[lttng-tools.git] / src / lib / lttng-ctl / filter / filter-parser.y

%{
/*
 * filter-parser.y
 *
 * LTTng filter expression parser
 *
 * Copyright 2012 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
 *
 * SPDX-License-Identifier: LGPL-2.1-only
 *
 * Grammar inspired from http://www.quut.com/c/ANSI-C-grammar-y.html
 */

#include <stdio.h>
#include <unistd.h>
#include <string.h>
#include <stdlib.h>
#include <assert.h>
#include <errno.h>
#include <inttypes.h>
#include "filter-ast.h"
#include "filter-parser.h"

#include <common/macros.h>

#define WIDTH_u64_SCANF_IS_A_BROKEN_API	"20"
#define WIDTH_o64_SCANF_IS_A_BROKEN_API	"22"
#define WIDTH_x64_SCANF_IS_A_BROKEN_API	"17"
#define WIDTH_lg_SCANF_IS_A_BROKEN_API	"4096"	/* Hugely optimistic approximation */

LTTNG_HIDDEN
int yydebug;
LTTNG_HIDDEN
int filter_parser_debug = 0;

LTTNG_HIDDEN
int yyparse(struct filter_parser_ctx *parser_ctx, yyscan_t scanner);
LTTNG_HIDDEN
int yylex(union YYSTYPE *yyval, yyscan_t scanner);
LTTNG_HIDDEN
int yylex_init_extra(struct filter_parser_ctx *parser_ctx, yyscan_t * ptr_yy_globals);
LTTNG_HIDDEN
int yylex_destroy(yyscan_t yyparser_ctx);
LTTNG_HIDDEN
void yyrestart(FILE * in_str, yyscan_t parser_ctx);

struct gc_string {
	struct cds_list_head gc;
	size_t alloclen;
	char s[];
};

static const char *node_type_to_str[] = {
	[ NODE_UNKNOWN ] = "NODE_UNKNOWN",
	[ NODE_ROOT ] = "NODE_ROOT",
	[ NODE_EXPRESSION ] = "NODE_EXPRESSION",
	[ NODE_OP ] = "NODE_OP",
	[ NODE_UNARY_OP ] = "NODE_UNARY_OP",
};

LTTNG_HIDDEN
const char *node_type(struct filter_node *node)
{
	if (node->type < NR_NODE_TYPES)
		return node_type_to_str[node->type];
	else
		return NULL;
}

static struct gc_string *gc_string_alloc(struct filter_parser_ctx *parser_ctx,
					 size_t len)
{
	struct gc_string *gstr;
	size_t alloclen;

	/* TODO: could be faster with find first bit or glib Gstring */
	/* sizeof long to account for malloc header (int or long ?) */
	for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + len;
	     alloclen *= 2);

	gstr = zmalloc(alloclen);
	if (!gstr) {
		goto end;
	}
	cds_list_add(&gstr->gc, &parser_ctx->allocated_strings);
	gstr->alloclen = alloclen;
end:
	return gstr;
}

/*
 * note: never use gc_string_append on a string that has external references.
 * gsrc will be garbage collected immediately, and gstr might be.
 * Should only be used to append characters to a string literal or constant.
 */
static
struct gc_string *gc_string_append(struct filter_parser_ctx *parser_ctx,
				   struct gc_string *gstr,
				   struct gc_string *gsrc)
{
	size_t newlen = strlen(gsrc->s) + strlen(gstr->s) + 1;
	size_t alloclen;

	/* TODO: could be faster with find first bit or glib Gstring */
	/* sizeof long to account for malloc header (int or long ?) */
	for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + newlen;
	     alloclen *= 2);

	if (alloclen > gstr->alloclen) {
		struct gc_string *newgstr;

		newgstr = gc_string_alloc(parser_ctx, newlen);
		strcpy(newgstr->s, gstr->s);
		strcat(newgstr->s, gsrc->s);
		cds_list_del(&gstr->gc);
		free(gstr);
		gstr = newgstr;
	} else {
		strcat(gstr->s, gsrc->s);
	}
	cds_list_del(&gsrc->gc);
	free(gsrc);
	return gstr;
}

LTTNG_HIDDEN
void setstring(struct filter_parser_ctx *parser_ctx, YYSTYPE *lvalp, const char *src)
{
	lvalp->gs = gc_string_alloc(parser_ctx, strlen(src) + 1);
	strcpy(lvalp->gs->s, src);
}

static struct filter_node *make_node(struct filter_parser_ctx *scanner,
				  enum node_type type)
{
	struct filter_ast *ast = filter_parser_get_ast(scanner);
	struct filter_node *node;

	node = zmalloc(sizeof(*node));
	if (!node)
		return NULL;
	memset(node, 0, sizeof(*node));
	node->type = type;
	cds_list_add(&node->gc, &ast->allocated_nodes);

	switch (type) {
	case NODE_ROOT:
		fprintf(stderr, "[error] %s: trying to create root node\n", __func__);
		break;

	case NODE_EXPRESSION:
		break;
	case NODE_OP:
		break;
	case NODE_UNARY_OP:
		break;

	case NODE_UNKNOWN:
	default:
		fprintf(stderr, "[error] %s: unknown node type %d\n", __func__,
			(int) type);
		break;
	}

	return node;
}

static struct filter_node *make_op_node(struct filter_parser_ctx *scanner,
			enum op_type type,
			struct filter_node *lchild,
			struct filter_node *rchild)
{
	struct filter_ast *ast = filter_parser_get_ast(scanner);
	struct filter_node *node;

	node = zmalloc(sizeof(*node));
	if (!node)
		return NULL;
	memset(node, 0, sizeof(*node));
	node->type = NODE_OP;
	cds_list_add(&node->gc, &ast->allocated_nodes);
	node->u.op.type = type;
	node->u.op.lchild = lchild;
	node->u.op.rchild = rchild;
	return node;
}

static
void yyerror(struct filter_parser_ctx *parser_ctx, yyscan_t scanner, const char *str)
{
	fprintf(stderr, "error %s\n", str);
}

