Re: Strataflash mapping problem

[Gary Thomas <gary at mlbassoc dot com>]

On Mon, 2004-02-16 at 14:15, Jonathan Larmour wrote:
> Enric Bel Prim wrote:
> > Hi, again.
> > Yes, I agree with Jeffrey Daly that CYGNUM_FLASH_DEVICES should be the real 
> > number of flash devices (4 in my platform), and CYGNUM_FLASH_WIDTH should 
> > be the number of bits in each device (16 in my platform).  But 
> > CYGNUM_FLASH_INTERLEAVE should be 1 for my platform (16 bits devices 
> > located in a 16 bit expansion bus of IXP425) and never equal to 
> > CYGNUM_FLASH_DEVICES  (as is defined in line 95 in strata.h)....
> > Finally I'm considering using the generic driver for the Intel 28fXXX 
> > devices instead of the Intel strata to acces the Intel Strata Flash devices 
> > of my platform.
> > 
> > The 28fXXX driver requires the following settings:
> > 
> > // Platform code must define the below
> > // #define CYGNUM_FLASH_INTERLEAVE      : Number of interleaved devices (in 
> > parallel)
> > // #define CYGNUM_FLASH_SERIES          : Number of devices in series
> > // #define CYGNUM_FLASH_WIDTH           : Width of devices on platform
> > // #define CYGNUM_FLASH_BASE            : Address of first device
> > 
> > For my platform I will set:  (CYGNUM_FLASH_INTERLEAVE = 1, 
> > CYGNUM_FLASH_SERIES = 4, CYGNUM_FLASH_WIDTH = 16, CYGNUM_FLASH_BASE = 
> > 0x50000000)
> > I think CYGNUM_FLASH_SERIES  will solve my flash mapping problems???  I 
> > wonder why the strata flash driver doesn't allow this setting!!.
> 
> Because no-one has implemented it :-).

Which mostly means that we (collectively) haven't seen a board with 
devices arranged this way.

> 
> > Does anybody know  if I can use the 28fXXX driver to acces Intel Strata 
> > Flash devices? I mean, will I loose some functionality/feature of the 
> > Strata devices if I use the generic 28fXXX driver?
> 
> I don't know, but if you look at where CYGNUM_FLASH_SERIES is mentioned in 
> the 28fxxx driver, you'll see it should be pretty easy to bring the same 
> ideas over into the strata driver if you needed to.
> 
> Given the buffered write support is broken, I don't perceive any real 
> difference in functionality between the drivers, but I haven't looked in 
> detail.

IIRC, the 28Fxxx driver is somewhat newer and has better device support.
If changes are to be made, it would be best to put them into this
driver.
-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates