        callback(symbolFromNList(symbolName, n_value, n_type, n_sect, n_desc), symbolIndex, stop);
+    });
 }
 
 void NListSymbolTable::forEachDefinedSymbol(void (^callback)(const Symbol& symbol, uint32_t symbolIndex, bool& stop)) const
@@ -253,30 +228,6 @@
         }
     }
     return stop;
-}
-
-bool NListSymbolTable::symbolAtIndex(uint32_t symbolIndex, Symbol& symbol) const
-{
-    if ( symbolIndex >= _nlistCount )
-        return false;
-
-    if ( _nlist64 ) {
-        const struct nlist_64& sym = _nlist64[symbolIndex];
-        if ( sym.n_un.n_strx > _stringPoolSize )
-            return false;
-        symbol = symbolFromNList(&_stringPool[sym.n_un.n_strx], sym.n_value, sym.n_type, sym.n_sect, sym.n_desc);
-        return true;
-    }
-
-    if ( _nlist32 ) {
-        const struct nlist& sym = _nlist32[symbolIndex];
-        if ( sym.n_un.n_strx > _stringPoolSize )
-            return false;
-        symbol = symbolFromNList(&_stringPool[sym.n_un.n_strx], sym.n_value, sym.n_type, sym.n_sect, sym.n_desc);
-        return true;
-    }
-
-    return false;
 }
 
 bool NListSymbolTable::findClosestDefinedSymbol(uint64_t unslidAddr, Symbol& sym) const
@@ -344,7 +295,13 @@
         return _nlist32[symbolIndex].n_value;
 }
 
-#if !TARGET_OS_EXCLAVEKIT
+
+//
+// MARK: --- NListSymbolTable building methods ---
+//
+
+#if BUILDING_MACHO_WRITER
+
 void NListSymbolTable::forEachDebugNote(bool freeFileInfo, void (^callback)(const DebugNote& note, bool& stop)) const
 {
     __block CString   currentSrcDir = nullptr;
@@ -422,18 +379,490 @@
         }
     });
 }
-#endif // !TARGET_OS_EXCLAVEKIT
+
+uint32_t NListSymbolTable::countDebugNoteNLists(std::span<const DebugBuilderNote> debugNotes)
+{
+    uint32_t debugStabNlists=0;
+    bool startedSO=false;
+
+    for ( const DebugBuilderNote& note : debugNotes ) {
+        if ( note.fileInfo->srcDir().empty() && note.fileInfo->srcName().empty() ) {
+            debugStabNlists += 1;
+        } else {
+            if ( !startedSO ) {
+                startedSO = true;
+                debugStabNlists += 1;
+            }
+            debugStabNlists += 4;
+            if ( note.fileInfo->hasOriginLibInfo() )
+                debugStabNlists += 1;
+
+            for ( const DebugBuilderNoteItem& item : note.items ) {
+                if ( item.type == N_FUN )
+                    debugStabNlists += 4;
+                else
+                    debugStabNlists += 1;
+            }
+        }
+    }
+    return debugStabNlists;
+}
+
+template <typename T>
+void NListSymbolTable::addStabsFromDebugNotes(std::span<const DebugBuilderNote> debugNotes, bool zeroModTimes, NListBuffer& nlists)
+{
+    typedef __typeof(T::n_value) V;
+
+    bool startedSOs = false;
+    for (const DebugBuilderNote& note : debugNotes) {
+        uint32_t mtime = (zeroModTimes ? 0 : note.fileInfo->objModTime());
+        if ( note.srcDirPoolOffset == 0 && note.srcNamePoolOffset == 0 ) {
+            nlists.add(T{{note.objPathPoolOffset},  N_AST,  0,                       0, (V)mtime});
+        }
+        else {
+            if ( !startedSOs )
+                nlists.add(T{{1}, N_SO, 1, 0, 0}); // match ld64 which always started debug notes with an "end SO"
+            // Put this before the other N_SO's.  We can't put it right before the N_OSO as lldb expects the N_OSO
+            // to be immediately preceded by the N_SO
+            if ( note.originLibPathPoolOffset != 0 ) {
+                nlists.add(T{{note.originLibPathPoolOffset}, N_LIB, 0, 0, 0});
+            }
