CAN controller

[Xanatos]

Gerd,

My Kernel is currently 3.10.25+.
I used the rpi_can_310x_spi_dma.tar.xz from your site and as far as I can see it includes the modules for this kernel too. Did I missunderstand that?

Thanks for your help!

Regards,
Marco

[bertr2d2]

Hi Marco,

Gerd,

My Kernel is currently 3.10.25+

So, you already upgraded the kernel.

I used the rpi_can_310x_spi_dma.tar.xz from your site and as far as I can see it includes the modules for this kernel too. Did I missunderstand that?

Yes, it should have both versions - the original 3.10.24+ and 310.25+. But I didn't test it myself. Do you get the same error as before ?

Regards

Gerd

[Xanatos]

Gerd,

I only tested with 3.10.25+. I will do a fresh img to an SD and don't do rpi-update and test again.
Did you tested youself with 3.10.24+?

Regards,
Marco

[Xanatos]

Gerd,

I repeated my steps without rpi-update and I get rid of the module problem. Thanks!

After extracting the archive and depmod -a I did:

Code: Select all

modprobe spi-bcm2708
modprobe can
modprobe can-dev
modprobe can-raw
modprobe can-bcm
modprobe spi-config devices=\ bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_release

But my dmesg don't look good.

Code: Select all

[  117.903868] bcm2708_spi bcm2708_spi.0: master is unqueued, this is deprecated
[  117.904799] bcm2708_spi bcm2708_spi.0: SPI Controller at 0x20204000 (irq 80)
[  117.965216] can: controller area network core (rev 20120528 abi 9)
[  117.965383] NET: Registered protocol family 29
[  118.033730] CAN device driver interface
[  118.074337] can: raw protocol (rev 20120528)
[  118.115628] can: broadcast manager protocol (rev 20120528 t)
[  118.299031] spi_config_register: device description:  bus=0:cs=0:modalias=mcp          2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_relea          se
[  118.300056]  spi_config_register: unsupported argument  bus - ignoring config

cat /sys/bus/spi/devices/*/modalias

Code: Select all

spi:spidev
spi:spidev

cat /proc/interrupts

Code: Select all

           CPU0
  3:       9294   ARMCTRL  BCM2708 Timer Tick
 32:     114742   ARMCTRL  dwc_otg, dwc_otg_pcd, dwc_otg_hcd:usb1
 52:          0   ARMCTRL  BCM2708 GPIO catchall handler
 65:          7   ARMCTRL  ARM Mailbox IRQ
 66:          1   ARMCTRL  VCHIQ doorbell
 75:          1   ARMCTRL
 77:       8097   ARMCTRL  bcm2708_sdhci (dma)
 80:          0   ARMCTRL  bcm2708_spi.0
 83:         19   ARMCTRL  uart-pl011
 84:      11347   ARMCTRL  mmc0
FIQ:              usb_fiq
Err:          0

ip -s -d link show can0

Code: Select all

Device "can0" does not exist.

You know what could be wrong?

Regards,
Marco

[bertr2d2]

Code: Select all

[  117.903868] bcm2708_spi bcm2708_spi.0: master is unqueued, this is deprecated
[  117.904799] bcm2708_spi bcm2708_spi.0: SPI Controller at 0x20204000 (irq 80)
[  117.965216] can: controller area network core (rev 20120528 abi 9)
[  117.965383] NET: Registered protocol family 29
[  118.033730] CAN device driver interface
[  118.074337] can: raw protocol (rev 20120528)
[  118.115628] can: broadcast manager protocol (rev 20120528 t)
[  118.299031] spi_config_register: device description:  bus=0:cs=0:modalias=mcp          2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_relea          se
[  118.300056]  spi_config_register: unsupported argument  bus - ignoring config

There seems to be a leading space before the bus argument.

