Marcus,

For what it is worth.
william@xanbustester:~$ cat /sys/devices/platform/bone_capemgr/slots
 0: P---L-   1 cape-CBB-Serial,r01,Logic Supply,cape-CBB-Serial
 1: PF----  -1
 2: PF----  -1
 3: PF----  -1

I've been using that same cape . . . with kernel 4.2.0* for about the last
month or maybe two. I'm not sure what you had in mind,
But I've been reading a live CANBUS network using it @250k /s. The
beaglebone, and the board are more than capable keeping up
without using a rt kernel.  The actual CANBUS protocol is a proprietary
NMEA 2000 fastpacket protocol, so I've been reading from the bus
in real time, while building variable length packets from the data - Using
socketCAN.

Anyway, I would not recommend using the same kernel I am, but just wanted
to give you some feedback as far as using a rt kernel.
Personally, I did want to experiment with a rt kernel, because I'm actually
rebuilding fastpackets from multiple socketcan frames,
also in realtime, as well as putting that data out over ethernet to a web
browser via websockets. As you might imagine, this is a ALOT
of work to get done on a single core 1Ghz. especially considering I've been
tracking ~20 or so PGN, re-constructing them, and then putting
that data out over ethernet via a web server.

I'd be interested in hearing what you plan on doing with that CAN cape
though ! Assuming you're even going to use the CAN portion of
the cape ;)



On Thu, Sep 24, 2015 at 2:40 PM, William Hermans <[email protected]> wrote:

> *Well there's a whole rt-test suite..*
>>
>> *https://git.kernel.org/cgit/utils/rt-tests/rt-tests.git/
>> <https://git.kernel.org/cgit/utils/rt-tests/rt-tests.git/>*
>>
>> * so it's gotta be good right?*
>>
>> * I've never compared..*
>
>
> Well I was just reading the real time Linux wiki . . . man their
> documentation sucks . . . But there was some claim under "how to write real
> time applications" which by the way has NOTHING to do with writing RT apps
> . . . and under the miscellaneous header it makes a claim that 10uS latency
> should be permanent. But what it is referring to . . . your guess would be
> as good as mine.
>
> There is also a ton of hoops one needs to jump through in order to achieve
> minimal latency. Booting with a USB stick for instance is said to increase
> latency to 500ms, but booting without, and inserting after boot is not a
> problem. Power management, and CPU on-demand CPU scaling are two thing the
> BBB uses by default that are supposedly detrimental to deterministic
> execution.  Probably a lot more that I'm unaware of too, but basically
> anything that generates SMI's or system calls are "bad".
>
> I think the best anyone can really hope for is to use the *rt* kernel
> image, and just be aware of the rest. I'm not so sure it is a good idea to
> disable power management, or one demand CPU scaling. .  .
>
> On Thu, Sep 24, 2015 at 2:06 PM, Robert Nelson <[email protected]>
> wrote:
>
>> On Thu, Sep 24, 2015 at 3:54 PM, William Hermans <[email protected]>
>> wrote:
>> > Robert, so what is the real difference between RT, and non RT ? I mean I
>> > know the scheduler should be faster / tighter, but has anyone bench
>> marked
>> > this ?
>>
>> Well there's a whole rt-test suite..
>>
>> https://git.kernel.org/cgit/utils/rt-tests/rt-tests.git/
>>
>> so it's gotta be good right?
>>
>> I've never compared..
>>
>> Regards,
>>
>> --
>> Robert Nelson
>> https://rcn-ee.com/
>>
>> --
>> For more options, visit http://beagleboard.org/discuss
>> ---
>> You received this message because you are subscribed to the Google Groups
>> "BeagleBoard" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [email protected].
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>

-- 
For more options, visit http://beagleboard.org/discuss
--- 
You received this message because you are subscribed to the Google Groups 
"BeagleBoard" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to