From Michael Mann via https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=6377, with the tshark man page update by me: tshark -z io,stat,interval,"[SUM|MIN|MAX|AVG](field)field [and filter]" should support floating point.

svn path=/trunk/; revision=39767

2 files changed, 139 insertions, 26 deletions

diff --git a/doc/tshark.pod b/doc/tshark.pod
index 8724f2c943..2be418c4c6 100644
--- a/doc/tshark.pod
+++ b/doc/tshark.pod
@@ -856,7 +856,7 @@ This will count the total number of SIDs seen in each 10ms interval.
 
 B<SUM(I<field>)I<field> [and I<filter>]> - Unlike COUNT, the I<values> of the
 specified field are summed per time interval.
-''I<field>'' can only be a named integer or relative time field.
+''I<field>'' can only be a named integer, float, double or relative time field.
 
 Example: B<-z io,stat,0.010,E<34>SUM(frame.len)frame.lenE<34>>
 
@@ -864,8 +864,8 @@ Reports the total number of bytes that were transmitted bidirectionally in
 all the packets within a 10 millisecond interval.
 
 B<MIN/MAX/AVG(I<field>)I<field> [and I<filter>]> - The minimum, maximum, or average field value
-in each interval is calculated.  The specified field must be a named integer
-or relative time field.  For relative time fields, the output is presented in
+in each interval is calculated.  The specified field must be a named integer,
+float, double or relative time field.  For relative time fields, the output is presented in
 seconds with six decimal digits of precision rounded to the nearest microsecond.
 
 In the following example, the time of the first Read_AndX call, the last Read_AndX
diff --git a/tap-iostat.c b/tap-iostat.c
index a70c8e84d4..1104edd6d6 100644
--- a/tap-iostat.c
+++ b/tap-iostat.c
@@ -67,6 +67,8 @@ typedef struct _io_stat_item_t {
 	guint64 frames;
 	guint64 num;
 	guint64 counter;
+	gfloat float_counter;
+	gdouble double_counter;
 } io_stat_item_t;
 
 #define NANOSECS_PER_SEC 1000000000
@@ -103,6 +105,8 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 		it->time    = it->prev->time + mit->parent->interval;
 		it->frames  = 0;
 		it->counter = 0;
+		it->float_counter = 0;
+		it->double_counter = 0;
 		it->num     = 0;
 		it->calc_type=it->prev->calc_type;
 		it->hf_index=it->prev->hf_index;
@@ -113,7 +117,7 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 
 	switch(it->calc_type){
 	case CALC_TYPE_BYTES:
-	case CALC_TYPE_FRAMES: 
+	case CALC_TYPE_FRAMES:
 	case CALC_TYPE_FRAMES_AND_BYTES:
 		it->counter+=pinfo->fd->pkt_len;
 		break;
@@ -149,6 +153,12 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 				case FT_INT64:
 					it->counter+=(gint64)fvalue_get_integer64(&((field_info *)gp->pdata[i])->value);
 					break;
+				case FT_FLOAT:
+					it->float_counter+=(gfloat)fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					break;
+				case FT_DOUBLE:
+					it->double_counter+=fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					break;
 				case FT_RELATIVE_TIME:
 					new_time = fvalue_get(&((field_info *)gp->pdata[i])->value);
 					val=(guint64)(new_time->secs) * NANOSECS_PER_SEC + new_time->nsecs;
@@ -163,6 +173,8 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 		if(gp){
 			int type;
 			guint64 val;
+			gfloat float_val;
+			gdouble double_val;
 			nstime_t *new_time;
 