#define parse_error(parser_ctx, str)				\
do {								\
	yyerror(parser_ctx, parser_ctx->scanner, YY_("parse error: " str "\n"));	\
	YYERROR;						\
} while (0)

static void free_strings(struct cds_list_head *list)
{
	struct gc_string *gstr, *tmp;

	cds_list_for_each_entry_safe(gstr, tmp, list, gc)
		free(gstr);
}

static struct filter_ast *filter_ast_alloc(void)
{
	struct filter_ast *ast;

	ast = zmalloc(sizeof(*ast));
	if (!ast)
		return NULL;
	memset(ast, 0, sizeof(*ast));
	CDS_INIT_LIST_HEAD(&ast->allocated_nodes);
	ast->root.type = NODE_ROOT;
	return ast;
}

static void filter_ast_free(struct filter_ast *ast)
{
	struct filter_node *node, *tmp;

	cds_list_for_each_entry_safe(node, tmp, &ast->allocated_nodes, gc)
		free(node);
	free(ast);
}

LTTNG_HIDDEN
int filter_parser_ctx_append_ast(struct filter_parser_ctx *parser_ctx)
{
	return yyparse(parser_ctx, parser_ctx->scanner);
}

LTTNG_HIDDEN
struct filter_parser_ctx *filter_parser_ctx_alloc(FILE *input)
{
	struct filter_parser_ctx *parser_ctx;
	int ret;

	yydebug = filter_parser_debug;

	parser_ctx = zmalloc(sizeof(*parser_ctx));
	if (!parser_ctx)
		return NULL;
	memset(parser_ctx, 0, sizeof(*parser_ctx));

	ret = yylex_init_extra(parser_ctx, &parser_ctx->scanner);
	if (ret) {
		fprintf(stderr, "yylex_init error\n");
		goto cleanup_parser_ctx;
	}
	/* Start processing new stream */
	yyrestart(input, parser_ctx->scanner);

	parser_ctx->ast = filter_ast_alloc();
	if (!parser_ctx->ast)
		goto cleanup_lexer;
	CDS_INIT_LIST_HEAD(&parser_ctx->allocated_strings);

	if (yydebug)
		fprintf(stdout, "parser_ctx input is a%s.\n",
			isatty(fileno(input)) ? "n interactive tty" :
						" noninteractive file");

	return parser_ctx;

cleanup_lexer:
	ret = yylex_destroy(parser_ctx->scanner);
	if (!ret)
		fprintf(stderr, "yylex_destroy error\n");
cleanup_parser_ctx:
	free(parser_ctx);
	return NULL;
}

LTTNG_HIDDEN
void filter_parser_ctx_free(struct filter_parser_ctx *parser_ctx)
{
	int ret;

	ret = yylex_destroy(parser_ctx->scanner);
	if (ret)
		fprintf(stderr, "yylex_destroy error\n");

	filter_ast_free(parser_ctx->ast);
	free_strings(&parser_ctx->allocated_strings);
	filter_ir_free(parser_ctx);
	free(parser_ctx->bytecode);
	free(parser_ctx->bytecode_reloc);

	free(parser_ctx);
}

%}

%code provides
{
#include "common/macros.h"

LTTNG_HIDDEN
void setstring(struct filter_parser_ctx *parser_ctx, YYSTYPE *lvalp, const char *src);
}

%define api.pure
	/* %locations */
%parse-param {struct filter_parser_ctx *parser_ctx}
%parse-param {yyscan_t scanner}
%lex-param {yyscan_t scanner}
%start translation_unit
%token CHARACTER_CONSTANT_START SQUOTE STRING_LITERAL_START DQUOTE
%token ESCSEQ CHAR_STRING_TOKEN
%token DECIMAL_CONSTANT OCTAL_CONSTANT HEXADECIMAL_CONSTANT FLOAT_CONSTANT
%token LSBRAC RSBRAC LPAREN RPAREN LBRAC RBRAC RARROW
%token STAR PLUS MINUS
%token MOD_OP DIV_OP RIGHT_OP LEFT_OP
%token EQ_OP NE_OP LE_OP GE_OP LT_OP GT_OP AND_OP OR_OP NOT_OP
%token ASSIGN COLON SEMICOLON DOTDOTDOT DOT EQUAL COMMA
%token XOR_BIN AND_BIN OR_BIN NOT_BIN

%token <gs> IDENTIFIER GLOBAL_IDENTIFIER
%token ERROR
%union
{
	long long ll;
	char c;
	struct gc_string *gs;
	struct filter_node *n;
}

%type <gs> s_char s_char_sequence c_char c_char_sequence

%type <n> primary_expression
%type <n> prefix_expression
%type <n> prefix_expression_rec
%type <n> postfix_expression
%type <n> unary_expression
%type <n> unary_operator
%type <n> multiplicative_expression
%type <n> additive_expression
%type <n> shift_expression
%type <n> relational_expression
%type <n> equality_expression
%type <n> and_expression
%type <n> exclusive_or_expression
%type <n> inclusive_or_expression
%type <n> logical_and_expression
%type <n> logical_or_expression
%type <n> expression
%type <n> identifiers

%%


/* 1.5 Constants */

c_char_sequence:
		c_char
		{	$$ = $1;					}
	|	c_char_sequence c_char
		{	$$ = gc_string_append(parser_ctx, $1, $2);		}
	;

c_char:
		CHAR_STRING_TOKEN
		{	$$ = yylval.gs;					}
	|	ESCSEQ
		{
			parse_error(parser_ctx, "escape sequences not supported yet");
		}
	;

/* 1.6 String literals */

s_char_sequence:
		s_char
		{	$$ = $1;					}
	|	s_char_sequence s_char
		{	$$ = gc_string_append(parser_ctx, $1, $2);		}
	;

s_char:
		CHAR_STRING_TOKEN
		{	$$ = yylval.gs;					}
	|	ESCSEQ
		{
			parse_error(parser_ctx, "escape sequences not supported yet");
		}
	;

primary_expression:
		DECIMAL_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_CONSTANT;
			if (sscanf(yylval.gs->s, "%" WIDTH_u64_SCANF_IS_A_BROKEN_API SCNu64,
					&$$->u.expression.u.constant) != 1) {
				parse_error(parser_ctx, "cannot scanf decimal constant");
			}
		}
	|	OCTAL_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_CONSTANT;
			if (!strcmp(yylval.gs->s, "0")) {
				$$->u.expression.u.constant = 0;
			} else if (sscanf(yylval.gs->s, "0%" WIDTH_o64_SCANF_IS_A_BROKEN_API SCNo64,
					&$$->u.expression.u.constant) != 1) {
				parse_error(parser_ctx, "cannot scanf octal constant");
			}
		}
	|	HEXADECIMAL_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_CONSTANT;
			if (sscanf(yylval.gs->s, "0x%" WIDTH_x64_SCANF_IS_A_BROKEN_API SCNx64,
					&$$->u.expression.u.constant) != 1) {
				parse_error(parser_ctx, "cannot scanf hexadecimal constant");
			}
		}
	|	FLOAT_CONSTANT
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_FLOAT_CONSTANT;
			if (sscanf(yylval.gs->s, "%" WIDTH_lg_SCANF_IS_A_BROKEN_API "lg",
					&$$->u.expression.u.float_constant) != 1) {
				parse_error(parser_ctx, "cannot scanf float constant");
			}
		}
	|	STRING_LITERAL_START DQUOTE
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_STRING;
			$$->u.expression.u.string = "";
		}
	|	STRING_LITERAL_START s_char_sequence DQUOTE
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_STRING;
			$$->u.expression.u.string = $2->s;
		}
	|	CHARACTER_CONSTANT_START c_char_sequence SQUOTE
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_STRING;
			$$->u.expression.u.string = $2->s;
		}
	|	LPAREN expression RPAREN
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_NESTED;
			$$->u.expression.u.child = $2;
		}
	;

identifiers
	:	IDENTIFIER
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_IDENTIFIER;
			$$->u.expression.u.identifier = yylval.gs->s;
		}
	|	GLOBAL_IDENTIFIER
		{
			$$ = make_node(parser_ctx, NODE_EXPRESSION);
			$$->u.expression.type = AST_EXP_GLOBAL_IDENTIFIER;
			$$->u.expression.u.identifier = yylval.gs->s;
		}
	;

prefix_expression_rec
	: LSBRAC unary_expression RSBRAC
		{
			$$ = $2;
		}
	| LSBRAC unary_expression RSBRAC prefix_expression_rec
		{
			$$ = $2;
			$$->u.expression.pre_op = AST_LINK_BRACKET;
			$$->u.expression.prev = $4;
		}
	;

prefix_expression
	: identifiers
		{
			$$ = $1;
		}
	| identifiers prefix_expression_rec
		{
			$$ = $1;
			$$->u.expression.pre_op = AST_LINK_BRACKET;
			$$->u.expression.next_bracket = $2;
		}
	;

postfix_expression
	: prefix_expression
		{
			$$ = $1;
		}
	| postfix_expression DOT prefix_expression
		{
			$$ = $3;
			$$->u.expression.post_op = AST_LINK_DOT;
			$$->u.expression.prev = $1;
		}
	| postfix_expression RARROW prefix_expression
		{
			$$ = $3;
			$$->u.expression.post_op = AST_LINK_RARROW;
			$$->u.expression.prev = $1;
		}
	;

unary_expression
	: postfix_expression
		{	$$ = $1;					}
	| primary_expression
		{	$$ = $1;					}
	| unary_operator unary_expression
		{
			$$ = $1;
			$$->u.unary_op.child = $2;
		}
	;

unary_operator
	: PLUS
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_PLUS;
		}
	| MINUS
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_MINUS;
		}
	| NOT_OP
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_NOT;
		}
	| NOT_BIN
		{
			$$ = make_node(parser_ctx, NODE_UNARY_OP);
			$$->u.unary_op.type = AST_UNARY_BIT_NOT;
		}
	;

multiplicative_expression
	: unary_expression
		{	$$ = $1;					}
	| multiplicative_expression STAR unary_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_MUL, $1, $3);
		}
	| multiplicative_expression DIV_OP unary_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_DIV, $1, $3);
		}
	| multiplicative_expression MOD_OP unary_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_MOD, $1, $3);
		}
	;

additive_expression
	: multiplicative_expression
		{	$$ = $1;					}
	| additive_expression PLUS multiplicative_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_PLUS, $1, $3);
		}
	| additive_expression MINUS multiplicative_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_MINUS, $1, $3);
		}
	;

shift_expression
	: additive_expression
		{	$$ = $1;					}
	| shift_expression LEFT_OP additive_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_LSHIFT, $1, $3);
		}
	| shift_expression RIGHT_OP additive_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_RSHIFT, $1, $3);
		}
	;

and_expression
	: shift_expression
		{	$$ = $1;					}
	| and_expression AND_BIN shift_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_AND, $1, $3);
		}
	;