[Xanatos]

Thanks Gerd! That did the trick CAN0 is now online. Will try with a real can device now to see if I can receive something

Regards,
Marco

[djrem]

Hi all =)

first of all i want to thank all the guys here. you did a great work and without you i would not start using RPI^^

i have a problem: my RPI with skpang extension board works only in loopback mode, but not with the real devices. After "candump can0" nothing happens. I tried to use it with http://www.peak-system.com/PCAN-USB.199.0.html.


dmesg:

Code: Select all

can: controller area network core (rev 20120528 abi 9)
NET: Registered protocoll family 29
CAN device driver interface
can: raw protocol (rev 20120528)
can: broadcast manager protocol (rev 20120528 t)
spi_config_register: device description: bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x002:force_release
spi_config_match_cs: SPI0: check CS=0 to be 0
spi_config_match_cs: SPI0.0: Found a device with modinfo spidev
spi_config_register:spi0.0:mcp2515: found already registered device
spi_config_register:spi0.0:mcp2515: forcefully releasing already registered device taints kernel
mcp251x spi0.0: CANSTAT 0x80 CANCTRL 0x07
mcp251x spi0.0: probed
spi_config_register:spi0.0: registering modalias=mcp2515 with max_speed_hz=10000000 mode =0 and gpio/irq=25/195
spi_config_register:spi0.0:platform data: daadbd60: 00 25 f4 00 02 20 00 00 00 00 00 00 00 00 00 00 .$... ..........
spi_config_register:spi0.0:platform data: daadbd70: 00 00 00 00                                                                  ....
mcp251x spi 0.0 can0: bit-timing not yet defined
mcp251x spi 0.0: unable to set initial baudrate!
...
a lot of this messages
...
mcp251x spi0.0: CNF: 0x00 0xb5 0x01

ifconfig can0:

Code: Select all

can0   Link encap:UNSPEC HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00-00
            UP RUNNING NOARP MTU:16 METRIC:1
            RX packets:0 errors:0 dropped:0 overruns:0 frame:0
            TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
            collisions:0 txqueuelen:10
            RX bytes:0 (0.0B)   TX bytes:0 (0.0B)

list of loaded modules:

Code: Select all

Module                  Size  Used by
fuse                   76145  1 
spi_config              9367  0 
mcp251x                 9557  0 
can_bcm                11476  0 
can_raw                 6183  0 
can_dev                 9884  1 mcp251x
can                    23591  2 can_bcm,can_raw
snd_bcm2835            16165  0 
snd_soc_bcm2708_i2s     5474  0 
regmap_mmio             2806  1 snd_soc_bcm2708_i2s
snd_soc_core          131268  1 snd_soc_bcm2708_i2s
snd_compress            8076  1 snd_soc_core
regmap_i2c              1645  1 snd_soc_core
regmap_spi              1897  1 snd_soc_core
snd_pcm                81593  2 snd_bcm2835,snd_soc_core
snd_page_alloc          5156  1 snd_pcm
snd_seq                53769  0 
snd_seq_device          6473  1 snd_seq
snd_timer              20133  2 snd_pcm,snd_seq
evdev                   9419  2 
leds_gpio               2059  0 
led_class               3688  1 leds_gpio
snd                    61291  7 snd_bcm2835,snd_soc_core,snd_timer,snd_pcm,snd_seq,snd_seq_device,snd_compress
spi_bcm2708             4728  0 

cat /proc/interrupts

Code: Select all

           CPU0       
  3:      74359   ARMCTRL  BCM2708 Timer Tick
 32:    2774379   ARMCTRL  dwc_otg, dwc_otg_pcd, dwc_otg_hcd:usb1
 52:          0   ARMCTRL  BCM2708 GPIO catchall handler
 65:          8   ARMCTRL  ARM Mailbox IRQ
 66:          1   ARMCTRL  VCHIQ doorbell
 75:          1   ARMCTRL
 77:      10794   ARMCTRL  bcm2708_sdhci (dma)
 80:         36   ARMCTRL  bcm2708_spi.0
 83:         19   ARMCTRL  uart-pl011
 84:      19964   ARMCTRL  mmc0
195:          0      GPIO  mcp251x
FIQ:              usb_fiq
Err:          0

cat /sys/bus/spi/devices/*/modalias

Code: Select all

spi:mcp2515
spi:spidev

You know what could be wrong?

Regards, Roman

[mroberto]

Hello everyone! I just joined this forum because it seems to be such a great resource with tons of great information. Thanks to everyone for all of the information that has been posted!