 			type=proto_registrar_get_ftype(it->hf_index);
@@ -206,6 +218,22 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 						it->counter=val;
 					}
 					break;
+				case FT_FLOAT:
+					float_val=(gfloat)fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					if((it->frames==1)&&(i==0)){
+						it->float_counter=float_val;
+					} else if(float_val<it->float_counter){
+						it->float_counter=float_val;
+					}
+					break;
+				case FT_DOUBLE:
+					double_val=fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					if((it->frames==1)&&(i==0)){
+						it->double_counter=double_val;
+					} else if(double_val<it->double_counter){
+						it->double_counter=double_val;
+					}
+					break;
 				case FT_RELATIVE_TIME:
 					new_time=fvalue_get(&((field_info *)gp->pdata[i])->value);
 					val=(guint64)(new_time->secs) * NANOSECS_PER_SEC + new_time->nsecs;
@@ -224,6 +252,8 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 		if(gp){
 			int type;
 			guint64 val;
+			gfloat float_val;
+			gdouble double_val;
 			nstime_t *new_time;
 
 			type=proto_registrar_get_ftype(it->hf_index);
@@ -267,6 +297,22 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 						it->counter=val;
 					}
 					break;
+				case FT_FLOAT:
+					float_val=(gfloat)fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					if((it->frames==1)&&(i==0)){
+						it->float_counter=float_val;
+					} else if(float_val>it->float_counter){
+						it->float_counter=float_val;
+					}
+					break;
+				case FT_DOUBLE:
+					double_val=fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					if((it->frames==1)&&(i==0)){
+						it->double_counter=double_val;
+					} else if(double_val>it->double_counter){
+						it->double_counter=double_val;
+					}
+					break;
 				case FT_RELATIVE_TIME:
 					new_time=fvalue_get(&((field_info *)gp->pdata[i])->value);
 					val=(guint64)(new_time->secs) * NANOSECS_PER_SEC + new_time->nsecs;
@@ -310,6 +356,12 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 					val=fvalue_get_sinteger(&((field_info *)gp->pdata[i])->value);
 					it->counter+=val;
 					break;
+				case FT_FLOAT:
+					it->float_counter+=(gfloat)fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					break;
+				case FT_DOUBLE:
+					it->double_counter+=fvalue_get_floating(&((field_info *)gp->pdata[i])->value);
+					break;
 				case FT_RELATIVE_TIME:
 					new_time=fvalue_get(&((field_info *)gp->pdata[i])->value);
 					val=(guint64)(new_time->secs) * NANOSECS_PER_SEC + new_time->nsecs;
@@ -353,7 +405,7 @@ iostat_packet(void *arg, packet_info *pinfo, epan_dissect_t *edt, const void *du
 					pit->counter += mit->parent->interval;
 					val -= mit->parent->interval;
 					pit = pit->prev;
-					
+
 				}
 			}
 		}
@@ -371,6 +423,8 @@ iostat_draw(void *arg)
 	io_stat_item_t **items;
 	guint64 *frames;
 	guint64 *counters;
+	gfloat *float_counters;
+	gdouble *double_counters;
 	guint64 *num;
 	guint32 i;
 	guint32 borderLen=68;
@@ -380,7 +434,7 @@ iostat_draw(void *arg)
 	iot=mit->parent;
 
 	printf("\n");
-	
+
 	/* Display the table border */
 	for(i=0;i<iot->num_items;i++){
 		if(iot->items[i].calc_type==CALC_TYPE_FRAMES_AND_BYTES)
@@ -394,12 +448,12 @@ iostat_draw(void *arg)
 		printf("=");
 	}
 	printf("\n");
-	
-	
+
+
 	printf("IO Statistics\n");
 	if(iot->interval!=G_MAXINT32)
-		printf("Interval: %3" G_GINT64_MODIFIER "u.%06" G_GINT64_MODIFIER "u secs\n", 
-				iot->interval/1000000, iot->interval%1000000);						
+		printf("Interval: %3" G_GINT64_MODIFIER "u.%06" G_GINT64_MODIFIER "u secs\n",
+				iot->interval/1000000, iot->interval%1000000);
 
