Added 29bit ID support to cansniffer #88

coryjfowler · 2018-08-03T15:09:09Z

Moved from CAN_BCM to CAN_RAW. I have most of the functionality restored. I just need to look into how CAN_RAW filtering works and add that ability.

marckleinebudde · 2018-08-06T08:03:45Z

cansniffer.c

@@ -60,7 +61,7 @@
 #include <net/if.h>

 #include <linux/can.h>
-#include <linux/can/bcm.h>
+//#include <linux/can/bcm.h>


Please remove and do not comment out.

marckleinebudde · 2018-08-06T08:03:55Z

cansniffer.c

-void rx_delete (int fd, int id);
-void print_snifline(int id);
+//void rx_setup (int fd, unsigned long id);
+//void rx_delete (int fd, unsigned long id);


marckleinebudde · 2018-08-06T08:04:12Z

cansniffer.c

 int handle_timeo(int fd, long currcms);
 void writesettings(char* name);
 void readsettings(char* name, int sockfd);

+int idx = 0;


static int idx;

as it's a global it will be initialized to 0 automatically.

marckleinebudde · 2018-08-06T08:04:50Z

cansniffer.c

+{
+    unsigned long f = ((struct snif*)elem1)->current.can_id;
+    unsigned long s = ((struct snif*)elem2)->current.can_id;
+    if (f > s) return  1;


don't put if and return in same line

Please use one Tab instead of 4 Spaces for indention.

marckleinebudde · 2018-08-06T08:05:08Z

cansniffer.c

+    return 0;
+}
+
+


one newline is enough

marckleinebudde · 2018-08-06T08:11:33Z

cansniffer.c


 	if (binary) {

-		for (i=0; i<sniftab[id].current.can_dlc; i++) {
+		for (i=0; i<sniftab[position].current.can_dlc; i++) {


Please fix the spaces here, too.

marckleinebudde · 2018-08-06T08:11:46Z

cansniffer.c

-		for (i=0; i<sniftab[id].current.can_dlc; i++)
-			if ((color) && (sniftab[id].marker.data[i]) && (!(sniftab[id].notch.data[i])))
-				printf("%s%02X%s ", ATTCOLOR, sniftab[id].current.data[i], ATTRESET);
+		for (i=0; i<sniftab[position].current.can_dlc; i++)


marckleinebudde · 2018-08-06T08:11:58Z

cansniffer.c

-			    (sniftab[id].current.data[i] < 0x7F))
-				if ((color) && (sniftab[id].marker.data[i]) && (!(sniftab[id].notch.data[i])))
-					printf("%s%c%s", ATTCOLOR, sniftab[id].current.data[i], ATTRESET);
+		for (i=0; i<sniftab[position].current.can_dlc; i++)


marckleinebudde · 2018-08-06T08:12:11Z

cansniffer.c

-						if (sockfd)
-							rx_setup(sockfd, i);
+//						if (sockfd)
+//							rx_setup(sockfd, i);


please remove

marckleinebudde · 2018-08-06T08:12:19Z

cansniffer.c

-						if (sockfd)
-							rx_delete(sockfd, i);
+//						if (sockfd)
+//							rx_delete(sockfd, i);


please remove

marckleinebudde

You probably want to squash this into the first commit.

marckleinebudde

Please don't comment code out, just remove it, if it's not used anymore.

marckleinebudde · 2018-08-06T08:16:00Z

cansniffer.c

@@ -814,3 +764,14 @@ void readsettings(char* name, int sockfd){
 	else
 		printf("unable to read setting file '%s'!\n", fname);
 };
+
+int sniftab_index(canid_t id){


static
and a space between ) {

Please move the function to the beginning of this file, so that you don'y need the forward declaration.

marckleinebudde · 2018-08-06T08:17:59Z

README.md

@@ -13,7 +13,7 @@ subsystem (aka SocketCAN):
 * canplayer : replay CAN logfiles
 * cansend : send a single frame
 * cangen : generate (random) CAN traffic
-* cansniffer : display CAN data content differences (just 11bit CAN IDs)
+* cansniffer : display CAN data content differences (~~just 11bit CAN IDs~~ Not any more!)


Just remove the comment in the about the 11 bit. We've git to have a history of the source code.

Moved from CAN_BCM to CAN_RAW. Restored basic functionality that BCM had natively. There are likely bugs due to some non functionality between how the original program operated with CAN_BCM and now, but it is working on my bench with a NMEA 2000 GPS. Update README.md Restored add/remove ID Restored ability to remove or add active CAN frames. The CAN_ID must already exist in sniftab for this to work. Update cansniffer.c Fixed a potential for a segmentation fault to occur with the sniftab array. Changed arbitrary max array value to defined name for flexibility with increasing the array size. Other polish I forgot about. Settings file operations work Aligned settings file operations for use with 29bit IDs. Instead of saving all 800 "slots" (or 11 bits IDs for the program as originally written), it now saves in use slots based on CAN ID and will load that ID into a slot for the program to potentially match upon. Truncate when overwritting settings Need to truncate settings file... Update README.md Update cansniffer.c Cleaned up some unused commented code.

Corrected formatting, removed old commented code, made suggested code changes. Update README.md

coryjfowler · 2018-08-07T00:37:57Z

I hope that works... I'm not very fluent with 'git'.

marckleinebudde

Please squash all commits into one and adjust the commit message of that commit.

marckleinebudde · 2018-08-17T08:12:38Z

cansniffer.c

+{
+    unsigned long f = ((struct snif*)elem1)->current.can_id;
+    unsigned long s = ((struct snif*)elem2)->current.can_id;
+    if (f > s) return  1;


Please use one Tab instead of 4 Spaces for indention.

marckleinebudde · 2018-08-17T08:13:10Z

cansniffer.c

+int idx = 0;
+
+
+int comp (const void * elem1, const void * elem2) 


Please remove the trailing whitespace

marckleinebudde · 2018-08-17T08:14:13Z

cansniffer.c

-					printf("was only able to read until index %d from setting file '%s'!\n",
-					       i, fname);
+					printf("was only able to read unti from setting file '%s'!\n", fname);
+


Please remove the trailing whitespace.

marckleinebudde · 2018-08-17T08:14:35Z

cansniffer.c

 			for (j=0; j<8 ; j++){
 				sprintf(buf, "%02X", sniftab[i].notch.data[j]);
 				write(fd, buf, 2);
 			}
 			write(fd, "\n", 1);
-			/* 7 + 16 + 1 = 24 bytes per entry */ 
+			/* 12 + 16 + 1 = 29 bytes per entry */ 


Please remove the trailing whitespace.

marckleinebudde · 2018-08-17T08:15:12Z

cansniffer.c

@@ -709,40 +696,50 @@ void readsettings(char* name, int sockfd){
 	int fd;
 	char fname[30] = SETFNAME;
 	char buf[25] = {0};
-	int i,j;
+	int pos,j;


Please add a space after the ,

marckleinebudde · 2018-08-17T08:19:12Z

cansniffer.c

@@ -754,3 +751,14 @@ void readsettings(char* name, int sockfd){
 	else


... and here, too.

marckleinebudde · 2018-08-17T08:19:33Z

cansniffer.c

@@ -754,3 +751,14 @@ void readsettings(char* name, int sockfd){
 	else
 		printf("unable to read setting file '%s'!\n", fname);
 };
+
+int sniftab_index(canid_t id){
+


Please remove that line.

marckleinebudde · 2018-08-17T08:20:26Z

cansniffer.c

@@ -814,3 +764,14 @@ void readsettings(char* name, int sockfd){
 	else
 		printf("unable to read setting file '%s'!\n", fname);
 };
+
+int sniftab_index(canid_t id){


Please move the function to the beginning of this file, so that you don'y need the forward declaration.

marckleinebudde · 2018-08-17T08:21:01Z

cansniffer.c

@@ -593,11 +578,12 @@ int handle_timeo(int fd, long currcms){

 };

-void print_snifline(int id){
+void print_snifline(int slot){


Please make it static

marckleinebudde · 2018-08-17T08:21:20Z

cansniffer.c

@@ -489,45 +451,68 @@ int handle_keyb(int fd){
 	return 1; /* ok */
 };

-int handle_bcm(int fd, long currcms){
+int handle_frame(int fd, long currcms){


Please make it static.

marckleinebudde

I'm sure you've fixed some of these in the second commit. If so, please feel free to ignore my rumblings on these. :D

marckleinebudde · 2018-08-17T09:02:33Z

cansniffer.c

@@ -200,21 +216,21 @@ int main(int argc, char **argv)
 	signal(SIGHUP, sigterm);
 	signal(SIGINT, sigterm);

