Modbus plugin: Add "exec. req. flag" and "write on change" features
authorMario de Sousa <msousa@fe.up.pt>
Thu, 28 May 2020 10:54:48 +0100
changeset 2647 990004083eb8
parent 2640 1b4b335e19ea
child 2648 e4ab768170f9
Modbus plugin: Add "exec. req. flag" and "write on change" features
modbus/mb_runtime.c
modbus/mb_runtime.h
modbus/mb_utils.py
modbus/modbus.py
--- a/modbus/mb_runtime.c	Wed Nov 13 11:21:04 2019 +0100
+++ b/modbus/mb_runtime.c	Thu May 28 10:54:48 2020 +0100
@@ -25,6 +25,8 @@
 
 #include <stdio.h>
 #include <string.h>  /* required for memcpy() */
+#include <time.h>
+#include <signal.h>
 #include "mb_slave_and_master.h"
 #include "MB_%(locstr)s.h"
 
@@ -299,10 +301,42 @@
 	// Enable thread cancelation. Enabled is default, but set it anyway to be safe.
 	pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL);
 	
-	// get the current time
-	clock_gettime(CLOCK_MONOTONIC, &next_cycle);
-
-	// loop the communication with the client
+	// configure the timer for periodic activation
+    {
+      struct itimerspec timerspec;
+      timerspec.it_interval.tv_sec  = period_sec;
+      timerspec.it_interval.tv_nsec = period_nsec;
+      timerspec.it_value            = timerspec.it_interval;
+      
+      if (timer_settime(client_nodes[client_node_id].timer_id, 0 /* flags */, &timerspec, NULL) < 0)
+        fprintf(stderr, "Modbus plugin: Error configuring periodic activation timer for Modbus client %%s.\n", client_nodes[client_node_id].location);          
+    }
+
+    /* loop the communication with the client
+     * 
+         * When the client thread has difficulty communicating with remote client and/or server (network issues, for example),
+         * then the communications get delayed and we will fall behind in the period. 
+         * 
+         * This is OK. Note that if the condition variable were to be signaled multiple times while the client thread is inside the same
+         * Modbus transaction, then all those signals would be ignored.
+         * However, and since we keep the mutex locked during the communication cycle, it is not possible to signal the condition variable
+         * during that time (it is only possible while the thread is blocked during the call to pthread_cond_wait().
+         * 
+         * This means that when network issues eventually get resolved, we will NOT have a bunch of delayed activations to handle
+         * in quick succession (which would goble up CPU time). 
+         * 
+         * Notice that the above property is valid whether the communication cycle is run with the mutex locked, or unlocked.
+         * Since it makes it easier to implement the correct semantics for the other activation methods if the communication cycle
+         * is run with the mutex locked, then that is what we do.
+         * 
+     * Note that during all the communication cycle we will keep locked the mutex 
+     * (i.e. the mutex used together with the condition variable that will activate a new communication cycle)
+     * 
+     * Note that we never get to explicitly unlock this mutex. It will only be unlocked by the pthread_cond_wait()
+     * call at the end of the cycle.
+     */
+    pthread_mutex_lock(&(client_nodes[client_node_id].mutex));
+
 	while (1) {
 		/*
 		struct timespec cur_time;
@@ -311,9 +345,22 @@
 		*/
 		int req;
 		for (req=0; req < NUMBER_OF_CLIENT_REQTS; req ++){
-			/*just do the requests belonging to the client */
+			/* just do the requests belonging to the client */
 			if (client_requests[req].client_node_id != client_node_id)
 				continue;
+            
+            /* only do the request if:
+             *   - this request was explictly asked to be executed by the client program
+             *  OR
+             *   - the client thread was activated periodically
+             *     (in which case we execute all the requests belonging to the client node)
+             */
+            if ((client_requests[req].flag_exec_req == 0) && (client_nodes[client_requests[req].client_node_id].periodic_act == 0))