I'd like to try contribute a bit on my first post. I recently purchased the PICAN CAN-bus board (http://skpang.co.uk/catalog/pican-canbu ... -1196.html). This might be the same board that Mayco and st1g have. Mine was purchased from SK Pang Electronics and the schematic that's provided from them shows three solder jumpers near the DB9 connector. These jumpers are used to connect the CAN H, CAN L and GND signals to the DB9 pins. The way my board was shipped, these jumpers were not connected. This might be why you're not receiving any data when connecting to your car's OBDII system.

[radman]

Hi at all
with an indispensable help of this post chain I was able to setup my pican board
http://skpang.co.uk/catalog/pican-canbu ... -1196.html
and get CAN going without patching the kernel (spi-config is really pretty cool!).
I had to recompile the kernel though to include all the necessary modules.

I run into an issue and I'm completely stuck:

1. The interface is started with following options: set can0 type can bitrate 125000 restart-ms 100 triple-sampling on
2. After starting can interface I start candump: candump -ta -c can0
3. I have an external hardware generating traffic on canbus and sending message every 10ms.
4. After starting message generation - candump starts capturing messages but after a while it stops and hangs.
5. The only way to get the interface back and running is to restart the hardware: ifconfig can0 down; ifconfig can0 up

After going deeper I noticed that it looks like the select() in candump is hanging.
Any ideas gentleman how to bite into it and figure out why it is happening?
I tried it with 3.10.y and 3.12.y kernels - the same result.

Code: Select all

pi@raspberrypi ~ $ dmesg | egrep -i "can|spi|mcp"
[    5.483414] bcm2708_spi bcm2708_spi.0: master is unqueued, this is deprecated
[    5.591924] bcm2708_spi bcm2708_spi.0: SPI Controller at 0x20204000 (irq 80)
[   11.306202] spi_config_register: device description: bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_release
[   11.328815] spi_config_register:spi0.0: registering modalias=mcp2515 with max_speed_hz=10000000 mode=0 and gpio/irq=25/195
[   11.358623] spi_config_register:spi0.0:platform data:da0963c0: 00 24 f4 00 02 20 00 00 00 00 00 00 00 00 00 00  .$... ..........
[   11.385710] spi_config_register:spi0.0:platform data:da0963d0: 00 00 00 00                                      ....
[   11.407797] CAN device driver interface
[   11.460230] mcp251x spi0.0: probed
[   19.236877] mcp251x spi0.0: CNF: 0x03 0xf5 0x01
[  161.444663] can: controller area network core (rev 20120528 abi 9)
[  161.464498] can: raw protocol (rev 20120528)

This is the info after running and stopping candump:

Code: Select all

pi@raspberrypi ~ $ ip -details -statistics link show can0
3: can0: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN mode DEFAULT qlen 10
    link/can
    can <TRIPLE-SAMPLING> state ERROR-ACTIVE restart-ms 100
    bitrate 125000 sample-point 0.875
    tq 500 prop-seg 6 phase-seg1 7 phase-seg2 2 sjw 1
    mcp251x: tseg1 3..16 tseg2 2..8 sjw 1..4 brp 1..64 brp-inc 1
    clock 8000000
    re-started bus-errors arbit-lost error-warn error-pass bus-off
    0          0          0          0          0          0
    RX: bytes  packets  errors  dropped overrun mcast
    2401       343      0       0       0       0
    TX: bytes  packets  errors  dropped carrier collsns
    0          0        0       0       0       0

[Xanatos]

Hi everybody,

I have another question as I'm still not able to read any packets from my CAN Device.
I think it has to do with the bitrate. In the docs it say 20kBit/s should be used for my device and when I test with my USB to CAN it works. But what is the correct parameter I need to use on the rpi?

Code: Select all

ip link set can0 up type can restart-ms 100 bitrate 20000

Regards,
Marco

[radman]

I have another question as I'm still not able to read any packets from my CAN Device.
I think it has to do with the bitrate. In the docs it say 20kBit/s should be used for my device and when I test with my USB to CAN it works. But what is the correct parameter I need to use on the rpi?

