[U-Boot] [PATCH 1/2] arm: imx6: Add DDR3 calibration code for MX6 Q/D/DL

[Marek Vasut]

On Monday, January 25, 2016 at 12:30:00 AM, Tom Rini wrote:
> On Mon, Jan 25, 2016 at 12:00:05AM +0100, Marek Vasut wrote:
> > On Sunday, January 24, 2016 at 11:21:51 PM, Tom Rini wrote:
> > > On Sun, Jan 24, 2016 at 11:07:30PM +0100, Marek Vasut wrote:
> > > > On Sunday, January 24, 2016 at 08:33:57 PM, Tom Rini wrote:
> > > > > On Sun, Jan 24, 2016 at 06:22:10PM +0100, Marek Vasut wrote:
> > > > > > On Sunday, January 24, 2016 at 06:18:23 PM, Stefano Babic wrote:
> > > > > > > On 24/01/2016 18:11, Marek Vasut wrote:
> > > > > > > > It is not clear when the wait_for_bit() will be applied, I am
> > > > > > > > certain there will be another round for so. I do not want to
> > > > > > > > wait for it and I don't see a reason why those patches should
> > > > > > > > block this if the conversion can be done afterward.
> > > > > > > 
> > > > > > > Just wait for a while - if it takes too much, I reconsider to
> > > > > > > apply this first and factorize wait_for_bit() in a follow-up
> > > > > > > patch.
> > > > > > 
> > > > > > I have waited for over a month and I fail to see a reason why
> > > > > > patches which will be applied at uncertain point in the future
> > > > > > shall block this patchset. The wait_for_bit() can be removed by
> > > > > > a subsequent patch, it is already pulled out explicitly in the
> > > > > > code, so I don't see a problem with applying this.
> > > > > 
> > > > > Did I miss something or isn't v4 of wait_for_bit good to go?
> > > > 
> > > > I don't really know if it's good to go, but this patch does not
> > > > depend on it in any way. A subsequent patch can drop the
> > > > wait_for_bit() from here, it's the same as the wait_for_bit() in
> > > > dwc2 and other wait_for_bit() anywhere else, but I don't see a
> > > > reason why this patch should not be applied now.
> > > > 
> > > > If I follow the logic in this thread, it would also be possible to
> > > > say that this patch should wait until Eric submits the MX6S DDR
> > > > support for example. We could indefinitelly wait for new and new
> > > > stuff which might possibly block this.
> > > 
> > > Why don't you ack/test/review the wait_for_bit series and post a
> > > follow-up to this one that uses the common function, which can be
> > > squashed into the original and then this gets picked up?
> > 
> > Why ? I can send subsequent patch which removes the duplicate
> > wait_for_bit() from this code once the wait_for_bit series is applied. I
> > don't see a problem with that and the wait_for_bit() being so explicitly
> > pulled out from the code is done with that in mind. But that does not
> > block this patch from being applied now. Does it?
> 
> If I tell you I'm probably going to have the wait_for_bit stuff applied
> before the next imx PR is ready... ?

Well so what ? I still do not see why these patches should be blocked by the
wait_for_bit() stuff. The conversion to wait_for_bit() can well be done after
that, so what is the problem ?

Best regards,
Marek Vasut