+                continue;
+            
+            //fprintf(stderr, "Modbus plugin: RUNNING<###> of Modbus request %%d  (periodic = %%d  flag_exec_req = %%d)\n", 
+            //        req, client_nodes[client_requests[req].client_node_id].periodic_act, client_requests[req].flag_exec_req );
+            
 			int res_tmp = __execute_mb_request(req);
 			switch (res_tmp) {
 			  case PORT_FAILURE: {
@@ -357,36 +404,40 @@
 				break;
 			  }
 			}
-		}
-		// Determine absolute time instant for starting the next cycle
-		struct timespec prev_cycle, now;
-		prev_cycle = next_cycle;
-		timespec_add(next_cycle, period_sec, period_nsec);
-		/* NOTE A:
-		 * When we have difficulty communicating with remote client and/or server, then the communications get delayed and we will
-		 * fall behind in the period. This means that when communication is re-established we may end up running this loop continuously
-		 * for some time until we catch up.
-		 * This is undesirable, so we detect it by making sure the next_cycle will start in the future.
-		 * When this happens we will switch from a purely periodic task _activation_ sequence, to a fixed task suspension interval.
-		 * 
-		 * NOTE B:
-		 * It probably does not make sense to check for overflow of timer - so we don't do it for now!
-		 * Even in 32 bit systems this will take at least 68 years since the computer booted
-		 * (remember, we are using CLOCK_MONOTONIC, which should start counting from 0
-		 * every time the system boots). On 64 bit systems, it will take over 
-		 * 10^11 years to overflow.
-		 */
-		clock_gettime(CLOCK_MONOTONIC, &now);
-		if (  ((now.tv_sec > next_cycle.tv_sec) || ((now.tv_sec == next_cycle.tv_sec) && (now.tv_nsec > next_cycle.tv_nsec)))
-		   /* We are falling behind. See NOTE A above */
-		   || (next_cycle.tv_sec < prev_cycle.tv_sec)
-		   /* Timer overflow. See NOTE B above */
-		   ) {
-			next_cycle = now;
-			timespec_add(next_cycle, period_sec, period_nsec);
-		}
-
-		clock_nanosleep(CLOCK_MONOTONIC, TIMER_ABSTIME, &next_cycle, NULL);
+        
+            /* We have just finished excuting a client transcation request.
+             * If the current cycle was activated by user request we reset the flag used to ask to run it
+             */
+            if (0 != client_requests[req].flag_exec_req) {
+                client_requests[req].flag_exec_req     = 0;
+                client_requests[req].flag_exec_started = 0;   
+            }
+            
+            //fprintf(stderr, "Modbus plugin: RUNNING<---> of Modbus request %%d  (periodic = %%d  flag_exec_req = %%d)\n", 
+            //        req, client_nodes[client_requests[req].client_node_id].periodic_act, client_requests[req].flag_exec_req );
+        }
+
+        // Wait for signal (from timer or explicit request from user program) before starting the next cycle
+        {
+            // No need to lock the mutex. Is is already locked just before the while(1) loop.
+            // Read the comment there to understand why.
+            // pthread_mutex_lock(&(client_nodes[client_node_id].mutex));
+            
+            /* the client thread has just finished a cycle, so all the flags used to signal an activation
+             * and specify the activation source (periodic, user request, ...)
+             * get reset here, before waiting for a new activation.
+             */
+            client_nodes[client_node_id].periodic_act = 0;            
+            client_nodes[client_node_id].execute_req  = 0;
+            
+            while (client_nodes[client_node_id].execute_req == 0)
+                pthread_cond_wait(&(client_nodes[client_node_id].condv),
+                                &(client_nodes[client_node_id].mutex)); 
+                            
+            // We run the communication cycle with the mutex locked.
+            // Read the comment just above the while(1) to understand why.
+            // pthread_mutex_unlock(&(client_nodes[client_node_id].mutex));
+        }
 	}
 
 	// humour the compiler.
@@ -394,6 +445,50 @@
 }
 
 
+
+/* Function to activate a client node's thread */
+/* returns -1 if it could not send the signal */
+static int __signal_client_thread(int client_node_id) {
+    /* We TRY to signal the client thread.
+     * We do this because this function can be called at the end of the PLC scan cycle
+     * and we don't want it to block at that time.
+     */
+    if (pthread_mutex_trylock(&(client_nodes[client_node_id].mutex)) != 0)
+        return -1;
+    client_nodes[client_node_id].execute_req = 1; // tell the thread to execute
+    pthread_cond_signal (&(client_nodes[client_node_id].condv));
+    pthread_mutex_unlock(&(client_nodes[client_node_id].mutex));
+    return 0;
+}
+
+
+
+/* Function that will be called whenever a client node's periodic timer expires. */
+/* The client node's thread will be waiting on a condition variable, so this function simply signals that 
+ * condition variable.
+ * 
+ * The same callback function is called by the timers of all client nodes. The id of the client node
+ * in question will be passed as a parameter to the call back function.
+ */
+void __client_node_timer_callback_function(union sigval sigev_value) {
+    /* signal the client node's condition variable on which the client node's thread should be waiting... */
+    /* Since the communication cycle is run with the mutex locked, we use trylock() instead of lock() */
+    //pthread_mutex_lock  (&(client_nodes[sigev_value.sival_int].mutex));
+    if (pthread_mutex_trylock (&(client_nodes[sigev_value.sival_int].mutex)) != 0)
+        /* we never get to signal the thread for activation. But that is OK.
+         * If it still in the communication cycle (during which the mutex is kept locked)
+         * then that means that the communication cycle is falling behing in the periodic 
+         * communication cycle, and we therefore need to skip a period.
+         */
+        return;
+    client_nodes[sigev_value.sival_int].execute_req  = 1; // tell the thread to execute
+    client_nodes[sigev_value.sival_int].periodic_act = 1; // tell the thread the activation was done by periodic timer   
+    pthread_cond_signal (&(client_nodes[sigev_value.sival_int].condv));
+    pthread_mutex_unlock(&(client_nodes[sigev_value.sival_int].mutex));
+}
+
+
+
 int __cleanup_%(locstr)s ();
 int __init_%(locstr)s (int argc, char **argv){
 	int index;
@@ -421,9 +516,14 @@
 		return -1;
 	}
 	
-	/* init the mutex for each client request */
+	/* init each client request */
 	/* Must be done _before_ launching the client threads!! */
 	for (index=0; index < NUMBER_OF_CLIENT_REQTS; index ++){
+        /* make sure flags connected to user program MB transaction start request are all reset */
+        client_requests[index].flag_exec_req     = 0;
+        client_requests[index].flag_exec_started = 0;        
+        /* init the mutex for each client request */
+        /* Must be done _before_ launching the client threads!! */
 		if (pthread_mutex_init(&(client_requests[index].coms_buf_mutex), NULL)) {
 			fprintf(stderr, "Modbus plugin: Error initializing request for modbus client node %%s\n", client_nodes[client_requests[index].client_node_id].location);
 			goto error_exit;
@@ -443,6 +543,39 @@
 		}
 		client_nodes[index].init_state = 1; // we have created the node 
 		