Code: Select all

ip link set can0 up type can restart-ms 100 bitrate 20000

Hey Marco,
I got my skpang board working just few days ago. The bitrate refers to CAN bus bitrate so if you are shooting for 20kbit/s bus speed - it should work. The typical series of values for busrates are: 10k, 20k, 50k, 125k, 250k, 500k, 1M. I tested the board 125k and 1M - works fine but is not stable.
I configure mine as follows and it works for some time (see my other post):

Code: Select all

sudo modprobe spi-config devices=bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_release
sudo ip link set can0 type can bitrate 125000
sudo ifconfig can0 up

Have you tried on your rpi to run in one terminal window:

Code: Select all

candump can0

and run in other terminal window:

Code: Select all

cansend can0 123#deadbeef

prior to actually using an external CAN device?

[Cky_2k]

Greetings,

I am having trouble getting my PiCan board from http://skpang.co.uk/catalog/pican-canbu ... -1196.html to work. Skpang has instructions on there website for a pre-compiled kernel image but it is out of date. Can someone please help me.

These are the steps I have taken so far:

1: Download fresh 2013-12-20-wheezy-raspbian.img
2: RPI: uname -a : Kernel verison 3.10.24+
3. Download rpi_can_310x_spi_dma.tar.xz
4. Copied it onto a usb stick from my mac
5. RPI: mkdir /media/usbstick
6. RPI: mount -t vfat -o uid=pi,gid=pi /dev/sda1 /media/usbstick
7. RPI: cd /media/usbstick
8. RPI: ls -l (Shows rpi_can_310x_spi_dma.tar.xz)

I'm new to Linux and would really appreciate it if some one could please walk me thought the next steps.

[radman]

Greetings,

I am having trouble getting my PiCan board from http://skpang.co.uk/catalog/pican-canbu ... -1196.html to work. Skpang has instructions on there website for a pre-compiled kernel image but it is out of date. Can someone please help me.

Current, updated instructions can be found here:
http://lnxpps.de/rpie/

[Cky_2k]

Greetings,

I am having trouble getting my PiCan board from http://skpang.co.uk/catalog/pican-canbu ... -1196.html to work. Skpang has instructions on there website for a pre-compiled kernel image but it is out of date. Can someone please help me.

Current, updated instructions can be found here:
http://lnxpps.de/rpie/

I have Raspbian 2013-12-20 and rpi_can_310x_spi.tar.xz downloaded. I just don't know how to transfer over the rpi_can_310x_spi.tar.xz onto the pi.

Copied it onto a usb stick from my mac
5. RPI: mkdir /media/usbstick
6. RPI: mount -t vfat -o uid=pi,gid=pi /dev/sda1 /media/usbstick
7. RPI: cd /media/usbstick
8. RPI: ls -l (Shows rpi_can_310x_spi_dma.tar.xz)

How do I transfer it over and where?

[radman]

Everything should be clear after reading posts in this thread and http://lnxpps.de/rpie/

[radman]

After going deeper I noticed that it looks like the select() in candump is hanging.
Any ideas gentleman how to bite into it and figure out why it is happening?
I tried it with 3.10.y and 3.12.y kernels - the same result.

I just tested the update posted on http://lnxpps.de/rpie/ (Update: Just use Raspbian 2013-12-20 with this update).
I see the same issue with candump hanging upon heavy traffic.
Has anybody noticed it or there is something wrong with my hardware?

I also have performed following test.

1. Terminal #1: candump -ta -c can0
2. Terminal #2: cangen can0 -g 10
3. Komodo CAN analizer also monitoring traffic on the bus.

As long as cangen was running candump and komodo were only listening on the bus everything was great.
As soon as I started generating additonal bus messages with komodo:
candump stopped receiving messages
cangen crashed with a message: write: No buffer space available
Ideas?

[cadsenthilkumar]

I prefer to have it more automatic,

This is the best method I found so far,

P.S. I'm log as root. If you are not , use the sudo command.

1- from a stock raspbian we need to change the kernel

Code: Select all

cd /tmp
wget http://lnxpps.de/rpie/rpi_can_3611_spi_dma.tar.xz
cd /
tar Jxvf /tmp/rpi_can_3611_spi_dma.tar.xz 
depmod -a

reboot

2- We need to update the /etc/modprobe.d/raspi-blacklist.conf file
Enable the spi-bcm2708 driver and blacklist the mcp251x driver

Code: Select all

#blacklist spi-bcm2708
blacklist mcp251x

3- Let's add the spi-config and the new mcp2515 modules in file /etc/modules

Code: Select all

# MCP2515 configuration with /INT on GPIO25 and 16MHz clock
spi-config devices=\
bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_release

#and of course the mcp2515 driver 
mcp2515

4- And now the network can0 auto connect. Add these in the /etc/network/interfaces

Code: Select all

auto can0
iface can0 inet manual
     pre-up /sbin/ip link set can0 type can bitrate 125000
     up /sbin/ifconfig can0 up
     down /sbin/ifconfig can0 down

and reboot!

if everything is ok , the command ifconfig should report something like

Code: Select all

root@RpiTest:~# ifconfig
can0      Link encap:UNSPEC  HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00-00  
          UP RUNNING NOARP  MTU:16  Metric:1
          RX packets:6674 errors:0 dropped:0 overruns:0 frame:0
          TX packets:58915 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 lg file transmission:10 
          RX bytes:6695 (6.5 KiB)  TX bytes:176745 (172.6 KiB)

I don't know if this is the best method but it is working fine for me.

Daniel

Hi Daniel,
I followed your post.Its really great.
I need to know what is loop back and how to enable loopback on, triple-sampling on..

[cadsenthilkumar]

Hi cadsenthilkumar,

I did check the method and didn't find any problem.

I changed the crystal to 8Mhz and everything was ok.

Maybe the best way to check is to use two MCP2515 on the raspberry pi and see if you can send from one to the other.

On the First MCP2515 I use the schematic from this site.

http://skpang.co.uk/catalog/pican-canbu ... -1196.html


And for the second MCP2515 I use CE1 instead of CE0 and GPIO24 instead of GPIO25

I add the second device into the /etc/modules

Code: Select all

root@RpiTest:~# cat /etc/modules
# /etc/modules: kernel modules to load at boot time.
#
# This file contains the names of kernel modules that should be loaded
# at boot time, one per line. Lines beginning with "#" are ignored.
# Parameters can be specified after the module name.

snd-bcm2835

# MCP2515 configuration w[code]

ith /INT on GPIO25 and 16MHz clock
spi-config devices=\
bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_release,bus=0:cs=1:modalias=mcp2515:speed=10000000:gpioirq=24:pd=20:pds32-0=8000000:pdu32-4=0x2002:force_release
#and of course the mcp2515 driver
mcp2515
root@RpiTest:~#[/code]

B.T.W. The second MCP2515 run with a 8Mhz crystal.


and into /etc/network/interfaces

Code: Select all

root@RpiTest:~# cat /etc/network/interfaces
auto lo

iface lo inet loopback
#iface eth0 inet dhcp

allow-hotplug wlan0
#iface wlan0 inet manual
#wpa-roam /etc/wpa_supplicant/wpa_supplicant.conf
iface wlan0 inet dhcp
wpa-conf /etc/wpa_supplicant/wpa_supplicant.conf
#pre-up ifdown eth0; echo -n '1-1.1:1.0' >/sys/bus/usb/drivers/smsc95xx/unbind
#post-down echo -n '1-1.1:1.0' >/sys/bus/usb/drivers/smsc95xx/bind
iface default inet dhcp

auto can0
iface can0 inet manual
     pre-up /sbin/ip link set can0 type can bitrate 125000
     up /sbin/ifconfig can0 up
     down /sbin/ifconfig can0 down

auto can1
iface can1 inet manual
     pre-up /sbin/ip link set can1 type can bitrate 125000
     up /sbin/ifconfig can1 up
     down /sbin/ifconfig can1 down
root@RpiTest:~# 

I open 2 ssh consoles

On the first one I type

Code: Select all

root@RpiTest:~# candump can0

On the second console I type

Code: Select all

root@RpiTest:~# cansend can1 7ef#0101050000000000

And I got a respond on the first console

Code: Select all

can0  7EF   [8]  01 01 05 00 00 00 00 00

This is a neat trick to check if the canbus is working on the Rpi side.



Daniel

Hi Daniel,
followed the same connection as given in http://skpang.co.uk/catalog/images/rasp ... _rev_B.pdf
and still candump is not showing anything,can you tell whats the problem
and follwed the same steps as your post in page 14 of this forum.
the virtual can is working.
Thanks.