-	for (i=0; i < 2048 ;i++) /* default: check all CAN-IDs */
+	for (i=0; i < MAX_SLOTS ;i++) /* default: enable all slots */


for (i=0; i < MAX_SLOTS; i++)

marckleinebudde · 2018-08-17T09:03:28Z

cansniffer.c

+		}
+	}
+	else {
+		if(frame.can_dlc == sniftab[pos].current.can_dlc)


marckleinebudde · 2018-08-17T09:04:03Z

cansniffer.c

+	}
+	else {
+		if(frame.can_dlc == sniftab[pos].current.can_dlc)
+			for(i=0; i<frame.can_dlc; i++){


for (i = 0; i < frame.can_dlc; i++) {

marckleinebudde · 2018-08-17T09:04:30Z

cansniffer.c

+	else {
+		if(frame.can_dlc == sniftab[pos].current.can_dlc)
+			for(i=0; i<frame.can_dlc; i++){
+				if(frame.data[i] != sniftab[pos].current.data[i] ){


if (frame.data[i] != sniftab[pos].current.data[i]) {

marckleinebudde · 2018-08-17T09:05:01Z

cansniffer.c


-	sniftab[id].timeout = (timeout)?(currcms + timeout):0;
+		sniftab[pos].current = frame;
+		for (i=0; i < 8; i++)


for (i = 0; i < 8; i++)

marckleinebudde · 2018-08-17T09:08:47Z

cansniffer.c

+				}
+				if (buf[10] & 1) {
+					if (is_clr(pos, ENABLE)) {
+						do_set(pos, ENABLE);


if (buf[10] & 1 && is_clr(pos, ENABLE)) {

I feel that this would create undesired behavior unless I modify the 'else' that follows it.

marckleinebudde · 2018-08-17T09:09:23Z

cansniffer.c

-						if (sockfd)
-							rx_delete(sockfd, i);
+					if (is_set(pos, ENABLE)) {
+						do_clr(pos, ENABLE);
 					}
 				for (j=7; j>=0 ; j--){


for (j = 7; j >= 0; j--) {

marckleinebudde · 2018-08-17T09:09:43Z

cansniffer.c

-					buf[2*j+7] = 0; /* cut off each time */
+					sniftab[pos].notch.data[j] =
+						(__u8) strtoul(&buf[2*j+12], (char **)NULL, 16) & 0xFF;
+					buf[2*j+12] = 0; /* cut off each time */


buf[2 * j + 12] = 0

marckleinebudde · 2018-08-17T09:10:45Z

cansniffer.c

-						(__u8) strtoul(&buf[2*j+7], (char **)NULL, 16) & 0xFF;
-					buf[2*j+7] = 0; /* cut off each time */
+					sniftab[pos].notch.data[j] =
+						(__u8) strtoul(&buf[2*j+12], (char **)NULL, 16) & 0xFF;


You can drop the casts and the & 0xff.
strtoul(&buf[2 * j + 12], NULL, 16);

marckleinebudde · 2018-08-17T09:12:15Z

cansniffer.c

+	int i;
+
+	for(i = 0; i <= idx; i++)
+		if(id == sniftab[i].current.can_id)


if (id == sniftab[i].current.can_id)

hartkopp · 2018-08-17T11:34:14Z

Dear @marckleinebudde and @coryjfowler ,
I've just followed your discussion and I'm personally pretty unsure, how to proceed with cansniffer.
The original approach was to use the entire CAN_BCM functionality to just display the differences filtered inside BCM socket (and therefore in kernel space).
With 29bit IDs this becomes unusable as it makes no sense to create 2^29 BCM-jobs in a socket in the kernel, right? Therefore I tried to merge some portions of candump (using CAN_RAW) and the output generation of cansniffer into a new implementation here: https://github.com/hartkopp/can-utils/commits/master/cansniffer.c
My idea was to support more than one CAN bus at a time, provide a different colorization for them, support CAN FD and sort the output by CAN interface OR by CAN ID.

Maybe my goals where just to ambitious that I got stuck with the implementation?!?