+		/* initialize the mutex variable that will be used by the thread handling the client node */
+        if (pthread_mutex_init(&(client_nodes[index].mutex), NULL) < 0) {
+			fprintf(stderr, "Modbus plugin: Error creating mutex for modbus client node %%s\n", client_nodes[index].location);
+			goto error_exit;                
+        }
+		client_nodes[index].init_state = 2; // we have created the mutex
+		
+		/* initialize the condition variable that will be used by the thread handling the client node */
+        if (pthread_cond_init(&(client_nodes[index].condv), NULL) < 0) {
+			fprintf(stderr, "Modbus plugin: Error creating condition variable for modbus client node %%s\n", client_nodes[index].location);
+			goto error_exit;                
+        }
+        client_nodes[index].execute_req = 0; //variable associated with condition variable
+		client_nodes[index].init_state = 3; // we have created the condition variable
+		
+		/* initialize the timer that will be used to periodically activate the client node */
+        {
+            // start off by reseting the flag that will be set whenever the timer expires
+            client_nodes[index].periodic_act = 0;
+
+            struct sigevent evp;
+            evp.sigev_notify            = SIGEV_THREAD; /* Notification method - call a function in a new thread context */
+            evp.sigev_value.sival_int   = index;        /* Data passed to function upon notification - used to indentify which client node to activate */
+            evp.sigev_notify_function   = __client_node_timer_callback_function; /* function to call upon timer expiration */
+            evp.sigev_notify_attributes = NULL;         /* attributes for new thread in which sigev_notify_function will be called/executed */
+            
+            if (timer_create(CLOCK_MONOTONIC, &evp, &(client_nodes[index].timer_id)) < 0) {
+                fprintf(stderr, "Modbus plugin: Error creating timer for modbus client node %%s\n", client_nodes[index].location);
+                goto error_exit;                
+            }
+        }
+        client_nodes[index].init_state = 4; // we have created the timer
+
 		/* launch a thread to handle this client node */
 		{
 			int res = 0;
@@ -450,11 +583,11 @@
 			res |= pthread_attr_init(&attr);
 			res |= pthread_create(&(client_nodes[index].thread_id), &attr, &__mb_client_thread, (void *)((char *)NULL + index));
 			if (res !=  0) {
-				fprintf(stderr, "Modbus plugin: Error starting modbus client thread for node %%s\n", client_nodes[index].location);
+				fprintf(stderr, "Modbus plugin: Error starting thread for modbus client node %%s\n", client_nodes[index].location);
 				goto error_exit;
 			}
 		}
-		client_nodes[index].init_state = 2; // we have created the node and a thread
+		client_nodes[index].init_state = 5; // we have created the thread
 	}
 
 	/* init each local server */
@@ -499,9 +632,26 @@
 	int index;
 
 	for (index=0; index < NUMBER_OF_CLIENT_REQTS; index ++){
-		/*just do the output requests */
+		/* synchronize the PLC and MB buffers only for the output requests */
 		if (client_requests[index].req_type == req_output){
+            
+            // lock the mutex brefore copying the data
 			if(pthread_mutex_trylock(&(client_requests[index].coms_buf_mutex)) == 0){
+                
+                // Check if user configured this MB request to be activated whenever the data to be written changes
+                if (client_requests[index].write_on_change) {
+                    // Let's check if the data did change...
+                    // compare the data in plcv_buffer to coms_buffer
+                    int res;
+                    res = memcmp((void *)client_requests[index].coms_buffer /* buf 1 */,
+                                 (void *)client_requests[index].plcv_buffer /* buf 2*/,
+                                 REQ_BUF_SIZE * sizeof(u16) /* size in bytes */);
+                    
+                    // if data changed, activate execution request 
+                    if (0 != res)
+                        client_requests[index].flag_exec_req = 1;
+                }
+                
                 // copy from plcv_buffer to coms_buffer
                 memcpy((void *)client_requests[index].coms_buffer /* destination */,
                        (void *)client_requests[index].plcv_buffer /* source */,
@@ -509,7 +659,33 @@
                 pthread_mutex_unlock(&(client_requests[index].coms_buf_mutex));
             }
 		}
-	}
+        /* if the user program set the execution request flag, then activate the thread
+         *  that handles this Modbus client transaction so it gets a chance to be executed
+         *  (but don't activate the thread if it has already been activated!)
+         * 
+         * NOTE that we do this, for both the IN and OUT mapped location, under this
+         *  __publish_() function. The scan cycle of the PLC works as follows:
+         *   - call __retrieve_()
+         *   - execute user programs
+         *   - call __publish_()
+         *   - insert <delay> until time to start next periodic/cyclic scan cycle
+         * 
+         *  In an attempt to be able to run the MB transactions during the <delay>
+         *  interval in which not much is going on, we handle the user program
+         *  requests to execute a specific MB transaction in this __publish_()
+         *  function.
+         */
+        if ((client_requests[index].flag_exec_req != 0) && (0 == client_requests[index].flag_exec_started)) {
+            int client_node_id = client_requests[index].client_node_id;
+            if (__signal_client_thread(client_node_id) >= 0) {
+                /* - upon success, set flag_exec_started
+                 * - both flags (flag_exec_req and flag_exec_started) will be reset
+                 *   once the transaction has completed.
+                 */
+                client_requests[index].flag_exec_started = 1;    
+            }
+        }                    
+    }
 }
 