[Cky_2k]

I am following the tutorial on "How to Build the Modules for adding an mcp251x to a Raspberry Pi" http://lnxpps.de/rpie/raspi-anleitung.txt but I keep receiving this message when I get to git checkout.

paul@KoolKat:~/raspi/linux$ git checkout 16eb921a96db3dc3555a65b53b99c15753e6b770
fatal: reference is not a tree: 16eb921a96db3dc3555a65b53b99c15753e6b770

Any help would be appreciated!

[bertr2d2]

Hi,

I just tested the update posted on http://lnxpps.de/rpie/ (Update: Just use Raspbian 2013-12-20 with this update).
I see the same issue with candump hanging upon heavy traffic.
Has anybody noticed it or there is something wrong with my hardware?

I also have performed following test.

1. Terminal #1: candump -ta -c can0
2. Terminal #2: cangen can0 -g 10
3. Komodo CAN analizer also monitoring traffic on the bus.

As long as cangen was running candump and komodo were only listening on the bus everything was great.
As soon as I started generating additonal bus messages with komodo:
candump stopped receiving messages
cangen crashed with a message: write: No buffer space available
Ideas?

the "crash" under heavy load is a known issue since 3.10.x. The DMA SPI driver from Martin Sperl using DMA
based on notros SPI driver improves the behaviour.

The MCP2515 looks easy to connect to the RPi, but the SPI driver latency is the bottleneck for high speed CAN.
If you need full speed CAN you should considering using a Beaglebone Black. The CAN controller is integrated
- you only need a tranceiver (eg MCP2551 or ISO1050).

Don't know if Martin is still working on the SPI driver.

Regards

Gerd

[radman]

Hi Gerd,
thank you for the pointer. I found this SPI driver but when I tried it with 3.12.y it turned out to be even less stable than the default driver from the rpi branch. I used following driver for the test:
https://github.com/notro/spi-bcm2708

As next steps I tried 3.9.y kernel version (default and DMA driver) - also less stable than the 3.12.y.
In a meantime I got it to work stable enough with default 3.12.y to play with it at 250kbit. The beagle bone black is on order.

If anybody has a recommendation for a kernel driver version along with SPI driver version (git links please), I'm willing to give it a quick test and feedback results to this forum. I put together small script on ubuntu for automated kernel cross-compilation and deployment for very quick turnaround.

[bertr2d2]

Hi radman,

Hi Gerd,
thank you for the pointer. I found this SPI driver but when I tried it with 3.12.y it turned out to be even less stable than the default driver from the rpi branch. I used following driver for the test:
https://github.com/notro/spi-bcm2708

As next steps I tried 3.9.y kernel version (default and DMA driver) - also less stable than the 3.12.y.
In a meantime I got it to work stable enough with default 3.12.y to play with it at 250kbit.

AFAIK the problem began with the kernel developer decision, that all drivers need to set IRQF_ONESHOT if you don't have a hardirq function call. Due to the architecture of the MCP2515 driver a hardirq function doesn't make any sense.
As far as i can remember the Raspbian dists based on 3.2.x+ didn't had the problem.

The beagle bone black is on order.

A good decision. BTW: If you give up cape management you are getting 2 CAN interfaces.

If anybody has a recommendation for a kernel driver version along with SPI driver version (git links please), I'm willing to give it a quick test and feedback results to this forum. I put together small script on ubuntu for automated kernel cross-compilation and deployment for very quick turnaround.

Regards

Gerd

[radman]

Hi Gerd,

After some more testing I have to say that for my testing methodology, I haven't found a stable enough SPI driver (regardless of distro version). It is probably related to the way how the driver has been tested. My current test which I perform goes as follows:

1. Bus analizer/exerciser, one external device and raspberry pi with skpang cape connected together.
2. SkPang board has a JP3 shorted to provide bus termination. http://skpang.co.uk/catalog/pican-canbu ... -1196.html
3. Bus is configured at 250kbit.
4. There is a can dump running on raspberry pi: candump -ta -c can0
5. Bus exerciser generates a 7 byte message every 50ms and external device reponds to it with 7 byte a message. Response time of external device is ~2ms.
6. After a short period of time (<1min) the candump stops and does not resume capture after restarting. The only recovery method is ifconfig can0 down/ifconfig can0 up.

