[U-Boot-Users] Running w/o flash

[Dave Ellis]

Andy Fleming wrote: 
> On 3/28/07, Wolfgang Denk <wd at denx.de> wrote:
> > Dear Dave,

> > > The only problem configurations are the MPC8xxx configurations.
> > > Many of them are based on MPC8540ADS.h and MPC8560ADS.h, which
> > > define CFG_NO_FLASH when CFG_RAMBOOT is defined, but don't
> > > disable CFG_CMD_FLASH. I would make those like MPC8641HPCN.h,
> > > which does disable CFG_CMD_FLASH.
> >
> > This is definitely not a good idea. One pretty common usage patttern
> > is to load anU-Boot image to RAM and use this to load and program the
> > "real" image to flash, so flash command support should be kept.
> >
> > For me the question is why the boards define CFG_NO_FLASH then.  You
> > will have to check this with the board maintainers and the custodian.
> 
> Sadly, we no longer remember why it was set.  I wouldn't object to
> someone removing the CFG_NO_FLASH definition if they could test it.
> 
> I might not even object that much if it broke RAM_BOOT.  Get a BDI and
> program the flash, people!  ;)
> 
> I would even agree to test that change on one of our boards if you
> remind me what I need to change to boot from RAM.  I've not booted
> from RAM since we were bringing up the 8540!

I don't have the hardware to test it, but apparently you build for 
RAM_BOOT by setting TEXT_BASE below the start of flash. I tried 
building MPC8540ADS with 'make TEXT_BASE=0xf0080000' (I don't know 
if that is the right address but at least it is in SDRAM) and it 
won't build because CFG_NO_FLASH and CFG_CMD_FLASH are incompatible.

If I remove CFG_NO_FLASH it almost builds but I get a relocation error
in cpu/mpc85xx/start.o (maybe 0xf0080000 isn't the best address?)
Also, the comment on CFG_NO_FLASH is 'Flash is not usable now', so I
don't liking removing it.

If I disable CFG_CMD_FLASH and CFG_CMD_IMLS like MPC8641HPCN does
for RAM_BOOT it builds clean. Jon, did you actually run the RAM_BOOT
configuration for the MPC8641HPCN? This is at least less broken
than it is now, so I'd like to take this approach, but I can't
test more than the build.

Dave

--
Dave Ellis
___________________________________________________
SIXNET - Truly Open Industrial Automation Solutions 
PO Box 767, 331 Ushers Road, Clifton Park, NY 12065 
Phone: +1(518)877-5173,  Facsimile: +1(518)877-8346 
E-mail: mailto:dge at sixnetio.com
Get product details at http://www.sixnetio.com