mirror of
https://github.com/BlackLight/Snort_AIPreproc.git
synced 2024-11-23 20:25:12 +01:00
Circular buffer for pkt history & more improvements
This commit is contained in:
parent
dbba991c47
commit
bd19aa77c7
8 changed files with 162 additions and 52 deletions
16
README
16
README
|
@ -188,12 +188,14 @@ preprocessor ai: \
|
|||
database_parsing_interval 30 \
|
||||
hashtable_cleanup_interval 300 \
|
||||
manual_correlations_parsing_interval 120 \
|
||||
max_hash_pkt_number 1000 \
|
||||
neural_clustering_interval 1200 \
|
||||
neural_network_training_interval 43200 \
|
||||
neural_train_steps 10 \
|
||||
output_database ( type="dbtype", name="snort", user="snortusr", password="snortpass", host="dbhost" ) \
|
||||
output_neurons_per_side 20 \
|
||||
tcp_stream_expire_interval 300 \
|
||||
use_stream_hash_table 1 \
|
||||
webserv_banner "Snort AIPreprocessor module" \
|
||||
webserv_dir "/prefix/share/htdocs" \
|
||||
webserv_port 7654
|
||||
|
@ -323,7 +325,13 @@ the alerts from database and the next one (default if not specified: 30 seconds)
|
|||
|
||||
- hashtable_cleanup_interval: The interval that should occur from the cleanup of
|
||||
the hashtable of TCP streams and the next one (default if not specified: 300
|
||||
seconds)
|
||||
seconds). Set this option to 0 for performing no cleanup on the stream hash
|
||||
table
|
||||
|
||||
|
||||
- max_hash_pkt_number: Maximum number of packets that each element of the stream
|
||||
hash table should hold, set it to 0 for no limit (default value if not
|
||||
specified: 1000)
|
||||
|
||||
|
||||
- manual_correlations_parsing_interval: Interval in seconds between an execution
|
||||
|
@ -370,6 +378,12 @@ stream as "expired", if no more packets are received inside of that and it's not
|
|||
"marked" as suspicious (default if not specified: 300 seconds)
|
||||
|
||||
|
||||
- use_stream_hash_table: Set this option to 0 if you do not want to use the hash
|
||||
table for storing the streams of packets associated to alerts, this is a good
|
||||
choice on a system where many alerts are triggered (default value if not
|
||||
specified: 1)
|
||||
|
||||
|
||||
- webserv_banner: Banner of the web server, to be placed on the error pages and
|
||||
in the "Server" HTTP reply header
|
||||
|
||||
|
|
|
@ -80,8 +80,6 @@ AI_serializer ( AI_snort_alert *alert )
|
|||
void*
|
||||
AI_alerts_pool_thread ( void *arg )
|
||||
{
|
||||
pthread_t serializer_thread;
|
||||
|
||||
while ( 1 )
|
||||
{
|
||||
sleep ( config->alertSerializationInterval );
|
||||
|
@ -138,8 +136,6 @@ AI_file_alertparser_thread ( void* arg )
|
|||
BOOL in_alert = false;
|
||||
|
||||
pthread_t alerts_pool_thread;
|
||||
pthread_t serializer_thread;
|
||||
pthread_t db_thread;
|
||||
|
||||
/* Initialize the mutex lock, so nobody can read the alerts while we write there */
|
||||
pthread_mutex_init ( &alert_mutex, NULL );
|
||||
|
|
|
@ -262,7 +262,6 @@ __AI_merge_alerts ( AI_snort_alert **log )
|
|||
{
|
||||
AI_snort_alert *tmp, *tmp2, *tmp3;
|
||||
AI_alerts_couple *alerts_couple;
|
||||
pthread_t db_thread;
|
||||
int count = 0;
|
||||
|
||||
for ( tmp = *log; tmp; tmp = tmp->next )
|
||||
|
|
|
@ -1238,8 +1238,7 @@ AI_alert_correlation_thread ( void *arg )
|
|||
AI_snort_alert *alert_iterator = NULL,
|
||||
*alert_iterator2 = NULL;
|
||||
|
||||
pthread_t db_thread,
|
||||
manual_corr_thread;
|
||||
pthread_t manual_corr_thread;
|
||||
|
||||
#ifdef HAVE_LIBGVC
|
||||
char corr_png_file[4096] = { 0 };
|
||||
|
|
1
db.c
1
db.c
|
@ -45,7 +45,6 @@ AI_db_alertparser_thread ( void *arg )
|
|||
int latest_cid = 0;
|
||||
time_t latest_time = time ( NULL );
|
||||
pthread_t alerts_pool_thread;
