[loong64][mips][regexp] Compact codegen for large character classes

Port commit 8bbb44e537

Bug: v8:11069
Change-Id: I66532e8410390bc220d7811e320bb44181b00d1f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3234303
Reviewed-by: Liu yu <liuyu@loongson.cn>
Commit-Queue: Zhao Jiazhong <zhaojiazhong-hf@loongson.cn>
Cr-Commit-Position: refs/heads/main@{#77468}

src/regexp/loong64/regexp-macro-assembler-loong64.cc

@@ -216,6 +216,21 @@ void RegExpMacroAssemblerLOONG64::CheckGreedyLoop(Label* on_equal) {
   BranchOrBacktrack(on_equal, eq, current_input_offset(), Operand(a0));
 }
 
+// Push (pop) caller-saved registers used by irregexp.
+void RegExpMacroAssemblerLOONG64::PushCallerSavedRegisters() {
+  RegList caller_saved_regexp =
+      current_input_offset().bit() | current_character().bit() |
+      end_of_input_address().bit() | backtrack_stackpointer().bit();
+  __ MultiPush(caller_saved_regexp);
+}
+
+void RegExpMacroAssemblerLOONG64::PopCallerSavedRegisters() {
+  RegList caller_saved_regexp =
+      current_input_offset().bit() | current_character().bit() |
+      end_of_input_address().bit() | backtrack_stackpointer().bit();
+  __ MultiPop(caller_saved_regexp);
+}
+
 void RegExpMacroAssemblerLOONG64::CheckNotBackReferenceIgnoreCase(
     int start_reg, bool read_backward, bool unicode, Label* on_no_match) {
   Label fallthrough;
@@ -297,11 +312,7 @@ void RegExpMacroAssemblerLOONG64::CheckNotBackReferenceIgnoreCase(
     }
   } else {
     DCHECK(mode_ == UC16);
-    // Put regexp engine registers on stack.
-    RegList regexp_registers_to_retain = current_input_offset().bit() |
-                                         current_character().bit() |
-                                         backtrack_stackpointer().bit();
-    __ MultiPush(regexp_registers_to_retain);
+    PushCallerSavedRegisters();
 
     int argument_count = 4;
     __ PrepareCallCFunction(argument_count, a2);
@@ -339,8 +350,7 @@ void RegExpMacroAssemblerLOONG64::CheckNotBackReferenceIgnoreCase(
       __ CallCFunction(function, argument_count);
     }
 
-    // Restore regexp engine registers.
-    __ MultiPop(regexp_registers_to_retain);
+    PopCallerSavedRegisters();
     __ li(code_pointer(), Operand(masm_->CodeObject()), CONSTANT_SIZE);
     __ Ld_d(end_of_input_address(), MemOperand(frame_pointer(), kInputEnd));
 
@@ -461,6 +471,41 @@ void RegExpMacroAssemblerLOONG64::CheckCharacterNotInRange(
   BranchOrBacktrack(on_not_in_range, hi, a0, Operand(to - from));
 }
 
+void RegExpMacroAssemblerLOONG64::CallIsCharacterInRangeArray(
+    const ZoneList<CharacterRange>* ranges) {
+  static const int kNumArguments = 3;
+  PushCallerSavedRegisters();
+  __ PrepareCallCFunction(kNumArguments, a0);
+
+  __ mov(a0, current_character());
+  __ li(a1, Operand(GetOrAddRangeArray(ranges)));
+  __ li(a2, Operand(ExternalReference::isolate_address(isolate())));
+
+  {
+    // We have a frame (set up in GetCode), but the assembler doesn't know.
+    FrameScope scope(masm_.get(), StackFrame::MANUAL);
+    __ CallCFunction(ExternalReference::re_is_character_in_range_array(),
+                     kNumArguments);
+  }
+
+  PopCallerSavedRegisters();
+  __ li(code_pointer(), Operand(masm_->CodeObject()));
+}
+
+bool RegExpMacroAssemblerLOONG64::CheckCharacterInRangeArray(
+    const ZoneList<CharacterRange>* ranges, Label* on_in_range) {
+  CallIsCharacterInRangeArray(ranges);
+  BranchOrBacktrack(on_in_range, ne, a0, Operand(zero_reg));
+  return true;
+}
+
+bool RegExpMacroAssemblerLOONG64::CheckCharacterNotInRangeArray(
+    const ZoneList<CharacterRange>* ranges, Label* on_not_in_range) {
+  CallIsCharacterInRangeArray(ranges);
+  BranchOrBacktrack(on_not_in_range, eq, a0, Operand(zero_reg));
+  return true;
+}
+
 void RegExpMacroAssemblerLOONG64::CheckBitInTable(Handle<ByteArray> table,
                                                   Label* on_bit_set) {
   __ li(a0, Operand(table));
@@ -878,12 +923,9 @@ Handle<HeapObject> RegExpMacroAssemblerLOONG64::GetCode(Handle<String> source) {
       // Put regexp engine registers on stack.
       StoreRegExpStackPointerToMemory(backtrack_stackpointer(), a1);
 
-      RegList regexp_registers_to_retain = current_input_offset().bit() |
-                                           current_character().bit() |
-                                           backtrack_stackpointer().bit();
-      __ MultiPush(regexp_registers_to_retain);
+      PushCallerSavedRegisters();
       CallCheckStackGuardState(a0);
-      __ MultiPop(regexp_registers_to_retain);
+      PopCallerSavedRegisters();
       // If returning non-zero, we should end execution with the given
       // result as return value.
       __ Branch(&return_v0, ne, a0, Operand(zero_reg));

src/regexp/loong64/regexp-macro-assembler-loong64.h

@@ -47,6 +47,10 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerLOONG64
                              Label* on_in_range) override;
   void CheckCharacterNotInRange(base::uc16 from, base::uc16 to,
                                 Label* on_not_in_range) override;
+  bool CheckCharacterInRangeArray(const ZoneList<CharacterRange>* ranges,
+                                  Label* on_in_range) override;
+  bool CheckCharacterNotInRangeArray(const ZoneList<CharacterRange>* ranges,
+                                     Label* on_not_in_range) override;
   void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set) override;
 
   // Checks whether the given offset from the current position is before
@@ -129,6 +133,9 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerLOONG64
   // Initial size of code buffer.
   static const int kRegExpCodeSize = 1024;
 
+  void PushCallerSavedRegisters();
+  void PopCallerSavedRegisters();
+
   // Check whether preemption has been requested.
   void CheckPreemption();
 
@@ -137,6 +144,7 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerLOONG64
 
   // Generate a call to CheckStackGuardState.
   void CallCheckStackGuardState(Register scratch);
+  void CallIsCharacterInRangeArray(const ZoneList<CharacterRange>* ranges);
 
   // The ebp-relative location of a regexp register.
   MemOperand register_location(int register_index);

src/regexp/mips/regexp-macro-assembler-mips.cc

@@ -223,6 +223,21 @@ void RegExpMacroAssemblerMIPS::CheckGreedyLoop(Label* on_equal) {
   BranchOrBacktrack(on_equal, eq, current_input_offset(), Operand(a0));
 }
 