+            startedSOs = true;
+            nlists.add(T{{note.srcDirPoolOffset},  N_SO,  0,                           0, 0});
+            nlists.add(T{{note.srcNamePoolOffset}, N_SO,  0,                           0, 0});
+            nlists.add(T{{note.objPathPoolOffset}, N_OSO, note.fileInfo->objSubType(), 1, (V)mtime});
+            for (const DebugBuilderNoteItem& item : note.items) {
+                uint32_t stringPoolOffset = item.stringPoolOffset;
+                switch ( item.type ) {
+                case N_FUN:
+                    // for functions, we use four symbols to record the name, address, size, and sectNum
+                    nlists.add(T{{1},                                N_BNSYM, item.sectNum, 0, (V)item.addr});
+                    nlists.add(T{{stringPoolOffset}, N_FUN,   item.sectNum, 0, (V)item.addr});
+                    nlists.add(T{{1},                                N_FUN,   0,            0, (V)item.size});
+                    nlists.add(T{{1},                                N_ENSYM, item.sectNum, 0, (V)item.addr});
+                    break;
+                case N_STSYM:
+                    // for static variables, we record the name, address, and sectNum
+                    nlists.add(T{{stringPoolOffset}, N_STSYM, item.sectNum, 0, (V)item.addr});
+                    break;
+                case N_GSYM:
+                    // for global variables, we record just the name
+                    nlists.add(T{{stringPoolOffset}, N_GSYM,  0,            0, 0});
+                    break;
+                default:
+                    assert(false && "invalid debug note item");
+                    break;
+                }
+            }
+            nlists.add(T{{1},                           N_SO, 1, 0, 0});
+        }
+    }
+}
+
+
+/*!
+ * @class NListStringPoolBuffer
+ *
+ * @abstract
+ *      Simple NList string pool buffer, used in unit-tests.
+ */
+struct NListStringPoolBuffer
+{
+    std::vector<char> buffer;
+    uint32_t          pos = 0;
+
+    NListStringPoolBuffer()
+    {
+        add(' ');
+        add('\0');
+    }
+
+    NListStringPoolBuffer(const NListStringPoolBuffer&) = delete;
+    NListStringPoolBuffer(NListStringPoolBuffer&&) = default;
+    NListStringPoolBuffer& operator=(const NListStringPoolBuffer&) = delete;
+    NListStringPoolBuffer& operator=(NListStringPoolBuffer&&) = default;
+
+    uint32_t add(CString str);
+    uint32_t add(std::span<const char> bytes);
+    uint32_t add(char ch);
+    uint32_t size() { return pos; }
+    char*    data() { return buffer.data(); }
+
+    void     finalize(bool is64)
+    {
+        uint32_t pointerSize = is64 ? 8 : 4;
+        while ( size() % pointerSize )
+            add('\0');
+    }
+
+    std::pair<uint32_t, char*> reserve(size_t);
+};
+
+NListSymbolTable::SymbolPartition::SymbolPartition(std::span<const Symbol> symbols, bool objectFile)
+{
+    for (const Symbol& symbol : symbols) {
+        int      libOrdinal;
+        bool     weakImport;
+        uint64_t size;
+        uint8_t  p2Align;
+        if ( symbol.isUndefined(libOrdinal, weakImport) || symbol.isTentativeDef(size, p2Align) )
+            undefs.push_back(symbol);
+        else if ( symbol.scope() == Symbol::Scope::global )
+            globals.push_back(symbol);
+        else if ( (symbol.scope() == Symbol::Scope::linkageUnit) && objectFile )
+            globals.push_back(symbol); // in .o files hidden symbols are in globals range
+        else if ( (symbol.scope() == Symbol::Scope::autoHide) && objectFile )