 	for(i=0;i<iot->num_items;i++){
 		printf("Column #%u: %s\n",i,iot->filters[i]?iot->filters[i]:"");
@@ -412,7 +466,7 @@ iostat_draw(void *arg)
 	for(i=0;i<iot->num_items;i++){
 		if(iot->items[i].calc_type==CALC_TYPE_FRAMES_AND_BYTES){
 			printf("            Column #%-2u           |",i);
-		} else {	
+		} else {
 			printf("    Column #%-2u  |",i);
 		}
 	}
@@ -459,6 +513,8 @@ iostat_draw(void *arg)
 	items=g_malloc(sizeof(io_stat_item_t *)*iot->num_items);
 	frames=g_malloc(sizeof(guint64)*iot->num_items);
 	counters=g_malloc(sizeof(guint64)*iot->num_items);
+	float_counters=g_malloc(sizeof(gfloat)*iot->num_items);
+	double_counters=g_malloc(sizeof(gdouble)*iot->num_items);
 	num=g_malloc(sizeof(guint64)*iot->num_items);
 	/* preset all items at the first interval */
 	for(i=0;i<iot->num_items;i++){
@@ -472,6 +528,8 @@ iostat_draw(void *arg)
 		for(i=0;i<iot->num_items;i++){
 			frames[i]=0;
 			counters[i]=0;
+			float_counters[i]=0;
+			double_counters[i]=0;
 			num[i]=0;
 		}
 		for(i=0;i<iot->num_items;i++){
@@ -482,6 +540,8 @@ iostat_draw(void *arg)
 			if(items[i] && (t<(items[i]->time+iot->interval)) && (t>=items[i]->time) ){
 				frames[i]=items[i]->frames;
 				counters[i]=items[i]->counter;
+				float_counters[i]=items[i]->float_counter;
+				double_counters[i]=items[i]->double_counter;
 				num[i]=items[i]->num;
 			}
 
@@ -515,14 +575,29 @@ iostat_draw(void *arg)
 					break;
 				case CALC_TYPE_SUM:
 					switch(proto_registrar_get_ftype(iot->items[i].hf_index)){
+					case FT_UINT8:
+					case FT_UINT16:
+					case FT_UINT24:
+					case FT_UINT32:
+					case FT_UINT64:
+					case FT_INT8:
+					case FT_INT16:
+					case FT_INT24:
+					case FT_INT32:
+					case FT_INT64:
+						printf(" %15" G_GINT64_MODIFIER "u ", counters[i]);
+						break;
+					case FT_FLOAT:
+						printf(" %f ", float_counters[i]);
+						break;
+					case FT_DOUBLE:
+						printf(" %f ", double_counters[i]);
+						break;
 					case FT_RELATIVE_TIME:
 						counters[i] = (counters[i]+500)/1000;
 						printf(" %8u.%06u ",
 								(int)(counters[i]/1000000), (int)(counters[i]%1000000));
 						break;
-					default:
-						printf(" %15" G_GINT64_MODIFIER "u ", counters[i]);
-	 					break;
 					}
 					break;
 				case CALC_TYPE_MIN:
@@ -541,6 +616,12 @@ iostat_draw(void *arg)
 					case FT_INT64:
 						printf(" %15" G_GINT64_MODIFIER "d ", counters[i]);
 						break;
+					case FT_FLOAT:
+						printf(" %f ", float_counters[i]);
+						break;
+					case FT_DOUBLE:
+						printf(" %f ", double_counters[i]);
+						break;
 					case FT_RELATIVE_TIME:
 						counters[i]=(counters[i]+500)/1000;
 						printf(" %8u.%06u ",
@@ -564,6 +645,12 @@ iostat_draw(void *arg)
 					case FT_INT64:
 						printf(" %15" G_GINT64_MODIFIER "d ", counters[i]);
 						break;
+					case FT_FLOAT:
+						printf(" %f ", float_counters[i]);
+						break;
+					case FT_DOUBLE:
+						printf(" %f ", double_counters[i]);
+						break;
 					case FT_RELATIVE_TIME:
 						counters[i]=(counters[i]+500)/1000;
 						printf(" %8u.%06u ",
@@ -590,6 +677,12 @@ iostat_draw(void *arg)
 					case FT_INT64:
 						printf(" %15" G_GINT64_MODIFIER "d ", counters[i]/num[i]);
 						break;
+					case FT_FLOAT:
+						printf(" %f ", float_counters[i]/num[i]);
+						break;
+					case FT_DOUBLE:
+						printf(" %f ", double_counters[i]/num[i]);
+						break;
 					case FT_RELATIVE_TIME:
 						counters[i]=((counters[i]/num[i])+500)/1000;
 						printf(" %8u.%06u ",
@@ -623,6 +716,8 @@ iostat_draw(void *arg)
 	g_free(items);
 	g_free(frames);
 	g_free(counters);
+	g_free(float_counters);
+	g_free(double_counters);
 	g_free(num);
 }
 