|
||||
pthread_t serializer_thread;
|
||||
|
||||
DB_result res, res2;
|
||||
DB_row row, row2;
|
||||
|
|
44
spp_ai.c
44
spp_ai.c
|
@ -218,11 +218,13 @@ static AI_config * AI_parse(char *args)
|
|||
correlation_graph_interval = 0,
|
||||
database_parsing_interval = 0,
|
||||
manual_correlations_parsing_interval = 0,
|
||||
max_hash_pkt_number = 0,
|
||||
neural_clustering_interval = 0,
|
||||
neural_network_training_interval = 0,
|
||||
neural_train_steps = 0,
|
||||
output_neurons_per_side = 0,
|
||||
stream_expire_interval = 0,
|
||||
use_stream_hash_table = 0,
|
||||
webserv_banner_len = 0,
|
||||
webserv_dir_len = 0;
|
||||
|
||||
|
@ -591,6 +593,48 @@ static AI_config * AI_parse(char *args)
|
|||
config->neural_train_steps = neural_train_steps;
|
||||
_dpd.logMsg( " Neural train steps: %u\n", config->neural_train_steps );
|
||||
|
||||
/* Parsing the max_hash_pkt_number option */
|
||||
if (( arg = (char*) strcasestr( args, "max_hash_pkt_number" ) ))
|
||||
{
|
||||
for ( arg += strlen("max_hash_pkt_number");
|
||||
*arg && (*arg < '0' || *arg > '9');
|
||||
arg++ );
|
||||
|
||||
if ( !(*arg) )
|
||||
{
|
||||
AI_fatal_err ( "max_hash_pkt_number option used but "
|
||||
"no value specified", __FILE__, __LINE__ );
|
||||
}
|
||||
|
||||
max_hash_pkt_number = strtoul ( arg, NULL, 10 );
|
||||
} else {
|
||||
max_hash_pkt_number = DEFAULT_MAX_HASH_PKT_NUMBER;
|
||||
}
|
||||
|
||||
config->max_hash_pkt_number = max_hash_pkt_number;
|
||||
_dpd.logMsg( " Maximum number of packets stored in the hash table: %u\n", config->max_hash_pkt_number );
|
||||
|
||||
/* Parsing the use_stream_hash_table option */
|
||||
if (( arg = (char*) strcasestr( args, "use_stream_hash_table" ) ))
|
||||
{
|
||||
for ( arg += strlen("use_stream_hash_table");
|
||||
*arg && (*arg < '0' || *arg > '9');
|
||||
arg++ );
|
||||
|
||||
if ( !(*arg) )
|
||||
{
|
||||
AI_fatal_err ( "use_stream_hash_table option used but "
|
||||
"no value specified", __FILE__, __LINE__ );
|
||||
}
|
||||
|
||||
use_stream_hash_table = strtoul ( arg, NULL, 10 );
|
||||
} else {
|
||||
use_stream_hash_table = DEFAULT_USE_STREAM_HASH_TABLE;
|
||||
}
|
||||
|
||||
config->use_stream_hash_table = use_stream_hash_table;
|
||||
_dpd.logMsg( " Using the stream hash table: %u\n", config->use_stream_hash_table );
|
||||
|
||||
/* Parsing the alert_correlation_weight option */
|
||||
if (( arg = (char*) strcasestr( args, "alert_correlation_weight" ) ))
|
||||
{
|
||||
|
|
17
spp_ai.h
17
spp_ai.h
|
@ -98,11 +98,18 @@
|
|||
/** Default number of steps used for training the neural network */
|
||||
#define DEFAULT_NEURAL_TRAIN_STEPS 10
|
||||
|
||||
/** Default maximum number of packets that an observed stream in the hash table should hold */
|
||||
#define DEFAULT_MAX_HASH_PKT_NUMBER 1000
|
||||
|
||||
/** Default number of alerts needed in the history file or database for letting a certain
|
||||
* heuristic correlation index weight be =~ 0.95 (the weight monotonically increases
|
||||
* with the number of alerts according to a hyperbolic tangent function) */
|
||||
#define DEFAULT_ALERT_CORRELATION_WEIGHT 5000
|
||||
|
||||
/** Default setting for the use of the hash table for holding streams of packets
|
||||
* associated to a certain alert (0 = do not use, 1 or any value != 0: use) */
|
||||
#define DEFAULT_USE_STREAM_HASH_TABLE 1
|
||||
|
||||
/** Default web server port */
|
||||
#define DEFAULT_WEBSERV_PORT 7654
|
||||
|
||||
|
@ -164,6 +171,9 @@ struct pkt_info
|
|||
/** Flag set if the packet is observed, i.e. associated to a security alert */
|
||||
BOOL observed;
|
||||
|
||||
/** Number of packets in the current flow, if available */
