[PATCH 0/9] OF: address and device related sync and cleanup

[Sebastian Hesselbarth]

On 06/27/13 20:19, Sascha Hauer wrote:
> On Thu, Jun 27, 2013 at 11:00:46AM +0200, Sebastian Hesselbarth wrote:
>> On 06/27/13 10:58, Sascha Hauer wrote:
>>> On Thu, Jun 27, 2013 at 09:50:10AM +0200, Sebastian Hesselbarth wrote:
>>>> On 06/27/13 08:51, Sascha Hauer wrote:
>>>>> Applied this series with the change that I removed the dead code
>>>>> ifdeffed with CONFIG_PPC_DCR.
>>>>
>>>> Ok, great! Next patch sets will be focused on actually using the
>>>> DT improvements. I have an SPI driver for MACH_MVEBU ready and
>>>> running, and I am working on importing generic SDHCI from Linux.
>>>
>>> The sdhc driver in Linux is a big piece of shit which uses the wrong
>>> abstraction model. This is shown by the fact that 32bits are not enough
>>> to abstract the quirks. Even with all these quirks the i.MX part of this
>>> driver has to implement driver logic in the register access functions.
>>> That just insane. I'm not really looking forward to add this driver
>>> to barebox.
>>
>> Ok, fair enough. Then I'll be adding the required callbacks to Dove
>> SDHCI driver directly based on current barebox i.MX driver.
>
> I think for the SDHC a library approach would be good: Add a new driver
> and share common code/helper functions. I recently factored out the
> register defines as a start so that these can be shared.

Yeah, I just working at SDHCI driver for Dove and wonder if I start
to wrap my brain around readl/writel or go with readb/w/l as it is
supposed to be.

As soon as I got some valid responses back from the usdcard, I have
another look at sdhci.h and a generic approach.

Sebastian