+            globals.push_back(symbol); // in .o files hidden symbols are in globals range
+        else
+            locals.push_back(symbol);
+    }
+
+    // for historical binary search reasons, globals are sorted by name
+    std::sort(globals.begin(), globals.end(), [&](const Symbol& a, const Symbol& b) {
+        return a.name() < b.name();
+    });
+    // undefs are sorted by name
+    std::sort(undefs.begin(), undefs.end(), [&](const Symbol& a, const Symbol& b) {
+        return a.name() < b.name();
+    });
+    // locals are already sorted by their position in their section.  We don't need to sort them again
+}
+
+NListSymbolTable::NListSymbolTable(std::span<const Symbol> symbols, uint64_t prefLoadAddr, bool is64, std::span<DebugBuilderNote> debugNotes,
+                                   bool zeroModTimes, bool objectFile)
+    : NListSymbolTable(SymbolPartition(symbols, objectFile), debugNotes, prefLoadAddr, is64, zeroModTimes)
+{}
+
+
+NListSymbolTable::NListSymbolTable(const SymbolPartition& partition, std::span<DebugBuilderNote> debugNotes,
+                 uint64_t prefLoadAddr, bool is64, bool zeroModTimes)
+    : NListSymbolTable(partition.globals, partition.undefs, partition.locals, debugNotes, prefLoadAddr, is64, zeroModTimes)
+{}
+
+NListSymbolTable::NListSymbolTable(std::span<const Symbol> globals, std::span<const Symbol> undefs,
+        std::span<const Symbol> locals, std::span<DebugBuilderNote> debugNotes, uint64_t prefLoadAddr, bool is64, bool zeroModTimes)
+{
+    uint32_t numDebugNlist = countDebugNoteNLists(debugNotes);
+    size_t nlistSize = (locals.size() + globals.size() + undefs.size() + numDebugNlist) * (is64 ? sizeof(nlist_64) : sizeof(struct nlist));
+
+    NListStringPoolBuffer stringPoolBuffer;
+
+    size_t strxAllCount = globals.size() * 2 + locals.size() + undefs.size();
+    std::vector<uint32_t> strxAll(strxAllCount);
+    std::span<uint32_t> globalsStrx = std::span(strxAll).subspan(0, globals.size());
+    std::span<uint32_t> reexportsStrx = std::span(globalsStrx.end().base(), globals.size());
+    std::span<uint32_t> undefsStrx = std::span(reexportsStrx.end().base(), undefs.size());
+    std::span<uint32_t> localsStrx = std::span(undefsStrx.end().base(), locals.size());
+    for ( size_t i = 0; i < globals.size(); ++i ) {
+        const Symbol& s = globals[i];
+        globalsStrx[i] = stringPoolBuffer.add(s.name());
+        int32_t ordinal;
+        const char* importName=nullptr;
+        if ( s.isReExport(ordinal, importName) )
+            reexportsStrx[i] = stringPoolBuffer.add(s.name());
+    }
+    for ( size_t i = 0; i < undefs.size(); ++i ) {
+        const Symbol& s = undefs[i];
+        undefsStrx[i] = stringPoolBuffer.add(s.name());
+    }
+    for ( size_t i = 0; i < locals.size(); ++i ) {
+        const Symbol& s = locals[i];
+        localsStrx[i] = stringPoolBuffer.add(s.name());
+    }
+
+    for ( DebugBuilderNote& debugNote : debugNotes ) {
+        if ( CString srcDir = debugNote.fileInfo->srcDir(); !srcDir.empty() )
+            debugNote.srcDirPoolOffset = stringPoolBuffer.add(srcDir);
+        if ( CString srcName = debugNote.fileInfo->srcName(); !srcName.empty() )
+            debugNote.srcNamePoolOffset = stringPoolBuffer.add(srcName);
+        if ( CString originLibPath = debugNote.fileInfo->originLibPath(); !originLibPath.empty() )
+            debugNote.originLibPathPoolOffset=stringPoolBuffer.add(originLibPath);
+        if ( CString objPath = debugNote.fileInfo->objPath(); !objPath.empty() )