|
||||
unsigned int n_packets;
|
||||
|
||||
/** Make the struct 'hashable' */
|
||||
UT_hash_handle hh;
|
||||
};
|
||||
|
@ -218,12 +228,19 @@ typedef struct
|
|||
* with the number of alerts according to a hyperbolic tangent function) */
|
||||
unsigned long alert_correlation_weight;
|
||||
|
||||
/** Maximum number of packets that an observed stream in the hash table should hold */
|
||||
unsigned long max_hash_pkt_number;
|
||||
|
||||
/** Number of steps used for training the neural network */
|
||||
unsigned long neural_train_steps;
|
||||
|
||||
/** Size of the alerts' buffer to be periodically sent to the serialization thread */
|
||||
unsigned long alert_bufsize;
|
||||
|
||||
/** Setting for the use of the hash table for holding streams of packets
|
||||
* associated to a certain alert (0 = do not use, 1 or any value != 0: use) */
|
||||
unsigned long use_stream_hash_table;
|
||||
|
||||
/** Correlation threshold coefficient for correlating two hyperalerts. Two hyperalerts
|
||||
* are 'correlated' to each other in a multi-step attack graph if and only if their
|
||||
* correlation value is >= m + ks, where m is the average correlation coefficient,
|
||||
|
|
102
stream.c
102
stream.c
|
@ -98,8 +98,16 @@ __AI_stream_free ( struct pkt_info* stream )
|
|||
void*
|
||||
AI_hashcleanup_thread ( void* arg )
|
||||
{
|
||||
struct pkt_info *h, *stream;
|
||||
struct pkt_info *h, *stream, *tmp;
|
||||
time_t max_timestamp;
|
||||
int pkt_count, pkt_rm;
|
||||
BOOL has_old_streams;
|
||||
|
||||
if ( config->hashCleanupInterval == 0 )
|
||||
{
|
||||
pthread_exit ((void*) 0);
|
||||
return (void*) 0;
|
||||
}
|
||||
|
||||
while ( 1 ) {
|
||||
/* Sleep for the specified number of seconds */
|
||||
|
@ -109,9 +117,51 @@ AI_hashcleanup_thread ( void* arg )
|
|||
if ( !hash || !HASH_COUNT(hash) )
|
||||
continue;
|
||||
|
||||
has_old_streams = true;
|
||||
|
||||
while ( has_old_streams )
|
||||
{
|
||||
has_old_streams = false;
|
||||
|
||||
/* Check all the streams in the hash */
|
||||
for ( h = hash; h; h = (struct pkt_info*) h->next ) {
|
||||
if ( h->observed ) continue;
|
||||
if ( h->observed )
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
if ( h->next )
|
||||
{
|
||||
if ( h->next->observed )
|
||||
{
|
||||
if ( config->max_hash_pkt_number != 0 )
|
||||
{
|
||||
if ( h->next->n_packets == 0 )
|
||||
{
|
||||
for ( stream = h->next, pkt_count=0; stream; stream = (struct pkt_info*) stream->next, pkt_count++ );
|
||||
h->next->n_packets = pkt_count;
|
||||
} else {
|
||||
pkt_count = h->next->n_packets;
|
||||
}
|
||||
|
||||
/* If this stream has too many packets inside, remove the oldest ones */
|
||||
if ( pkt_count > config->max_hash_pkt_number )
|
||||
{
|
||||
for ( stream = h->next, pkt_rm = 0;
|
||||
stream && pkt_rm < pkt_count - config->max_hash_pkt_number;
|
||||
stream = stream->next, pkt_rm++ )
|
||||
{
|
||||
tmp = stream->next;
|
||||
__AI_stream_free ( stream );
|
||||
stream = tmp;
|
||||
}
|
||||
|
||||
h->next = stream;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
max_timestamp = 0;
|
||||
|
||||
/* Find the maximum timestamp in the flow */
|
||||
|
@ -121,19 +171,15 @@ AI_hashcleanup_thread ( void* arg )
|
|||
}
|
||||
|
||||
/* If the most recent packet in the stream is older than the specified threshold, remove that stream */
|
||||
if ( time(NULL) - max_timestamp > config->streamExpireInterval ) {
|
||||
|
||||
if ( time (NULL) - max_timestamp > config->streamExpireInterval ) {
|
||||
has_old_streams = true;
|
||||
stream = h;
|
||||
|
||||
if ( stream )
|
||||
{
|
||||
/* XXX This, sometimes, randomly, leads to the crash of the module.