exclusive_or_expression
	: and_expression
		{	$$ = $1;					}
	| exclusive_or_expression XOR_BIN and_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_XOR, $1, $3);
		}
	;

inclusive_or_expression
	: exclusive_or_expression
		{	$$ = $1;					}
	| inclusive_or_expression OR_BIN exclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_BIT_OR, $1, $3);
		}
	;

relational_expression
	: inclusive_or_expression
		{	$$ = $1;					}
	| relational_expression LT_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_LT, $1, $3);
		}
	| relational_expression GT_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_GT, $1, $3);
		}
	| relational_expression LE_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_LE, $1, $3);
		}
	| relational_expression GE_OP inclusive_or_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_GE, $1, $3);
		}
	;

equality_expression
	: relational_expression
		{	$$ = $1;					}
	| equality_expression EQ_OP relational_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_EQ, $1, $3);
		}
	| equality_expression NE_OP relational_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_NE, $1, $3);
		}
	;

logical_and_expression
	: equality_expression
		{	$$ = $1;					}
	| logical_and_expression AND_OP equality_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_AND, $1, $3);
		}
	;

logical_or_expression
	: logical_and_expression
		{	$$ = $1;					}
	| logical_or_expression OR_OP logical_and_expression
		{
			$$ = make_op_node(parser_ctx, AST_OP_OR, $1, $3);
		}
	;

expression
	: logical_or_expression
		{	$$ = $1;					}
	;

