added a mmap based ipc feature to prevent duplicates

configure

@@ -621,6 +621,7 @@ ac_includes_default="\
 
 ac_subst_vars='LTLIBOBJS
 LIBOBJS
+DATAROOTDIR
 LIBEXECDIR
 DATADIR
 CFGDIR
@@ -4610,6 +4611,30 @@ _ACEOF
 
 
 
+dataroot_dir=`echo $datarootdir | grep prefix`
+
+if test -n "$dataroot_dir"; then
+    if test "$prefix" = "NONE"
+    then
+        dataroot_dir="$ac_default_prefix/share"
+    else
+        dataroot_dir="$prefix/share"
+    fi
+else
+    dataroot_dir="$datarootdir"
+fi
+
+
+DATAROOTDIR=$dataroot_dir
+
+
+cat >>confdefs.h <<_ACEOF
+#define DATAROOTDIR "$dataroot_dir"
+_ACEOF
+
+
+
+
 cat >>confdefs.h <<_ACEOF
 #define VIRUS_TEMPLATE "$my_prefix/share/clapf/template.virus"
 _ACEOF

configure.in

@@ -340,6 +340,27 @@ AC_SUBST(LIBEXECDIR)
 AC_DEFINE_UNQUOTED(LIBEXECDIR,"$libexec_dir",[where to look for the piler helpers])
 
 
+dnl configure dataroot directory
+
+dataroot_dir=`echo $datarootdir | grep prefix`
+
+if test -n "$dataroot_dir"; then
+    if test "$prefix" = "NONE"
+    then
+        dataroot_dir="$ac_default_prefix/share"
+    else
+        dataroot_dir="$prefix/share"
+    fi
+else
+    dataroot_dir="$datarootdir"
+fi
+
+
+DATAROOTDIR=$dataroot_dir
+AC_SUBST(DATAROOTDIR)
+AC_DEFINE_UNQUOTED(DATAROOTDIR,"$dataroot_dir",[where to look for the share data files])
+
+
 AC_DEFINE_UNQUOTED(VIRUS_TEMPLATE, "$my_prefix/share/clapf/template.virus", [where the virus template is])
 AC_DEFINE_UNQUOTED(ZOMBIE_NET_REGEX, "$my_prefix/share/clapf/zombienets.regex", [where the virus template is])
 

piler-config.h.in

@@ -4,9 +4,12 @@
 
 #define CONFDIR "/usr/local/etc"
 #define DATADIR "/usr/local/var"
+#define DATAROOTDIR "/usr/local/share"
 
 #define KEYFILE CONFDIR "/piler.key"
 
+#define MESSAGE_ID_DEDUP_FILE DATAROOTDIR "/piler/deduphelper"
+
 #define HAVE_DAEMON 1
 
 #undef TIMEOUT_BINARY

src/cfg.c

@@ -83,6 +83,7 @@ struct _parse_rule config_parse_rules[] =
    { "memcached_to_db_interval", "integer", (void*) int_parser, offsetof(struct __config, memcached_to_db_interval), "900", sizeof(int)},
    { "memcached_ttl", "integer", (void*) int_parser, offsetof(struct __config, memcached_ttl), "86400", sizeof(int)},
    { "min_word_len", "integer", (void*) int_parser, offsetof(struct __config, min_word_len), "1", sizeof(int)},
+   { "mmap_dedup_test", "integer", (void*) int_parser, offsetof(struct __config, mmap_dedup_test), "0", sizeof(int)},
    { "mysqlhost", "string", (void*) string_parser, offsetof(struct __config, mysqlhost), "", MAXVAL-1},
    { "mysqlport", "integer", (void*) int_parser, offsetof(struct __config, mysqlport), "", sizeof(int)},
    { "mysqlsocket", "string", (void*) string_parser, offsetof(struct __config, mysqlsocket), "/tmp/mysql.sock", MAXVAL-1},

src/cfg.h

@@ -89,6 +89,8 @@ struct __config {
 
    int syslog_recipients;
 
+   int mmap_dedup_test;
+
    int debug;
 };
 

src/defs.h

@@ -69,6 +69,7 @@ typedef void signal_func (int);
 
 struct child {
    pid_t pid;
+   int serial;
    int messages;
    int status;
 };
@@ -288,6 +289,8 @@ struct __data {
    struct node *mydomains[MAXHASH];
    struct node *imapfolders[MAXHASH];
    struct import *import;
+   char *dedup;
+   int child_serial;
 
 #ifdef NEED_MYSQL
    MYSQL_STMT *stmt_generic;

src/errmsg.h

@@ -18,6 +18,7 @@
 #define ERR_OPEN_TMP_FILE "ERR: opening a tempfile"
 #define ERR_TIMED_OUT "ERR: timed out"
 #define ERR_FORK_FAILED "ERR: cannot fork()"
+#define ERR_OPEN_DEDUP_FILE "ERR: cannot open dedup file"
 
 #define ERR_MYSQL_CONNECT "Cannot connect to mysql server"
 #define ERR_PSQL_CONNECT "Cannot connect to PSql server"

src/message.c

@@ -263,6 +263,24 @@ int process_message(struct session_data *sdata, struct _state *state, struct __d
 
    if(cfg->verbosity >= _LOG_DEBUG) syslog(LOG_PRIORITY, "%s: touch %s OK (%s)", sdata->ttmpfile, state->message_id_hash, state->message_id);
 
