[poky] "core" explosion?

[Tom Zanussi]

On Mon, 2011-04-25 at 05:14 -0700, Gary Thomas wrote:
> I've been building from a Poky tree, keeping up to date with
> the master branch for quite some time.  Recently, I used this
> tree to build to do some testing on my [emenlow-like] Atom.
> 
> Using the external meta-intel, I've built for atom-pc since it's
> supported by core (but sadly doesn't work on my Atom PC!) and
> emenlow (which does work).  I've noticed quite a change in the
> size of images over the past month:
> 
> ls -l /work/tmp/*.hddimg
> -rw-r--r-- 1 gthomas gthomas  521244672 Apr 22 15:00 /work/tmp/core-image-sato-live-atom-pc.hddimg
> -rw-r--r-- 1 gthomas gthomas 2056809472 Apr 22 22:28 /work/tmp/core-image-sato-live-emenlow.hddimg
> -rw-r--r-- 1 gthomas gthomas  508077056 Mar 21 12:37 /work/tmp/poky-image-sato-live-atom-pc.hddimg
> -rw-r--r-- 1 gthomas gthomas 1032937472 Mar 22 05:40 /work/tmp/poky-image-sato-live-emenlow.hddimg
> 
> Notice that with the new "core" images, the size is bigger, but
> why is the emenlow image _doubled_ whereas the atom-pc version only
> grew by around 4%?

Previous to the below commit, the IMAGE_ROOTFS_SIZE overrides were being
ignored and the rootfs size defaulted to the actual image size + 20%.
Now IMAGE_ROOTFS_SIZE is driving the image sizes as expected.

commit adac5a690fd5b4239bb59ec54e3e5174dbcf29e9
Author: Richard Purdie <richard.purdie at linuxfoundation.org>
Date:   Thu Mar 24 12:29:19 2011 +0000

    image.bbclass: Move ROOTFS_SIZE into runimagecmd() function