Use std::shared_ptr for lexer capture buffers

include/asm/fstack.hpp

@@ -64,14 +64,16 @@ std::optional<std::string> fstk_FindFile(std::string const &path);
 bool yywrap();
 void fstk_RunInclude(std::string const &path, bool updateStateNow);
 void fstk_RunMacro(std::string const &macroName, std::shared_ptr<MacroArgs> macroArgs);
-void fstk_RunRept(uint32_t count, int32_t reptLineNo, char const *body, size_t size);
+void fstk_RunRept(
+    uint32_t count, int32_t reptLineNo, std::shared_ptr<char const[]> body, size_t size
+);
 void fstk_RunFor(
     std::string const &symName,
     int32_t start,
     int32_t stop,
     int32_t step,
     int32_t reptLineNo,
-    char const *body,
+    std::shared_ptr<char const[]> body,
     size_t size
 );
 void fstk_StopRept();

include/asm/lexer.hpp

@@ -52,21 +52,19 @@ struct BufferedContent {
 };
 
 struct MmappedContent {
-	char *ptr;
+	std::shared_ptr<char[]> ptr;
 	size_t size;
 	size_t offset = 0;
-	bool isReferenced = false; // If a macro in this file requires not unmapping it
 
-	MmappedContent(char *ptr_, size_t size_) : ptr(ptr_), size(size_) {}
-	~MmappedContent();
+	MmappedContent(std::shared_ptr<char[]> ptr_, size_t size_) : ptr(ptr_), size(size_) {}
 };
 
 struct ViewedContent {
-	char const *ptr;
+	std::shared_ptr<char const[]> ptr;
 	size_t size;
 	size_t offset = 0;
 
-	ViewedContent(char const *ptr_, size_t size_) : ptr(ptr_), size(size_) {}
+	ViewedContent(std::shared_ptr<char const[]> ptr_, size_t size_) : ptr(ptr_), size(size_) {}
 };
 
 struct LexerState {
@@ -80,9 +78,10 @@ struct LexerState {
 
 	std::deque<IfStackEntry> ifStack;
 
-	bool capturing;                // Whether the text being lexed should be captured
-	size_t captureSize;            // Amount of text captured
-	std::vector<char> *captureBuf; // Buffer to send the captured text to if non-null
+	bool capturing;     // Whether the text being lexed should be captured
+	size_t captureSize; // Amount of text captured
+	std::shared_ptr<std::vector<char>>
+	    captureBuf; // Buffer to send the captured text to if non-null
 
 	bool disableMacroArgs;
 	bool disableInterpolation;
@@ -96,7 +95,9 @@ struct LexerState {
 
 	void setAsCurrentState();
 	bool setFileAsNextState(std::string const &filePath, bool updateStateNow);
-	void setViewAsNextState(char const *name, char const *buf, size_t size, uint32_t lineNo_);
+	void setViewAsNextState(
+	    char const *name, std::shared_ptr<char const[]> ptr, size_t size, uint32_t lineNo_
+	);
 
 	void clear(uint32_t lineNo_);
 };
@@ -136,7 +137,7 @@ void lexer_DumpStringExpansions();
 
 struct Capture {
 	uint32_t lineNo;
-	char const *body;
+	std::shared_ptr<char const[]> body;
 	size_t size;
 };
 

include/asm/symbol.hpp

@@ -10,6 +10,7 @@
 #include <string>
 #include <string_view>
 #include <time.h>
+#include <utility>
 #include <variant>
 
 #include "asm/section.hpp"
@@ -36,10 +37,10 @@ struct Symbol {
 	uint32_t fileLine;                  // Line where the symbol was defined
 
 	std::variant<
-	    int32_t,                     // If isNumeric()
-	    int32_t (*)(),               // If isNumeric() and has a callback
-	    std::string_view,            // For SYM_MACRO
-	    std::shared_ptr<std::string> // For SYM_EQUS
+	    int32_t,                                          // If isNumeric()
+	    int32_t (*)(),                                    // If isNumeric() and has a callback
+	    std::pair<std::shared_ptr<char const[]>, size_t>, // For SYM_MACRO
+	    std::shared_ptr<std::string>                      // For SYM_EQUS
 	    >
 	    data;
 
