This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
RE: eCos flash-device support questions
- To: Andreas Bürgel <ab at genologic dot de>
- Subject: RE: [ECOS] eCos flash-device support questions
- From: Gary Thomas <gthomas at cambridge dot redhat dot com>
- Date: Fri, 18 May 2001 07:12:08 -0600 (MDT)
- Cc: ecos-discuss at sources dot redhat dot com
- Organization: Red Hat, Inc.
On 18-May-2001 Andreas Bürgel wrote:
> After loosing lots of sweat I made it to program my onboard flash-device
> of type Fujitsu 29LV800B - compatible to the AMD am29LV800 types.
You can compare your work with the recently released driver at:
devs/flash/powerpc/viper
> But there are two things making me little headache.
> 1.) I`ve read the device-id 0x22B5 with eCos flash support, but my
> flash-programmer(-hardware) reads 0x005B as device-id.
Possibly the difference between AMD & Fujistu. A data sheet could confirm.
> 2.) It does not play a big role, but eCos-flash-support assumes that the
> sector/block-size of the flash-chip is consistent over the whole
> address-range. But the 29LV800B has 4 different block-sizes dependent of
> the address range I want to write to.
Indeed. Since the major reason for the Flash drivers is use by RedBoot's
Flash Image System, a very simplified approach has been taken to date. This
approach is to treat flash devices as if they are composed of a set of
equal sized blocks, with the block size set to the largest block. While
this does work, it is certainly sub-optimal. In the case of this particular
part, the first four sectors are grouped together to make up a single "super"
block.
We are continuing to think about how Flash can be handled well. Any comments
or suggestions are welcome.