xstrstr: speed up analysis by avoiding string comparisons

With large input one should see cut 15% for overall run time.

Signed-off-by: Sami Kerola <kerolasa@iki.fi>

src/defaults.h

@@ -52,32 +52,4 @@ static const size_t MAXLEN = 1024;
  * \brief Maximum number of different shared networks in dhcpd.conf file. */
 static const unsigned int SHARED_NETWORKS = 8192;
 
-/*! \var prefixes[2][NUM_OF_PREFIX]
- * \brief ISC lease file formats for IPv4 and IPv6.
- *
- * The .indent.pro in use will mess formatting of array below.
- * Please do not commit less readable indentation. */
-static const char *prefixes[2][NUM_OF_PREFIX] = {
-	[VERSION_4] = {
-		       [PREFIX_LEASE]                   = "lease ",
-		       [PREFIX_BINDING_STATE_FREE]      = "  binding state free",
-		       [PREFIX_BINDING_STATE_ABANDONED] = "  binding state abandoned",
-		       [PREFIX_BINDING_STATE_EXPIRED]   = "  binding state expired",
-		       [PREFIX_BINDING_STATE_RELEASED]  = "  binding state released",
-		       [PREFIX_BINDING_STATE_ACTIVE]    = "  binding state active",
-		       [PREFIX_BINDING_STATE_BACKUP]    = "  binding state backup",
-		       [PREFIX_HARDWARE_ETHERNET]       = "  hardware ethernet"
-	},
-	[VERSION_6] = {
-		       [PREFIX_LEASE]                   = "  iaaddr ",
-		       [PREFIX_BINDING_STATE_FREE]      = "    binding state free",
-		       [PREFIX_BINDING_STATE_ABANDONED] = "    binding state abandoned",
-		       [PREFIX_BINDING_STATE_EXPIRED]   = "    binding state expired",
-		       [PREFIX_BINDING_STATE_RELEASED]  = "    binding state released",
-		       [PREFIX_BINDING_STATE_ACTIVE]    = "    binding state active",
-		       [PREFIX_BINDING_STATE_BACKUP]    = "    binding state backup",
-		       [PREFIX_HARDWARE_ETHERNET]       = "    hardware ethernet"
-	}
-};
-
 #endif				/* DEFAULTS_H */

src/dhcpd-pools.c

@@ -272,13 +272,6 @@ int main(int argc, char **argv)
  * FIXME: This function should return void. */
 int prepare_memory(void)
 {
-	/* Fill in prefix length cache */
-	int i, j;
-	for (i = 0; i < 2; i++) {
-		for (j = 0; j < NUM_OF_PREFIX; j++) {
-			prefix_length[i][j] = strlen(prefixes[i][j]);
-		}
-	}
 	config.dhcp_version = VERSION_UNKNOWN;
 	RANGES = 64;
 	num_ranges = num_shared_networks = 0;

src/dhcpd-pools.h

@@ -210,8 +210,7 @@ void copy_ipaddr(union ipaddr_t *restrict dst,
 		 const union ipaddr_t *restrict src);
 const char *ntop_ipaddr(const union ipaddr_t *ip);
 double get_range_size(const struct range_t *r);
-int xstrstr(const char *__restrict a, const char *__restrict b, int len)
+int xstrstr(const char *__restrict str)
-    __attribute__ ((nonnull(1, 2)))
 # if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)
     __attribute__ ((__hot__))
 # endif