@@ -61,7 +62,7 @@ struct Symbol {
 
 	int32_t getValue() const;
 	int32_t getOutputValue() const;
-	std::string_view getMacro() const;
+	std::pair<std::shared_ptr<char const[]>, size_t> getMacro() const;
 	std::shared_ptr<std::string> getEqus() const;
 	uint32_t getConstantValue() const;
 };
@@ -88,7 +89,9 @@ Symbol *sym_FindScopedSymbol(std::string const &symName);
 // Find a scoped symbol by name; do not return `@` or `_NARG` when they have no value
 Symbol *sym_FindScopedValidSymbol(std::string const &symName);
 Symbol const *sym_GetPC();
-Symbol *sym_AddMacro(std::string const &symName, int32_t defLineNo, char const *body, size_t size);
+Symbol *sym_AddMacro(
+    std::string const &symName, int32_t defLineNo, std::shared_ptr<char const[]> body, size_t size
+);
 Symbol *sym_Ref(std::string const &symName);
 Symbol *sym_AddString(std::string const &symName, std::shared_ptr<std::string> value);
 Symbol *sym_RedefString(std::string const &symName, std::shared_ptr<std::string> value);

src/asm/fstack.cpp

@@ -281,11 +281,13 @@ static void newMacroContext(Symbol const &macro, std::shared_ptr<MacroArgs> macr
 	    .macroArgs = macroArgs,
 	});
 
-	std::string_view view = macro.getMacro();
-	context.lexerState.setViewAsNextState("MACRO", view.data(), view.size(), macro.fileLine);
+	auto [body, size] = macro.getMacro();
+	context.lexerState.setViewAsNextState("MACRO", body, size, macro.fileLine);
 }
 
-static Context &newReptContext(int32_t reptLineNo, char const *body, size_t size, uint32_t count) {
+static Context &newReptContext(
+    int32_t reptLineNo, std::shared_ptr<char const[]> body, size_t size, uint32_t count
+) {
 	checkRecursionDepth();
 
 	Context &oldContext = contextStack.top();
@@ -347,7 +349,9 @@ void fstk_RunMacro(std::string const &macroName, std::shared_ptr<MacroArgs> macr
 	newMacroContext(*macro, macroArgs);
 }
 
-void fstk_RunRept(uint32_t count, int32_t reptLineNo, char const *body, size_t size) {
+void fstk_RunRept(
+    uint32_t count, int32_t reptLineNo, std::shared_ptr<char const[]> body, size_t size
+) {
 	if (count == 0)
 		return;
 
@@ -360,7 +364,7 @@ void fstk_RunFor(
     int32_t stop,
     int32_t step,
     int32_t reptLineNo,
-    char const *body,
+    std::shared_ptr<char const[]> body,
     size_t size
 ) {
 	if (Symbol *sym = sym_AddVar(symName, start); sym->type != SYM_VAR)

src/asm/lexer.cpp

@@ -69,9 +69,11 @@ static void mapFile(void *&mappingAddr, int fd, std::string const &path, size_t)
 	}
 }
 
-static int munmap(void *mappingAddr, size_t) {
-	return UnmapViewOfFile(mappingAddr) == 0 ? -1 : 0;
-}
+struct MunmapDeleter {
+	MunmapDeleter(size_t) {}
+
+	void operator()(char *mappingAddr) { UnmapViewOfFile(mappingAddr); }
+};
 
 #else // defined(_MSC_VER) || defined(__MINGW32__)
 	#include <sys/mman.h>
@@ -88,6 +90,14 @@ static void mapFile(void *&mappingAddr, int fd, std::string const &path, size_t
 	}
 }
 
+struct MunmapDeleter {
+	size_t mappingSize;
+
+	MunmapDeleter(size_t mappingSize_) : mappingSize(mappingSize_) {}
+
+	void operator()(char *mappingAddr) { munmap(mappingAddr, mappingSize); }
+};
+
 #endif // !( defined(_MSC_VER) || defined(__MINGW32__) )
 
 using namespace std::literals;
@@ -415,7 +425,12 @@ bool LexerState::setFileAsNextState(std::string const &filePath, bool updateStat
 
 			if (mappingAddr != MAP_FAILED) {
 				close(fd);
-				content.emplace<MmappedContent>((char *)mappingAddr, (size_t)statBuf.st_size);
+				content.emplace<MmappedContent>(
+				    std::shared_ptr<char[]>(
+				        (char *)mappingAddr, MunmapDeleter((size_t)statBuf.st_size)
+				    ),
+				    (size_t)statBuf.st_size
+				);
 				if (verbose)
 					printf("File \"%s\" is mmap()ped\n", path.c_str());
 				isMmapped = true;
@@ -446,10 +461,10 @@ bool LexerState::setFileAsNextState(std::string const &filePath, bool updateStat
 }
 
 void LexerState::setViewAsNextState(
-    char const *name, char const *buf, size_t size, uint32_t lineNo_
+    char const *name, std::shared_ptr<char const[]> ptr, size_t size, uint32_t lineNo_
 ) {
 	path = name; // Used to report read errors in `peekInternal`
-	content.emplace<ViewedContent>(buf, size);
+	content.emplace<ViewedContent>(ptr, size);
 	clear(lineNo_);
 	lexerStateEOL = this;
 }
@@ -482,12 +497,6 @@ BufferedContent::~BufferedContent() {
 	close(fd);
 }
 