+            debugNote.objPathPoolOffset=stringPoolBuffer.add(objPath);
+
+        for ( DebugBuilderNoteItem& item : debugNote.items ) {
+            item.stringPoolOffset=stringPoolBuffer.add(item.name);
+        }
+    }
+    stringPoolBuffer.finalize(is64);
+
+    *this = NListSymbolTable(NListLayout{ globals, globalsStrx, reexportsStrx, undefs, undefsStrx, locals, localsStrx, debugNotes, numDebugNlist }, NListBuffer(nlistSize), std::move(stringPoolBuffer.buffer), prefLoadAddr, is64, zeroModTimes);
+}
+
+NListSymbolTable::NListSymbolTable(NListLayout layout, std::span<uint8_t> nlistBuffer, uint64_t prefLoadAddr, bool is64, bool zeroModTimes): NListSymbolTable(layout, NListBuffer(nlistBuffer), {}, prefLoadAddr, is64, zeroModTimes) {}
+
+NListSymbolTable::NListSymbolTable(NListLayout layout, NListBuffer nlist, std::vector<char> stringPoolBuffer, uint64_t prefLoadAddr, bool is64, bool zeroModTimes)
+    : _nlistBuffer(std::move(nlist)), _stringPoolBuffer(std::move(stringPoolBuffer))
+{
+    // partition symbols into locals, globals, and undefs
+    _localsCount    = (uint32_t)layout.locals.size() + layout.debugNotesNListCount;
+    _globalsCount   = (uint32_t)layout.globals.size();
+    _undefsCount    = (uint32_t)layout.undefs.size();
+    _nlistCount     = _localsCount + _globalsCount + _undefsCount;
+
+    assert(layout.globals.size()   == layout.globalsStrx.size());
+    assert(layout.globals.size()   == layout.reexportStrx.size());
+    assert(layout.undefs.size()    == layout.undefsStrx.size());
+    assert(layout.locals.size()    == layout.localsStrx.size());
+    std::span<uint8_t> nlistBuffer = _nlistBuffer.buffer;
+
+    // convert each symbol to nlist
+    _preferredLoadAddress = prefLoadAddr;
+    if ( is64 ) {
+        assert(nlistBuffer.size() == (_localsCount + _globalsCount + _undefsCount) * sizeof(nlist_64));
+
+        // symbol table strings are added in the order of globals, imports, locals
+        // but the nlist itself is emitted as locals, globals, imports.
+        // So we'll walk in the string order, and then create the nlist after
+
+        std::span<nlist_64> nlist64Buffer = std::span<nlist_64>((nlist_64*)nlistBuffer.data(), nlistBuffer.size() / sizeof(nlist_64));
+        std::span<nlist_64> globalsBuffer(nlist64Buffer.subspan(_localsCount, _globalsCount));
+        std::span<nlist_64> undefsBuffer(nlist64Buffer.subspan(_localsCount + _globalsCount, _undefsCount));
+        std::span<nlist_64> localsBuffer(nlist64Buffer.subspan(0, _localsCount));
+
+        dispatchForEach(layout.globals, [this, globalsBuffer, &layout](size_t i, const Symbol& sym) {
+            globalsBuffer[i] = nlist64FromSymbol(sym, layout.globalsStrx[i], layout.reexportStrx[i]);
+        });
+        dispatchForEach(layout.undefs, [this, undefsBuffer, &layout](size_t i, const Symbol& sym) {
+            undefsBuffer[i] = nlist64FromSymbol(sym, layout.undefsStrx[i], 0);
+        });
+        dispatchForEach(layout.locals, [this, localsBuffer, &layout](size_t i, const Symbol& sym) {
+            localsBuffer[i] = nlist64FromSymbol(sym, layout.localsStrx[i], 0);
+        });
+
+        NListBuffer stabsBuffer = localsBuffer.subspan(layout.locals.size());
+        assert((stabsBuffer.buffer.size() / sizeof(nlist_64)) == layout.debugNotesNListCount);
+        addStabsFromDebugNotes<nlist_64>(layout.debugNotes, zeroModTimes, stabsBuffer);
+    }
+    else {
+        // symbol table strings are added in the order of globals, imports, locals
+        // but the nlist itself is emitted as locals, globals, imports.
