PHPOpcode kernel implementation

Implementation of catalogue in the PHPOpcode kernel

1. Opcode Introduction 2. Opcode3. opcode translation execution in PHP (instant interpretation execution)

1. Opcode introduction

Opcode is a part of a computer instruction. it is used to specify the operation to be executed. the instruction format and specification are specified by the instruction specification of the processor. In addition to the instruction itself, there are usually operations required by the instruction. some commands may not need explicit operations. These operands may be values in registers, values in stacks, memory values, or values in I/O ports.

Opcode also has another title: byte codes ). For example, Java virtual machine (JVM) and. NET Common Intermeditate Language

The opcode in PHP belongs to the previous section. PHP is built on Zend VM (Zend VM. PHP opcode is the command in Zend virtual machine (based on Zend intermediate code)

Relevant Link:

http://www.luocong.com/learningopcode/doc/1._%E4%BB%80%E4%B9%88%E6%98%AFOpCode%EF%BC%9F.htm

2. Opcode in PHP

0x1: Data structure

In PHP implementation, opcode is represented by the following struct \ php-5.6.17 \ Zend \ zend_compile.h

Struct _ zend_op {opcode_handler_t handler; // The processing function znode_op op1 called when the opcode is executed; // The operand znode_op op2 operated by opcode; // The operand znode_op result operated by opcode; ulong extended_value; uint lineno; zend_uchar opcode; // opcode zend_uchar op1_type; zend_uchar op2_type; zend_uchar result_type ;};

Similar to CPU commands, there is an opcode field indicating commands and the operands operated by this opcode. PHP is not as underlying as assembly, when the script is actually executed, more information may be required. the extended_value field stores this type of information, and the result field stores the result after the command is executed.

For example, the following code is a function compiled when the compiler encounters a print statement.

Php-5.6.17 \ Zend \ zend_compile.c

Void merge (znode * result, const znode * arg TSRMLS_DC)/* {*/{// create a new zend_op * opline = get_next_op (CG (active_op_array) TSRMLS_CC ); // Set the returned value type of the newly created zend_op to a temporary variable (IS_TMP_VAR), because the memory in print is only for temporary output and does not need to save opline-> result_type = IS_TMP_VAR; // Apply for space opline for temporary variables-> result. var = get_temporary_variable (CG (active_op_array); // specify opcode as ZEND_PRINT opline-> opcode = ZEND_PRINT; // assign the passed parameter to SET_NODE (opline-> op1, arg), SET_UNUSED (opline-> op2), and GET_NODE (result, opline-> result );}

0x2: opcode type: zend_op-> zend_uchar opcode

Compared with the concept of assembly language, each opcode corresponds to a type, indicating the "operation command" of the opcpvdf. the opcode type is zend_uchar, and zend_uchar is actually unsigned char, the integer value stored in this field is the op number, which is used to distinguish different op types. the opcode values are defined as macros/Zend/zend_vm_opcodes.h.

