[U-Boot] Update and Cut down mach types

[Michael Schwingen]

On 04/20/2011 10:58 AM, Igor Grinberg wrote:
> Hi Sandeep, Albert, Wolfgang,
>
> On 04/19/11 15:42, Paulraj, Sandeep wrote:
>> Wolfgang, Albert,
>>
>> Russell King sent some updates to the linux kernel for mach-types.
>>
>> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=6f82f4db80189281a8ac42f2e72396accb719b57
>>
>> He also removed a lot of entries which never made it to mainline.
> Well, as I understood from Russell, the main purpose of this "cut down"
> is to make "du -s linux/arch/arm" smaller, because there is no real need in
> all those boards listed in mach-types.h unless there is a support for them
> in mainline Linux kernel.
> Nevertheless the real ARM registry remains untouched - meaning that all
> board ids remain the same and no board is removed from the registry.
Correct.
> This is the place where U-Boot board support diverges from Linux...
>
> Are we obliged to follow the Linux mach-types.h?
> Can't we just adopt Russell's "cut down" script to boards supported by U-Boot?
> Or will it harden the mach-types.h future updates?
>
> Have you thought of getting rid of mach-types.h completely?
> Making every board define its ARM registry id can work and will
> eliminate the need for mach-types.h update every couple of months.
Also, why do we need to pull mach-types.h from Linux at all?

Why don't we pull the original master mach-types file, and generate the 
required .h file(s) during make using the same (or a similar) script 
Linux uses?


>> I have a patch and it is the branch below
>>
>> http://git.denx.de/?p=u-boot/u-boot-ti.git;a=shortlog;h=refs/heads/update-mach-types
> Have you checked that none of the removed boards are in U-Boot tree?
> Because if there are some, then their build will be broken...
It will break ACTUX1-ACTUX4 (which are in-tree, and work fine as soon as 
the relocation-breakage-patch is accepted), plus  DVLHOST, for which I 
have patches submitted to add support.

For my own boards, I can go to the ARM machine database, touch the 
entry, and wait until the define re-emerges in Linux, and await until 
that is marged back to u-boot, but this is plain silly. However, for 
DVLHOST, I am not the registered maintainer in the machine database, so 
I would have to create a duplicate entry for this to work.

cu
Michael