PROCESSOR-SDK-AM437X: Getting timeout message when we enable CAN interface

Please find the below response,

[TI] Which version of the kernel are you using?

[Monish] Kernel version: 4.19

[TI] Could you please post the output for each interface?

[Monish] We are doing loopback between CAN0 and CAN1. First, we are transmitting messages from CAN0 and receiving them through CAN1 and then vice-versa. Please find the below output.

CAN0 Transmit.....CAN1 Receive

interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1
interface = can0, family = 29, type = 3, proto = 1

Received Messages from CAN0

read: Network is down
interface = can1, family = 29, type = 3, proto = 1
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01
<0x00a> [8] 08 07 06 05 04 03 02 01

./pre_compliance_i2c_lcd_remove.sh: line 847: 23340 Killed ( while true; do
msg_candump="$( candump can1 )"; if [[ ${#msg_candump} > 1 ]]; then
echo "$msg_candump";
fi; sleep 0.025;
done )
CAN1 Transmit.....CAN0 Receive

interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1
interface = can1, family = 29, type = 3, proto = 1

Received Messages from CAN1

read: Network is down
interface = can0, family = 29, type = 3, proto = 1
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08
<0x005> [8] 01 02 03 04 05 06 07 08

read: Network is down
./pre_compliance_i2c_lcd_remove.sh: line 847: 23376 Killed ( while true; do
msg_candump="$( candump can0 )"; if [[ ${#msg_candump} > 1 ]]; then
echo "$msg_candump";
fi; sleep 0.025;
done )

[TI] ip -d -s link show <can interface>

[Monish] Please find the below output image.

Regards,

Monish P

Hi Schuyler,

Thanks for the response. 

While transmitting a message in the loop we are having a 0.5s delay between the messages. Whether the delay between messages need to be increased to 1 sec? 

Also, find the below CAN test script.

echo " "
echo "CAN0 Transmit.....CAN1 Receive"
echo " "

ip link set can0 type can bitrate 1000000	# set bit rate
ip link set can0 up				# set can0 up

ip link set can1 type can bitrate 1000000	# set bit rate
ip link set can1 up				# set can1 up

sleep 1
a=0;                                                          
(       
  while true;                                                 
  do                                                          
    msg_candump="$( candump can1 )" 		# read CAN    
    if [[ ${#msg_candump} > 1 ]]; then 		#received msg          
      echo "$msg_candump"                                     
    fi                                                       
    sleep 0.025       
  done                                       
) &                                                     
PID=$! 

echo 0 > /sys/class/gpio/gpio160/value		# Enable CAN0 Transeiver sleep pin
echo 0 > /sys/class/gpio/gpio161/value		# Enable CAN1 Transeiver sleep pin

sleep 0.5

while true;
do
    cansend can0 --identifier=10 8 7 6 5 4 3 2 1 	#Transmit CAN message
    sleep 0.5
    if [ $a -eq 10 ]
    then
      break
    fi
    ((a=a+1))
done
echo " "
echo "Received Messages from CAN0 "
echo " "
ip link set can0 down                                         
ip link set can1 down

kill -9 $PID

echo " "
echo " "
sleep 2
echo "CAN1 Transmit.....CAN0 Receive"

echo " "

ip link set can0 type can bitrate 1000000		# set bit rate
ip link set can0 up					# set can0 up

ip link set can1 type can bitrate 1000000		# set bit rate
ip link set can1 up					# set can1 up

sleep 1 

a=0;                                                          
(     
  while true;                                                 
  do                                                          
    msg_candump="$( candump can0 )" 			# read CAN   
    if [[ ${#msg_candump} > 1 ]]; then 			#received msg          
      echo "$msg_candump"                                     
    fi                                                       
    sleep 0.025       
  done                                       
) &                                                     
PID=$!

sleep 0.5

while true;
do
    cansend can1 --identifier=5 1 2 3 4 5 6 7 8		#Transmit CAN message
    sleep 0.5
    if [ $a -eq 10 ]
    then
      break
    fi
    ((a=a+1))
done

echo " "                              
echo "Received Messages from CAN1 "
echo " " 

ip link set can0 down				#Down can 0
ip link set can1 down				#Down can 1

echo 1 > /sys/class/gpio/gpio160/value		# Disable CAN0 Transeiver sleep pin
echo 1 > /sys/class/gpio/gpio161/value		# Disable CAN1 Transeiver sleep pin
kill -9 $PID

Hi,

Could you try reducing the message count in the loops to 1 as a means of identifying if the issue is an overrun condition. How long has the test been running before seeing the first timeout? Could describe how you are bringing up the interfaces and when the test starts? Is the test continuously running or is being stopped and restarted?

Best Regards,

Schuyler

Hi,

Thank you for the update. The error occurring on disabling and enabling the interface I will look into with the development team. Also are you seeing any message loss?

Best Regards,

Schuyler

Hi,

I need the results of uname -a too please. 

Thanks and Best Regards,

Schuyler

Hi Schuyler,

We are not observing any message loss. Actually, we are transmitting 11 data from CAN0 to CAN1, and vice-versa, we observe the expected output on both cases.

Regards,

Monish P

Hi, 

Please find the result of uname -a.

Hi,

I apologize, I should have asked for the results of uname -a when I asked for kernel version. That kernel you are running is not a TI released kernel. Is there a reason this kernel version is being used? Where is the kernel source being cloned from?

I would recommend switcing back to using the kernel that is in the TI SDK v6.03 which is kernel Version: 4.19.94-rt39. The reason I recommend this is because pulling from mainline and compiling for RT is likely to miss patches that TI applies and tests. After the SDK is released testing and auto-merge stop on RT kernels.

Here is the link to ti kernel tree for the 6.03 RT Kernel. 

Best Regards,

Schuyler

Hi Schuyler,

Our image is based on ti 4.19.94-rt39

On this version, we have also kept ourselves updated on a few of the patches that have been applied on upstream 4.19-rt-kernel and also few patches for CVE vulnerabilities.

Regards,

Monish P

Hi Monish,

If possible I would like to recommend as a means of problem isolation trying the pre-built TI kernel & drivers and see if you see the same issue.  

Best Regards,

Schuyler