#define ZEND_NOP                               0#define ZEND_ADD                               1#define ZEND_SUB                               2#define ZEND_MUL                               3#define ZEND_DIV                               4#define ZEND_MOD                               5#define ZEND_SL                                6#define ZEND_SR                                7#define ZEND_CONCAT                            8#define ZEND_BW_OR                             9#define ZEND_BW_AND                           10#define ZEND_BW_XOR                           11#define ZEND_BW_NOT                           12#define ZEND_BOOL_NOT                         13#define ZEND_BOOL_XOR                         14#define ZEND_IS_IDENTICAL                     15#define ZEND_IS_NOT_IDENTICAL                 16#define ZEND_IS_EQUAL                         17#define ZEND_IS_NOT_EQUAL                     18#define ZEND_IS_SMALLER                       19#define ZEND_IS_SMALLER_OR_EQUAL              20#define ZEND_CAST                             21#define ZEND_QM_ASSIGN                        22#define ZEND_ASSIGN_ADD                       23#define ZEND_ASSIGN_SUB                       24#define ZEND_ASSIGN_MUL                       25#define ZEND_ASSIGN_DIV                       26#define ZEND_ASSIGN_MOD                       27#define ZEND_ASSIGN_SL                        28#define ZEND_ASSIGN_SR                        29#define ZEND_ASSIGN_CONCAT                    30#define ZEND_ASSIGN_BW_OR                     31#define ZEND_ASSIGN_BW_AND                    32#define ZEND_ASSIGN_BW_XOR                    33#define ZEND_PRE_INC                          34#define ZEND_PRE_DEC                          35#define ZEND_POST_INC                         36#define ZEND_POST_DEC                         37#define ZEND_ASSIGN                           38#define ZEND_ASSIGN_REF                       39#define ZEND_ECHO                             40#define ZEND_PRINT                            41#define ZEND_JMP                              42#define ZEND_JMPZ                             43#define ZEND_JMPNZ                            44#define ZEND_JMPZNZ                           45#define ZEND_JMPZ_EX                          46#define ZEND_JMPNZ_EX                         47#define ZEND_CASE                             48#define ZEND_SWITCH_FREE                      49#define ZEND_BRK                              50#define ZEND_CONT                             51#define ZEND_BOOL                             52#define ZEND_INIT_STRING                      53#define ZEND_ADD_CHAR                         54#define ZEND_ADD_STRING                       55#define ZEND_ADD_VAR                          56#define ZEND_BEGIN_SILENCE                    57#define ZEND_END_SILENCE                      58#define ZEND_INIT_FCALL_BY_NAME               59#define ZEND_DO_FCALL                         60#define ZEND_DO_FCALL_BY_NAME                 61#define ZEND_RETURN                           62#define ZEND_RECV                             63#define ZEND_RECV_INIT                        64#define ZEND_SEND_VAL                         65#define ZEND_SEND_VAR                         66#define ZEND_SEND_REF                         67#define ZEND_NEW                              68#define ZEND_INIT_NS_FCALL_BY_NAME            69#define ZEND_FREE                             70#define ZEND_INIT_ARRAY                       71#define ZEND_ADD_ARRAY_ELEMENT                72#define ZEND_INCLUDE_OR_EVAL                  73#define ZEND_UNSET_VAR                        