[U-Boot-Users] current simple 8260 platform as start...

[Wolfgang Denk]

In message <16302.24038.849399.138394 at gargle.gargle.HOWL> you wrote:
> 
>     Wolfgang> Like the fact that CFG_DEFAULT_IMMR definition in
>     Wolfgang> MPC8260ADS.h will bring you nasty kernel crashes the first
>     Wolfgang> time you want to boot your Linux kernel.
> 
> CFG_DEFAULT_IMMR can't cause any crash. It only tells to U-Boot
> initialisation code where to find IMMR to change it to
> CFG_IMMR. CFG_IMMR was unusable, it's right (its value was copied from
> the board's manual just like the entire memory map), but it was fixed
> long time ago by my patches.

You are right; sorry I confused this (I just  remembered  the  _many_
emais I received about Linux crashing on the MPC8260ADS).

>     Wolfgang> Like the fact that there is seems to be need for a
>     Wolfgang> proprietary version of jffs2_part_info() (because of a
>     Wolfgang> broken flash layout?)
> 
> No, no, the flash is OK. At least, I think it's OK. I introduced JFFS2
> support so I have to explain. The board provides two sources of HRCW:
> the FPGA (BCSR) and the flash. Many boards have got two flashes: small
> one for boot and large one for filesystem, etc. MPC826xADS/PQ2FADS has

Actually in my experience these are only very few boards - less  than
10%. Most use just "big" flash banks.

> only one, and it's flash SIMM, so 8MB, 16MB, and 32MB flash can be
> supported by simple replacement of the SIMM. HRCW occupies only first
> bytes of the SIMM but since flash is erased by blocks, entire block (on
> 4x8bit SIMM it's 256KB) wasted. I prefer to use BCSR-supplied HRCW but
> didn't want to touch Motorola-provided value in the flash either. So I
> burn the U-Boot at 0xFFF00000 (8MB SIMM starts at 0xFF800000). In such a
> configuration, I have the range 0xFF840000-0xFFEFFFFF available for
> JFFS2. Standard jffs2_part_info() allows to specify first filesystem

But that wastes the whole last megabyte in flash, while U-boot easily
fits in 256 kB. Such a fragmentation is IMHO worse.

> sector but not last. It "thinks" that the filesystem spans to the flash
> end. This is the reason why proprietary version of jffs2_part_info()
> appeared. Of course, different flash layout can be chosen.


Well, as a summary my statement is still standing: the MPC8260ADS  is
not  an "easy" board to start with. I do NOT recommend to use it as a
model. YMMV.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
The only person who always got his work done by Friday
                                                 was Robinson Crusoe.