- Added makefile_slinux

- Changed mesure to be a silver bullet

el737hpdriv.c

@@ -37,6 +37,10 @@ typedef struct {
                       NoBeam and Counting status if the beam goes off
 		   */
   char *badReply;
+  int readErrorCount; /* need to remember failed reads: on RDOE, upper level
+                         code will see busy's and thus not catch the case of
+                         multiple failures
+                       */
 } EL737hp, *pEL737hp;
 /*--------------------- ERROR CODES -------------------------------------*/
 #define OFFLINE -1
@@ -49,6 +53,7 @@ typedef struct {
 #define BADREPLY  -10
 #define SELECTFAIL -11
 #define TIMEOUT737    -12
+#define TOMANYREADERRORS -23
 /*---------------------------------------------------------------------*/
 static void setBadReply(pEL737hp self, char *reply){
   if(self->badReply != NULL){
@@ -133,20 +138,24 @@ static int readRS(pEL737hp pPriv, int *RS){
   status = readRS232TillTerm(pPriv->controller,
 			 reply,&len);
   if(status < 0) {
+    pPriv->readErrorCount++;
     pPriv->errorCode = status;
     return 0;
   }
   status = checkEL737Error(reply);
   if(status < 0){
+    pPriv->readErrorCount++;
     pPriv->errorCode = status;
     return 0;
   }
   status = sscanf(reply,"%d",RS);
   if(status < 1){
+    pPriv->readErrorCount++;
     pPriv->errorCode = BADREPLY;
     setBadReply(pPriv,reply);
     return 0;
   }
+  pPriv->readErrorCount = 0;
   return 1;
 }
 /*-----------------------------------------------------------------*/
@@ -239,7 +248,11 @@ static int EL737Status(struct __COUNTER *self, float *fControl){
     handle STATSEND mode
   */
   if(pPriv->statusMode == STATSEND){
-    writeRS232(pPriv->controller,"RS\r",3);
+    status = writeRS232(pPriv->controller,"RS\r",3);
+    if(status < 0){
+      pPriv->errorCode = status;
+      return HWFault;
+    }
     pPriv->statusMode = STATRECEIVE;
     pPriv->startRequest = time(NULL);
     *fControl = pPriv->cachedControl;
@@ -253,6 +266,7 @@ static int EL737Status(struct __COUNTER *self, float *fControl){
   if(time(NULL) > pPriv->startRequest + 10){
     pPriv->statusMode = STATSEND;
     pPriv->errorCode = TIMEOUT737;
+    pPriv->readErrorCount++;
     return HWFault;
   }
 
@@ -285,6 +299,14 @@ static int EL737Status(struct __COUNTER *self, float *fControl){
   returnValue = decodeRS(pPriv,RS);
   pPriv->lastStatus = returnValue;
 
+  /*
+    check for excessive failed reads
+  */
+  if(pPriv->readErrorCount > 3){
+    pPriv->errorCode = TOMANYREADERRORS;
+    return HWFault;
+  }
+
   /*
     check if we update the monitors and do it 
   */
@@ -310,6 +332,7 @@ static int EL737Start(struct __COUNTER *self){
   pPriv = (pEL737hp)self->pData;
 
   fixMode(pPriv);
+  pPriv->readErrorCount = 0;
   if(self->eMode == ePreset){
     snprintf(pCommand,49,"MP %d\r",(int)self->fPreset);
   } else {
@@ -414,8 +437,13 @@ static int EL737GetError(struct __COUNTER *self, int *iCode,
     break;
   case SELECTFAIL:
     strncpy(pError,"select system call failed, network trouble",errLen);
+    break;
   case TIMEOUT737:
-    strncpy(pError,"timeout waiting for status repsonse",errLen);
+    strncpy(pError,"timeout or network problem while waiting for status repsonse",errLen);
+    break;
+  case TOMANYREADERRORS:
+    strncpy(pError,"Failed more then three times to read counter box",errLen); 
+    break;
   default:
     getRS232Error(pPriv->errorCode,pError,errLen);
   }
@@ -428,29 +456,39 @@ static int EL737FixIt(struct __COUNTER *self, int iCode){
   char pReply[50];
   char buffer[256];
   int dataLen = 255;
+  int i;
 
   assert(self);
   pPriv = (pEL737hp)self->pData;
 
   switch(iCode){
-  case TIMEOUT:
   case BADPARAM:
   case NOPARAM:
   case BADRANGE:
   case BADTRANGE:
-  case TIMEOUT737:
     return COREDO;
     break;
   case BADREPLY:
-    /*
-      try to read away all the garbage which may still be in the line
-    */
-    if(availableRS232(pPriv->controller) > 0){
-      memset(buffer,0,256);
-      readRS232(pPriv->controller,buffer,&dataLen);
-      printf("EL737hpdriv dumped %s after bad reply\n", buffer);
+  case TIMEOUT737:
+  case TIMEOUT:
+    for(i = 0; i < 3; i++){
+      status = readRS232TillTerm(pPriv->controller,buffer,&dataLen);
+      if(status == 1){
+	return COREDO;
+      }
+    }
+    /*
+      If nothing can be read, the only fixable cause is a network breakdown
+      Try to fix this. If this does not work: give up
+    */
+    closeRS232(pPriv->controller);
+    SicsWait(60);
+    status = initRS232(pPriv->controller);
+    if(status != 1){
+      return COTERM;
+    } else {
+      return COREDO;
     }
-    return COREDO;
     break;
   case OFFLINE:
     EL737Command(pPriv,"RMT 1\r",pReply,49);
@@ -463,12 +501,13 @@ static int EL737FixIt(struct __COUNTER *self, int iCode){
     break;
   case TOMANYCOUNTS:
   case SYSERROR:
-    return COTERM;
     break;
   default:
     /*
       network problem; try to reopen
     */
+    closeRS232(pPriv->controller);
+    SicsWait(60); 
     status = initRS232(pPriv->controller);
     if(status != 1){
       return COTERM;