Hi Ulrich,

On 08/02/2011 05:50 PM, uprinz wrote:
> 
> Hi all, back again :)
> 
> 
> Wolfgang Grandegger wrote:
>>
>> Hi Ulirich
>>
>> On 01/26/2011 05:02 PM, uprinz wrote:
>>>
>>>
>>> Wolfgang Grandegger wrote:
>>>>
>>>>> CAN_FLEXCAN=y
>>>>> FLEXCAN_MXC_SELECT1=y
>>>>> are set.
>>>>> Kernel is 2.6.34.7
>>>>
>>
>> Well, I doubt the i.MX will ever be fully supported by the kernel but I
>> think that the recent kernel version (2.6.37+) does have reasonable
>> support for the i.MX25. That's what I would start with. Anyway, you may
>> need to port the BSP for your board if it's not already supported in the
>> mainline kernel. It seems that you are using a LTIB-kernel provided by
>> Freescale.
>>
>> Wolfgang.
>>
> 
> Well, Wolfgang, finally we got a pretty fine working kernel 2.6.39 including
> full flexcan support.
> Both channels work smoothly. All Ip commands are supported and I think I
> have to thank not only our supplier but some folks in this forum too.
> 
> But there is one question left that was mentioned in this thread a while
> back.
> 
> Scenario is that I have two physical CAN channels. To both of them might be
> some devices connected or, what is important, may be not connected.
> The i.MX25 features two separate masters, each one responsible for only one
> bus.
> Additional it features one slave that hooks on one of the channels. But this
> slave is just a piece of software, not a physical device.
> 
> Now what happens is, that the physical channels might be unconnected. On one
> of the channels there is the slave hooked. But the interface of both
> channels is flooding my syslog with 
> "SC2:write error; : No buffer space available"
> 
> Is there a chance to get rid of this message?

You could handle the errno ENOBUFS and simply retry after some short
delay without printing a message.

> (Besides the one from throwing this error out of the source code)

> Just toi make clear the differences to the scenario we had with the older
> kernels:
> We do not have any problem with the bus! As soon as I connect any kind of
> device to the physical bus, the error message disappears and the bus is up
> and running. 
> 
> I configured the bus with 
> 
>         ifconfig $1 down
>         echo "$1: Setting bitrate $BITRATE, restart $RESTART"
>         ip link set $1 type can bitrate $BITRATE triple-sampling on

BTW, triple-sampling with 125 kb/s is pretty unusual.

>         ip link set $1 type can restart-ms $RESTART
>         ifconfig $1 up
> 
> where $RESTART is a value of 100 and $1 ist can0 or can1, $BITRATE is 125000

What does "ip -d -s link" display? What type of errors and state changes
do you see?

Wolfgang.
_______________________________________________
Socketcan-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-users

Reply via email to