1 // Copyright 2021 the V8 project authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4 
5 #ifndef V8_REGEXP_RISCV64_REGEXP_MACRO_ASSEMBLER_RISCV64_H_
6 #define V8_REGEXP_RISCV64_REGEXP_MACRO_ASSEMBLER_RISCV64_H_
7 
8 #include "src/base/strings.h"
9 #include "src/codegen/macro-assembler.h"
10 #include "src/codegen/riscv64/assembler-riscv64.h"
11 #include "src/regexp/regexp-macro-assembler.h"
12 
13 namespace v8 {
14 namespace internal {
15 
16 class V8_EXPORT_PRIVATE RegExpMacroAssemblerRISCV
17     : public NativeRegExpMacroAssembler {
18  public:
19   RegExpMacroAssemblerRISCV(Isolate* isolate, Zone* zone, Mode mode,
20                             int registers_to_save);
21   ~RegExpMacroAssemblerRISCV() override;
22   int stack_limit_slack() override;
23   void AdvanceCurrentPosition(int by) override;
24   void AdvanceRegister(int reg, int by) override;
25   void Backtrack() override;
26   void Bind(Label* label) override;
27   void CheckAtStart(int cp_offset, Label* on_at_start) override;
28   void CheckCharacter(uint32_t c, Label* on_equal) override;
29   void CheckCharacterAfterAnd(uint32_t c, uint32_t mask,
30                               Label* on_equal) override;
31   void CheckCharacterGT(base::uc16 limit, Label* on_greater) override;
32   void CheckCharacterLT(base::uc16 limit, Label* on_less) override;
33   // A "greedy loop" is a loop that is both greedy and with a simple
34   // body. It has a particularly simple implementation.
35   void CheckGreedyLoop(Label* on_tos_equals_current_position) override;
36   void CheckNotAtStart(int cp_offset, Label* on_not_at_start) override;
37   void CheckNotBackReference(int start_reg, bool read_backward,
38                              Label* on_no_match) override;
39   void CheckNotBackReferenceIgnoreCase(int start_reg, bool read_backward,
40                                        bool unicode,
41                                        Label* on_no_match) override;
42   void CheckNotCharacter(uint32_t c, Label* on_not_equal) override;
43   void CheckNotCharacterAfterAnd(uint32_t c, uint32_t mask,
44                                  Label* on_not_equal) override;
45   void CheckNotCharacterAfterMinusAnd(base::uc16 c, base::uc16 minus,
46                                       base::uc16 mask,
47                                       Label* on_not_equal) override;
48   void CheckCharacterInRange(base::uc16 from, base::uc16 to,
49                              Label* on_in_range) override;
50   void CheckCharacterNotInRange(base::uc16 from, base::uc16 to,
51                                 Label* on_not_in_range) override;
52   bool CheckCharacterInRangeArray(const ZoneList<CharacterRange>* ranges,
53                                   Label* on_in_range) override;
54   bool CheckCharacterNotInRangeArray(const ZoneList<CharacterRange>* ranges,
55                                      Label* on_not_in_range) override;
56   void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set) override;
57 
58   // Checks whether the given offset from the current position is before
59   // the end of the string.
60   void CheckPosition(int cp_offset, Label* on_outside_input) override;
61   bool CheckSpecialCharacterClass(StandardCharacterSet type,
62                                   Label* on_no_match) override;
63   void Fail() override;
64   Handle<HeapObject> GetCode(Handle<String> source) override;
65   void GoTo(Label* label) override;
66   void IfRegisterGE(int reg, int comparand, Label* if_ge) override;
67   void IfRegisterLT(int reg, int comparand, Label* if_lt) override;
68   void IfRegisterEqPos(int reg, Label* if_eq) override;
69   IrregexpImplementation Implementation() override;
70   void LoadCurrentCharacterUnchecked(int cp_offset,
71                                      int character_count) override;
72   void PopCurrentPosition() override;
73   void PopRegister(int register_index) override;
74   void PushBacktrack(Label* label) override;
75   void PushCurrentPosition() override;
76   void PushRegister(int register_index,
77                     StackCheckFlag check_stack_limit) override;
78   void ReadCurrentPositionFromRegister(int reg) override;
79   void ReadStackPointerFromRegister(int reg) override;
80   void SetCurrentPositionFromEnd(int by) override;
81   void SetRegister(int register_index, int to) override;
82   bool Succeed() override;
83   void WriteCurrentPositionToRegister(int reg, int cp_offset) override;
84   void ClearRegisters(int reg_from, int reg_to) override;
85   void WriteStackPointerToRegister(int reg) override;
86 #ifdef RISCV_HAS_NO_UNALIGNED
87   bool CanReadUnaligned() const override;
88 #endif
89   // Called from RegExp if the stack-guard is triggered.
90   // If the code object is relocated, the return address is fixed before
91   // returning.
92   // {raw_code} is an Address because this is called via ExternalReference.
93   static int64_t CheckStackGuardState(Address* return_address, Address raw_code,
94                                       Address re_frame);
95 
96   void print_regexp_frame_constants();
97 
98  private:
99   // Offsets from frame_pointer() of function parameters and stored registers.
100   static const int kFramePointer = 0;
101 
102   // Above the frame pointer - Stored registers and stack passed parameters.
103   // Registers s1 to s8, fp, and ra.
104   static const int kStoredRegisters = kFramePointer;
105   // Return address (stored from link register, read into pc on return).
106 
107   // This 9 is 8 s-regs (s1..s8) plus fp.