@@ -679,22 +774,22 @@ register_io_tap(io_stat_t *io, int i, const char *filter)
 					fprintf(stderr, "\ntshark: Closing parenthesis missing from calculated expression.\n");
 					exit(10);
 				}
-				
-				
+
+
 				if(io->items[i].calc_type==CALC_TYPE_FRAMES || io->items[i].calc_type==CALC_TYPE_BYTES){
-					if(parenp!=p) { 
-						fprintf(stderr, "\ntshark: %s does require or allow a field name within the parens.\n", 
+					if(parenp!=p) {
+						fprintf(stderr, "\ntshark: %s does require or allow a field name within the parens.\n",
 							calc_type_table[j].func_name);
 						exit(10);
 					}
 				} else {
-					if(parenp==p) { 
+					if(parenp==p) {
 							/* bail out if a field name was not specified */
-							fprintf(stderr, "\ntshark: You didn't specify a field name for %s(*).\n", 
+							fprintf(stderr, "\ntshark: You didn't specify a field name for %s(*).\n",
 								calc_type_table[j].func_name);
 							exit(10);
 					}
-				}				
+				}
 
 				field=g_malloc(parenp-p+1);
 				memcpy(field, p, parenp-p);
@@ -718,8 +813,8 @@ register_io_tap(io_stat_t *io, int i, const char *filter)
 				flt="";
 		}
 	}
-	if(hfi && !(io->items[i].calc_type==CALC_TYPE_BYTES || 
-			    io->items[i].calc_type==CALC_TYPE_FRAMES || 
+	if(hfi && !(io->items[i].calc_type==CALC_TYPE_BYTES ||
+			    io->items[i].calc_type==CALC_TYPE_FRAMES ||
 			    io->items[i].calc_type==CALC_TYPE_FRAMES_AND_BYTES)){
 		/* check that the type is compatible */
 		switch(hfi->type){
@@ -735,8 +830,26 @@ register_io_tap(io_stat_t *io, int i, const char *filter)
 		case FT_INT64:
 			/* these types support all calculations */
 			break;
+		case FT_FLOAT:
+		case FT_DOUBLE:
+			/* these types only support SUM, COUNT, MAX, MIN, AVG */
+			switch(io->items[i].calc_type){
+			case CALC_TYPE_SUM:
+			case CALC_TYPE_COUNT:
+			case CALC_TYPE_MAX:
+			case CALC_TYPE_MIN:
+			case CALC_TYPE_AVG:
+				break;
+			default:
+				fprintf(stderr,
+					"\ntshark: %s is a float field, so %s(*) calculations are not supported on it.",
+					field,
+					calc_type_table[j].func_name);
+				exit(10);
+			}
+			break;
 		case FT_RELATIVE_TIME:
-			/* this type only supports SUM, COUNT, MAX, MIN, AVG */
+			/* this type only supports SUM, COUNT, MAX, MIN, AVG, LOAD */
 			switch(io->items[i].calc_type){
 			case CALC_TYPE_SUM:
 			case CALC_TYPE_COUNT:
@@ -811,11 +924,11 @@ iostat_init(const char *optarg, void* userdata _U_)
 		interval=G_MAXINT32;
 	} else {
 		/* make interval be number of us rounded to the nearest integer*/
-		interval=(gint64)(interval_float*1000000.0+0.5); 
+		interval=(gint64)(interval_float*1000000.0+0.5);
 	}
 
 	if(interval<1){
-		fprintf(stderr, 
+		fprintf(stderr,
 			"\ntshark: \"-z\" interval must be >=0.000001 seconds or \"0\" for the entire capture duration.\n");
 		exit(10);
 	}