 
@@ -544,12 +720,39 @@
 	/* kill thread and close connections of each modbus client node */
 	for (index=0; index < NUMBER_OF_CLIENT_NODES; index++) {
 		close = 0;
-		if (client_nodes[index].init_state >= 2) {
+		if (client_nodes[index].init_state >= 5) {
 			// thread was launched, so we try to cancel it!
 			close  = pthread_cancel(client_nodes[index].thread_id);
 			close |= pthread_join  (client_nodes[index].thread_id, NULL);
 			if (close < 0)
-				fprintf(stderr, "Modbus plugin: Error closing thread for modbus client %%s\n", client_nodes[index].location);
+				fprintf(stderr, "Modbus plugin: Error closing thread for modbus client node %%s\n", client_nodes[index].location);
+		}
+		res |= close;
+
+		close = 0;
+		if (client_nodes[index].init_state >= 4) {
+			// timer was created, so we try to destroy it!
+			close  = timer_delete(client_nodes[index].timer_id);
+			if (close < 0)
+				fprintf(stderr, "Modbus plugin: Error destroying timer for modbus client node %%s\n", client_nodes[index].location);
+		}
+		res |= close;
+
+		close = 0;
+		if (client_nodes[index].init_state >= 3) {
+			// condition variable was created, so we try to destroy it!
+			close  = pthread_cond_destroy(&(client_nodes[index].condv));
+			if (close < 0)
+				fprintf(stderr, "Modbus plugin: Error destroying condition variable for modbus client node %%s\n", client_nodes[index].location);
+		}
+		res |= close;
+
+		close = 0;
+		if (client_nodes[index].init_state >= 2) {
+			// mutex was created, so we try to destroy it!
+			close  = pthread_mutex_destroy(&(client_nodes[index].mutex));
+			if (close < 0)
+				fprintf(stderr, "Modbus plugin: Error destroying mutex for modbus client node %%s\n", client_nodes[index].location);
 		}
 		res |= close;
 
--- a/modbus/mb_runtime.h	Wed Nov 13 11:21:04 2019 +0100
+++ b/modbus/mb_runtime.h	Thu May 28 10:54:48 2020 +0100
@@ -54,11 +54,32 @@
 typedef struct{
 	    const char *location;
 	    node_addr_t	node_address;
-	    int		mb_nd;
+	    int		mb_nd;      // modbus library node used for this client
 	    int		init_state; // store how far along the client's initialization has progressed
-	    u64		comm_period;
+	    u64		comm_period;// period to use when periodically sending requests to remote server
 	    int		prev_error; // error code of the last printed error message (0 when no error) 
-	    pthread_t	thread_id;  // thread handling all communication with this client
+	    pthread_t   thread_id;  // thread handling all communication for this client node
+	    timer_t      timer_id;  // timer used to periodically activate this client node's thread
+	    pthread_mutex_t mutex;  // mutex to be used with the following condition variable
+        pthread_cond_t  condv;  // used to signal the client thread when to start new modbus transactions
+        int       execute_req;  /* used, in association with condition variable,  
+                                 *   to signal when to send the modbus request to the server
+                                 * Note that we cannot simply rely on the condition variable to signal
+                                 *   when to activate the client thread, as the call to 
+                                 *   pthread_cond_wait() may return without having been signaled!
+                                 *   From the manual:
+                                 *      Spurious  wakeups  from  the
+                                 *      pthread_cond_timedwait() or pthread_cond_wait()  functions  may  occur.
+                                 *      Since  the  return from pthread_cond_timedwait() or pthread_cond_wait()
+                                 *      does not imply anything about the value of this predicate,  the  predi-
+                                 *      cate should be re-evaluated upon such return.
+                                 */
+        int      periodic_act;  /* (boolen) flag will be set when the client node's thread was activated 
+                                 * (by signaling the above condition variable) by the periodic timer.
+                                 * Note that this same thread may also be activated (condition variable is signaled)
+                                 * by other sources, such as when the user program requests that a specific 
+                                 * client MB transation be executed (flag_exec_req in client_request_t)
+                                 */
 	} client_node_t;
 
