[libvalor] [dbgen]: add warning to dbgen about small capacity, add mo…

…re checks and asserts for memory-related issues, fix memory leak database_read function

dbgen/src/dbgen.c

@@ -18,7 +18,7 @@ int main(int argc, const char *argv[]) {
     arguments_set_usage("--db FILE [-h|--help] [--check-entry|--increment-version|--add-name NAME|--check-name NAME|--add-threat FILE|--check-threat FILE|--set-version VERSION]");
     argument_add_compulsory("--db", "Path to database file", ARG_STR);
     argument_add("--capacity", "Capacity to pre-allocate in database structures", ARG_INT,
-                 (argvalue)(int64_t)1, true, false);
+                 (argvalue)(int64_t)128, true, false);
     argument_add("--filename", "Threat file", ARG_STR, (argvalue) NULL,
                  false, false);
     argument_add("--check-entry", "Check whether file in database and exit.", ARG_BOOL,
@@ -62,6 +62,9 @@ int main(int argc, const char *argv[]) {
         if(capacity < 0){
             die("Invalid capacity");
         }
+	if(capacity < 16){
+	    warning("Capacity is too small. This may lead to performance and stability issues");
+	}
         db = create_database(capacity);
         fclose(db_file);
     }

libvalor/src/db/db.c

@@ -60,6 +60,7 @@ void database_add_name(database_t *db, const char *name) {
 }
 
 bool database_check_name(database_t *db, const char *name) {
+    assert(db);
     if(name == NULL){
         fprintf(stderr, "Name passed to database_check_name is NULL!");
         return false;
@@ -69,7 +70,6 @@ bool database_check_name(database_t *db, const char *name) {
 
 void database_save(FILE *file, database_t *db) {
     size_t i = 0;
-    //printf("database_save: entered\n");
     // Write metadata
     fwrite(&db->version, sizeof(db->version), 1, file);
     fwrite(&db->modulo, sizeof(db->modulo), 1, file);
@@ -82,10 +82,8 @@ void database_save(FILE *file, database_t *db) {
         fwrite(str, sizeof(char), len, file);
     }
     // Write hashtree
-    //printf("Writing hashtree\n");
     uint64_t tree_sz = tree_size(db->hash_tree);
     uint8_t* serialized_tree = serialize_ed_tree(db->hash_tree);
-    //printf("Serialize OK\n");
     fwrite(serialized_tree, sizeof(uint8_t) * tree_sz, 1, file);
     // Clean-up
     free(serialized_tree);
@@ -111,6 +109,7 @@ void database_read(FILE *file, database_t *db) {
         str[len] = (char) 0;
         fread(str, sizeof(char), len, file);
         database_add_name(db, str);
+	free(str); // String copied in add_name, so we do not need anymore
     }
     // Read hashtree
     uint8_t* serialized_tree = (uint8_t*)malloc(sizeof(uint8_t) * tree_size_for_depth(FUZZY_MAX_RESULT));
@@ -122,4 +121,4 @@ database_t *database_from_file(FILE *file) {
     database_t *db = (database_t *) malloc(sizeof(database_t));
     database_read(file, db);
     return db;
-}
+}

libvalor/src/hashset.c

@@ -4,7 +4,7 @@
 #include <string.h>
 #include <assert.h>
 
-unsigned long hash_djb2(const unsigned char *str) //djb2 hash by Dan Bernstein
+static unsigned long hash_djb2(const unsigned char *str) //djb2 hash by Dan Bernstein
 {
     unsigned long hash = 5381;
     int c;
@@ -104,6 +104,8 @@ stringset_t* create_stringset(uint32_t capacity){
 
 
 void stringset_add(stringset_t* set, const char* str){
+    assert(set);
+    assert(str);
     uint32_t hash = hash_djb2((unsigned char*) str);
     assert(set->capacity > 0);
     uint32_t index = hash % set->capacity;
@@ -126,6 +128,8 @@ void stringset_add(stringset_t* set, const char* str){
 }
 
 bool stringset_check(stringset_t* set, const char* str){
+    assert(set);
+    assert(str);
     uint32_t hash = hash_djb2((unsigned char*) str);
     uint32_t index = hash % set->capacity;
     stringset_node_t* current_node = set->node_table[index];
@@ -153,4 +157,4 @@ void destroy_stringset(stringset_t * set){
         }
     }
     free(set);
-}
+}