Re: [microblaze-uclinux] 1st SUZAKU patch for microblaze uClinux

[John Williams <jwilliams@xxxxxxxxxxxxxx>]

Hi Yashi,

Yasushi SHOJI wrote:
> This is the first patch for SUZAKU.
> 
> There is nothing special in this patch.  All it does is
> - create new paltform in menuconfig
> - copy asm-microblaze/mbvanilla.h to asm-microblaze/suzaku.h
> - copy arch/microblaze/kernel/mbvanilla.c arch/microblaze/kernel/suzaku.c

Excellent, thanks for that.  We'll get it into the CVS.

> no contents has been changed.
> 
> regarding to the port, John, are you planing to create def-configs dir
> in arch/microblaze like arm? if not, where should we keep default
> configs?

I modelled microblaze on the m68knommu way of doing this, which is to 
create a subdir for each platform in /vendors, containing the files 
config.linux-2.4.x, config.arch and config.uclibc, which are the default 
configs for the kernel, platform, and uClibc respectively.

Is there something else about the arm def-config approach I should know 
about?

There's definitely a weakness, since everything has grown out of the 
mbvanilla platform, it's not as clean as it could be.  I'd welcome 
suggestions or contributions on cleaning it up!

> other note, we, as in Atmark Techno, have been using Hermit[1] as our
> product's downloader/bootloader for a while.  And I have a microblaze
> port for that.  It's still in alpha stage, but it's working for me on
> my SUZAKU board.
> 
> If anyone is interested, it's here.

Sounds great.  The current bootloader is just a hangover from my early 
hacking, if we can get something a bit more robust with better features, 
I'm all for it.

> a sad thing is that EMC activates WE wrong number of times if we use
> bus width matching.  that force us to choose either '32bit fetch' or
> 'ability to send command to flash'.  we've chosen not to execute in
> flash. :< that is, some program in bram need to copy hermit from flash
> to dram.

Did you notice that in EDK6.1, the EMC is distributed in EDK with full 
source (/edk/hw/XilinxProcessoIPLib/pcores/....)?

So in principle you could change this behaviour to work as you need it. 
  Also the license header in the sources asserts Xilinx copyright, but 
does not prohibit reproduction or modification.

My personal interpretation is that you could modify this controller, 
note those changes in the header, and leave the Xilinx copyright notice, 
and not be breaking any rules.

Regards,

John

___________________________
microblaze-uclinux mailing list
microblaze-uclinux@itee.uq.edu.au
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/