+        // So we'll walk in the string order, and then create the nlist after
+
+        std::span<struct nlist> nlist32Buffer = std::span<struct nlist>((struct nlist*)nlistBuffer.data(), nlistBuffer.size() / sizeof(struct nlist));
+        std::span<struct nlist> globalsBuffer(nlist32Buffer.subspan(_localsCount, _globalsCount));
+        std::span<struct nlist> undefsBuffer(nlist32Buffer.subspan(_localsCount + _globalsCount, _undefsCount));
+        std::span<struct nlist> localsBuffer(nlist32Buffer.subspan(0, _localsCount));
+
+        dispatchForEach(layout.globals, [this, globalsBuffer, &layout](size_t i, const Symbol& sym) {
+            globalsBuffer[i] = nlistFromSymbol(sym, layout.globalsStrx[i], layout.reexportStrx[i]);
+        });
+        dispatchForEach(layout.undefs, [this, undefsBuffer, &layout](size_t i, const Symbol& sym) {
+            undefsBuffer[i] = nlistFromSymbol(sym, layout.undefsStrx[i], 0);
+        });
+        dispatchForEach(layout.locals, [this, localsBuffer, &layout](size_t i, const Symbol& sym) {
+            localsBuffer[i] = nlistFromSymbol(sym, layout.localsStrx[i], 0);
+        });
+
+        NListBuffer stabsBuffer = localsBuffer.subspan(layout.locals.size());
+        assert((stabsBuffer.buffer.size() / sizeof(struct nlist)) == layout.debugNotesNListCount);
+        addStabsFromDebugNotes<struct nlist>(layout.debugNotes, zeroModTimes, stabsBuffer);
+    }
+
+    // fill in all ivars as if this came from a mach-o file
+    _preferredLoadAddress = prefLoadAddr;
+    _stringPool           = _stringPoolBuffer.data();
+    _nlist32              = is64 ? nullptr : (struct nlist*)nlistBuffer.data();
+    _nlist64              = is64 ? (nlist_64*)nlistBuffer.data() : nullptr;
+    _stringPoolSize       = (uint32_t)_stringPoolBuffer.size();
+}
+
+static uint8_t ntypeFromSymbol(const Symbol& symbol)
+{
+    switch ( symbol.scope() ) {
+        case Symbol::Scope::global:
+        case Symbol::Scope::globalNeverStrip:
+        case Symbol::Scope::autoHide:
+            return N_EXT;
+        case Symbol::Scope::linkageUnit:
+            return N_EXT | N_PEXT;
+        case Symbol::Scope::translationUnit:
+            return 0;
+        case Symbol::Scope::wasLinkageUnit:
+            return N_PEXT;
+   }
+}
+
+struct nlist_64 NListSymbolTable::nlist64FromSymbol(const Symbol& symbol, uint32_t strx, uint32_t reexportStrx)
+{
+    struct nlist_64 result;
+    int             libOrdinal;
+    uint64_t        absAddress;
+    bool            weakImport;
+    uint64_t        implOffset;
+    uint64_t        size;
+    uint64_t        stubOffset;
+    uint8_t         p2align;
+    const char*     importName;
+    if ( symbol.isTentativeDef(size, p2align) ) {
+        result.n_un.n_strx = strx;
+        result.n_type      = N_UNDF | ntypeFromSymbol(symbol);
+        result.n_sect      = 0;
+        result.n_desc      = 0;
+        result.n_value     = size;
+        SET_COMM_ALIGN(result.n_desc,p2align);
+    }
+    else if ( symbol.isUndefined(libOrdinal, weakImport) ) {
+        result.n_un.n_strx = strx;
+        result.n_type      = N_UNDF | N_EXT;
+        result.n_sect      = 0;
+        result.n_desc      = (libOrdinal << 8) | (weakImport ? N_WEAK_REF : 0);
+        result.n_value     = 0;
+    }
+    else if ( symbol.isAbsolute(absAddress) )  {
+        result.n_un.n_strx = strx;
+        result.n_type      = N_ABS | ntypeFromSymbol(symbol);
+        result.n_sect      = 0;
+        result.n_desc      = 0;
+        result.n_value     = absAddress;
+    }
+    else if ( symbol.isRegular(implOffset) || symbol.isThreadLocal(implOffset) ) {
