[poky] [PATCH 1/4] crownbay: remove custom BASE_PACKAGE_ARCH and PACKAGE_EXTRA_ARCHS

Tom Zanussi tom.zanussi at intel.com
Tue May 17 10:59:14 PDT 2011


On Tue, 2011-05-17 at 10:25 -0700, Richard Purdie wrote:
> On Mon, 2011-05-16 at 09:17 -0500, Tom Zanussi wrote:
> > On Mon, 2011-05-16 at 01:37 -0700, Richard Purdie wrote:
> > > On Sun, 2011-05-15 at 00:13 -0500, Tom Zanussi wrote:
> > > > On Sat, 2011-05-14 at 21:18 -0700, Wold, Saul wrote:
> > > > > On 05/14/2011 08:36 PM, Tom Zanussi wrote:
> > > > > > From: Tom Zanussi<tom.zanussi at intel.com>
> > > > > >
> > > > > > PACKAGE_EXTRA_ARCHS in the conf file gets overwritten by the tune-atom
> > > > > > include following it, preventing the atom BASE_PACKAGE_ARCH from being
> > > > > > included in the final PACKAGE_ARCHS and causing do_rootfs to fail.
> > > > > > Remove the atom BASE_PACKAGE_ARCH and PACKAGE_EXTRA_ARCHS in the conf
> > > > > > file and allow the values in tune-atom to be used as intended.
> > > > > >
> > > > > I think we may still want to change the tune-atom PACKAGE_EXTRA_ARCHS to ?=.
> > > > > 
> > > > 
> > > > Yeah, I'll change that along with a couple other changes to that file,
> > > > such as switching -march=core2 to -march=atom, soon...
> > > 
> > > Does that actually make any difference to gcc? It never used to but I'm
> > > wondering if anything has changed?
> > 
> > I don't know yet, which is why I didn't just make the change
> > straightaway - it would need some benchmarking first, at least.
> > 
> > I would be kind of surprised though if it didn't have some effect -
> > otherwise why would it exist at all?  It sounds like you've tested this
> > before and it didn't make any difference - how did you test, and do you
> > have any benchmarking suggestions that would help decide one way or the
> > other?
> 
> I think I looked at the gcc codebase and found it didn't do anything.
> Its more common that you'd have thought in gcc :/
> 

OK, good to know.  I'll take a look when I get a chance and see if
anything's changed in that area lately...

Tom

> Cheers,
> 
> Richard
> 





More information about the poky mailing list