+// Push (pop) caller-saved registers used by irregexp.
+void RegExpMacroAssemblerMIPS::PushCallerSavedRegisters() {
+  RegList caller_saved_regexp =
+      current_input_offset().bit() | current_character().bit() |
+      end_of_input_address().bit() | backtrack_stackpointer().bit();
+  __ MultiPush(caller_saved_regexp);
+}
+
+void RegExpMacroAssemblerMIPS::PopCallerSavedRegisters() {
+  RegList caller_saved_regexp =
+      current_input_offset().bit() | current_character().bit() |
+      end_of_input_address().bit() | backtrack_stackpointer().bit();
+  __ MultiPop(caller_saved_regexp);
+}
+
 void RegExpMacroAssemblerMIPS::CheckNotBackReferenceIgnoreCase(
     int start_reg, bool read_backward, bool unicode, Label* on_no_match) {
   Label fallthrough;
@@ -303,10 +318,7 @@ void RegExpMacroAssemblerMIPS::CheckNotBackReferenceIgnoreCase(
     }
   } else {
     DCHECK_EQ(UC16, mode_);
-    // Put regexp engine registers on stack.
-    RegList regexp_registers_to_retain = current_input_offset().bit() |
-        current_character().bit() | backtrack_stackpointer().bit();
-    __ MultiPush(regexp_registers_to_retain);
+    PushCallerSavedRegisters();
 
     int argument_count = 4;
     __ PrepareCallCFunction(argument_count, a2);
@@ -345,7 +357,7 @@ void RegExpMacroAssemblerMIPS::CheckNotBackReferenceIgnoreCase(
     }
 
     // Restore regexp engine registers.
-    __ MultiPop(regexp_registers_to_retain);
+    PopCallerSavedRegisters();
     __ li(code_pointer(), Operand(masm_->CodeObject()), CONSTANT_SIZE);
     __ lw(end_of_input_address(), MemOperand(frame_pointer(), kInputEnd));
 
@@ -476,6 +488,41 @@ void RegExpMacroAssemblerMIPS::CheckCharacterNotInRange(
   BranchOrBacktrack(on_not_in_range, hi, a0, Operand(to - from));
 }
 
+void RegExpMacroAssemblerMIPS::CallIsCharacterInRangeArray(
+    const ZoneList<CharacterRange>* ranges) {
+  static const int kNumArguments = 3;
+  PushCallerSavedRegisters();
+  __ PrepareCallCFunction(kNumArguments, a0);
+
+  __ mov(a0, current_character());
+  __ li(a1, Operand(GetOrAddRangeArray(ranges)));
+  __ li(a2, Operand(ExternalReference::isolate_address(isolate())));
+
+  {
+    // We have a frame (set up in GetCode), but the assembler doesn't know.
+    FrameScope scope(masm_.get(), StackFrame::MANUAL);
+    __ CallCFunction(ExternalReference::re_is_character_in_range_array(),
+                     kNumArguments);
+  }
+
+  PopCallerSavedRegisters();
+  __ li(code_pointer(), Operand(masm_->CodeObject()));
+}
+
+bool RegExpMacroAssemblerMIPS::CheckCharacterInRangeArray(
+    const ZoneList<CharacterRange>* ranges, Label* on_in_range) {
+  CallIsCharacterInRangeArray(ranges);
+  BranchOrBacktrack(on_in_range, ne, v0, Operand(zero_reg));
+  return true;
+}
+
+bool RegExpMacroAssemblerMIPS::CheckCharacterNotInRangeArray(
+    const ZoneList<CharacterRange>* ranges, Label* on_not_in_range) {
+  CallIsCharacterInRangeArray(ranges);
+  BranchOrBacktrack(on_not_in_range, eq, v0, Operand(zero_reg));
+  return true;
+}
+
 void RegExpMacroAssemblerMIPS::CheckBitInTable(
     Handle<ByteArray> table,
     Label* on_bit_set) {
@@ -886,12 +933,9 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
     if (check_preempt_label_.is_linked()) {
       SafeCallTarget(&check_preempt_label_);
       StoreRegExpStackPointerToMemory(backtrack_stackpointer(), a0);
-      // Put regexp engine registers on stack.
-      RegList regexp_registers_to_retain = current_input_offset().bit() |
-          current_character().bit() | backtrack_stackpointer().bit();
-      __ MultiPush(regexp_registers_to_retain);
+      PushCallerSavedRegisters();
       CallCheckStackGuardState(a0);
-      __ MultiPop(regexp_registers_to_retain);
+      PopCallerSavedRegisters();
       // If returning non-zero, we should end execution with the given
       // result as return value.
       __ Branch(&return_v0, ne, v0, Operand(zero_reg));