src/getdata.c

@@ -101,62 +101,52 @@ int parse_leases(void)
 		ethernets = true;
 	}
 
-	const char **p = prefixes[config.dhcp_version];
-	int *l = prefix_length[config.dhcp_version];
-
-/*! \def HAS_PREFIX(line, type)
- * \brief A macro to match IPv4 and IPv6 lease lines.
- *
- * FIXME: This macro should have better name. The HAS_PREFIX sounds like
- * some sort of prefix length test. */
-#define HAS_PREFIX(line, type) xstrstr((line), p[type], l[type])
-
 	while (!feof(dhcpd_leases)) {
 		if (!fgets(line, MAXLEN, dhcpd_leases) && ferror(dhcpd_leases)) {
 			err(EXIT_FAILURE, "parse_leases: %s",
 			    config.dhcpdlease_file);
 		}
+		switch(xstrstr(line)) {
 		/* It's a lease, save IP */
-		if (HAS_PREFIX(line, PREFIX_LEASE)) {
+		case PREFIX_LEASE:
-			nth_field(ipstring, line + l[PREFIX_LEASE]);
+			nth_field(ipstring, line + (config.dhcp_version == VERSION_4 ? 6 : 9));
 			parse_ipaddr(ipstring, &addr);
-			continue;
+			break;
-		}
+		case PREFIX_BINDING_STATE_FREE:
-		if (HAS_PREFIX(line, PREFIX_BINDING_STATE_FREE) ||
+		case PREFIX_BINDING_STATE_ABANDONED:
-		    HAS_PREFIX(line, PREFIX_BINDING_STATE_ABANDONED) ||
+		case PREFIX_BINDING_STATE_EXPIRED:
-		    HAS_PREFIX(line, PREFIX_BINDING_STATE_EXPIRED) ||
+		case PREFIX_BINDING_STATE_RELEASED:
-		    HAS_PREFIX(line, PREFIX_BINDING_STATE_RELEASED)) {
-			/* remove old entry, if exists */
 			if ((lease = find_lease(&addr)) != NULL) {
 				delete_lease(lease);
 			}
 			add_lease(&addr, FREE);
-			continue;
+			break;
-		}
+		case PREFIX_BINDING_STATE_ACTIVE:
-		/* Copy IP to correct array */
-		if (HAS_PREFIX(line, PREFIX_BINDING_STATE_ACTIVE)) {
 			/* remove old entry, if exists */
 			if ((lease = find_lease(&addr)) != NULL) {
 				delete_lease(lease);
 			}
 			add_lease(&addr, ACTIVE);
-			continue;
+			break;
-		}
+		case PREFIX_BINDING_STATE_BACKUP:
-		if (HAS_PREFIX(line, PREFIX_BINDING_STATE_BACKUP)) {
 			/* remove old entry, if exists */
 			if ((lease = find_lease(&addr)) != NULL) {
 				delete_lease(lease);
 			}
 			add_lease(&addr, BACKUP);
 			config.backups_found = true;
-			continue;
+			break;
-		}
+		case PREFIX_HARDWARE_ETHERNET:
-		if (ethernets && (xstrstr(line, "  hardware ethernet", 19))) {
+			if (ethernets == false)
+				break;
 			nth_field(macstring, line + 20);
 			macstring[17] = '\0';
 			if ((lease = find_lease(&addr)) != NULL) {
 				lease->ethernet = xstrdup(macstring);
 			}
+			break;
+		default:
+			/* do nothing */;
 		}
 	}
 #undef HAS_PREFIX