74#define ZEND_UNSET_DIM                        75#define ZEND_UNSET_OBJ                        76#define ZEND_FE_RESET                         77#define ZEND_FE_FETCH                         78#define ZEND_EXIT                             79#define ZEND_FETCH_R                          80#define ZEND_FETCH_DIM_R                      81#define ZEND_FETCH_OBJ_R                      82#define ZEND_FETCH_W                          83#define ZEND_FETCH_DIM_W                      84#define ZEND_FETCH_OBJ_W                      85#define ZEND_FETCH_RW                         86#define ZEND_FETCH_DIM_RW                     87#define ZEND_FETCH_OBJ_RW                     88#define ZEND_FETCH_IS                         89#define ZEND_FETCH_DIM_IS                     90#define ZEND_FETCH_OBJ_IS                     91#define ZEND_FETCH_FUNC_ARG                   92#define ZEND_FETCH_DIM_FUNC_ARG               93#define ZEND_FETCH_OBJ_FUNC_ARG               94#define ZEND_FETCH_UNSET                      95#define ZEND_FETCH_DIM_UNSET                  96#define ZEND_FETCH_OBJ_UNSET                  97#define ZEND_FETCH_DIM_TMP_VAR                98#define ZEND_FETCH_CONSTANT                   99#define ZEND_GOTO                            100#define ZEND_EXT_STMT                        101#define ZEND_EXT_FCALL_BEGIN                 102#define ZEND_EXT_FCALL_END                   103#define ZEND_EXT_NOP                         104#define ZEND_TICKS                           105#define ZEND_SEND_VAR_NO_REF                 106#define ZEND_CATCH                           107#define ZEND_THROW                           108#define ZEND_FETCH_CLASS                     109#define ZEND_CLONE                           110#define ZEND_RETURN_BY_REF                   111#define ZEND_INIT_METHOD_CALL                112#define ZEND_INIT_STATIC_METHOD_CALL         113#define ZEND_ISSET_ISEMPTY_VAR               114#define ZEND_ISSET_ISEMPTY_DIM_OBJ           115#define ZEND_PRE_INC_OBJ                     132#define ZEND_PRE_DEC_OBJ                     133#define ZEND_POST_INC_OBJ                    134#define ZEND_POST_DEC_OBJ                    135#define ZEND_ASSIGN_OBJ                      136#define ZEND_INSTANCEOF                      138#define ZEND_DECLARE_CLASS                   139#define ZEND_DECLARE_INHERITED_CLASS         140#define ZEND_DECLARE_FUNCTION                141#define ZEND_RAISE_ABSTRACT_ERROR            142#define ZEND_DECLARE_CONST                   143#define ZEND_ADD_INTERFACE                   144#define ZEND_DECLARE_INHERITED_CLASS_DELAYED 145#define ZEND_VERIFY_ABSTRACT_CLASS           146#define ZEND_ASSIGN_DIM                      147#define ZEND_ISSET_ISEMPTY_PROP_OBJ          148#define ZEND_HANDLE_EXCEPTION                149#define ZEND_USER_OPCODE                     150#define ZEND_JMP_SET                         152#define ZEND_DECLARE_LAMBDA_FUNCTION         153#define ZEND_ADD_TRAIT                       154#define ZEND_BIND_TRAITS                     155#define ZEND_SEPARATE                        156#define ZEND_QM_ASSIGN_VAR                   157#define ZEND_JMP_SET_VAR                     158#define ZEND_DISCARD_EXCEPTION               159#define ZEND_YIELD                           160#define ZEND_GENERATOR_RETURN                161#define ZEND_FAST_CALL                       162#define ZEND_FAST_RET                        163#define ZEND_RECV_VARIADIC                   164#define ZEND_SEND_UNPACK                     165#define ZEND_POW                             166#define ZEND_ASSIGN_POW                      167