src/regexp/mips/regexp-macro-assembler-mips.h

@@ -47,6 +47,10 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerMIPS
                              Label* on_in_range) override;
   void CheckCharacterNotInRange(base::uc16 from, base::uc16 to,
                                 Label* on_not_in_range) override;
+  bool CheckCharacterInRangeArray(const ZoneList<CharacterRange>* ranges,
+                                  Label* on_in_range) override;
+  bool CheckCharacterNotInRangeArray(const ZoneList<CharacterRange>* ranges,
+                                     Label* on_not_in_range) override;
   void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set) override;
 
   // Checks whether the given offset from the current position is before
@@ -125,6 +129,9 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerMIPS
   // Initial size of code buffer.
   static const int kRegExpCodeSize = 1024;
 
+  void PushCallerSavedRegisters();
+  void PopCallerSavedRegisters();
+
   // Check whether preemption has been requested.
   void CheckPreemption();
 
@@ -133,6 +140,7 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerMIPS
 
   // Generate a call to CheckStackGuardState.
   void CallCheckStackGuardState(Register scratch);
+  void CallIsCharacterInRangeArray(const ZoneList<CharacterRange>* ranges);
 
   // The ebp-relative location of a regexp register.
   MemOperand register_location(int register_index);

src/regexp/mips64/regexp-macro-assembler-mips64.cc

@@ -261,6 +261,21 @@ void RegExpMacroAssemblerMIPS::CheckGreedyLoop(Label* on_equal) {
   BranchOrBacktrack(on_equal, eq, current_input_offset(), Operand(a0));
 }
 
+// Push (pop) caller-saved registers used by irregexp.
+void RegExpMacroAssemblerMIPS::PushCallerSavedRegisters() {
+  RegList caller_saved_regexp =
+      current_input_offset().bit() | current_character().bit() |
+      end_of_input_address().bit() | backtrack_stackpointer().bit();
+  __ MultiPush(caller_saved_regexp);
+}
+
+void RegExpMacroAssemblerMIPS::PopCallerSavedRegisters() {
+  RegList caller_saved_regexp =
+      current_input_offset().bit() | current_character().bit() |
+      end_of_input_address().bit() | backtrack_stackpointer().bit();
+  __ MultiPop(caller_saved_regexp);
+}
+
 void RegExpMacroAssemblerMIPS::CheckNotBackReferenceIgnoreCase(
     int start_reg, bool read_backward, bool unicode, Label* on_no_match) {
   Label fallthrough;
@@ -341,10 +356,7 @@ void RegExpMacroAssemblerMIPS::CheckNotBackReferenceIgnoreCase(
     }
   } else {
     DCHECK(mode_ == UC16);
-    // Put regexp engine registers on stack.
-    RegList regexp_registers_to_retain = current_input_offset().bit() |
-        current_character().bit() | backtrack_stackpointer().bit();
-    __ MultiPush(regexp_registers_to_retain);
+    PushCallerSavedRegisters();
 
     int argument_count = 4;
     __ PrepareCallCFunction(argument_count, a2);
@@ -382,8 +394,7 @@ void RegExpMacroAssemblerMIPS::CheckNotBackReferenceIgnoreCase(
       __ CallCFunction(function, argument_count);
     }
 
-    // Restore regexp engine registers.
-    __ MultiPop(regexp_registers_to_retain);
+    PopCallerSavedRegisters();
     __ li(code_pointer(), Operand(masm_->CodeObject()), CONSTANT_SIZE);
     __ Ld(end_of_input_address(), MemOperand(frame_pointer(), kInputEnd));
 
@@ -508,6 +519,41 @@ void RegExpMacroAssemblerMIPS::CheckCharacterNotInRange(
   BranchOrBacktrack(on_not_in_range, hi, a0, Operand(to - from));
 }
 
