[oe] RFC; Building multiple u-boot images for a single board

[Tom Rini]

On Fri, 2010-02-26 at 20:27 +0100, Koen Kooi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On 26-02-10 19:45, Tom Rini wrote:
> > On Fri, 2010-02-26 at 13:57 +0100, Koen Kooi wrote:
> >> -----BEGIN PGP SIGNED MESSAGE-----
> >> Hash: SHA1
> >>
> >> On 26-02-10 12:37, Ulf Samuelsson wrote:
> >>> If you want to support multiple boot memories, you have to
> >>> have multiple boards in the conf/machine directory
> >>>
> >>> I currently testing a change to at91bootstrap, where a defconfig
> >>> is not provided by openembedded.
> >>> Instead you provide a list of defconfig's in your machine description:
> >>>
> >>> I.E: in conf/machien/at91sam9g45ek.conf you have:
> >>>
> >>> AT91BOOTSTRAP_BOARD = "at91sam9g45df at91sam9g45ek at91sam9g45nf"
> >>>
> >>> when at91bootstrap is built it will loop through all the boards.
> >>> and build three versions.
> >>>
> >>> I think it would make sense to do the same for u-boot,
> >>> so that you can build u-boot for several configurations.
> >>
> >> Not about u-boot, but slightly related:
> >>
> >> I need something similar for uImages due to hardware limitations
> >> (daughtercards requiring conflicting pinmux) and I have created this for
> >> kernel builds:
> >>
> >> http://dominion.thruhere.net/git/cgit.cgi/openembedded/tree/recipes/linux/multi-kernel.inc?h=ti/staging
> >>
> >> It's working nicely for my usecase (validate hardware configs with
> >> multiple kernels, validate software with only one kernel).
> > 
> > That indeed looks useful.  Any chance it'll go in soon?  Or what's left
> > to cleanup before it can?  Aside from some shellfile for pstaging that
> > is.
> 
> The goal is "next wednesday" provided the tests on monday/tuesday pass,
> but see Denys' respone about shellfile assumptions.

Yeah, DEPLOY_DIR outside of TMPDIR needs some special thought.
DEPLOY_DIR_PSTAGING outside of TMPDIR is fine, happily.

-- 
Tom Rini <tom_rini at mentor.com>
Mentor Graphics Corporation