+        uint16_t desc = 0;
+        if ( symbol.isWeakDef() ) {
+            switch ( symbol.scope() ) {
+                case Symbol::Scope::globalNeverStrip:
+                case Symbol::Scope::global:
+                case Symbol::Scope::linkageUnit:
+                case Symbol::Scope::wasLinkageUnit:
+                    desc = N_WEAK_DEF;
+                    break;
+                case Symbol::Scope::autoHide:
+                    desc = N_WEAK_DEF | N_WEAK_REF;
+                    break;
+                case Symbol::Scope::translationUnit:
+                    break;
+            }
+        }
+        if ( symbol.dontDeadStrip() )
+            desc |= N_NO_DEAD_STRIP;
+        if ( symbol.cold() )
+            desc |= N_COLD_FUNC;
+        if ( symbol.scope() == Symbol::Scope::globalNeverStrip )
+            desc |= REFERENCED_DYNAMICALLY;
+        result.n_un.n_strx = strx;
+        result.n_type      = N_SECT | ntypeFromSymbol(symbol);
+        result.n_sect      = symbol.sectionOrdinal();
+        result.n_desc      = desc;
+        result.n_value     = _preferredLoadAddress + implOffset;
+    }
+    else if ( symbol.isAltEntry(implOffset) ) {
+        uint64_t desc = N_ALT_ENTRY;
+        if ( symbol.dontDeadStrip() )
+            desc |= N_NO_DEAD_STRIP;
+        result.n_un.n_strx = strx;
+        result.n_type      = N_SECT | ntypeFromSymbol(symbol);
+        result.n_sect      = symbol.sectionOrdinal();
+        result.n_desc      = desc;
+        result.n_value     = _preferredLoadAddress + implOffset;
+    }
+    else if ( symbol.isReExport(libOrdinal, importName) ) {
+        // re-exports can't be local, they're always global in linked images,
+        // in object files they can have global/linkage unit scope or be undefined.
+        assert(symbol.scope() != Symbol::Scope::translationUnit && "re-exports can't have a translation unit");
+        result.n_un.n_strx = strx;
+        result.n_type      = N_INDR | ntypeFromSymbol(symbol);
+        result.n_sect      = 0;
+        result.n_desc      = 0;
+        result.n_value     = reexportStrx;
+    }
+    else if ( symbol.isDynamicResolver(stubOffset) ) {
+        result.n_un.n_strx = strx;
+        result.n_type      = N_SECT | ntypeFromSymbol(symbol);
+        result.n_sect      = symbol.sectionOrdinal();
+        result.n_desc      = N_SYMBOL_RESOLVER;
+        result.n_value     = _preferredLoadAddress + symbol.implOffset();
+    }
+    else {
+        assert(false && "unhandled symbol kind");
+    }
+
+    return result;
+}
+
+// avoid duplicating code by filling in nlist_64 and converting to nlist
+struct nlist NListSymbolTable::nlistFromSymbol(const Symbol& symbol, uint32_t strx, uint32_t reexportStrx)
+{
+    struct nlist_64 result64 = nlist64FromSymbol(symbol, strx, reexportStrx);
+    struct nlist result;
+    result.n_un.n_strx = result64.n_un.n_strx;
+    result.n_type      = result64.n_type;
+    result.n_sect      = result64.n_sect;
+    result.n_desc      = result64.n_desc;
+    result.n_value     = (uint32_t)result64.n_value;
+    return result;
+}
+
+std::pair<uint32_t, char*> NListStringPoolBuffer::reserve(size_t size)
+{
+    size_t startPos = pos;
+    pos += size;
+    buffer.resize(buffer.size() + size);
+    return std::make_pair(startPos, buffer.data() + startPos);
+}
+
+uint32_t NListStringPoolBuffer::add(std::span<const char> bytes)
+{
+    auto [startPos, ptr] = reserve(bytes.size());
+    memcpy(ptr, bytes.data(), bytes.size());
+    return startPos;
+}
+
+uint32_t NListStringPoolBuffer::add(char ch)
+{
+    auto [startPos, ptr] = reserve(1);
+    *ptr = ch;
+    return startPos;