+
+
+   if(cfg->mmap_dedup_test == 1 && data->dedup != MAP_FAILED && data->child_serial >= 0 && data->child_serial < MAXCHILDREN){
+
+      if(strstr(data->dedup, state->message_id_hash)){
+         if(cfg->verbosity >= _LOG_DEBUG) syslog(LOG_INFO, "%s: dedup string: %s", sdata->ttmpfile, data->dedup);
+         if(cfg->verbosity >= _LOG_DEBUG) syslog(LOG_INFO, "%s: message-id-hash=%s, serial=%d", sdata->ttmpfile, state->message_id_hash, data->child_serial);
+
+         remove_stripped_attachments(state);
+         return ERR_EXISTS;
+      }
+
+      memcpy(data->dedup + data->child_serial*DIGEST_LENGTH*2, state->message_id_hash, DIGEST_LENGTH*2);
+   }
+
+
+
+
    /* store base64 encoded file attachments */
 
    if(state->n_attachments > 0){

src/piler.c

@@ -8,6 +8,7 @@
 #include <strings.h>
 #include <sys/types.h>
 #include <sys/socket.h>
+#include <sys/mman.h>
 #include <netdb.h>
 #include <netinet/in.h>
 #include <arpa/inet.h>
@@ -79,6 +80,7 @@ static void takesig(int sig){
                    if(quit == 0){
                       i = search_slot_by_pid(pid);
                       if(i >= 0){
+                         children[i].serial = i;
                          children[i].status = READY;
                          children[i].pid = child_make(&children[i]);
                       }
@@ -108,7 +110,7 @@ static void child_main(struct child *ptr){
 
    ptr->messages = 0;
 
-   if(cfg.verbosity >= _LOG_DEBUG) syslog(LOG_PRIORITY, "child (pid: %d) started main()", getpid());
+   if(cfg.verbosity >= _LOG_DEBUG) syslog(LOG_PRIORITY, "child (pid: %d, serial: %d) started main()", getpid(), ptr->serial);
 
    while(1){
       if(received_sighup == 1){
@@ -129,6 +131,7 @@ static void child_main(struct child *ptr){
 
       syslog(LOG_PRIORITY, "connection from %s", s);
 
+      data.child_serial = ptr->serial;
 
       sig_block(SIGHUP);
       ptr->messages += handle_smtp_session(new_sd, &data, &cfg);
@@ -137,7 +140,7 @@ static void child_main(struct child *ptr){
       close(new_sd);
 
       if(cfg.max_requests_per_child > 0 && ptr->messages >= cfg.max_requests_per_child){
-         if(cfg.verbosity >= _LOG_DEBUG) syslog(LOG_PRIORITY, "child (pid: %d) served enough: %d", getpid(), ptr->messages);
+         if(cfg.verbosity >= _LOG_DEBUG) syslog(LOG_PRIORITY, "child (pid: %d, serial: %d) served enough: %d", getpid(), ptr->messages, ptr->serial);
          break;
       }
 
@@ -184,10 +187,12 @@ int child_pool_create(){
       children[i].pid = 0;
       children[i].messages = 0;
       children[i].status = UNDEF;
+      children[i].serial = -1;
    }
 
    for(i=0; i<cfg.number_of_worker_processes; i++){
       children[i].status = READY;
+      children[i].serial = i;
       children[i].pid = child_make(&children[i]);
 
       if(children[i].pid == -1){
@@ -237,6 +242,8 @@ void p_clean_exit(){
 
    unlink(cfg.pidfile);
 
+   if(data.dedup != MAP_FAILED) munmap(data.dedup, MAXCHILDREN*DIGEST_LENGTH*2);
+
 #ifdef HAVE_STARTTLS
    if(data.ctx){
       SSL_CTX_free(data.ctx);
@@ -345,7 +352,7 @@ void initialise_configuration(){
 
 
 int main(int argc, char **argv){
-   int i, rc, yes=1, daemonise=0;
+   int i, rc, yes=1, daemonise=0, dedupfd;
    char port_string[8];
    struct addrinfo hints, *res;
 
@@ -384,6 +391,7 @@ int main(int argc, char **argv){
    initrules(data.retention_rules);
    data.ctx = NULL;
    data.ssl = NULL;
+   data.dedup = MAP_FAILED;
    memset(data.starttls, 0, sizeof(data.starttls));
 
 
@@ -425,6 +433,13 @@ int main(int argc, char **argv){
 
    if(drop_privileges(pwd)) fatal(ERR_SETUID);
 
+   dedupfd = open(MESSAGE_ID_DEDUP_FILE, O_RDWR);
+   if(dedupfd == -1) fatal(ERR_OPEN_DEDUP_FILE);
+
+   data.dedup = mmap(NULL, MAXCHILDREN*DIGEST_LENGTH*2, PROT_READ|PROT_WRITE, MAP_SHARED, dedupfd, 0);
+   close(dedupfd);
+
+   if(data.dedup == MAP_FAILED) syslog(LOG_INFO, "cannot mmap() %s, errno=%d", MESSAGE_ID_DEDUP_FILE, errno);
 
    syslog(LOG_PRIORITY, "%s %s, build %d starting", PROGNAME, VERSION, get_build());
 

util/Makefile.in

@@ -42,6 +42,7 @@ install:
 
 	$(INSTALL) -m 0755 $(srcdir)/db-mysql.sql $(DESTDIR)$(datarootdir)/piler
 	$(INSTALL) -m 0755 $(srcdir)/db-mysql-root.sql.in $(DESTDIR)$(datarootdir)/piler
+	$(INSTALL) -m 0644 -o $(RUNNING_USER) -g $(RUNNING_GROUP) $(srcdir)/deduphelper $(DESTDIR)$(datarootdir)/piler
 
 clean:
 	rm -f postinstall.sh

util/deduphelper

@@ -0,0 +1 @@
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