 
@@ -82,11 +103,37 @@
 	    u8		error_code; // modbus error code (if any) of current request
 	    int		prev_error; // error code of the last printed error message (0 when no error) 
 	    struct timespec resp_timeout;
+	    u8		write_on_change; // boolean flag. If true => execute MB request when data to send changes
 	      // buffer used to store located PLC variables
 	    u16		plcv_buffer[REQ_BUF_SIZE];
 	      // buffer used to store data coming from / going to server
 	    u16		coms_buffer[REQ_BUF_SIZE]; 
 	    pthread_mutex_t coms_buf_mutex; // mutex to access coms_buffer[]
+          /* boolean flag that will be mapped onto a (BOOL) located variable 
+           * (u16 because IEC 61131-3 BOOL are mapped onto u16 in C code! )
+           *    -> allow PLC program to request when to start the MB transaction
+           *    -> will be reset once the MB transaction has completed
+           */
+        u16     flag_exec_req;  
+          /* flag that works in conjunction with flag_exec_req
+           * (does not really need to be u16 as it is not mapped onto a located variable. )
+           *    -> used by internal logic to indicate that the client thread 
+           *       that will be executing the MB transaction
+           *       requested by flag exec_req has already been activated.
+           *    -> will be reset once the MB transaction has completed
+           */
+        u16     flag_exec_started;  
+          /* flag that will be mapped onto a (WORD) located variable 
+           * (u16 because the flag is a word! )
+           *    -> MSByte will store the result of the last executed MB transaction
+           *         1 -> error accessing IP network, or serial interface
+           *         2 -> reply received from server was an invalid frame
+           *         3 -> server did not reply before timeout expired
+           *         4 -> server returned a valid error frame
+           *    -> if the MSByte is 4, the LSByte will store the MB error code returned by the server
+           *    -> will be reset (set to 0) once this MB transaction has completed sucesfully
+           */
+        u16     flag_exec_status;  
 	} client_request_t;
 
 
--- a/modbus/mb_utils.py	Wed Nov 13 11:21:04 2019 +0100
+++ b/modbus/mb_utils.py	Thu May 28 10:54:48 2020 +0100
@@ -184,7 +184,7 @@
 
     req_init_template = '''/*request %(locreqstr)s*/
 {"%(locreqstr)s", %(nodeid)s, %(slaveid)s, %(iotype)s, %(func_nr)s, %(address)s , %(count)s,
-DEF_REQ_SEND_RETRIES, 0 /* error_code */, 0 /* prev_code */, {%(timeout_s)d, %(timeout_ns)d} /* timeout */,
+DEF_REQ_SEND_RETRIES, 0 /* error_code */, 0 /* prev_code */, {%(timeout_s)d, %(timeout_ns)d} /* timeout */, %(write_on_change)d /* write_on_change */,
 {%(buffer)s}, {%(buffer)s}}'''
 