108   static const int kNumCalleeRegsToRetain = 9;
109   static const int kReturnAddress =
110       kStoredRegisters + kNumCalleeRegsToRetain * kSystemPointerSize;
111 
112   // Stack frame header.
113   static const int kStackFrameHeader = kReturnAddress;
114   // Below the frame pointer.
115   // Register parameters stored by setup code.
116   static const int kIsolate = kFramePointer - kSystemPointerSize;
117   static const int kDirectCall = kIsolate - kSystemPointerSize;
118   static const int kNumOutputRegisters = kDirectCall - kSystemPointerSize;
119   static const int kRegisterOutput = kNumOutputRegisters - kSystemPointerSize;
120   static const int kInputEnd = kRegisterOutput - kSystemPointerSize;
121   static const int kInputStart = kInputEnd - kSystemPointerSize;
122   static const int kStartIndex = kInputStart - kSystemPointerSize;
123   static const int kInputString = kStartIndex - kSystemPointerSize;
124   // When adding local variables remember to push space for them in
125   // the frame in GetCode.
126   static const int kSuccessfulCaptures = kInputString - kSystemPointerSize;
127   static const int kStringStartMinusOne =
128       kSuccessfulCaptures - kSystemPointerSize;
129   static const int kBacktrackCount = kStringStartMinusOne - kSystemPointerSize;
130   // Stores the initial value of the regexp stack pointer in a
131   // position-independent representation (in case the regexp stack grows and
132   // thus moves).
133   static const int kRegExpStackBasePointer =
134       kBacktrackCount - kSystemPointerSize;
135   static constexpr int kNumberOfStackLocals = 4;
136   // First register address. Following registers are below it on the stack.
137   static const int kRegisterZero = kRegExpStackBasePointer - kSystemPointerSize;
138 
139   // Initial size of code buffer.
140   static const int kRegExpCodeSize = 1024;
141 
142   void PushCallerSavedRegisters();
143   void PopCallerSavedRegisters();
144 
145   // Check whether preemption has been requested.
146   void CheckPreemption();
147 
148   // Check whether we are exceeding the stack limit on the backtrack stack.
149   void CheckStackLimit();
150 
151   void CallCheckStackGuardState(Register scratch);
152   void CallIsCharacterInRangeArray(const ZoneList<CharacterRange>* ranges);
153 
154   // The ebp-relative location of a regexp register.
155   MemOperand register_location(int register_index);
156 
157   // Register holding the current input position as negative offset from
158   // the end of the string.
current_input_offset()159   static constexpr Register current_input_offset() { return s1; }
160 
161   // The register containing the current character after LoadCurrentCharacter.
current_character()162   static constexpr Register current_character() { return s2; }
163 
164   // Register holding address of the end of the input string.
end_of_input_address()165   static constexpr Register end_of_input_address() { return t2; }
166 
167   // Register holding the frame address. Local variables, parameters and
168   // regexp registers are addressed relative to this.
frame_pointer()169   static constexpr Register frame_pointer() { return fp; }
170 
171   // The register containing the backtrack stack top. Provides a meaningful
172   // name to the register.
backtrack_stackpointer()173   static constexpr Register backtrack_stackpointer() { return t0; }
174 
175   // Register holding pointer to the current code object.
code_pointer()176   static constexpr Register code_pointer() { return s4; }
177 
178   // Byte size of chars in the string to match (decided by the Mode argument).
char_size() const179   inline int char_size() const { return static_cast<int>(mode_); }
180 
181   // Equivalent to a conditional branch to the label, unless the label
182   // is nullptr, in which case it is a conditional Backtrack.
183   void BranchOrBacktrack(Label* to, Condition condition, Register rs,
184                          const Operand& rt);
185 
186   // Call and return internally in the generated code in a way that
187   // is GC-safe (i.e., doesn't leave absolute code addresses on the stack)
188   inline void SafeCall(Label* to, Condition cond, Register rs,
189                        const Operand& rt);
190   inline void SafeReturn();
191   inline void SafeCallTarget(Label* name);
192 
193   // Pushes the value of a register on the backtrack stack. Decrements the
194   // stack pointer by a word size and stores the register's value there.
195   inline void Push(Register source);
196 
197   // Pops a value from the backtrack stack. Reads the word at the stack pointer
198   // and increments it by a word size.
199   inline void Pop(Register target);
200 
201   void LoadRegExpStackPointerFromMemory(Register dst);
202   void StoreRegExpStackPointerToMemory(Register src, Register scratch);
203   void PushRegExpBasePointer(Register scratch1, Register scratch2);
204   void PopRegExpBasePointer(Register scratch1, Register scratch2);
205 
isolate() const206   Isolate* isolate() const { return masm_->isolate(); }
207 
208   const std::unique_ptr<MacroAssembler> masm_;
209   const NoRootArrayScope no_root_array_scope_;
210 
211   // Which mode to generate code for (Latin1 or UC16).
212   const Mode mode_;
213 
214   // One greater than maximal register index actually used.
215   int num_registers_;
216 
217   // Number of registers to output at the end (the saved registers
218   // are always 0..num_saved_registers_-1).
219   const int num_saved_registers_;
220 
221   // Labels used internally.
222   Label entry_label_;
223   Label start_label_;
224   Label success_label_;
225   Label backtrack_label_;
226   Label exit_label_;
227   Label check_preempt_label_;
228   Label stack_overflow_label_;
229   Label internal_failure_label_;
230   Label fallback_label_;
231 };
232 
233 }  // namespace internal
234 }  // namespace v8
235 
236 #endif  // V8_REGEXP_RISCV64_REGEXP_MACRO_ASSEMBLER_RISCV64_H_
237