Re: I2C module problem on 8260

On Wed, 13 Feb 2002 00:18:08 +0100, Wolfgang Denk [off-list ref] writes:

I've sent such patches to Tom Rini many months ago (starting  29  Aug
2001),  and  in some iterations again and again. It's for 8xx only so
far, but adaption for 82xx is just a small exercise.

Well, I have an i2c driver for the CPM which supports both 8xx and 8260
(and potentially anything else that has a CPM) in the same source (although
I still haven't got around to testing the 8xx code). I avoid the issue raised
here by using kmalloc(). The code works fine as a loadable module, and it
only allocates the dual port ram once, the first time the module is loaded.
Of course, it is called drivers/i2c/i2c{-algo,}cpm.c.

These patches (amongst a whole lot of others) were also submitted around that
same time. After much back and forth I was asked to post the patches to this
list, which I have been meaning to do for many months now.

Initially, I was asked to submit the i2c changes via the i2c driver people,
but I countered this by pointing out that they really couldn't care less
about driver code for embedded powerpc (CPM), since it will never run on
anything else (as opposed to a driver for a PCI card which may work on
multiple platforms). We really only submit our code back to them for the
sake of completeness.

Unfortunately, this patch was always ignored (as some  others,  too),
so I finally gave up.

Hmm.. I wonder if we are being too careful about allowing stuff into the
_devel tree - isn't that what its for? Development stuff that might break
things? Besides, there are already some changes that make it in that severely
break things for some platforms - we all just put up with it and fix the
problems as they arise (witness the recent change in operation of the get_irq
function - the 8260 platform hung during boot the first time interrupts were
enabled, and it stayed that way for many months until I finally decided to
track the problem down).

Now, I'm not criticising anyone's effort, decisions or contributions here,
simply discussing our operational policy in a calm manner in the hope that
things might be freed up a bit (of course I recognise the chaos that could
arise if all changes were simply immediately incorporated without question,
but there has to be a happy medium somewhere).

Anyway, I should get off my arse and post those patches - I think there
is some good stuff in them (but then I would, wouldn't I). Cheers!
								Murray...
--
Murray Jensen, CSIRO Manufacturing Sci & Tech,         Phone: +61 3 9662 7763
Locked Bag No. 9, Preston, Vic, 3072, Australia.         Fax: +61 3 9662 7853
Internet: Murray.Jensen@cmst.csiro.au  (old address was mjj@mlb.dmt.csiro.au)


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/