     timeout = int(GetCTVal(child, 4))
@@ -198,6 +198,7 @@
         "slaveid": GetCTVal(child, 1),
         "address": GetCTVal(child, 3),
         "count": GetCTVal(child, 2),
+        "write_on_change": GetCTVal(child, 5),
         "timeout": timeout,
         "timeout_s": timeout_s,
         "timeout_ns": timeout_ns,
@@ -222,5 +223,11 @@
         self.GetCTRoot().logger.write_error(
             "Modbus plugin: Invalid number of channels in TCP client request node %(locreqstr)s (start_address + nr_channels must be less than 65536)\nModbus plugin: Aborting C code generation for this node\n" % request_dict)
         return None
+    if (request_dict["write_on_change"] and (request_dict["iotype"] == 'req_input')):
+        self.GetCTRoot().logger.write_error(
+            "Modbus plugin: (warning) MB client request node %(locreqstr)s has option 'write_on_change' enabled.\nModbus plugin: This option will be ignored by the Modbus read function.\n" % request_dict)
+        # NOTE: this is only a warning (we don't wish to abort code generation) so following line must be left commented out!
+        # return None
+    
 
     return req_init_template % request_dict
--- a/modbus/modbus.py	Wed Nov 13 11:21:04 2019 +0100
+++ b/modbus/modbus.py	Thu May 28 10:54:48 2020 +0100
@@ -83,6 +83,7 @@
                 </xsd:restriction>
             </xsd:simpleType>
           </xsd:attribute>
+          <xsd:attribute name="Write_on_change" type="xsd:boolean" use="optional" default="false"/>
         </xsd:complexType>
       </xsd:element>
     </xsd:schema>
@@ -115,7 +116,29 @@
         datatacc = modbus_function_dict[function][6]
         # 'Coil', 'Holding Register', 'Input Discrete' or 'Input Register'
         dataname = modbus_function_dict[function][7]
+        # start off with a boolean entry
+        # This is a flag used to allow the user program to control when to 
+        # execute the Modbus request.
+        # NOTE: If the Modbus request has a 'current_location' of
+        #          %QX1.2.3
+        #       then the execution control flag will be
+        #          %QX1.2.3.0.0
+        #       and all the Modbus registers/coils will be
+        #          %QX1.2.3.0
+        #          %QX1.2.3.1
+        #          %QX1.2.3.2
+        #            ..
+        #          %QX1.2.3.n
         entries = []
+        entries.append({
+            "name": "Exec. request flag",
+            "type": LOCATION_VAR_MEMORY,
+            "size": 1,           # BOOL flag
+            "IEC_type": "BOOL",  # BOOL flag
+            "var_name": "var_name",
+            "location": "X" + ".".join([str(i) for i in current_location]) + ".0.0",
+            "description": "MB request execution control flag",
+            "children": []})        
         for offset in range(address, address + count):
             entries.append({
                 "name": dataname + " " + str(offset),
@@ -270,7 +293,7 @@
           <xsd:attribute name="Invocation_Rate_in_ms" use="optional" default="100">
             <xsd:simpleType>
                 <xsd:restriction base="xsd:unsignedLong">
-                    <xsd:minInclusive value="1"/>
+                    <xsd:minInclusive value="0"/>
                     <xsd:maxInclusive value="2147483647"/>
                 </xsd:restriction>
             </xsd:simpleType>
@@ -388,7 +411,7 @@
           <xsd:attribute name="Invocation_Rate_in_ms" use="optional" default="100">
             <xsd:simpleType>
                 <xsd:restriction base="xsd:integer">
-                    <xsd:minInclusive value="1"/>
+                    <xsd:minInclusive value="0"/>
                     <xsd:maxInclusive value="2147483647"/>
                 </xsd:restriction>
             </xsd:simpleType>
@@ -720,12 +743,32 @@
                     for iecvar in subchild.GetLocations():
                         # absloute address - start address
                         relative_addr = iecvar["LOC"][3] - int(GetCTVal(subchild, 3))
-                        # test if relative address in request specified range
-                        if relative_addr in xrange(int(GetCTVal(subchild, 2))):
+                        # test if the located variable 
+                        #    (a) has relative address in request specified range
+                        #  AND is NOT
+                        #    (b) is a control flag added by this modbus plugin
+                        #        to control its execution at runtime.
+                        #        Currently, we only add the "Execution Control Flag"
+                        #        to each client request (one flag per request)
