[yocto] Boot failure when adding the meta-oe layer

[Paul Eggleton]

On Tuesday 11 December 2012 12:54:57 Zach Tibbitts wrote:
> I'm working on an embedded Linux platform for work to run an app that's
> written with Gtk+3. The hardware in question is a Sandy Bridge Core i3
> platform. With a vanilla Yocto build, I can boot an image just fine (I've
> tested with core-image-basic, core-image-minimal, and core-image-x11) and
> everything seems to work; I can even install it to the internal drive
> without issue.
> 
> I found recipes for Gtk+3, as well as some of our other dependencies in the
> meta-oe layer from the meta-openembedded repository. However, just adding
> these layers and rebuilding any of the images that I've built (without
> actually adding any new recipes) cause the image to hang when run on
> hardware (but oddly enough not in QEMU).
> 
> The last line printed before the bootup apparently stops is along the lines
> of: "sd 4:0:0:0: [sdb] Attached SCSI removable disk", and the line that
> seems most at fault is a little farther up, "/init: line 69: udevd: not
> found". If needed, I can try to capture more of the output, maybe with a
> camera, I'm not sure of an easy way to pastebin the output from hardware...
> 
> What I'm wondering is if this is a known issue, or if meta-openembedded
> isn't meant to be used in conjunction with Yocto. My assumption is that
> something from meta-oe is superceeding a recipe in Yocto and causing this
> failure.

This is a known issue in meta-oe unfortunately; the udev version there 
apparently does not work outside of use with systemd. I believe the plan is to 
drop udev from meta-oe fairly soon.

The workaround is to force the system to use the version of udev in meta/ - 
I'm guessing you are using the 1.3 release version, in which case you can add 
the following to your local.conf:

PREFERRED_VERSION_udev = "164"

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre