Implemented regex for locations and httptables

Made regex configuration same way as in Nginx.

Author: supervisor

Makefile

@@ -146,7 +146,7 @@ KERNEL = /lib/modules/$(shell uname -r)/build
 
 export KERNEL TFW_CFLAGS AVX2 BMI2 ADX TFW_GCOV
 
-obj-m	+= lib/ db/core/ fw/ tls/
+obj-m	+= lib/ db/core/ regex/ fw/ tls/
 
 all: build
 

fw/cfg.c

@@ -109,6 +109,9 @@
  * them. Helpers below facilitate that.
  */
 
+unsigned short number_of_regex = 0;
+unsigned short number_of_db_regex = 0;
+
 static const char *
 __alloc_and_copy_literal(const char *src, size_t len, bool keep_bs)
 {
@@ -397,6 +400,9 @@ typedef enum {
 	TOKEN_SEMICOLON,
 	TOKEN_LITERAL,
 	TOKEN_ARROW,
+	TOKEN_TILDA,
+	TOKEN_REGEX,
+	TOKEN_REGEX_CI,
 	_TOKEN_COUNT,
 } token_t;
 
@@ -588,9 +594,12 @@ read_next_token(TfwCfgParserState *ps)
 				    TOKEN_NEQSIGN);
 		TFSM_COND_MOVE_EXIT(ps->c == '>' && ps->prev_c == '-',
 				    TOKEN_ARROW);
+		TFSM_COND_MOVE_EXIT(ps->c == '*' && ps->prev_c == '~',
+		                    TOKEN_REGEX_CI);
 
 		/* Special case to differ single equal sign from double one. */
 		TFSM_COND_MOVE(ps->c == '=', TS_EQSIGN);
+		TFSM_COND_MOVE(ps->c == '~', TS_TILDA);
 
 		/* Everything else is not a special character and therefore
 		 * it starts a literal. */
@@ -619,6 +628,14 @@ read_next_token(TfwCfgParserState *ps)
 		TFSM_JMP_EXIT(TOKEN_EQSIGN);
 	}
 
+	FSM_STATE(TS_TILDA) {
+		TFSM_COND_JMP_EXIT(!ps->c, TOKEN_REGEX);
+
+		/* If this is double equal sign, eat second sign and exit. */
+		TFSM_COND_MOVE_EXIT(ps->c == '*', TOKEN_REGEX_CI);
+		TFSM_JMP_EXIT(TOKEN_REGEX);
+	}
+
 	FSM_STATE(TS_COMMENT) {
 		TFSM_COND_JMP_EXIT(!ps->c, TOKEN_NA);
 
@@ -732,7 +749,21 @@ entry_set_cond(TfwCfgEntry *e, token_t cond_type, const char *src, int len)
 	if (!(e->name = alloc_and_copy_literal(name, name_len)))
 		return -ENOMEM;
 
-	rule->inv = cond_type == TOKEN_DEQSIGN ? false : true;
+	switch (cond_type) {
+	case TOKEN_REGEX:
+		rule->regex = TFW_REGEX_REGULAR;
+		rule->inv = false;
+		break;
+	case TOKEN_REGEX_CI:
+		rule->regex = TFW_REGEX_CI;
+		rule->inv = false;
+		break;
+	default:
+		rule->regex = TFW_REGEX_NO;
+		rule->inv = cond_type == TOKEN_DEQSIGN ? false : true;
+		break;
+	}
+
 	return 0;
 }
 
@@ -806,8 +837,10 @@ parse_cfg_entry(TfwCfgParserState *ps)
 
 	FSM_STATE(PS_PLAIN_OR_RULE) {
 		PFSM_COND_MOVE(ps->t == TOKEN_DEQSIGN ||
-			       ps->t == TOKEN_NEQSIGN,
-			       PS_RULE_COND);
+		               ps->t == TOKEN_NEQSIGN ||
+		               ps->t == TOKEN_REGEX ||
+		               ps->t == TOKEN_REGEX_CI,
+		               PS_RULE_COND);
 		PFSM_COND_MOVE(ps->t == TOKEN_LITERAL, PS_PLAIN_OR_LONG_RULE);
 
 		/* Jump to plain val/attr scheme to make remained checks
@@ -819,35 +852,41 @@ parse_cfg_entry(TfwCfgParserState *ps)
 
 	FSM_STATE(PS_PLAIN_OR_LONG_RULE) {
 		FSM_COND_JMP(ps->t == TOKEN_DEQSIGN ||
-			     ps->t == TOKEN_NEQSIGN,
-			     PS_LONG_RULE_COND);
+		             ps->t == TOKEN_NEQSIGN ||
+		             ps->t == TOKEN_REGEX ||
+		             ps->t == TOKEN_REGEX_CI,
+		             PS_LONG_RULE_COND);
 