translation_unit
	: expression
		{
			parser_ctx->ast->root.u.root.child = $1;
		}
	;
Commit	Line	Data
953192ba MD	1	%{
	2	/*
	3	* filter-parser.y
	4	*
	5	* LTTng filter expression parser
	6	*
ab5be9fa	7	* Copyright 2012 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
953192ba	8	*
ab5be9fa	9	* SPDX-License-Identifier: LGPL-2.1-only
953192ba MD	10	*
	11	* Grammar inspired from http://www.quut.com/c/ANSI-C-grammar-y.html
	12	*/
	13
	14	#include <stdio.h>
	15	#include <unistd.h>
	16	#include <string.h>
	17	#include <stdlib.h>
	18	#include <assert.h>
	19	#include <errno.h>
	20	#include <inttypes.h>
953192ba	21	#include "filter-ast.h"
95b9bd90	22	#include "filter-parser.h"
953192ba	23
a187da1a DG	24	#include <common/macros.h>
a187da1a DG	25
8ab7c0d9 MD	26	#define WIDTH_u64_SCANF_IS_A_BROKEN_API "20"
	27	#define WIDTH_o64_SCANF_IS_A_BROKEN_API "22"
	28	#define WIDTH_x64_SCANF_IS_A_BROKEN_API "17"
	29	#define WIDTH_lg_SCANF_IS_A_BROKEN_API "4096" /* Hugely optimistic approximation */
	30
a187da1a	31	LTTNG_HIDDEN
953192ba	32	int yydebug;
a187da1a	33	LTTNG_HIDDEN
953192ba MD	34	int filter_parser_debug = 0;
953192ba MD	35
a187da1a	36	LTTNG_HIDDEN
9039edd4	37	int yyparse(struct filter_parser_ctx *parser_ctx, yyscan_t scanner);
a187da1a	38	LTTNG_HIDDEN
9039edd4	39	int yylex(union YYSTYPE *yyval, yyscan_t scanner);
a187da1a	40	LTTNG_HIDDEN
953192ba	41	int yylex_init_extra(struct filter_parser_ctx parser_ctx, yyscan_t ptr_yy_globals);
a187da1a	42	LTTNG_HIDDEN
953192ba	43	int yylex_destroy(yyscan_t yyparser_ctx);
a187da1a	44	LTTNG_HIDDEN
953192ba MD	45	void yyrestart(FILE * in_str, yyscan_t parser_ctx);
	46
	47	struct gc_string {
	48	struct cds_list_head gc;
	49	size_t alloclen;
	50	char s[];
	51	};
	52
	53	static const char *node_type_to_str[] = {
	54	[ NODE_UNKNOWN ] = "NODE_UNKNOWN",
	55	[ NODE_ROOT ] = "NODE_ROOT",
	56	[ NODE_EXPRESSION ] = "NODE_EXPRESSION",
	57	[ NODE_OP ] = "NODE_OP",
	58	[ NODE_UNARY_OP ] = "NODE_UNARY_OP",
	59	};
	60
a187da1a	61	LTTNG_HIDDEN
953192ba MD	62	const char node_type(struct filter_node node)
	63	{
	64	if (node->type < NR_NODE_TYPES)
	65	return node_type_to_str[node->type];
	66	else
	67	return NULL;
	68	}
	69
	70	static struct gc_string gc_string_alloc(struct filter_parser_ctx parser_ctx,
	71	size_t len)
	72	{
	73	struct gc_string *gstr;
	74	size_t alloclen;
	75
	76	/* TODO: could be faster with find first bit or glib Gstring */
	77	/* sizeof long to account for malloc header (int or long ?) */
	78	for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + len;
	79	alloclen *= 2);
	80
be61ee34 MD	81	gstr = zmalloc(alloclen);
	82	if (!gstr) {
	83	goto end;
	84	}
953192ba MD	85	cds_list_add(&gstr->gc, &parser_ctx->allocated_strings);
953192ba MD	86	gstr->alloclen = alloclen;
be61ee34	87	end:
953192ba MD	88	return gstr;
	89	}
	90
	91	/*
	92	* note: never use gc_string_append on a string that has external references.
	93	* gsrc will be garbage collected immediately, and gstr might be.
	94	* Should only be used to append characters to a string literal or constant.
	95	*/
1a0b02f1	96	static
953192ba MD	97	struct gc_string gc_string_append(struct filter_parser_ctx parser_ctx,
	98	struct gc_string *gstr,
	99	struct gc_string *gsrc)
	100	{
	101	size_t newlen = strlen(gsrc->s) + strlen(gstr->s) + 1;
	102	size_t alloclen;
	103
	104	/* TODO: could be faster with find first bit or glib Gstring */
	105	/* sizeof long to account for malloc header (int or long ?) */
	106	for (alloclen = 8; alloclen < sizeof(long) + sizeof(*gstr) + newlen;
	107	alloclen *= 2);
	108
	109	if (alloclen > gstr->alloclen) {
	110	struct gc_string *newgstr;
	111
	112	newgstr = gc_string_alloc(parser_ctx, newlen);
	113	strcpy(newgstr->s, gstr->s);
	114	strcat(newgstr->s, gsrc->s);
	115	cds_list_del(&gstr->gc);
	116	free(gstr);
	117	gstr = newgstr;
	118	} else {
	119	strcat(gstr->s, gsrc->s);
	120	}
	121	cds_list_del(&gsrc->gc);
	122	free(gsrc);
	123	return gstr;
	124	}
	125
a187da1a	126	LTTNG_HIDDEN
953192ba MD	127	void setstring(struct filter_parser_ctx parser_ctx, YYSTYPE lvalp, const char *src)
	128	{
	129	lvalp->gs = gc_string_alloc(parser_ctx, strlen(src) + 1);
	130	strcpy(lvalp->gs->s, src);
	131	}
	132
	133	static struct filter_node make_node(struct filter_parser_ctx scanner,
	134	enum node_type type)
	135	{
	136	struct filter_ast *ast = filter_parser_get_ast(scanner);
	137	struct filter_node *node;
	138
be61ee34	139	node = zmalloc(sizeof(*node));
953192ba MD	140	if (!node)
	141	return NULL;
	142	memset(node, 0, sizeof(*node));
	143	node->type = type;
	144	cds_list_add(&node->gc, &ast->allocated_nodes);
	145
	146	switch (type) {
	147	case NODE_ROOT:
	148	fprintf(stderr, "[error] %s: trying to create root node\n", __func__);
	149	break;
	150
	151	case NODE_EXPRESSION:
	152	break;
	153	case NODE_OP:
	154	break;
	155	case NODE_UNARY_OP:
	156	break;
	157
	158	case NODE_UNKNOWN:
	159	default:
