[U-Boot] [PATCH v5] arm: Add sata support on Layerscape ARMv8 board

[Yuantian Tang]

Hi Sinan,

> -----Original Message-----
> From: Sinan Akman [mailto:sinan at writeme.com]
> Sent: Monday, December 07, 2015 2:04 PM
> To: Tang Yuantian-B29983 <Yuantian.Tang at freescale.com>; Sun York-R58495
> <yorksun at freescale.com>
> Cc: u-boot at lists.denx.de
> Subject: Re: [U-Boot] [PATCH v5] arm: Add sata support on Layerscape
> ARMv8 board
> 
> 
>    Hi Yuantian
> 
> On 06/12/15 10:09 PM, Yuantian Tang wrote:
> > Hi York,
> >
> > Please see explanation inline.
> > [...]
> > I was trying to use one function for all, but I found separating them is
> better.
> > Take ls1043a and ls2080a as an example, ls2080a has two controllers, while
> ls1043a has one.
> > Ls2080a has two registers that need to be updated while ls1043a has four.
> > A lot of #ifdef are needed if we unify them, not mention that in the future,
> changing one of the platforms' register will affect the other.
> 
>     You might want to take into consideration that in the near future we will be
> moving this to dm. In that respect having all that in one file already will
> probably make things much easier. If you consider this, perhaps you will have
> a different view.
> 
They are in the same file but different functions.

> > Maybe I am not thinking it through.  If you can give me more detail that
> viable, I can give a try.
> >
> >> [...]
> >> ports, so we have to choice one. In this case I choice the first one
> >> which is SATA1.
> >>
> >> This should be put into comment, or README if you have one.
> > This phenomenon is not LS platform specific, that's uboot's issue which
> needs another patch to fix.
> > I think uboot know that and choice to not fix it because for uboot
> supporting two sata port is not that significant.
> 
>    Again, with dm and reading all the hardware properties from device tree
> will also change this. If both device nodes are enabled we will have to
> support both as long as there is no hardware limitation. So I think there is no
> reason why having both SATA and PCIe would not be significant. It is just that
> the current implementation has this limitation and there is already some
> timeline for removing these limitations.
> 
I am not seeing what we are arguing here? 
Are we talking about if this limitation is important?
Please point out what's wrong with this patch.

Regards,
Yuantian

>    Regards
>    Sinan Akman