Re: [Socketcan-users] Message stalls in SocketCan Layer?

["Michael Jäntsch" <michael.jaentsch@in.tum.de>]

Hi,

thanks Wolfgang, I didn't know the socket-can mailinglist is deprecated.

On 02.02.2012 16:03, Wolfgang Grandegger wrote:
> On 02/01/2012 10:42 AM, Michael Jäntsch wrote:
>> Hi everyone,
>>
>> for about a month now, I have a problem with messages that get stalled,
>> causing the select system call on several CAN sockets to time out. We're
>> using the PEAK PCAN-USB interface on a Ubuntu 11.10 (kernel:
>> 3.0.0-15-generic). I've been using the system for about 2 years now and
>> there were no revelant changes to the protocol or software lately. This
>> problem occurs quite frequently now.
> When did the reported problems start to show up? After switching to a
> new kernel?

yes, it all started when I switched to kernel 3.0.0-15-generic. However,
weirdly enough there is one computer with this kernel where the problem
doesn't appear and two where it does. I think I will confirm this by
going back to an older kernel and seeing if the problem disappears.
>> What happens is, that our master sends out a message and receives 13
>> reply messages from 13 different nodes on the bus, at a frequency of
>> 50Hz. This works for some time (minutes) and then a timeout on the
>> select system call occurs that reads from the 13 sockets. Wireshark
>> shows that there is an unusual time delay between the messages. They get
>> sent at the same time and are normally received within the time of
>> 2-4ms. When the timeout occurs, this time is 30ms. When I turn on the
>> debug information in the peak driver I see debug output with time
>> stamps. On this layer the times are still ok when the timeout occurs.
>> Talking to the peak support, I got pointed at the socket can layer which
>> in this case might be causing the problem. However, I have no idea how
>> to debug this and find out more about the problem.
> Hm, is there some other activity blocking the kernel? You could use
> ftrace to find out what's going on.

I don't know, it does look like something is blocking my kernel.
However, I didn't really want to go into kernel debugging/tracing.
If you can give me some hints, as to where I should start I could give
it a shot...

thanks a lot
Michael

-- 
Technische Universität München
Michael Jäntsch
Fakultät für Informatik
Robotics and Embedded Systems
Parkring 13
85748 Garching bei München
Tel: + 49.89.289.17626
Fax: + 49.89.289.17637
michael.jaentsch@in.tum.de
www6.in.tum.de