Hi Wolfram,

Le 09/12/2012 12:16 PM, Wolfram Sang a ?crit :

On Wed, Sep 12, 2012 at 10:03:59AM +0200, Nicolas Ferre wrote:

quoted

On 09/12/2012 08:42 AM, ludovic.desroches at atmel.com :

quoted

From: Ludovic Desroches <redacted>

Hi,

This set of patches is based on Nikolaus at91_i2c driver.

Changes:
v3:
   - only put multi-drive lines in the if...else statement (suggested
   by Warner Losh)

Hi Wolfram,

As said by Ludovic, this series goes on top of Nikolaus' one.
My Acked-by is already set on this one, so I think that I have nothing
more to do ;-)

BTW, in case you need help to sort all this, do not hesitate to contact
us... we can setup a git tree for this...

I think I am fine. Patches look good. I wondered a bit about first
removing the old driver, then adding the new one with regard to
bisectability. But as the old driver depends on BROKEN, I think this is
OK to do.

One thing I'd like to make, though. I'd like to squash the following
patches into one:

	drivers/i2c/busses/i2c-at91.c: add new driver
	i2c: at91: use managed resources
	i2c: at91: add warning about transmission issues for some devices
	i2c: at91: use an id table for SoC dependent parameters

It is especially the last patch I am mostly interested in. The id_table
approach is what I like, while the original id_entry mechanism looks
fishy. I'd was good for reviewing to have the patches split like this;
yet for hitting mainline, I'd prefer to have the driver proper on first
occasion. I already did the squashing in a test-branch and the result
looks good to me.

Nikolaus, Ludovic: Are you fine with this?

No problem on my side.

Regards

Ludovic

Thanks,

    Wolfram