[OLPC-devel] Re: Not so good news (Was: good news)

[Stefan Reinauer]

* Eric W. Biederman <ebiederm at xmission.com> [060807 18:50]:
> > You can flash the payload independently as before. You just have to make
> > sure its either compressed or not so.
> 
> With which algorithm?
> 
> There is a coupling here.  As long as the LinuxBIOS tooks take care and handle
> everything we are fine.
 
At the moment we're supporting nrv2b but we want to exchange this to
something "better" in the future. 

> My concern is simply with the compression option we have the potential to make
> things more complicated for users of LinuxBIOS.  I don't want to do that.
> Saying don't flash the payload separately is an easy way to ensure we don't get
> that extra complexity.
 
The old way works fine: Have a payload that takes care of compression
itself and don't enable CONFIG_COMPRESSED_ROM_STREAM.

So for the user we dont break anything.

Changing the compression algorithm would definitely be a candidate for
pushging the version number of LinuxBIOS. We might even want to make it
build time configurable but then again, I'd prefer a single "good for most
situations" algorithm and keep that.

nrv2b was great because its simple. Especially when uncompressing
linuxbios_ram from assembler code was needed.

but if we have plenty of "components/modules" we want to have them
compressed all the same way. Still not all together because we
want to be able to exchange them seperately.

Stefan

-- 
coresystems GmbH • Brahmsstr. 16 • D-79104 Freiburg i. Br.
      Tel.: +49 761 7668825 • Fax: +49 761 7664613
Email: info at coresystems.de  • http://www.coresystems.de/