But we need a complete new concept to handle, sort (by ID) and compare incoming frames from CAN_RAW - and I wonder if using the CAN_RAW filters makes sense at all :-/

Any thoughts?

coryjfowler · 2018-08-17T15:54:14Z

Hello @hartkopp,

Right now this implementation checks to see if an ID exists in the sniftab array. It will then either add it to an empty slot or check the existing slot for changes in the frame. If is a new ID or there are changes to the existing, it updates the struct in that slot like the original code. It will then sort the sniftab array by current.ID using 'qsort' so that it is displayed nicely in order with a simple for loop.

My aim for socket filtering is to help alleviate the program from handling a bus with potentially more than 2048 IDs in active use, but the direction here could also just be to change the value of MAX_SLOTS to something sane and recompile.

ChevySSinSD · 2019-02-06T17:34:27Z

I am very interested in using cansniffer for some reverse engineering on my Chevy SS and the GMLAN uses 29 bit extended IDs. Any chance this update to support extended IDs will be merged into the master anytime soon?

brandonros · 2019-04-20T20:29:35Z

This isn't clear: does candump support 29-bit messages as it stands now? Not cansniffer, but candump?

hartkopp · 2019-04-22T18:32:31Z

candump can deal with all kinds for CAN traffic (11bit / 29bit Identifiers, Classic CAN and CAN FD). cansniffer can only cope with 11bit IDs and Classic CAN (8 bytes payload).

mitchdetailed · 2019-12-27T05:21:09Z

The method I used to get around this is to set up a vcan network and mask the id to the 11b value, and send it onto vcan0.

Example shown here: https://github.com/mitchdetailed/python-can-re/blob/master/can0ext-vcan1masked.py

peteg944 · 2020-03-17T03:30:10Z

Still no progress? It's been a year and a half... RIP. Are there any alternatives to cansniffer that work with the full-length IDs?

coryjfowler · 2020-03-17T16:17:50Z

Still no progress? It's been a year and a half... RIP. Are there any alternatives to cansniffer that work with the full-length IDs?

The version on my github has the functionality. However, it has not been accepted here because of styling issues that existed before I even made changes and I simply do not have the time to update those. So its being held back by purely petty reasons...

hartkopp · 2020-04-27T06:52:00Z

Would you like to test some 29bit cansniffer rework?
See #181

BR,
Oliver

hartkopp · 2020-04-30T09:12:25Z

Followed up in issue #181
Integrated this suggested slot approach from @coryjfowler with some more fixes and improvements.
@coryjfowler @SSinSD @brandonros @mitchdetailed @peteg944 :
You are invited to test (and review) this pull request: #200
The latest code can be found here: https://github.com/hartkopp/can-utils/tree/29bit-cansniffer
The single steps of development and improvements can be found here: https://github.com/hartkopp/can-utils/tree/snif

marckleinebudde reviewed Aug 6, 2018

View reviewed changes

coryjfowler force-pushed the master branch from 3914dfc to 384ed8d Compare August 6, 2018 23:22

Formatting, Cleanup, & Suggested Changes

042038f

Corrected formatting, removed old commented code, made suggested code changes. Update README.md

coryjfowler force-pushed the master branch from c56b616 to 042038f Compare August 7, 2018 00:33

marckleinebudde requested changes Aug 17, 2018

View reviewed changes

asbachb mentioned this pull request Aug 28, 2018

cansniffer filtering #30

Closed

hartkopp closed this Apr 30, 2020

		int idx = 0;


		int comp (const void * elem1, const void * elem2)

		@@ -754,3 +751,14 @@ void readsettings(char* name, int sockfd){
		else

Added 29bit ID support to cansniffer #88

Added 29bit ID support to cansniffer #88

Conversation

coryjfowler commented Aug 3, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

marckleinebudde left a comment

Choose a reason for hiding this comment

marckleinebudde left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

coryjfowler commented Aug 7, 2018

marckleinebudde left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

marckleinebudde left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hartkopp commented Aug 17, 2018

coryjfowler commented Aug 17, 2018 • edited Loading

ChevySSinSD commented Feb 6, 2019

brandonros commented Apr 20, 2019

hartkopp commented Apr 22, 2019

mitchdetailed commented Dec 27, 2019

peteg944 commented Mar 17, 2020

coryjfowler commented Mar 17, 2020

hartkopp commented Apr 27, 2020

hartkopp commented Apr 30, 2020

coryjfowler commented Aug 17, 2018 •

edited

Loading