+}
+
+uint32_t NListStringPoolBuffer::add(CString cstr)
+{
+    return add(std::span(cstr.c_str(), cstr.size() + 1));
+}
+
+
+#endif // BUILDING_MACHO_WRITER
+
 
 const DebugNoteFileInfo* DebugNoteFileInfo::make(CString srcDir, CString srcName, CString objPath, uint32_t objModTime, uint8_t objSubType, CString libPath, CString originLibPath)
 {
-    DebugNoteFileInfo* result = (DebugNoteFileInfo*)calloc(1, sizeof(DebugNoteFileInfo));
+    const size_t srcDirOffset  = sizeof(DebugNoteFileInfo);
+    const size_t srcNameOffset = srcDirOffset  + srcDir.size() + 1;
+    const size_t objPathOffset = srcNameOffset + srcName.size() + 1;
+    size_t lastOffset   = objPathOffset + objPath.size() + 1;
+    const size_t libPathOffset = libPath.size() > 0 ? lastOffset : 0;
+    lastOffset = libPathOffset > 0 ? libPathOffset : lastOffset;
+    const size_t originLibPathOffset = originLibPath.size() > 0 ? lastOffset + libPath.size() + 1 : 0;
+    lastOffset = originLibPathOffset > 0 ? originLibPathOffset : lastOffset + libPath.size() + 1;
+    // Update finalOffset if there is dynamic library information
+    const size_t finalOffset  = originLibPathOffset > 0 ?  lastOffset + originLibPath.size() + 1 : lastOffset;
+    const size_t contentSize   = (finalOffset + 7) & (-8); // 8-byte align
+
+    DebugNoteFileInfo* result = (DebugNoteFileInfo*)calloc(contentSize, sizeof(uint8_t));
+    result->_size          = (uint32_t)contentSize;
+    result->_version       = 1;
+    result->_srcDirOffset  = srcDirOffset;
+    result->_srcDirSize    = (uint32_t)srcDir.size();
+    result->_srcNameOffset = (uint32_t)srcNameOffset;
+    result->_srcNameSize   = (uint32_t)srcName.size();
+    result->_objPathOffset = (uint32_t)objPathOffset;
+    result->_objPathSize   = (uint32_t)objPath.size();
     result->_objModTime    = objModTime;
     result->_objSubType    = objSubType;
-    result->_srcDir        = srcDir;
-    result->_srcName       = srcName;
-    result->_objPath       = objPath;
-    result->_libPath       = libPath;
-    result->_originLibPath = originLibPath;
+    result->_libPathOffset = (uint32_t)libPathOffset;
+    result->_libPathSize   = (uint32_t)libPath.size();
+    result->_originLibPathOffset = (uint32_t)originLibPathOffset;
+    result->_originLibPathSize   = (uint32_t)originLibPath.size();
+
+    if ( srcDir.size() )
+        memcpy((char*)result + srcDirOffset,  srcDir.c_str(),  srcDir.size());
+    if ( srcName.size() )
+        memcpy((char*)result + srcNameOffset, srcName.c_str(), srcName.size());
+    if ( objPath.size() )
+        memcpy((char*)result + objPathOffset, objPath.c_str(), objPath.size());
+    if ( libPath.size() )
+        memcpy((char*)result + libPathOffset, libPath.c_str(), libPath.size());
+    if ( originLibPath.size() )
+        memcpy((char*)result + originLibPathOffset, originLibPath.c_str(), originLibPath.size());
 
     return result;
 }
@@ -445,7 +874,9 @@
 
 const DebugNoteFileInfo* DebugNoteFileInfo::copy() const
 {
-    return DebugNoteFileInfo::make(_srcDir, _srcName, _objPath, _objModTime, _objSubType, _libPath, _originLibPath);
+    DebugNoteFileInfo* result = (DebugNoteFileInfo*)malloc(_size);
+    memcpy(result, this, _size);
+    return result;
 }
 bool DebugNoteFileInfo::shouldbeUpdated(CString libPath) const
 {
@@ -462,6 +893,7 @@
 __attribute__((used))
 void DebugNoteFileInfo::dump() const
 {
+    fprintf(stdout, "size:        %u\n", this->_size);
     fprintf(stdout, "scrDir:      %s\n", srcDir().c_str());
     fprintf(stdout, "scrName:     %s\n", srcName().c_str());
     fprintf(stdout, "objPath:     %s\n", objPath().c_str());