+void RegExpMacroAssemblerMIPS::CallIsCharacterInRangeArray(
+    const ZoneList<CharacterRange>* ranges) {
+  static const int kNumArguments = 3;
+  PushCallerSavedRegisters();
+  __ PrepareCallCFunction(kNumArguments, a0);
+
+  __ mov(a0, current_character());
+  __ li(a1, Operand(GetOrAddRangeArray(ranges)));
+  __ li(a2, Operand(ExternalReference::isolate_address(isolate())));
+
+  {
+    // We have a frame (set up in GetCode), but the assembler doesn't know.
+    FrameScope scope(masm_.get(), StackFrame::MANUAL);
+    __ CallCFunction(ExternalReference::re_is_character_in_range_array(),
+                     kNumArguments);
+  }
+
+  PopCallerSavedRegisters();
+  __ li(code_pointer(), Operand(masm_->CodeObject()));
+}
+
+bool RegExpMacroAssemblerMIPS::CheckCharacterInRangeArray(
+    const ZoneList<CharacterRange>* ranges, Label* on_in_range) {
+  CallIsCharacterInRangeArray(ranges);
+  BranchOrBacktrack(on_in_range, ne, v0, Operand(zero_reg));
+  return true;
+}
+
+bool RegExpMacroAssemblerMIPS::CheckCharacterNotInRangeArray(
+    const ZoneList<CharacterRange>* ranges, Label* on_not_in_range) {
+  CallIsCharacterInRangeArray(ranges);
+  BranchOrBacktrack(on_not_in_range, eq, v0, Operand(zero_reg));
+  return true;
+}
+
 void RegExpMacroAssemblerMIPS::CheckBitInTable(
     Handle<ByteArray> table,
     Label* on_bit_set) {
@@ -925,12 +971,9 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
       SafeCallTarget(&check_preempt_label_);
       StoreRegExpStackPointerToMemory(backtrack_stackpointer(), a0);
 
-      // Put regexp engine registers on stack.
-      RegList regexp_registers_to_retain = current_input_offset().bit() |
-          current_character().bit() | backtrack_stackpointer().bit();
-      __ MultiPush(regexp_registers_to_retain);
+      PushCallerSavedRegisters();
       CallCheckStackGuardState(a0);
-      __ MultiPop(regexp_registers_to_retain);
+      PopCallerSavedRegisters();
       // If returning non-zero, we should end execution with the given
       // result as return value.
       __ Branch(&return_v0, ne, v0, Operand(zero_reg));

src/regexp/mips64/regexp-macro-assembler-mips64.h

@@ -47,6 +47,10 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerMIPS
                              Label* on_in_range) override;
   void CheckCharacterNotInRange(base::uc16 from, base::uc16 to,
                                 Label* on_not_in_range) override;
+  bool CheckCharacterInRangeArray(const ZoneList<CharacterRange>* ranges,
+                                  Label* on_in_range) override;
+  bool CheckCharacterNotInRangeArray(const ZoneList<CharacterRange>* ranges,
+                                     Label* on_not_in_range) override;
   void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set) override;
 
   // Checks whether the given offset from the current position is before
@@ -130,6 +134,9 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerMIPS
   // Initial size of code buffer.
   static const int kRegExpCodeSize = 1024;
 
+  void PushCallerSavedRegisters();
+  void PopCallerSavedRegisters();
+
   // Check whether preemption has been requested.
   void CheckPreemption();
 
@@ -139,6 +146,7 @@ class V8_EXPORT_PRIVATE RegExpMacroAssemblerMIPS
 
   // Generate a call to CheckStackGuardState.
   void CallCheckStackGuardState(Register scratch);
+  void CallIsCharacterInRangeArray(const ZoneList<CharacterRange>* ranges);
 
   // The ebp-relative location of a regexp register.
   MemOperand register_location(int register_index);