	160	fprintf(stderr, "[error] %s: unknown node type %d\n", __func__,
	161	(int) type);
	162	break;
	163	}
	164
	165	return node;
	166	}
	167
	168	static struct filter_node make_op_node(struct filter_parser_ctx scanner,
	169	enum op_type type,
	170	struct filter_node *lchild,
	171	struct filter_node *rchild)
	172	{
	173	struct filter_ast *ast = filter_parser_get_ast(scanner);
	174	struct filter_node *node;
	175
be61ee34	176	node = zmalloc(sizeof(*node));
953192ba MD	177	if (!node)
	178	return NULL;
	179	memset(node, 0, sizeof(*node));
	180	node->type = NODE_OP;
	181	cds_list_add(&node->gc, &ast->allocated_nodes);
	182	node->u.op.type = type;
	183	node->u.op.lchild = lchild;
	184	node->u.op.rchild = rchild;
	185	return node;
	186	}
	187
1a0b02f1	188	static
9039edd4	189	void yyerror(struct filter_parser_ctx parser_ctx, yyscan_t scanner, const char str)
953192ba MD	190	{
	191	fprintf(stderr, "error %s\n", str);
	192	}
953192ba MD	193
	194	#define parse_error(parser_ctx, str) \
	195	do { \
9039edd4	196	yyerror(parser_ctx, parser_ctx->scanner, YY_("parse error: " str "\n")); \
953192ba MD	197	YYERROR; \
	198	} while (0)
	199
	200	static void free_strings(struct cds_list_head *list)
	201	{
	202	struct gc_string gstr, tmp;
	203
	204	cds_list_for_each_entry_safe(gstr, tmp, list, gc)
	205	free(gstr);
	206	}
	207
	208	static struct filter_ast *filter_ast_alloc(void)
	209	{
	210	struct filter_ast *ast;
	211
be61ee34	212	ast = zmalloc(sizeof(*ast));
953192ba MD	213	if (!ast)
	214	return NULL;
	215	memset(ast, 0, sizeof(*ast));
	216	CDS_INIT_LIST_HEAD(&ast->allocated_nodes);
	217	ast->root.type = NODE_ROOT;
	218	return ast;
	219	}
	220
	221	static void filter_ast_free(struct filter_ast *ast)
	222	{
	223	struct filter_node node, tmp;
	224
	225	cds_list_for_each_entry_safe(node, tmp, &ast->allocated_nodes, gc)
	226	free(node);
37600d79	227	free(ast);
953192ba MD	228	}
953192ba MD	229
a187da1a	230	LTTNG_HIDDEN
953192ba MD	231	int filter_parser_ctx_append_ast(struct filter_parser_ctx *parser_ctx)
953192ba MD	232	{
9039edd4	233	return yyparse(parser_ctx, parser_ctx->scanner);
953192ba MD	234	}
953192ba MD	235
a187da1a	236	LTTNG_HIDDEN
953192ba MD	237	struct filter_parser_ctx filter_parser_ctx_alloc(FILE input)
	238	{
	239	struct filter_parser_ctx *parser_ctx;
	240	int ret;
	241
	242	yydebug = filter_parser_debug;
	243
be61ee34	244	parser_ctx = zmalloc(sizeof(*parser_ctx));
953192ba MD	245	if (!parser_ctx)
	246	return NULL;
	247	memset(parser_ctx, 0, sizeof(*parser_ctx));
	248
	249	ret = yylex_init_extra(parser_ctx, &parser_ctx->scanner);
	250	if (ret) {
	251	fprintf(stderr, "yylex_init error\n");
	252	goto cleanup_parser_ctx;
	253	}
	254	/* Start processing new stream */
	255	yyrestart(input, parser_ctx->scanner);
	256
	257	parser_ctx->ast = filter_ast_alloc();
	258	if (!parser_ctx->ast)
	259	goto cleanup_lexer;
	260	CDS_INIT_LIST_HEAD(&parser_ctx->allocated_strings);
	261
	262	if (yydebug)
	263	fprintf(stdout, "parser_ctx input is a%s.\n",
	264	isatty(fileno(input)) ? "n interactive tty" :
	265	" noninteractive file");
	266
	267	return parser_ctx;
	268
	269	cleanup_lexer:
	270	ret = yylex_destroy(parser_ctx->scanner);
	271	if (!ret)
	272	fprintf(stderr, "yylex_destroy error\n");
	273	cleanup_parser_ctx:
	274	free(parser_ctx);
	275	return NULL;
	276	}
	277
a187da1a	278	LTTNG_HIDDEN
953192ba MD	279	void filter_parser_ctx_free(struct filter_parser_ctx *parser_ctx)
	280	{
	281	int ret;
	282
953192ba MD	283	ret = yylex_destroy(parser_ctx->scanner);
	284	if (ret)
	285	fprintf(stderr, "yylex_destroy error\n");
d59903e9 SM	286
	287	filter_ast_free(parser_ctx->ast);
	288	free_strings(&parser_ctx->allocated_strings);
	289	filter_ir_free(parser_ctx);
	290	free(parser_ctx->bytecode);
	291	free(parser_ctx->bytecode_reloc);
	292
953192ba MD	293	free(parser_ctx);
	294	}
	295
	296	%}
	297
1a0b02f1 SM	298	%code provides
	299	{
	300	#include "common/macros.h"
	301
	302	LTTNG_HIDDEN
	303	void setstring(struct filter_parser_ctx parser_ctx, YYSTYPE lvalp, const char *src);
	304	}
	305
953192ba MD	306	%define api.pure
	307	/* %locations */
	308	%parse-param {struct filter_parser_ctx *parser_ctx}
9039edd4 ZT	309	%parse-param {yyscan_t scanner}
9039edd4 ZT	310	%lex-param {yyscan_t scanner}
953192ba MD	311	%start translation_unit
	312	%token CHARACTER_CONSTANT_START SQUOTE STRING_LITERAL_START DQUOTE
	313	%token ESCSEQ CHAR_STRING_TOKEN
e90d8561	314	%token DECIMAL_CONSTANT OCTAL_CONSTANT HEXADECIMAL_CONSTANT FLOAT_CONSTANT
953192ba MD	315	%token LSBRAC RSBRAC LPAREN RPAREN LBRAC RBRAC RARROW
	316	%token STAR PLUS MINUS
	317	%token MOD_OP DIV_OP RIGHT_OP LEFT_OP
	318	%token EQ_OP NE_OP LE_OP GE_OP LT_OP GT_OP AND_OP OR_OP NOT_OP
	319	%token ASSIGN COLON SEMICOLON DOTDOTDOT DOT EQUAL COMMA
	320	%token XOR_BIN AND_BIN OR_BIN NOT_BIN
	321
586dc72f	322	%token <gs> IDENTIFIER GLOBAL_IDENTIFIER
953192ba MD	323	%token ERROR
	324	%union
	325	{