-MmappedContent::~MmappedContent() {
-	// FIXME: This never unmaps a referenced file!
-	if (!isReferenced)
-		munmap(ptr, size);
-}
-
 void lexer_SetMode(LexerMode mode) {
 	lexerState->mode = mode;
 }
@@ -2174,17 +2183,14 @@ static Capture startCapture() {
 	Capture capture = {.lineNo = lexer_GetLineNo(), .body = nullptr, .size = 0};
 	if (auto *mmap = std::get_if<MmappedContent>(&lexerState->content);
 	    mmap && lexerState->expansions.empty()) {
-		capture.body = &mmap->ptr[mmap->offset];
+		capture.body = std::shared_ptr<char const[]>(mmap->ptr, &mmap->ptr[mmap->offset]);
 	} else if (auto *view = std::get_if<ViewedContent>(&lexerState->content);
 	           view && lexerState->expansions.empty()) {
-		capture.body = &view->ptr[view->offset];
+		capture.body = std::shared_ptr<char const[]>(view->ptr, &view->ptr[view->offset]);
 	} else {
 		// `capture.body == nullptr`; indicates to retrieve the capture buffer when done capturing
 		assert(lexerState->captureBuf == nullptr);
-		// FIXME: This leaks the captured text!
-		lexerState->captureBuf = new (std::nothrow) std::vector<char>();
-		if (!lexerState->captureBuf)
-			fatalerror("Failed to allocate capture buffer: %s\n", strerror(errno));
+		lexerState->captureBuf = std::make_shared<std::vector<char>>();
 	}
 	return capture;
 }
@@ -2193,7 +2199,8 @@ static void endCapture(Capture &capture) {
 	// This being `nullptr` means we're capturing from the capture buffer, which is reallocated
 	// during the whole capture process, and so MUST be retrieved at the end
 	if (!capture.body)
-		capture.body = lexerState->captureBuf->data();
+		capture.body =
+		    std::shared_ptr<char const[]>(lexerState->captureBuf, lexerState->captureBuf->data());
 	capture.size = lexerState->captureSize;
 
 	// ENDR/ENDM or EOF puts us past the start of the line
@@ -2262,10 +2269,6 @@ Capture lexer_CaptureMacro() {
 
 	Defer reenableExpansions = scopedDisableExpansions();
 
-	// If the file is `mmap`ed, we need not to unmap it to keep access to the macro
-	if (auto *mmap = std::get_if<MmappedContent>(&lexerState->content); mmap)
-		mmap->isReferenced = true;
-
 	int c = EOF;
 
 	for (;;) {

src/asm/symbol.cpp

@@ -74,9 +74,9 @@ int32_t Symbol::getOutputValue() const {
 	}
 }
 
-std::string_view Symbol::getMacro() const {
-	assert(std::holds_alternative<std::string_view>(data));
-	return std::get<std::string_view>(data);
+std::pair<std::shared_ptr<char const []>, size_t> Symbol::getMacro() const {
+	assert((std::holds_alternative<std::pair<std::shared_ptr<char const[]>, size_t>>(data)));
+	return std::get<std::pair<std::shared_ptr<char const[]>, size_t>>(data);
 }
 
 std::shared_ptr<std::string> Symbol::getEqus() const {
@@ -489,14 +489,16 @@ void sym_Export(std::string const &symName) {
 }
 
 // Add a macro definition
-Symbol *sym_AddMacro(std::string const &symName, int32_t defLineNo, char const *body, size_t size) {
+Symbol *sym_AddMacro(
+    std::string const &symName, int32_t defLineNo, std::shared_ptr<char const[]> body, size_t size
+) {
 	Symbol *sym = createNonrelocSymbol(symName, false);
 
 	if (!sym)
 		return nullptr;
 
 	sym->type = SYM_MACRO;
-	sym->data = std::string_view(body, size);
+	sym->data = std::make_pair(body, size);
 
 	sym->src = fstk_GetFileStack();
 	// The symbol is created at the line after the `ENDM`,