0x3: opcode execution handle: zend_op-> handler

Op execution handle, whose type is opcode_handler_t

typedef int (ZEND_FASTCALL *opcode_handler_t) (ZEND_OPCODE_HANDLER_ARGS);

This function pointer defines the execution method for op. Each opcode field corresponds to a kind of handler. for example, if $ a = 1; such code generates op, the operands are const and cv. Finally, we can determine that handler is the ZEND_ASSIGN_SPEC_CV_CONST_HANDLER/Zend/zend_vm_execute.h function.

void zend_init_opcodes_handlers(void){  static const opcode_handler_t labels[] = {    ..    ZEND_ASSIGN_SPEC_CV_CONST_HANDLER,    ..    }}

0x4: opcpvdf operand znode

The operand field is an important part of the _ zend_op type. The three operands op1, op2, and result are defined as znode type \ php-5.6.17 \ Zend \ zend_compile.h.

Typedef struct _ znode {/* used only during compilation * // * This int type field defines the type of the znode operand # define IS_CONST (1 <0) // represents a constant, for example, $ a = 123; $ B = "hello"; after the code is generated, both 123 and "hello" exist as constants. # define IS_TMP_VAR (1 <1) // indicates a temporary variable. the temporary variable is usually prefixed ~ This is an intermediate variable used during OP execution. for example, when initializing an array, a temporary variable is required to temporarily store the array zval, then assign the array value to the variable # define IS_VAR (1 <2) // variable in the general sense, expressed in $ development # define IS_UNUSED (1 <3) // Unused variable # define IS_CV (1 <4) // Compiled variable. the operands of this type are important. This type is in a later PHP version (about 5.1) in this example, CV indicates compiled variable, which is the compiled variable. all the variables are saved in a symbol table. this symbol table is a hash table, if you need to search in the hash table for each read/write variable, the efficiency will be affected. Therefore, some variables generated during compilation will be cached in the execution context environment. The operands of this type are generally! For example, $ a = 123; $ B = "hello", the corresponding operands of $ a and $ B may be! 0 and! 1, 0, and 1 are equivalent to an index number. The index number is used to obtain the corresponding value */int op_type from the cache./* This field is a consortium. according to op_type, u takes different values 1. when op_type = IS_CONST, constant in u stores the zval structure corresponding to the operand 2. for example, when $ a = 123, in the 123 operand, constant in u is an IS_LONG type zval, and its value of lval is 123 */union {znode_op op; zval constant; /* replaced by literal/zv */zend_op_array * op_array; zend_ast * ast;} u; zend_uint EA;/* extended attributes */} znode;

0x5: opcode compiled array op_array

In the first line of the zend_do_print function, we noticed the following line of code:

zend_op *opline = get_next_op(CG(active_op_array) TSRMLS_CC);

The opcode generated after PHP script code is compiled is stored in op_array, and its internal storage structure is as follows \ php-5.6.17 \ Zend \ zend_compile.h

Struct _ zend_op_array {/* Common elements */zend_uchar type; const char * function_name; // if it is a user-defined function, the function name zend_class_entry * scope will be saved here; zend_uint fn_flags; union _ zend_function * prototype; zend_uint num_args; zend_uint limit; limit * arg_info;/* END of common elements */limit * refcount; zend_op * opcodes; // opcode array returns last; optional * vars; int last_var; zend_uint T; zend_uint values; zend_uint used_stack; optional * brk_cont_array; int values; optional * values; int values; zend_bool values; /* static variables support */HashTable * static_variables; zend_uint this_var; const char * filename; zend_uint line_start; zend_uint line_end; const char * doc_comment; zend_uint begin; zend_uint early_binding; /* the linked list of delayed declarations */zend_literal * literals; int last_literal; void ** run_time_cache; int last_cache_slot; void * reserved [region];};

The compiled opcodes of the PHP script code is stored here and executed by the following execute function during execution.

ZEND_API void execute (zend_op_array * op_array TSRMLS_DC) {//... cyclically execute opcode in op_array or execute opcode in other op_array}

Each opcode has an opcode_handler_t function pointer field for executing this opcode. PHP can process opcode in three ways.

1. CALL: PHP uses the CALL method by default, that is, the function CALL method 2. SWITCH: because opcode execution is frequently required by every PHP program, you can use SWITCH or GOTO to for distribution. GOTO: Generally, the GOTO efficiency is relatively higher, but whether the efficiency is improved depends on different CPUs.

In fact, we will find that in/zend/zend_language_parser.c, Zend's opcode translation and interpretation execution process is, it includes three opcode execution methods: call, switch, and goto. This is why PHP calls it the kernel principle of interpreted language. after PHP completes Lex lexical parsing, the zend api is called directly through call, switch, and goto in syntax parsing, even if the interpretation is executed.

Relevant Link:

http://www.nowamagic.net/librarys/veda/detail/1325http://php.net/manual/zh/internals2.opcodes.list.phphttp://www.nowamagic.net/librarys/veda/detail/1543http://www.nowamagic.net/librarys/veda/detail/1324http://www.nowamagic.net/librarys/veda/detail/1543 http://www.laruence.com/2008/06/18/221.htmlhttp://www.php-internals.com/book/?p=chapt02/02-03-02-opcode

3. opcode translation execution (instant interpretation execution)

Relevant Link:

http://www.php-internals.com/book/?p=chapt02/02-03-03-from-opcode-to-handler

Copyright (c) 2016 Little5ann All rights reserved