[OLPC-devel] Re: OLPC development board

David Woodhouse dwmw2 at infradead.org
Sat Jun 10 18:33:28 EDT 2006


On Sat, 2006-06-10 at 18:17 -0400, Michail Bletsas wrote:
> I have a feeling that the Flash interface on the 88W8388 will not meet our 
> performance requirements since its purpose is to access firmware.
> 
> A solution using the 88W8388's flash controller  will have a relative long 
> path from the main CPU to the flash (GX2 <-> 5536 <-USB-> 88W8388 <-> 
> Flash).
> Such a chain will make it hard to run JFFS2 on, am I missing something?

I don't think we actually _do_ use it for firmware, do we? We're loading
firmware from the host side instead? But yes, you're probably right --
the latency isn't going to be great.

> Mark Foster has already started looking for an ASIC solution for 
> interfacing the flash (separately from the DCON chip) via PCI.

OK, that's good. Have you spoken to Thomas about the working CPLD
implementation we already have working with Linux, and which we have
permission to relicense?

> Should we really pursue the 88W8388 option with Marvell? 

Since they seem to have gone silent, I suppose there's no point in
_chasing_ them to talk to us... :)

-- 
dwmw2




More information about the libertas-dev mailing list