	326	long long ll;
	327	char c;
	328	struct gc_string *gs;
	329	struct filter_node *n;
	330	}
	331
	332	%type <gs> s_char s_char_sequence c_char c_char_sequence
	333
	334	%type <n> primary_expression
bff988fa MD	335	%type <n> prefix_expression
bff988fa MD	336	%type <n> prefix_expression_rec
953192ba MD	337	%type <n> postfix_expression
	338	%type <n> unary_expression
	339	%type <n> unary_operator
	340	%type <n> multiplicative_expression
	341	%type <n> additive_expression
	342	%type <n> shift_expression
	343	%type <n> relational_expression
	344	%type <n> equality_expression
	345	%type <n> and_expression
	346	%type <n> exclusive_or_expression
	347	%type <n> inclusive_or_expression
	348	%type <n> logical_and_expression
	349	%type <n> logical_or_expression
	350	%type <n> expression
bff988fa	351	%type <n> identifiers
953192ba MD	352
	353	%%
	354
	355
	356	/* 1.5 Constants */
	357
	358	c_char_sequence:
	359	c_char
	360	{ $$ = $1; }
	361	\| c_char_sequence c_char
	362	{ $$ = gc_string_append(parser_ctx, $1, $2); }
	363	;
	364
	365	c_char:
	366	CHAR_STRING_TOKEN
	367	{ $$ = yylval.gs; }
	368	\| ESCSEQ
	369	{
	370	parse_error(parser_ctx, "escape sequences not supported yet");
	371	}
	372	;
	373
	374	/* 1.6 String literals */
	375
	376	s_char_sequence:
	377	s_char
	378	{ $$ = $1; }
	379	\| s_char_sequence s_char
	380	{ $$ = gc_string_append(parser_ctx, $1, $2); }
	381	;
	382
	383	s_char:
	384	CHAR_STRING_TOKEN
	385	{ $$ = yylval.gs; }
	386	\| ESCSEQ
	387	{
	388	parse_error(parser_ctx, "escape sequences not supported yet");
	389	}
	390	;
	391
bff988fa MD	392	primary_expression:
bff988fa MD	393	DECIMAL_CONSTANT
953192ba MD	394	{
	395	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	396	$$->u.expression.type = AST_EXP_CONSTANT;
8ab7c0d9 MD	397	if (sscanf(yylval.gs->s, "%" WIDTH_u64_SCANF_IS_A_BROKEN_API SCNu64,
	398	&$$->u.expression.u.constant) != 1) {
	399	parse_error(parser_ctx, "cannot scanf decimal constant");
	400	}
953192ba MD	401	}
	402	\| OCTAL_CONSTANT
	403	{
	404	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	405	$$->u.expression.type = AST_EXP_CONSTANT;
8ab7c0d9 MD	406	if (!strcmp(yylval.gs->s, "0")) {
	407	$$->u.expression.u.constant = 0;
	408	} else if (sscanf(yylval.gs->s, "0%" WIDTH_o64_SCANF_IS_A_BROKEN_API SCNo64,
	409	&$$->u.expression.u.constant) != 1) {
	410	parse_error(parser_ctx, "cannot scanf octal constant");
	411	}
953192ba MD	412	}
	413	\| HEXADECIMAL_CONSTANT
	414	{
	415	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	416	$$->u.expression.type = AST_EXP_CONSTANT;
8ab7c0d9 MD	417	if (sscanf(yylval.gs->s, "0x%" WIDTH_x64_SCANF_IS_A_BROKEN_API SCNx64,
	418	&$$->u.expression.u.constant) != 1) {
	419	parse_error(parser_ctx, "cannot scanf hexadecimal constant");
	420	}
953192ba	421	}
e90d8561 MD	422	\| FLOAT_CONSTANT
	423	{
	424	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	425	$$->u.expression.type = AST_EXP_FLOAT_CONSTANT;
8ab7c0d9 MD	426	if (sscanf(yylval.gs->s, "%" WIDTH_lg_SCANF_IS_A_BROKEN_API "lg",
	427	&$$->u.expression.u.float_constant) != 1) {
	428	parse_error(parser_ctx, "cannot scanf float constant");
	429	}
e90d8561	430	}
953192ba MD	431	\| STRING_LITERAL_START DQUOTE
	432	{
	433	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	434	$$->u.expression.type = AST_EXP_STRING;
	435	$$->u.expression.u.string = "";
	436	}
	437	\| STRING_LITERAL_START s_char_sequence DQUOTE
	438	{
	439	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	440	$$->u.expression.type = AST_EXP_STRING;
	441	$$->u.expression.u.string = $2->s;
	442	}
	443	\| CHARACTER_CONSTANT_START c_char_sequence SQUOTE
	444	{
	445	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	446	$$->u.expression.type = AST_EXP_STRING;
	447	$$->u.expression.u.string = $2->s;
	448	}
	449	\| LPAREN expression RPAREN
	450	{
	451	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	452	$$->u.expression.type = AST_EXP_NESTED;
	453	$$->u.expression.u.child = $2;
	454	}
	455	;
	456
bff988fa MD	457	identifiers
	458	: IDENTIFIER
	459	{
	460	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	461	$$->u.expression.type = AST_EXP_IDENTIFIER;
	462	$$->u.expression.u.identifier = yylval.gs->s;
	463	}
	464	\| GLOBAL_IDENTIFIER
	465	{
	466	$$ = make_node(parser_ctx, NODE_EXPRESSION);
	467	$$->u.expression.type = AST_EXP_GLOBAL_IDENTIFIER;
	468	$$->u.expression.u.identifier = yylval.gs->s;
	469	}
	470	;
	471
	472	prefix_expression_rec
	473	: LSBRAC unary_expression RSBRAC
	474	{
	475	$$ = $2;
	476	}
	477	\| LSBRAC unary_expression RSBRAC prefix_expression_rec
	478	{
	479	$$ = $2;
	480	$$->u.expression.pre_op = AST_LINK_BRACKET;
	481	$$->u.expression.prev = $4;
	482	}
	483	;
	484
	485	prefix_expression
	486	: identifiers
	487	{
	488	$$ = $1;
	489	}
	490	\| identifiers prefix_expression_rec
661dfdd1 MD	491	{
	492	$$ = $1;
	493	$$->u.expression.pre_op = AST_LINK_BRACKET;
bff988fa	494	$$->u.expression.next_bracket = $2;
