CAN Subsystem and MPC52xx onboard controller

CAN Subsystem and MPC52xx onboard controller

[Roman Fietze]

Hello List Members,

I have tried multiple versions/branches and git repos (torvalds,
benh/{next,master}, denx, pengutronix, ...) to get a 2.6.34 or HEAD
version of that repos that compiles w/o errors when the CAN subsystem
is enabled and the MPC5xxx onboard driver is selected starting with
the lite5200b_defconfig.

Until now I'm out of luck. If other's also have compile problems I can
of course offer to try to dig into the sources.

Q0:

I'm somewhat unsure what repos to use in general to develop using the
2.6 on a Lite5200B compatible board (for the 2.4 we always used the
DENX repos).

Q1:

And which one is "good" or "better" for CAN+MPC52xx if that's
different?


Thanks for any enlightenment


Roman

=2D-=20
Roman Fietze                Telemotive AG B=FCro M=FChlhausen
Breitwiesen                              73347 M=FChlhausen
Tel.: +49(0)7335/18493-45        http://www.telemotive.de

Re: CAN Subsystem and MPC52xx onboard controller

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 574 bytes --]

> And which one is "good" or "better" for CAN+MPC52xx if that's
> different?

The mainline kernel works fine here with Phytec based MPC5xxx-boards. Some
custom boards, too. You probably already know that it is always best to
stay as close to mainline as possible ;) Maybe just the lite-support is
slightly bit-rotten? What kind of build-errors do you get with the
mainline-kernel?

Regards,

   Wolfram

-- 
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: CAN Subsystem and MPC52xx onboard controller

[Wolfgang Grandegger]

On 06/10/2010 10:41 AM, Roman Fietze wrote:
> Hello List Members,
> 
> I have tried multiple versions/branches and git repos (torvalds,
> benh/{next,master}, denx, pengutronix, ...) to get a 2.6.34 or HEAD
> version of that repos that compiles w/o errors when the CAN subsystem
> is enabled and the MPC5xxx onboard driver is selected starting with
> the lite5200b_defconfig.
> 
> Until now I'm out of luck. If other's also have compile problems I can
> of course offer to try to dig into the sources.

Hm, kernel.org's "linux-2.6.34.tar.bz2" builds here just fine with
lite5200b_defconfig and Socket-CAN support enabled for the MSCAN. I can
imaging that there are issues with more recent "-rc" versions due to
Grant's OF platform device generalization. Also the DTS entries entries
for MSCAN in lite5200b.dts seem OK.

> Q0:
> 
> I'm somewhat unsure what repos to use in general to develop using the
> 2.6 on a Lite5200B compatible board (for the 2.4 we always used the
> DENX repos).

Please use the mainline kernel for that board.

Wolfgang.

Re: CAN Subsystem and MPC52xx onboard controller

[Roman Fietze]

Hello Wolfram,

On Thursday 10 June 2010 10:59:23 Wolfram Sang wrote:

> The mainline kernel works fine here with Phytec based MPC5xxx-boards.

Reading your answer made me hope again, and I just pulled the newest
HEAD from the mainline kernel and tried it once more. Now it compiles.
Thanks for retriggering me again.

All the other's sill have problems, except pengutronix, most of them
with an inconsistency (with the socket CAN files?), which shows up as

  'struct can_bittime_std' declared inside parameter list

in net/can/dev.c:69

So I assume we will switch to the mainline kernel, which we already
use for Atom based x86 development on the MEN boards.


Thanks once more

Roman

=2D-=20
Roman Fietze                Telemotive AG B=FCro M=FChlhausen
Breitwiesen                              73347 M=FChlhausen
Tel.: +49(0)7335/18493-45        http://www.telemotive.de

Re: CAN Subsystem and MPC52xx onboard controller

[Wolfgang Grandegger]

On 06/10/2010 11:29 AM, Roman Fietze wrote:
> Hello Wolfram,
> 
> On Thursday 10 June 2010 10:59:23 Wolfram Sang wrote:
> 
>> The mainline kernel works fine here with Phytec based MPC5xxx-boards.
> 
> Reading your answer made me hope again, and I just pulled the newest
> HEAD from the mainline kernel and tried it once more. Now it compiles.
> Thanks for retriggering me again.
> 
> All the other's sill have problems, except pengutronix, most of them
> with an inconsistency (with the socket CAN files?), which shows up as
> 
>   'struct can_bittime_std' declared inside parameter list
> 
> in net/can/dev.c:69

With all other you mean the linux-2.6-denx tree!? These seem to be some
reminders of the old Socket-CAN support merged into that tree (master
branch). I will clean that up a.s.a.p. Nevertheless, whenever possible,
you should use the mainline kernel.

> So I assume we will switch to the mainline kernel, which we already
> use for Atom based x86 development on the MEN boards.

DENX tries to push patches upstream as fast as possible and new projects
are usually based on the mainline kernel tree.

Wolfgang.


> 
> Thanks once more
> 
> Roman
>