Got past the libjpeg-turbo problem, but now it’s failing on ltp-ddt_0.0.4.bb:
`
ERROR: Function failed: do_compile (see /home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/work/beaglebone-angstrom-linux-gnueabi/ltp-ddt/0.0.4-r1/temp/log.do_compile.20093 for further information)
(NOTE: I’m not 100% sure the GCC command is the one causing the error … but it’s the only one in the log near to the errors)
In file included from /home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/sysroots/beaglebone/usr/src/kernel/include/linux/i2c.h:29:0,
from src/interface/common/st_i2c_common.h:54,
from src/testcases/st_i2c_eeprom.c:51:
/home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/sysroots/beaglebone/usr/src/kernel/include/linux/mod_devicetable.h:18:2: error: unknown type name ‘__u32’
/home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-
…
In file included from /home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/sysroots/beaglebone/usr/src/kernel/include/linux/kobject.h:27:0,
from /home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/sysroots/beaglebone/usr/src/kernel/include/linux/device.h:17,
from /home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/sysroots/beaglebone/usr/src/kernel/include/linux/i2c.h:30,
from src/interface/common/st_i2c_common.h:54,
from src/testcases/st_i2c_eeprom.c:51:
/home/winston/Work/bbb/setup-scripts/build/tmp-angstrom_v2013_06-eglibc/sysroots/beaglebone/usr/src/kernel/include/linux/wait.h:8:25: fatal error: asm/current.h: No such file or directory
compilation terminated.
`
Looks like some issues with includes, missing linux/types.h?
Did you have a branch mismatch when you did your initial setup?
I’m seeing meta-ti being pulled from git.yoctoproject.org/meta-ti, and configured with branch: angstrom-staging-yocto1.4
However that repository doesn’t have that branch (see below), however github.com/Angstrom-distribution/meta-ti does have the angstrom-staging-yocto1.4 branch.
Here are my steps:
git clone git://github.com/Angstrom-distribution/setup-scripts.git
cd setup-scripts
git checkout angstrom-v2013.06-yocto1.4
MACHINE=beaglebone ./oebb.sh config beaglebone << error in output setting meta-ti branch
Here are the valid remote branches for meta-ti:
$ cd setup-scripts/sources/meta-ti
Did you have a branch mismatch when you did your initial setup?
Yes. But the errors aren’t “fatal” as they are if you use the 2012.12 (master) branch of setup-scripts (which is what the documentation says to use) due to the missing ‘danny’ branch.
I’m seeing meta-ti being pulled from git.yoctoproject.org/meta-ti, and configured with branch: angstrom-staging-yocto1.4
However that repository doesn’t have that branch (see below), however github.com/Angstrom-distribution/meta-ti does have the angstrom-staging-yocto1.4 branch.
Here are my steps:
git clone git://github.com/Angstrom-distribution/setup-scripts.git
cd setup-scripts
git checkout angstrom-v2013.06-yocto1.4
MACHINE=beaglebone ./oebb.sh config beaglebone << error in output setting meta-ti branch
Here are the valid remote branches for meta-ti:
$ cd setup-scripts/sources/meta-ti
Also looks like this commit fixed the repo location for meta-ti: ae663d02a24e123418c7a57b018c183f8b9fc53b
This fixed my missing branch message.
Maybe I do understand this?
Next question is how to create the .img.xz image files. After running the bitbake cloud9-image, I have the following files in ./build/tmp-angstrom_v2012_12-eglibdeploy/images/beaglebone:
Next question is how to create the .img.xz image files. After running the bitbake cloud9-image, I have the following files in ./build/tmp-angstrom_v2012_12-eglibdeploy/images/beaglebone:
I did get this to work, but only by accident [because the eMMC on my BBB had already been set up]. I used the 8-21-13 version of the flasher, which might not have been a good idea, but I was able to mount the img, update ID.txt, replace the rootfs with mine and update the md5 hashes in the script to match.
When I booted it, the flashing process failed (the leds went out), but although I had the serial console connected, there was no “output” to indicate it had failed. Next, I logged in and manually ran the [modified] /usr/bin/emmc.sh script and found that there were some errors. Notably, the blockdev and kpartx tools were missing from the flasher linux image. However, the unpack of the rootfs succeeded and it did copy in the updated MLO and u-boot.bin files.
One thing that would make rolling-your-own flasher image easier if the following values were read from a config file rather than being hard coded into the emmc.sh script:
MD5 hash for MLO
MD5 hash for u-boot.bin
Root filesystem archive filename
ID/dogtag value
Extra stuff to copy/validate (such as the BBB getting started stuff, wallpaper etc … I didn’t need that so I commented it out)
Logging any errors to the console/syslog during flashing (just in case anyone is watching with a serial console)
Ultimately, I got what I needed, thanks to everyone for their help!
FYI it seems that cloud9’s pointer to sass is no longer around…this should only matter if you’re using the 2013 branch since the build process does some caching via sources.angstrom-distribution.org.
Should you want to build 2013 with the cloud9 image, you’ll need to mod this file:
setup-scripts/sources/meta-openembedded/meta-oe/recipes-devtools/cloud9/cloud9_0.6.bb