src/other.c

@@ -148,53 +148,103 @@ double get_range_size(const struct range_t *r)
 	}
 }
 
-/*! \fn xstrstr(const char *restrict a, const char *restrict b, const int len)
+/*! \fn xstrstr(const char *restrict str)
- * \brief Compare two strings.  Similar to strcmp, but tuned to be
+ * \brief Categorize dhcpd.leases line.
- * quicker which is possible because input data is known to have certain
- * structure.
  *
- * \param a String which is been compared, e.g., a haystack.
+ * \param str A line from dhcpd.conf
- * \param b Constant string which is hoped to found, e.g., a needle.
+ * \return prefix_t enum value
- * \param len Stop point in characters when comparison must be ended.
+ */
- * \return Zero if strings differ, one if they are the same. */
 int
 #if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)
     __attribute__ ((hot))
 #endif
-    xstrstr(const char *restrict a, const char *restrict b, const int len)
+    xstrstr(const char *restrict str)
 {
-	int i;
+	size_t len;
-
 	/* Needed when dhcpd.conf has zero range definitions.  */
 	if (config.dhcp_version == VERSION_UNKNOWN) {
-		if (!strcmp(prefixes[VERSION_4][PREFIX_LEASE], a)) {
+		if (strncmp("lease ", str, 6)) {
 			config.dhcp_version = VERSION_4;
-			return true;
+			return PREFIX_LEASE;
-		}
+		} else if (strncmp("  iaaddr ", str, 9)) {
-		if (!strcmp(prefixes[VERSION_6][PREFIX_LEASE], a)) {
 			config.dhcp_version = VERSION_6;
-			return true;
+			return PREFIX_LEASE;
 		}
-		return false;
+		return NUM_OF_PREFIX;
 	}
-
+	if ((config.dhcp_version == VERSION_4 && str[2] == 'b')
-	/* two spaces are very common in lease file, after them
+	    || (config.dhcp_version == VERSION_6 && str[4] == 'b')
-	 * nearly everything differs */
+	    || str[2] == 'h') {
-	if (likely(a[2] != b[2])) {
+		len = strlen(str);
-		return false;
+	} else {
+		len = 0;
 	}
-	/* "  binding state " == 16 chars, this will skip right
+	if (15 < len && config.dhcp_version == VERSION_4) {
-	 * to first differing line. */
+		switch (str[16]) {
-	if (17 < len && a[17] != b[17]) {
+		case 'f':
-		return false;
+			if (!strncmp("  binding state free;", str, 21))
-	}
+				return PREFIX_BINDING_STATE_FREE;
-	/* looking good, double check the whole thing... */
+			break;
-	for (i = 0; a[i] != '\0' && b[i] != '\0'; i++) {
+		case 'a':
-		if (a[i] != b[i]) {
+			if (!strncmp("  binding state active;", str, 23))
-			return false;
+				return PREFIX_BINDING_STATE_ACTIVE;
+			if (!strncmp("  binding state abandoned;", str, 25))
+				return PREFIX_BINDING_STATE_ABANDONED;
+			break;
+		case 'e':
+			if (!strncmp("  binding state expired;", str, 24))
+				return PREFIX_BINDING_STATE_EXPIRED;
+			break;
+		case 'r':
+			if (!strncmp("  binding state released;", str, 25))
+				return PREFIX_BINDING_STATE_RELEASED;
+			break;
+		case 'b':
+			if (!strncmp("  binding state backup;", str, 23))
+				return PREFIX_BINDING_STATE_BACKUP;
+			break;
+		case 'n':
+			if (!strncmp("  hardware ethernet", str, 19))
+				return PREFIX_HARDWARE_ETHERNET;
+			break;
+		}
+	} else if (17 < len /* && config.dhcp_version == VERSION_6 */ ) {
+		switch (str[18]) {
+		case 'f':
+			if (!strncmp("    binding state free;", str, 23))
+				return PREFIX_BINDING_STATE_FREE;
+			break;
+		case 'a':
+			if (!strncmp("    binding state active;", str, 25))
+				return PREFIX_BINDING_STATE_ACTIVE;
+			if (!strncmp("    binding state abandoned;", str, 27))
+				return PREFIX_BINDING_STATE_ABANDONED;
+			break;
+		case 'e':
+			if (!strncmp("    binding state expired;", str, 26))
+				return PREFIX_BINDING_STATE_EXPIRED;
+			break;
+		case 'r':
+			if (!strncmp("    binding state released;", str, 27))
+				return PREFIX_BINDING_STATE_RELEASED;
+			break;
+		case 'b':
+			if (!strncmp("    binding state backup;", str, 25))
+				return PREFIX_BINDING_STATE_BACKUP;
+			break;
+		case 'n':
+			if (!strncmp("  hardware ethernet", str, 19))
+				return PREFIX_HARDWARE_ETHERNET;
+			break;
 		}
 	}
-	return true;
+	if (config.dhcp_version == VERSION_4 && !strncmp("lease ", str, 6)) {
+		return PREFIX_LEASE;
+	} else if (config.dhcp_version == VERSION_6
+		   && !strncmp("  iaaddr ", str, 9)) {
+		return PREFIX_LEASE;
+	}
+	return NUM_OF_PREFIX;
 }
 
 /*! \brief Return a double floating point value.