 		/* This is not rule (simple or extended), so jump to
 		 * plain val/attr scheme. */
 		ps->err = entry_set_name(&ps->e);
 		FSM_COND_JMP(ps->err, PS_EXIT);
 		FSM_COND_JMP(ps->t == TOKEN_EQSIGN, PS_STORE_ATTR_PREV);
 		FSM_COND_JMP(ps->t == TOKEN_LITERAL ||
-			     ps->t == TOKEN_SEMICOLON ||
-			     ps->t == TOKEN_LBRACE,
-			     PS_STORE_VAL_PREV);
+		             ps->t == TOKEN_SEMICOLON ||
+		             ps->t == TOKEN_LBRACE,
+		             PS_STORE_VAL_PREV);
 
 		ps->err = -EINVAL;
 		FSM_JMP(PS_EXIT);
 	}
 
 	FSM_STATE(PS_LONG_RULE_COND) {
 		ps->err = entry_add_rule_param(&ps->e.rule.fst_ext,
-					       ps->prev_lit,
-					       ps->prev_lit_len);
+		                               ps->prev_lit,
+		                               ps->prev_lit_len);
 		FSM_COND_JMP(ps->err, PS_EXIT);
 		PFSM_MOVE(PS_RULE_COND);
 	}
 
 	FSM_STATE(PS_RULE_COND) {
+		FSM_COND_JMP(ps->prev_t == TOKEN_REGEX ||
+		             ps->prev_t == TOKEN_REGEX_CI,
+		             PS_STORE_VAL_PREV_REGEX);
+
 		PFSM_COND_JMP_EXIT_ERROR(ps->t != TOKEN_LITERAL);
 		ps->err = entry_set_cond(&ps->e, ps->prev_t, ps->lit,
-					 ps->lit_len);
+		                         ps->lit_len);
 		FSM_COND_JMP(ps->err, PS_EXIT);
 		PFSM_MOVE(PS_RULE_COND_END);
 	}
@@ -866,7 +905,7 @@ parse_cfg_entry(TfwCfgParserState *ps)
 	FSM_STATE(PS_RULE_ACTION) {
 		PFSM_COND_JMP_EXIT_ERROR(ps->t != TOKEN_LITERAL);
 		ps->err = entry_add_rule_param(&ps->e.rule.act, ps->lit,
-					       ps->lit_len);
+		                               ps->lit_len);
 		FSM_COND_JMP(ps->err, PS_EXIT);
 		PFSM_MOVE(PS_RULE_ACTION_VAL);
 	}
@@ -878,7 +917,7 @@ parse_cfg_entry(TfwCfgParserState *ps)
 		PFSM_COND_JMP_EXIT_ERROR(ps->t != TOKEN_LITERAL);
 
 		ps->err = entry_add_rule_param(&ps->e.rule.val, ps->lit,
-					       ps->lit_len);
+		                               ps->lit_len);
 		FSM_COND_JMP(ps->err, PS_EXIT);
 
 		read_next_token(ps);
@@ -914,6 +953,38 @@ parse_cfg_entry(TfwCfgParserState *ps)
 		FSM_JMP(PS_STORE_VAL_PREV);
 	}
 
+	FSM_STATE(PS_STORE_VAL_PREV_REGEX) {
+		/* name val1 val2;
+		 *           ^
+		 *           We are here (but still need to store val1)
+		 *           and name or condition.
+		*/
+		T_DBG3("add value: %.*s\n", ps->prev_lit_len, ps->prev_lit);
+
+		if (ps->e.ftoken && !strcmp(ps->e.ftoken, "location")) {
+			ps->err = entry_set_name(&ps->e);
+
+			if (!ps->err) {
+				if (ps->prev_t == TOKEN_REGEX)
+					ps->err = entry_add_val(&ps->e, "regex",
+					                        sizeof("regex"));
+				if (ps->prev_t == TOKEN_REGEX_CI)
+					ps->err = entry_add_val(&ps->e,
+					                        "regex_ci",
+					                        sizeof("regex_ci"));
+			}
+			FSM_COND_JMP(ps->err, PS_EXIT);
+			FSM_JMP(PS_VAL_OR_ATTR);
+		}
+
+		/*If it is not location*/
+		ps->err = entry_set_cond(&ps->e, ps->prev_t,
+		                                         ps->lit, ps->lit_len);
+		FSM_COND_JMP(ps->err, PS_EXIT);
+		PFSM_MOVE(PS_RULE_COND_END);
+
+	}
+
 	FSM_STATE(PS_STORE_VAL_PREV) {
 		/* name val1 val2;
 		 *           ^

fw/cfg.h

@@ -149,6 +149,7 @@ typedef struct {
 	const char *act;
 	const char *val;
 	bool inv;
+	int regex;
 } TfwCfgRule;
 
 typedef struct {
@@ -336,6 +337,9 @@ struct TfwCfgSpec {
 	void (*cleanup)(TfwCfgSpec *self);
 };
 
+extern unsigned short number_of_regex;
+extern unsigned short number_of_db_regex;
+
 /**
  * Walks over a NULL-terminated array of TfwCfgSpec structures.
  */
@@ -412,6 +416,12 @@ enum {
 	TFW_CFG_B_KEEP,		/* Keep an entry */
 };
 
+enum {
+	TFW_REGEX_NO = 0,
+	TFW_REGEX_REGULAR,
+	TFW_REGEX_CI,
+};
+
 #define TFW_CFG_F_ADD		(1 << TFW_CFG_B_ADD)
 #define TFW_CFG_F_DEL		(1 << TFW_CFG_B_DEL)
 #define TFW_CFG_F_MOD		(1 << TFW_CFG_B_MOD)