Merge pull request #5339 from rocallahan/fast-rtlil-parser

Rewrite the RTLIL parser for efficiency

kernel/rtlil.cc

@@ -264,10 +264,10 @@ std::string& Const::get_str() {
 	return *get_if_str();
 }
 
-RTLIL::Const::Const(const std::string &str)
+RTLIL::Const::Const(std::string str)
 {
 	flags = RTLIL::CONST_FLAG_STRING;
-	new ((void*)&str_) std::string(str);
+	new ((void*)&str_) std::string(std::move(str));
 	tag = backing_tag::string;
 }
 
@@ -3050,7 +3050,7 @@ void RTLIL::Module::fixup_ports()
 RTLIL::Wire *RTLIL::Module::addWire(RTLIL::IdString name, int width)
 {
 	RTLIL::Wire *wire = new RTLIL::Wire;
-	wire->name = name;
+	wire->name = std::move(name);
 	wire->width = width;
 	add(wire);
 	return wire;
@@ -3058,7 +3058,7 @@ RTLIL::Wire *RTLIL::Module::addWire(RTLIL::IdString name, int width)
 
 RTLIL::Wire *RTLIL::Module::addWire(RTLIL::IdString name, const RTLIL::Wire *other)
 {
-	RTLIL::Wire *wire = addWire(name);
+	RTLIL::Wire *wire = addWire(std::move(name));
 	wire->width = other->width;
 	wire->start_offset = other->start_offset;
 	wire->port_id = other->port_id;
@@ -3073,7 +3073,7 @@ RTLIL::Wire *RTLIL::Module::addWire(RTLIL::IdString name, const RTLIL::Wire *oth
 RTLIL::Cell *RTLIL::Module::addCell(RTLIL::IdString name, RTLIL::IdString type)
 {
 	RTLIL::Cell *cell = new RTLIL::Cell;
-	cell->name = name;
+	cell->name = std::move(name);
 	cell->type = type;
 	add(cell);
 	return cell;
@@ -3081,7 +3081,7 @@ RTLIL::Cell *RTLIL::Module::addCell(RTLIL::IdString name, RTLIL::IdString type)
 
 RTLIL::Cell *RTLIL::Module::addCell(RTLIL::IdString name, const RTLIL::Cell *other)
 {
-	RTLIL::Cell *cell = addCell(name, other->type);
+	RTLIL::Cell *cell = addCell(std::move(name), other->type);
 	cell->connections_ = other->connections_;
 	cell->parameters = other->parameters;
 	cell->attributes = other->attributes;
@@ -3091,7 +3091,7 @@ RTLIL::Cell *RTLIL::Module::addCell(RTLIL::IdString name, const RTLIL::Cell *oth
 RTLIL::Memory *RTLIL::Module::addMemory(RTLIL::IdString name, const RTLIL::Memory *other)
 {
 	RTLIL::Memory *mem = new RTLIL::Memory;
-	mem->name = name;
+	mem->name = std::move(name);
 	mem->width = other->width;
 	mem->start_offset = other->start_offset;
 	mem->size = other->size;
@@ -3103,7 +3103,7 @@ RTLIL::Memory *RTLIL::Module::addMemory(RTLIL::IdString name, const RTLIL::Memor
 RTLIL::Process *RTLIL::Module::addProcess(RTLIL::IdString name)
 {
 	RTLIL::Process *proc = new RTLIL::Process;
-	proc->name = name;
+	proc->name = std::move(name);
 	add(proc);
 	return proc;
 }
@@ -3111,7 +3111,7 @@ RTLIL::Process *RTLIL::Module::addProcess(RTLIL::IdString name)
 RTLIL::Process *RTLIL::Module::addProcess(RTLIL::IdString name, const RTLIL::Process *other)
 {
 	RTLIL::Process *proc = other->clone();
-	proc->name = name;
+	proc->name = std::move(name);
 	add(proc);
 	return proc;
 }

kernel/rtlil.h

@@ -209,10 +209,15 @@ struct RTLIL::IdString
 	}
 
 	static int get_reference(const char *p)
+	{
+		return get_reference(std::string_view(p));
+	}
+
+	static int get_reference(std::string_view p)
 	{
 		log_assert(destruct_guard_ok);
 
