[U-Boot] [RFC PATCH 1/6] Mips: Move content of arch/mips/cpu to arch/mips/cpu/mips32

[Aaron Williams]

Seperating the SOCs solves most of my problems with the exception of board.c, 
but then again, we're rather a special case there due to some of the stuff we 
do.

Not counting our SDK we have around 27Kloc for our platform, a large chunk of 
that being DDR2/3 memory initialization code. Our SDK currently weighs in at 
764Kloc, most of which isn't needed by U-Boot (lots of autogenerated hardware 
definition files for all our chips).

-Aaron

On Saturday, March 12, 2011 04:48:58 PM Daniel Schwierzeck wrote:
> Hi Shinya,
> 
> On 03/12/2011 03:43 PM, Shinya Kuribayashi wrote:
> > On 3/9/11 10:16 PM, daniel.schwierzeck at googlemail.com wrote:
> >> All current CPUs and SoCs are based on Mips32 arch. The complete
> > 
> > Is that true? What about purple SoC? IIUC It's based on MIPS 5Kc
> > and capable of 64-bit, which MIPS32 is 32-bit only architecture.
> 
> that was an assumption. I could not find any information about Purple.
> This patch series is a suggestion. Maybe there are better approaches
> like arch/mips/cpu/{mips4k|mips5k|mips24k|...|octeon}.
> 
> >> code resides in the global arch/mips/cpu directory. This is not
> >> suitable if other Mips architectures like Mips64 or Octeon should
> >> be supported in the future.
> > 
> > Just for the record.
> > 
> > Personally, Octeon is sort of a special case. It's based on MIPS64
> > architecture, and definitely a MIPS SoC, but not usual MIPS machine.
> > As Aaron already mentioned before, we have very few files sharable
> > with other MIPS machines even in the Linux kernel case. I think it
> > will take long time the Octeon port gets merged to upstream, regard-
> > less of my opinions, because they have a lot of things need to be
> > resolved prior to Octeon itself.
> 
> My original intention was cleaning up the MIPS CPU directory as
> preparation to support new SoCs. I included this patch because Aaron
> asked for this separation some weeks ago but no one answered or commented
> ;) However I posted this series to get some feedback if such a change is
> wanted or acceptable. If so I prepare a new patch series. At least I hope
> the SoC separation is acceptable.
> 
> Cheers,
> Daniel