[meta-ti] has anyone actually tested sdcard_image.bbclass?

[Chris Larson]

On Thu, Jan 10, 2013 at 3:50 PM, Trevor Woerner <twoerner at gmail.com> wrote:

> On Thu, Jan 10, 2013 at 10:07 AM, Gary Thomas <gary at mlbassoc.com> wrote:
> > Interestingly enough, the meta-fsl-arm layer also has an SD card image
> > layout (class) which works perfectly and needs no magic to make it work.
>
>
> Very interesting! Thanks so much for posting this information.
>
> The meta-ti method uses a combination of losetup, sfdisk, dd, and
> others to create its images (which is what I have traditionally used
> to create my own images). losetup (for sure and probably sfdisk too)
> requires privilege in order to work (which probably accounts for all
> the messy setup).
>
> The meta-fsl-arm method, on the other hand, uses GNU parted which
> doesn't appear to require privilege and does away with losetup and
> sfdisk.
>
> I'm glad you posted this information as I'll now be converting my
> scripts to use parted as well.


We started looking at a parted based pandaboard sdcard image creation setup
for Mentor, but haven't completed it, so we'd be interested in seeing what
people do in this regard. Obviously it'll all probably go away in favor of
the image creation/deployment rework in the long term, but we need
something in the short term.

One issue we ran into is, my panda seems to require a particular alignment
(e.g. first partition at 63*255) based on geometry, but parted's alignment
options are rather limited. Either one can use pyparted (but that's GPL,
and the metadata is MIT), or use sfdisk (which has its own issues, but lets
you specify geometry), or manually calculate the alignment (theoretical).
-- 
Christopher Larson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.yoctoproject.org/pipermail/meta-ti/attachments/20130111/5d6efc83/attachment.html>