[U-Boot] [PATCH] arm64: mvebu: use single defconfig for Armada8K development boards

[Igal Liberman]

Hi Stefan

Regards,
Igal


> -----Original Message-----
> From: Stefan Roese [mailto:sr at denx.de]
> Sent: Thursday, June 08, 2017 09:10
> To: Igal Liberman; u-boot at lists.denx.de
> Cc: Kostya Porotchkin; Nadav Haklai; Neta Zur Hershkovits
> Subject: Re: [PATCH] arm64: mvebu: use single defconfig for Armada8K
> development boards
> 
> Hi Igal,
> 
> On 08.06.2017 08:06, Igal Liberman wrote:
> >> On 07.06.2017 14:16, igall at marvell.com wrote:
> >>> From: Igal Liberman <igall at marvell.com>
> >>>
> >>> Currently, Marvell Armada8k development board use 3 different
> >>> defconfigs:
> >>>           mvebu_db-88f7040-nand_defconfig
> >>>           mvebu_db-88f7040_defconfig
> >>>           mvebu_db-88f8040_defconfig Having 3 different defconfigs
> >>> makes maintenance difficult.
> >>>
> >>> This patch removes the defconfigs mentioned above and introduce a
> >>> new defconfig which represents the Armada8k family.
> >>>
> >>> With this patch, no device-tree is set in the defconfig, so the
> >>> device-tree should be set during compilation, using:
> >>>
> >>> Configure u-boot:
> >>>
> >>>           make mvebu_db_armada8k_defconfig
> >>>
> >>> Select the desired device-tree during compilation:
> >>>
> >>>           make DEVICE_TREE=armada-7040-db
> >>>           make DEVICE_TREE=armada-8040-db
> >>>           make DEVICE_TREE=armada-7040-db-nand
> >>
> >> I like the idea to consolidate the defconfigs in general. One question
> though.
> >> This patch results in one build target for the automated tests
> >> (Travis). Does compiling without the additional DEVICE_TREE= work
> >> without generating compilation errors? Otherwise the Travis tests will fail.
> >>
> >
> > If you compile without DEVICE_TREE= it won't compile (unless you
> > exported DEVICE_TREE before).
> > Basically, I did it on purpose - to make sure that user doesn't rely
> > on some possible incorrect default settings.
> 
> Makes sense.
> 
> > If it causes a compilation issues for the automated tests - I guess we
> > can set a default value after all.
> > What do you think?
> 
> Yes, please do. We need to compile clean for the automated tests, otherwise
> this patch simply can't be accepted (sorry).
> 

I'll submit a new patch with default device-tree.
Thank you for the review.

> Thanks,
> Stefan