Re: [RFC 0/4] [flexcan] Add support for powerpc (freescale p1010) -V5

[Robin Holt <holt-sJ/iWh9BUns@public.gmane.org>]

On Sat, Aug 06, 2011 at 06:06:02AM -0500, Robin Holt wrote:
> On Fri, Aug 05, 2011 at 11:05:00PM -0500, Robin Holt wrote:
> > Marc, Wolfgang or U Bhaskar,
> > 
> > This patch set should have all your comments included.
> > 
> > I did implement a very simple clock source in the p1010rdb.c file, which,
> > unfortunately, your tree will not have so please do not apply the last
> > patch in the series.  That will need to go to the powerpc folks and
> > follow the p1010rdb patch from freescale.
> > 
> > Could you please apply the first three patches to a test branch, compile
> > and test them on an arm based system?  I would like to at least feel
> > comfortable that I have not broken anything there.
> > 
> > I have tested the full set on a p1010rdb with an external PSOC based
> > can communicator.  That PSOC code has a bunch of erroneous can comms it
> > can generate, but I do not know how the developer of that code injects
> > those errors.  As a result, no error handling from the can input has been
> > tested.  I have tested both flexcan interfaces on the board and both work
> > with these patches in addition to the other p1010rdb patches not included.
> 
> ARGH!
> 
> I just did a quick look back at my git log, and I have one other patch
> earlier in the series where I committed a one-line change to flexcan.c
> which is probably very relevant to you, but not so much to me.  I removed
> the mach/clock.h which does not seem to exist for powerpc.
> 
> Can any of you tell me if that is relevant for the arm flexcan build?
> If not, does it seem reasonable to just remove it early on?

It looks like the more-nearly right thing to do is to #include
<linux/clkdev.h> but powerpc does not implement one.

Thanks,
Robin