Re: [PATCH 1/2] mmc: sdhci: add quirk SDHCI_QUIRK2_NEED_DELAY_AFTER_INT_CLK_RST

[Ludovic Desroches <ludovic.desroches@atmel.com>]

On Thu, Oct 08, 2015 at 08:01:23PM +0200, Ulf Hansson wrote:
> +Russell
> 
> On 17 September 2015 at 10:16, Ludovic Desroches
> <ludovic.desroches@atmel.com> wrote:
> > The Atmel sdhci device needs a new quirk. sdhci_set_clock set the Clock
> > Control Register to 0 before computing the new value and writing it.
> > It disables the internal clock which causes a reset mecanism. If we
> > write the new value before this reset mecanism is done, it will prevent
> > the stabilisation of the internal clock, so a delay is needed. This
> > delay is about 2-3 cycles of the base clock. To be safe, a 1 ms delay is
> > used.
> >
> > Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
> > ---
> > Hi Ulf,
> >
> > Can you take these patches as fixes? I thought it was better to introduce a
> > quirk instead of duplicating a part of the sdhci_set_clock function to
> > implement my own set_clock function.
> 
> Hi Ludovic,
> 
> Sorry for the delay!
> 
> I have been thinking of how to address patches for sdhci that adds
> more quirks/callbacks. To be honest, I am fed up with such patches as
> I think it keeps the development of sdhci going in the wrong
> direction.
> 
> Instead, I have been hoping for someone to have stab into turning
> sdhci into a library as Russell King pointed out several times as
> well.
> 
> Now, as my response too you had a big delay for these patches and
> because it's a fix intended for the rc, I am going to apply patch1 and
> patch2 anyway. Although, you and everybody else that contributes to
> the sdhci development, please consider this as the *last* time for
> adding quirks/callbacks, unless you have *very good* reasons.

Thanks Ulf, it seems it is my lucky day! I understand and share your
point of view about quirks but it was so easy to manage controller
issues with them...

Regards

Ludovic