-		auto it = global_id_index_.find((char*)p);
+		auto it = global_id_index_.find(p);
 		if (it != global_id_index_.end()) {
 	#ifndef YOSYS_NO_IDS_REFCNT
 			global_refcount_storage_.at(it->second)++;
@@ -226,14 +231,13 @@ struct RTLIL::IdString
 
 		ensure_prepopulated();
 
-		if (!p[0])
+		if (p.empty())
 			return 0;
 
 		log_assert(p[0] == '$' || p[0] == '\\');
-		log_assert(p[1] != 0);
-		for (const char *c = p; *c; c++)
-			if ((unsigned)*c <= (unsigned)' ')
-				log_error("Found control character or space (0x%02x) in string '%s' which is not allowed in RTLIL identifiers\n", *c, p);
+		for (char ch : p)
+			if ((unsigned)ch <= (unsigned)' ')
+				log_error("Found control character or space (0x%02x) in string '%s' which is not allowed in RTLIL identifiers\n", ch, std::string(p).c_str());
 
 	#ifndef YOSYS_NO_IDS_REFCNT
 		if (global_free_idx_list_.empty()) {
@@ -245,8 +249,11 @@ struct RTLIL::IdString
 
 		int idx = global_free_idx_list_.back();
 		global_free_idx_list_.pop_back();
-		global_id_storage_.at(idx) = strdup(p);
-		global_id_index_[global_id_storage_.at(idx)] = idx;
+		char* buf = static_cast<char*>(malloc(p.size() + 1));
+		memcpy(buf, p.data(), p.size());
+		buf[p.size()] = 0;
+		global_id_storage_.at(idx) = buf;
+		global_id_index_.insert(it, {std::string_view(buf, p.size()), idx});
 		global_refcount_storage_.at(idx)++;
 	#else
 		int idx = global_id_storage_.size();
@@ -255,7 +262,7 @@ struct RTLIL::IdString
 	#endif
 
 		if (yosys_xtrace) {
-			log("#X# New IdString '%s' with index %d.\n", p, idx);
+			log("#X# New IdString '%s' with index %d.\n", global_id_storage_.at(idx), idx);
 			log_backtrace("-X- ", yosys_xtrace-1);
 		}
 
@@ -322,7 +329,8 @@ struct RTLIL::IdString
 	inline IdString(const char *str) : index_(get_reference(str)) { }
 	inline IdString(const IdString &str) : index_(get_reference(str.index_)) { }
 	inline IdString(IdString &&str) : index_(str.index_) { str.index_ = 0; }
-	inline IdString(const std::string &str) : index_(get_reference(str.c_str())) { }
+	inline IdString(const std::string &str) : index_(get_reference(std::string_view(str))) { }
+	inline IdString(std::string_view str) : index_(get_reference(str)) { }
 	inline IdString(StaticId id) : index_(static_cast<short>(id)) {}
 	inline ~IdString() { put_reference(index_); }
 
@@ -331,6 +339,12 @@ struct RTLIL::IdString
 		index_ = get_reference(rhs.index_);
 	}
 
+	inline void operator=(IdString &&rhs) {
+		put_reference(index_);
+		index_ = rhs.index_;
+		rhs.index_ = 0;
+	}
+
 	inline void operator=(const char *rhs) {
 		IdString id(rhs);
 		*this = id;
@@ -859,7 +873,7 @@ private:
 
 public:
 	Const() : flags(RTLIL::CONST_FLAG_NONE), tag(backing_tag::bits), bits_(std::vector<RTLIL::State>()) {}
-	Const(const std::string &str);
+	Const(std::string str);
 	Const(long long val); // default width is 32
 	Const(long long val, int width);
 	Const(RTLIL::State bit, int width = 1);