[meta-freescale] Unable to get u-boot running on i.MX6 SABRE for Smart Devices Eval Board

Daiane Angolini daiane.list at gmail.com
Wed Nov 5 06:56:54 PST 2014


On Wed, Nov 5, 2014 at 12:38 PM, Guillaume Fournier
<gfournier at brioconcept.com> wrote:
> Fabio,
>
> The ultimate goal is to have a cross-compiler build environment that works for both u-boot and kernel development.  For now, I used Yocto to generate the toolchain (using meta-toolchain as specified in my first post) as described in section 3.4 of the Yocto Project Application Developer's Guide.  Following your lead I did use "bitbake core-image-base" to generate an image and burned core-image-base-imx6dlsabresd.sdcard to an SD card using:
> cfimager -f core-image-base-imx6dlsabresd.sdcard -raw -d d
>
> I used cfimager which is a tool that comes with IMX_6DL_6S_MFG_TOOL since I don't have physical access to a linux box.  The board boots up fine but this is not the goal I seek.  I want to get a working cross-compiling toolchain so I can build u-boot and the kernel outside of Yocto.  I understand the subject of this post is therefore a bit misleading since u-boot does boot when using Yocto build output...  It just doesn't when using the cross-compiling toolchain.
>

use the cfimage to burn the u-boot binary in your sdcard. I understand
it would be the similar to the dd command. And, at least, you're going
to test if the u-boot image is OK or not.

mfgtool does not work the way you're expecting. You must have an
u-boot + uImage + initramfs to *start* the burning cycle of your own
image. And I have never tested to burn only one new u-boot with
mfgtools.

For me it looks like a mfgtools "misuse" or "misbehavior" more than
toolchain problem.

Oh, and only one more thing. The shell you use to bitbake your
toolchain is not the shell you use to export that toolchain.

Everytime you export the setup environment to use the toolchain, use a
clean shell (I faced weird errors in the past because of forgetting to
clean the environment variables)

Daiane


More information about the meta-freescale mailing list