+                        #        to control when to execute the request (if not executed periodically)
+                        #        While all Modbus registers/coils are mapped onto a location
+                        #        with 4 numbers (e.g. %QX0.1.2.55), this control flag is mapped
+                        #        onto a location with 4 numbers (e.g. %QX0.1.2.0.0), where the last
+                        #        two numbers are always '0.0', and the first two identify the request.
+                        #        In the following if, we check for this condition by checking
+                        #        if their are at least 4 or more number in the location's address.
+                        if (        relative_addr in xrange(int(GetCTVal(subchild, 2)))  # condition (a) explained above
+                            and len(iecvar["LOC"]) < 5):                                  # condition (b) explained above
                             if str(iecvar["NAME"]) not in loc_vars_list:
                                 loc_vars.append(
                                     "u16 *" + str(iecvar["NAME"]) + " = &client_requests[%d].plcv_buffer[%d];" % (client_requestid, relative_addr))
                                 loc_vars_list.append(str(iecvar["NAME"]))
+                        # Now add the located variable in case it is a flag (condition (b) above
+                        if  len(iecvar["LOC"]) >= 5:       # condition (b) explained above
+                            if str(iecvar["NAME"]) not in loc_vars_list:
+                                loc_vars.append(
+                                    "u16 *" + str(iecvar["NAME"]) + " = &client_requests[%d].flag_exec_req;" % (client_requestid))
+                                loc_vars_list.append(str(iecvar["NAME"]))
                     client_requestid += 1
                 tcpclient_node_count += 1
                 client_nodeid += 1
@@ -745,12 +788,32 @@
                     for iecvar in subchild.GetLocations():
                         # absloute address - start address
                         relative_addr = iecvar["LOC"][3] - int(GetCTVal(subchild, 3))
-                        # test if relative address in request specified range
-                        if relative_addr in xrange(int(GetCTVal(subchild, 2))):
+                        # test if the located variable 
+                        #    (a) has relative address in request specified range
+                        #  AND is NOT
+                        #    (b) is a control flag added by this modbus plugin
+                        #        to control its execution at runtime.
+                        #        Currently, we only add the "Execution Control Flag"
+                        #        to each client request (one flag per request)
+                        #        to control when to execute the request (if not executed periodically)
+                        #        While all Modbus registers/coils are mapped onto a location
+                        #        with 4 numbers (e.g. %QX0.1.2.55), this control flag is mapped
+                        #        onto a location with 4 numbers (e.g. %QX0.1.2.0.0), where the last
+                        #        two numbers are always '0.0', and the first two identify the request.
+                        #        In the following if, we check for this condition by checking
+                        #        if their are at least 4 or more number in the location's address.
+                        if (        relative_addr in xrange(int(GetCTVal(subchild, 2)))  # condition (a) explained above
+                            and len(iecvar["LOC"]) < 5):                                  # condition (b) explained above
                             if str(iecvar["NAME"]) not in loc_vars_list:
                                 loc_vars.append(
                                     "u16 *" + str(iecvar["NAME"]) + " = &client_requests[%d].plcv_buffer[%d];" % (client_requestid, relative_addr))
                                 loc_vars_list.append(str(iecvar["NAME"]))
+                        # Now add the located variable in case it is a flag (condition (b) above
+                        if  len(iecvar["LOC"]) >= 5:       # condition (b) explained above
+                            if str(iecvar["NAME"]) not in loc_vars_list:
+                                loc_vars.append(
+                                    "u16 *" + str(iecvar["NAME"]) + " = &client_requests[%d].flag_exec_req;" % (client_requestid))
+                                loc_vars_list.append(str(iecvar["NAME"]))
                     client_requestid += 1
                 rtuclient_node_count += 1
                 client_nodeid += 1