RE: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

RE: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Daniel Schnell]

Syed Amer Gilani wrote:
 
> I had similar problems, but with a 2.6.18 Kernel. Check you Kernel
> configurations of mscan. The default clock of 66Mhz is possibly
> wrong, on my mpc5200b board it is 33Mhz. And also check the Pin
> Configuration. I had also no termination on board, so i had to add it
> to the cable.   


Hi,

This fixed it: setting it from 66MHz to 33MHz in the MSCAN
configuration. At least the IOUnit gets messages now. Thanks a lot !!

If now only the clock_nanosleep() would wait as long as expected and
would not return after 1/4 of the time ....



Best regards,

Daniel.

Re: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Wolfgang Grandegger]

Daniel Schnell wrote:
> Syed Amer Gilani wrote:
>  
>> I had similar problems, but with a 2.6.18 Kernel. Check you Kernel
>> configurations of mscan. The default clock of 66Mhz is possibly
>> wrong, on my mpc5200b board it is 33Mhz. And also check the Pin
>> Configuration. I had also no termination on board, so i had to add it
>> to the cable.   
> 
> 
> Hi,
> 
> This fixed it: setting it from 66MHz to 33MHz in the MSCAN
> configuration. At least the IOUnit gets messages now. Thanks a lot !!

I will make this the default on MPC5200B. You are not the first on 
stumbling over this quirk.

> If now only the clock_nanosleep() would wait as long as expected and
> would not return after 1/4 of the time ....

Do you have a little demo program demonstrating the problem?

Wolfgang.

[Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Daniel Schnell]

[-- Attachment #1: Type: text/plain, Size: 645 bytes --]

Hi,
 
I am struggling with Xenomai latest SVN and RT-Socket-CAN on a MPC5200B
with MSCAN driver inside a Denx 2.4.25 kernel.
 
Does anybody have this configuration actively running ? How do you use
the provided rtcansend/rtcanreceive functions, so they actually transfer
data over the wires ?
 
I was doing the following:
rtcanconfig rtcan0 -b 125000 start
rtcansend rtcan0 -i 0 0x1 0x0
(returns)
rtcansend rtcan0 -i 0x1 0x0
(hangs, never comes back)
 
 
Actually the first rtcansend command does not seem to work, because a
connected CAN IO unit doesn't show any reaction.
 
 
Any ideas ?
 
Regards,
 
Daniel Schnell.

[-- Attachment #2: Type: text/html, Size: 2735 bytes --]

Re: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Wolfgang Grandegger]

Daniel Schnell wrote:
> Hi,
>  
> I am struggling with Xenomai latest SVN and RT-Socket-CAN on a MPC5200B 
> with MSCAN driver inside a Denx 2.4.25 kernel.
>  
> Does anybody have this configuration actively running ? How do you use 

Yes.

> the provided rtcansend/rtcanreceive functions, so they actually transfer 
> data over the wires ?
>  
> I was doing the following:
> rtcanconfig rtcan0 -b 125000 start
> rtcansend rtcan0 -i 0 0x1 0x0
> (returns)
> rtcansend rtcan0 -i 0x1 0x0
> (hangs, never comes back)
>  
> Actually the first rtcansend command does not seem to work, because a 
> connected CAN IO unit doesn't show any reaction.

The first message seems not to go out to the wire and the second send 
blocks because the TX done interrupt of the first message has not yet 
occured (no TX hardware slot free). Is there a CAN node able to receive 
the message?

> Any ideas ?

Likely, you have a problem with your clock setting, TX/RX routing or 
cabeling (termination etc.).

Wolfgang.

RE: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Daniel Schnell]

Wolfgang Grandegger wrote:
 
> The first message seems not to go out to the wire and the second send
> blocks because the TX done interrupt of the first message has not yet
> occured (no TX hardware slot free). Is there a CAN node able to
> receive the message?   

Yes, there is an IO unit which normally reacts on that message
 
> Likely, you have a problem with your clock setting, TX/RX routing or
> cabeling (termination etc.). 


I would suspect it is the clock setting, as we have also problems with
clock_nanosleep() running 4 times too fast. I assume then that the IPB
clock frequency is wrong ?


How do you setup your CAN network for a test ? Which commands do you use
?

Best regards,

Daniel Schnell.

Re: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Wolfgang Grandegger]

Daniel Schnell wrote:
> Wolfgang Grandegger wrote:
>  
>> The first message seems not to go out to the wire and the second send
>> blocks because the TX done interrupt of the first message has not yet
>> occured (no TX hardware slot free). Is there a CAN node able to
>> receive the message?   
> 
> Yes, there is an IO unit which normally reacts on that message
>  
>> Likely, you have a problem with your clock setting, TX/RX routing or
>> cabeling (termination etc.). 
> 
> 
> I would suspect it is the clock setting, as we have also problems with
> clock_nanosleep() running 4 times too fast. I assume then that the IPB
> clock frequency is wrong ?

On a Rev. B you normally set the CAN system clock to 33MHz via kernel 
configuration parameter. Is CAN TX/RX on PSC2?


> How do you setup your CAN network for a test ? Which commands do you use
> ?

Your setup with rtcanconfig was OK, or what do you mean.

Do you have terminators on both ends of the cable?

Wolfgang.

RE: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Daniel Schnell]

Wolfgang Grandegger wrote:

> On a Rev. B you normally set the CAN system clock to 33MHz via kernel
> configuration parameter. Is CAN TX/RX on PSC2? 

Yes 33 MHz was the magic number. Hmmm, shouldn't that not be set
automatically if MPC5200B is selected ?
The CAN in our case is on I2C1/TMR01


>> How do you setup your CAN network for a test ? Which commands do you
>> use ?
> 
> Your setup with rtcanconfig was OK, or what do you mean.

I mean physically. E.g. for a MPC5200B with two CAN ports you could make
a send/receive ping/pong with a sort of nullmodem cable. Additionally
you need a test case. Just wanted to know how your test setup looks
like.


> Do you have terminators on both ends of the cable?

Yes everything is fine here.


Best regards,

Daniel.

Re: [Xenomai-help] RT-SocketCan + Xenomai on MPC5200B

[Wolfgang Denk]

In message <DD39B5C3F4963040ADC9768BE7E430CB014DEFF0@domain.hidl.net> you wrote:
>
> I would suspect it is the clock setting, as we have also problems with
> clock_nanosleep() running 4 times too fast. I assume then that the IPB
> clock frequency is wrong ?

As you can see from the U-Boot startup messages,  the  IPB  clock  on
your board is set to 66 MHz.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@domain.hid
Success in marriage is not so much finding the right person as it  is
being the right person.