|
||||
* WHY??? Why can't computer science be deterministic, and if a
|
||||
* certain not-NULL stream exists in a not-NULL hash table why
|
||||
* should there be a crash, one day yes and the next one no? Until
|
||||
* I won't find an answer to these enigmatic questions, I will leave
|
||||
* this code commented, so if a certain stream goes timeout it won't
|
||||
* be removed. I'm sorry but it's not my fault. Ask the karma about this */
|
||||
/* __AI_stream_free ( stream ); */
|
||||
__AI_stream_free ( stream );
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
@ -163,12 +209,18 @@ AI_pkt_enqueue ( SFSnortPacket* pkt )
|
|||
if ( start_time == 0 )
|
||||
start_time = time (NULL);
|
||||
|
||||
/* If we are not using the stream hash table, just return */
|
||||
if ( config->use_stream_hash_table == 0 )
|
||||
return;
|
||||
|
||||
/* If this is not an IP and/or TCP packet, it's not for me */
|
||||
if ( !( pkt->ip4_header && pkt->tcp_header ))
|
||||
return;
|
||||
|
||||
if ( !( info = (struct pkt_info*) malloc( sizeof(struct pkt_info) )) )
|
||||
{
|
||||
AI_fatal_err ( "Fatal dynamic memory allocation error", __FILE__, __LINE__ );
|
||||
}
|
||||
|
||||
memset ( &key, 0, sizeof(struct pkt_key));
|
||||
key.src_ip = pkt->ip4_header->source.s_addr;
|
||||
|
@ -177,10 +229,13 @@ AI_pkt_enqueue ( SFSnortPacket* pkt )
|
|||
info->key = key;
|
||||
info->timestamp = time(NULL);
|
||||
info->observed = false;
|
||||
info->n_packets = 0;
|
||||
info->next = NULL;
|
||||
|
||||
if ( !( info->pkt = (SFSnortPacket*) malloc ( sizeof (SFSnortPacket) )) )
|
||||
{
|
||||
AI_fatal_err ( "Fatal dynamic memory allocation error", __FILE__, __LINE__ );
|
||||
}
|
||||
|
||||
memcpy ( info->pkt, pkt, sizeof (SFSnortPacket) );
|
||||
|
||||
|
@ -193,8 +248,12 @@ AI_pkt_enqueue ( SFSnortPacket* pkt )
|
|||
/* If there is already an element of this traffic stream in my hash table,
|
||||
* append the packet just received to this stream*/
|
||||
if ( found ) {
|
||||
/* If the current packet contains a RST, just deallocate the stream */
|
||||
if ( info->pkt->tcp_header->flags & TCPHEADER_RST ) {
|
||||
/* If the current packet contains a RST or a FIN, just deallocate the stream */
|
||||
if (
|
||||
( info->pkt->tcp_header->flags & TCPHEADER_RST ) ||
|
||||
(( info->pkt->tcp_header->flags & TCPHEADER_FIN ) &&
|
||||
( info->pkt->tcp_header->flags & TCPHEADER_ACK ))
|
||||
) {
|
||||
pthread_mutex_lock ( &hash_mutex );
|
||||
HASH_FIND ( hh, hash, &key, sizeof(struct pkt_key), found );
|
||||
pthread_mutex_unlock ( &hash_mutex );
|
||||
|
@ -227,23 +286,6 @@ AI_pkt_enqueue ( SFSnortPacket* pkt )
|
|||
if ( !tmp ) {
|
||||
found->next = info;
|
||||
}
|
||||
|
||||
/* If the current packet contains an ACK and the latest one
|
||||
* in this stream contained a FIN, then the communication
|
||||
* on this stream is over */
|
||||
if ( found->pkt->tcp_header->flags & TCPHEADER_FIN ) {
|
||||
if ( info->pkt->tcp_header->flags & TCPHEADER_ACK ) {
|
||||
pthread_mutex_unlock ( &hash_mutex );
|
||||
HASH_FIND ( hh, hash, &key, sizeof(struct pkt_key), found );
|
||||
pthread_mutex_unlock ( &hash_mutex );
|
||||
|
||||
if ( found ) {
|
||||
if ( !found->observed ) {
|
||||
__AI_stream_free ( found );
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
} else {
|
||||
/* If the packet contains the ACK flag, no payload and it is
|
||||
|
|
Loading…
Reference in a new issue