661dfdd1	495	}
bff988fa MD	496	;
	497
	498	postfix_expression
	499	: prefix_expression
953192ba	500	{
bff988fa MD	501	$$ = $1;
	502	}
	503	\| postfix_expression DOT prefix_expression
	504	{
	505	$$ = $3;
953192ba	506	$$->u.expression.post_op = AST_LINK_DOT;
953192ba MD	507	$$->u.expression.prev = $1;
953192ba MD	508	}
bff988fa	509	\| postfix_expression RARROW prefix_expression
953192ba	510	{
bff988fa	511	$$ = $3;
953192ba	512	$$->u.expression.post_op = AST_LINK_RARROW;
953192ba MD	513	$$->u.expression.prev = $1;
	514	}
	515	;
	516
	517	unary_expression
	518	: postfix_expression
	519	{ $$ = $1; }
bff988fa MD	520	\| primary_expression
bff988fa MD	521	{ $$ = $1; }
953192ba MD	522	\| unary_operator unary_expression
	523	{
	524	$$ = $1;
	525	$$->u.unary_op.child = $2;
	526	}
	527	;
	528
	529	unary_operator
	530	: PLUS
	531	{
	532	$$ = make_node(parser_ctx, NODE_UNARY_OP);
	533	$$->u.unary_op.type = AST_UNARY_PLUS;
	534	}
	535	\| MINUS
	536	{
	537	$$ = make_node(parser_ctx, NODE_UNARY_OP);
	538	$$->u.unary_op.type = AST_UNARY_MINUS;
	539	}
	540	\| NOT_OP
	541	{
	542	$$ = make_node(parser_ctx, NODE_UNARY_OP);
	543	$$->u.unary_op.type = AST_UNARY_NOT;
	544	}
ab78f161 CB	545	\| NOT_BIN
	546	{
	547	$$ = make_node(parser_ctx, NODE_UNARY_OP);
bff988fa	548	$$->u.unary_op.type = AST_UNARY_BIT_NOT;
ab78f161	549	}
953192ba MD	550	;
	551
	552	multiplicative_expression
	553	: unary_expression
	554	{ $$ = $1; }
	555	\| multiplicative_expression STAR unary_expression
	556	{
	557	$$ = make_op_node(parser_ctx, AST_OP_MUL, $1, $3);
	558	}
	559	\| multiplicative_expression DIV_OP unary_expression
	560	{
	561	$$ = make_op_node(parser_ctx, AST_OP_DIV, $1, $3);
	562	}
	563	\| multiplicative_expression MOD_OP unary_expression
	564	{
	565	$$ = make_op_node(parser_ctx, AST_OP_MOD, $1, $3);
	566	}
	567	;
	568
	569	additive_expression
	570	: multiplicative_expression
	571	{ $$ = $1; }
	572	\| additive_expression PLUS multiplicative_expression
	573	{
	574	$$ = make_op_node(parser_ctx, AST_OP_PLUS, $1, $3);
	575	}
	576	\| additive_expression MINUS multiplicative_expression
	577	{
	578	$$ = make_op_node(parser_ctx, AST_OP_MINUS, $1, $3);
	579	}
	580	;
	581
	582	shift_expression
	583	: additive_expression
	584	{ $$ = $1; }
	585	\| shift_expression LEFT_OP additive_expression
	586	{
116d3c01	587	$$ = make_op_node(parser_ctx, AST_OP_BIT_LSHIFT, $1, $3);
953192ba MD	588	}
	589	\| shift_expression RIGHT_OP additive_expression
	590	{
116d3c01	591	$$ = make_op_node(parser_ctx, AST_OP_BIT_RSHIFT, $1, $3);
953192ba MD	592	}
	593	;
	594
831b702b	595	and_expression
953192ba MD	596	: shift_expression
953192ba MD	597	{ $$ = $1; }
831b702b MD	598	\| and_expression AND_BIN shift_expression
	599	{
	600	$$ = make_op_node(parser_ctx, AST_OP_BIT_AND, $1, $3);
	601	}
	602	;
	603
	604	exclusive_or_expression
	605	: and_expression
	606	{ $$ = $1; }
	607	\| exclusive_or_expression XOR_BIN and_expression
	608	{
	609	$$ = make_op_node(parser_ctx, AST_OP_BIT_XOR, $1, $3);
	610	}
	611	;
	612
	613	inclusive_or_expression
	614	: exclusive_or_expression
	615	{ $$ = $1; }
	616	\| inclusive_or_expression OR_BIN exclusive_or_expression
	617	{
	618	$$ = make_op_node(parser_ctx, AST_OP_BIT_OR, $1, $3);
	619	}
	620	;
	621
	622	relational_expression
	623	: inclusive_or_expression
	624	{ $$ = $1; }
	625	\| relational_expression LT_OP inclusive_or_expression
953192ba MD	626	{
	627	$$ = make_op_node(parser_ctx, AST_OP_LT, $1, $3);
	628	}
831b702b	629	\| relational_expression GT_OP inclusive_or_expression
953192ba MD	630	{
	631	$$ = make_op_node(parser_ctx, AST_OP_GT, $1, $3);
	632	}
831b702b	633	\| relational_expression LE_OP inclusive_or_expression
953192ba MD	634	{
	635	$$ = make_op_node(parser_ctx, AST_OP_LE, $1, $3);
	636	}
831b702b	637	\| relational_expression GE_OP inclusive_or_expression
953192ba MD	638	{
	639	$$ = make_op_node(parser_ctx, AST_OP_GE, $1, $3);
	640	}
	641	;
	642
	643	equality_expression
	644	: relational_expression
	645	{ $$ = $1; }
	646	\| equality_expression EQ_OP relational_expression
	647	{
	648	$$ = make_op_node(parser_ctx, AST_OP_EQ, $1, $3);
	649	}
	650	\| equality_expression NE_OP relational_expression
	651	{
	652	$$ = make_op_node(parser_ctx, AST_OP_NE, $1, $3);
	653	}
	654	;
	655
953192ba	656	logical_and_expression
831b702b	657	: equality_expression
953192ba	658	{ $$ = $1; }
831b702b	659	\| logical_and_expression AND_OP equality_expression
953192ba MD	660	{
	661	$$ = make_op_node(parser_ctx, AST_OP_AND, $1, $3);
	662	}
	663	;
	664
	665	logical_or_expression
	666	: logical_and_expression
	667	{ $$ = $1; }
	668	\| logical_or_expression OR_OP logical_and_expression
	669	{
	670	$$ = make_op_node(parser_ctx, AST_OP_OR, $1, $3);
	671	}
	672	;
	673
	674	expression
	675	: logical_or_expression
	676	{ $$ = $1; }
	677	;
	678
	679	translation_unit
	680	: expression
	681	{
	682	parser_ctx->ast->root.u.root.child = $1;
	683	}
	684	;