I assume that this behavior indicates crash or malfunction of the driver. The instability increases with increasing the bitrate of the bus: 10kbit is the most stable and 1Mbit is almost unusable. Correct me if I'm wrong but shouldn't 700MHz cpu with 10MHz SPI bus and MCP2515 with 16MHz quartz be sufficient to get 1Mbit out of CAN bus?
I think it is good to know current limitations in CAN bus area while playing with raspberry pi.

[bertr2d2]

Hi radman,

Hi Gerd,

After some more testing I have to say that for my testing methodology, I haven't found a stable enough SPI driver (regardless of distro version). It is probably related to the way how the driver has been tested. My current test which I perform goes as follows:

1. Bus analizer/exerciser, one external device and raspberry pi with skpang cape connected together.
2. SkPang board has a JP3 shorted to provide bus termination. http://skpang.co.uk/catalog/pican-canbu ... -1196.html
3. Bus is configured at 250kbit.
4. There is a can dump running on raspberry pi: candump -ta -c can0
5. Bus exerciser generates a 7 byte message every 50ms and external device reponds to it with 7 byte a message. Response time of external device is ~2ms.

That's 40 packets per second, right ? I wouldn't call this a high rate.

[*]After a short period of time (<1min) the candump stops and does not resume capture after restarting. The only recovery method is ifconfig can0 down/ifconfig can0 up.[/list]

Here is my setup (old Raspbian 2013-09-25 Kernel 3.6.11+):
- USB2CAN interface from 8devices (BTW: my reference, a good catch) on can0
- PiCAN as can1 using async mcp2515 module
running at 1Mbps with this test code http://lnxpps.de/rpie/pican_test.c
Result:

Code: Select all

root@raspberrypi ~ # ./pican_test
CAN Socket can0 opened
CAN Socket can1 opened
Send 100000 in 101.656 sec (0.984 tpackets/sec 0.078 Mbit)
./can_bounce  1.39s user 15.05s system 16% cpu 1:41.67 total

Roughly 500 sent and 500 received packets per second.

I assume that this behavior indicates crash or malfunction of the driver. The instability increases with increasing the bitrate of the bus: 10kbit is the most stable and 1Mbit is almost unusable. Correct me if I'm wrong but shouldn't 700MHz cpu with 10MHz SPI bus and MCP2515 with 16MHz quartz be sufficient to get 1Mbit out of CAN bus?

Yeah, I took a long time to get the point. At the beginning of the adventure connecting the mcp2515 to the RPi I was standing at the same point as you: The RPi should have enough power to get a good and reliable CAN interface with the simple MCP2515. Days later after spending hours in front of my oscilloscope and digging into the code I must admit that the combination of Linux, RPi, the SPI and MCP2515 driver isn't a good combination: the problem is the latency - see many posts before. There are some oscilloscope pictures, which might get your an idea.
To get things worse the MCP2515 must be read via SPI three times to get a single frame. Have a look at the interrupts:

Code: Select all

root@raspberrypi ~ # cat /proc/interrupts 
           CPU0       
  3:      57529   ARMCTRL  BCM2708 Timer Tick
 16:     449173   ARMCTRL  bcm2708_spi.0
 32:   19745859   ARMCTRL  dwc_otg, dwc_otg_pcd, dwc_otg_hcd:usb1
 52:      99984   ARMCTRL  BCM2708 GPIO catchall handler
 65:       5673   ARMCTRL  ARM Mailbox IRQ
 66:          1   ARMCTRL  VCHIQ doorbell
 75:          1   ARM CTRL
 77:       7240   ARMCTRL  bcm2708_sdhci (dma)
 80:         27   ARMCTRL  bcm2708_spi.0
 83:         19   ARMCTRL  uart-pl011
 84:      13719   ARMCTRL  mmc0
195:      99984      GPIO  can1
FIQ:              usb_fiq
Err:          0

The MCP2515 interrupts (GPIO can1) generates a 4 time fold number of interrupts at the SPI (bcm2708_spi.0).
Every interrupt means a context switch which generates additional latency.
Linux isn't a realtime OS and the RPi does the rest. Don't get me wrong, I like my RPi, but it's not designed to be
a good performer in regard to devices. For example the network interfaces generates 8000 FIQ which isn't shown
interrupt statistics.
Notro and Martin did a good job improving the SPI driver (which is the key factor), but it isn't perfect.

