Re: [RFCv2: PATCH 1/2] mtd: mediatek: device tree docs for MTK Smart Device Gen1 NAND

[Boris Brezillon <boris.brezillon-wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8@public.gmane.org>]

On Sat, 26 Mar 2016 14:38:31 -0400
Jorge Ramirez-Ortiz <jorge.ramirez-ortiz-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:

> On 03/22/2016 09:52 AM, Boris Brezillon wrote:
> >> > +		compatible = "mediatek,mt2701-nfc";
> >> > +		reg = <0 0x1100d000 0 0x1000>;
> >> > +		interrupts = <GIC_SPI 56 IRQ_TYPE_LEVEL_LOW>;
> >> > +		clocks = <&pericfg CLK_PERI_NFI>,
> >> > +			 <&pericfg CLK_PERI_NFI_PAD>;
> >> > +		clock-names = "nfi_clk", "pad_clk";
> >> > +		nand-on-flash-bbt;
> >> > +		status = "disabled";
> >> > +		mediatek,ecc-controller = <&bch>;
> > Now that 2 different drivers use the same way to link the ECC engine
> > and the NAND controller we can think about defining a generic property
> > (ecc-engine ?), and provide a generic framework.
> >
> > The generic framework part is not something I'm asking right now, but I
> > think we should start using a generic property here.
> >
> 
> we have done all the changes required for v3 except this one.
> so please let me check my understanding before going ahead: are you suggesting
> that we replace "mediatek,ecc-controller" for "ecc-engine" (even though this
> generic property doesn't exist yet)?
> 
> and then afterwards, generate another patch-set set to define and document
> "ecc-engine"?

Yes, that's the idea, though the NAND controller aspect is not yet
documented in Documentation/devicetree/bindings/mtd/nand.txt, and
ecc-engine is supposed to be attached to the NAND controller node.

I'll send a patch to further document the NAND chip, NAND controller
description concepts in this generic NAND DT bindings doc (as suggested
by Brian), and let you document this ecc-engine property.



-- 
Boris Brezillon, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com