The problem you are facing today (hang in candump) might be, that the kernel missed the interrupts from the CAN controller. Have a look a the MCP interrupt pin - I'm quite sure that it is at 0V. The kernel missed some edge interrupts from the controller and the buffer from the controller is full. The MCP2515 is still waiting for the host (RPi) to get the interrupt cleared. This is normally done by reading RXB0 or RXB1 thru SPI.
Nevertheless check your cabling - I didn't saw this kind of problem at the rate you are using CAN before.

I think it is good to know current limitations in CAN bus area while playing with raspberry pi.

Regards

Gerd

[cadsenthilkumar]

Hi cadsenthilkumar,

I did check the method and didn't find any problem.

I changed the crystal to 8Mhz and everything was ok.

Maybe the best way to check is to use two MCP2515 on the raspberry pi and see if you can send from one to the other.

On the First MCP2515 I use the schematic from this site.

http://skpang.co.uk/catalog/pican-canbu ... -1196.html


And for the second MCP2515 I use CE1 instead of CE0 and GPIO24 instead of GPIO25

I add the second device into the /etc/modules

Code: Select all

root@RpiTest:~# cat /etc/modules
# /etc/modules: kernel modules to load at boot time.
#
# This file contains the names of kernel modules that should be loaded
# at boot time, one per line. Lines beginning with "#" are ignored.
# Parameters can be specified after the module name.

snd-bcm2835

# MCP2515 configuration w[code]

ith /INT on GPIO25 and 16MHz clock
spi-config devices=\
bus=0:cs=0:modalias=mcp2515:speed=10000000:gpioirq=25:pd=20:pds32-0=16000000:pdu32-4=0x2002:force_release,bus=0:cs=1:modalias=mcp2515:speed=10000000:gpioirq=24:pd=20:pds32-0=8000000:pdu32-4=0x2002:force_release
#and of course the mcp2515 driver
mcp2515
root@RpiTest:~#[/code]

B.T.W. The second MCP2515 run with a 8Mhz crystal.


and into /etc/network/interfaces

Code: Select all

root@RpiTest:~# cat /etc/network/interfaces
auto lo

iface lo inet loopback
#iface eth0 inet dhcp

allow-hotplug wlan0
#iface wlan0 inet manual
#wpa-roam /etc/wpa_supplicant/wpa_supplicant.conf
iface wlan0 inet dhcp
wpa-conf /etc/wpa_supplicant/wpa_supplicant.conf
#pre-up ifdown eth0; echo -n '1-1.1:1.0' >/sys/bus/usb/drivers/smsc95xx/unbind
#post-down echo -n '1-1.1:1.0' >/sys/bus/usb/drivers/smsc95xx/bind
iface default inet dhcp

auto can0
iface can0 inet manual
     pre-up /sbin/ip link set can0 type can bitrate 125000
     up /sbin/ifconfig can0 up
     down /sbin/ifconfig can0 down

auto can1
iface can1 inet manual
     pre-up /sbin/ip link set can1 type can bitrate 125000
     up /sbin/ifconfig can1 up
     down /sbin/ifconfig can1 down
root@RpiTest:~# 

I open 2 ssh consoles

On the first one I type

Code: Select all

root@RpiTest:~# candump can0

On the second console I type

Code: Select all

root@RpiTest:~# cansend can1 7ef#0101050000000000

And I got a respond on the first console

Code: Select all

can0  7EF   [8]  01 01 05 00 00 00 00 00

This is a neat trick to check if the canbus is working on the Rpi side.



Daniel

Hi Daniel,
followed the same connection as given in http://skpang.co.uk/catalog/images/rasp ... _rev_B.pdf
and still candump is not showing anything,can you tell whats the problem
and follwed the same steps as your post in page 14 of this forum.
the virtual can is working.
Thanks.

Hi Daniel,
for cansend can0 123#deadbeef
output using above circuit i got as,
candump can0
can0 1FFFFFFF [8] remote request
whats wrong?how to get dataframe output
Thanks.