gmane.linux.distributions.angstrom.devel

Re: Trouble building Angstrom for Beaglebone

Joseph Kubicky — 2012-05-26T02:37:30

I didn't know about .done files.  If I create a .done file for a
package will bitbake not try to download it?

- Jay


On Fri, May 25, 2012 at 7:10 PM, Denys Dmytriyenko <denis-hqcsM7nSK1nYtjvyW6yDsg< at >public.gmane.org> wrote:

Re: Trouble building Angstrom for Beaglebone

Denys Dmytriyenko — 2012-05-26T02:10:32

And did you have the corresponding *.done file for it?

Re: Trouble building Angstrom for Beaglebone

Joseph Kubicky — 2012-05-26T00:03:30

I tried your idea first, but when I reran the build it just pulled down the earlier (bad) version and wrote over the one I had downloaded. That's why I tried the approach I described. 

Jay 



On May 25, 2012, at 4:17 PM, Denys Dmytriyenko <denis-hqcsM7nSK1nYtjvyW6yDsg< at >public.gmane.org> wrote:

Re: Trouble building Angstrom for Beaglebone

Denys Dmytriyenko — 2012-05-25T23:17:52

Well, the suggestion was actually to download the copy off YP manually and 
place it in your downloads/ directory - you don't need to change any recipes.

As of it downloading "something" - debian servers have bad habit to serve an 
HTML page with instructions to where to go look for archived packages, instead 
of serving the regular 404 error. Due to that, the fetcher doesn't see the 
download as a failure and saves the above HTML message as a tarball in 
question. Just do a "file" inspection on that file with bad checksum...

Trouble building Angstrom for Beaglebone

Joseph Kubicky — 2012-05-25T23:01:49

Thanks.  In setup-scripts/sources/openembedded-core/meta/recipes-core/netbase,
I changed the SRC_URI (in netbase_4.47.bb) to use the address you
provided and now the build completes.  I still get a checksum error
(similar to before - see below), but now it is listed as a warning
instead of an error.

Previously, netbase_4.47.bb had  SRC_URI =
"${DEBIAN_MIRROR}/main/n/netbase/netbase_${PV}.tar.gz...", which
apparently was finding something - there was a netbase_4.47.tar.gz in
setup-scripts/sources/downloads, but it was only about half the size
of the one you pointed me to (and apparently invalid, as it caused the
build to stop).

We'll see if what I just built actually works on a beaglebone.

- Jay


NOTE: Running task 3031 of 3323 (ID: 1373,
/home/jayk/ang/setup-scripts/sources/openembedded-core/meta/recipes-core/netbase/netbase_4.47.bb,
do_fetch)
NOTE: Running task 3072 of 3323 (ID: 1239,
/home/jayk/ang/setup-scripts/sources/openembedded-core/meta/recipes-extended/tzdata/tzdata_2012b.bb,
do_unpack)
NOTE: pa

Re: Trouble building Angstrom for Beaglebone

Gary Thomas — 2012-05-25T21:33:49

In the meantime, try picking up a copy from
   http://downloads.yoctoproject.org/mirror/sources/netbase_4.47.tar.gz

Re: Trouble building Angstrom for Beaglebone

Jack Mitchell — 2012-05-25T21:12:52

Netbase 4.47 has just been removed from the Debian ftp, there has been a 
patch posted already updating to 5.xx and I believe that it will be 
pulled soon so it's just a matter of waiting, or merging the patch 
posted to the oe-dev mailing list.

Regards,
Jack.

Trouble building Angstrom for Beaglebone

Joseph Kubicky — 2012-05-25T17:48:43

I'm trying to build Angstrom for the Beaglebone.  I'm pretty sure I'm using
the most recent version of everything.  It almost works, but when I run
'MACHINE=beaglebone ./oebb.sh bitbake systemd-image', I get an error:

| ERROR: Function failed: Fetcher failure for URL: '
ftp://ftp.debian.org/debian/pool/main/n/netbase/netbase_4.47.tar.gz'.
Checksum mismatch!
| File: '/home/jayk/ang/setup-scripts/
sources/downloads/netbase_4.47.tar.gz' has md5 checksum
4c27c16e983f681a197a04bf74523440 when fcd87a62f177592724a8066e3774ae97 was
expected
| File:
'/home/jayk/ang/setup-scripts/sources/downloads/netbase_4.47.tar.gz' has
sha256 checksum
333f1a6c1181c6fbccc8f81ac6aeafac4d8f727cf4e9e9247c9b8bfe0d369c22 when
a7e6525c9a09b00eff3395d09739b3c70a18bff8d13f7daa21fdf81989387726 was
expected
NOTE: package netbase-4.47-r2: task do_fetch: Failed
ERROR: Task 1373
(/home/jayk/ang/setup-scripts/sources/openembedded-core/meta/recipes-core/netbase/
netbase_4.47.bb, do_fetch) failed with exit code '1'
Waiting for 1 running tasks to f

Re: Kernel 3.1.14 usb mount issue on BeagleBone

Mattias Millbro — 2012-05-25T15:01:02

The kernel version I'm talking about is of course 3.2.14.

Sorry for that.
/ M

On 05/25/2012 04:39 PM, Mattias Millbro wrote:

Kernel 3.1.14 usb mount issue on BeagleBone

Mattias Millbro — 2012-05-25T14:39:00

Hi all,

I'm glad that my previous email to this list lead to a fix of the 
missing hashes in python-crypt. Unfortunately I'm experiencing a new issue.

I'm using my BeagleBone for logging serial data, using the usbserial and 
ftdi-sio kernel modules. Installing them with opkg will update the 
kernel to 3.1.14, which is fine, apart from the fact that the serial 
data is to be saved on a usb-drive. As you probably know, usb-drives is 
not mountable in 3.1.14 as they do not show up as a mountable device in 
/dev/ (as discussed here: 
https://groups.google.com/forum/?fromgroups#!topic/beagleboard/IwzUbg3JY3c 
).

Im working against a very tight deadline, so I have the following question:

- Where can I find older versions of the packages 
kernel-module-usbserial and kernel-module-ftdi-sio that has a lower 
kernel version dependency? I've spent hours in the feed browser without 
sucess.

- Do you know what is causing the usb-drive issue?

Grateful for any kind of feedback and keep up the good work!
/Mattias

Re: compiling a kernel module [solved]

CB — 2012-05-25T13:24:58

Hi again

addition to my previous message :

bitbake -b linux-mainline_3.2.bb -c compile_kernelmodules -f

... seems a better approach to recompile modules after modification

HTH
Chris

Le 23/05/2012 13:09, chrbruno a écrit :

problem with 'init'

Gyorgy Kovesdi — 2012-05-23T13:46:07

Hi,

I have a Beagleboard C4

I tried to build systemd-gnome-image by Narcissus and on my machine too. Some other
packages are tried as well with the same result, but i would like to use this one.
Using the latest version i have problem with 'init'. I found that no init scripts are executed (at lest
for levels 3 and 5) and therefore e.g. the X is not started after boot.
Executing 'init 5' or 'init 3' does nothing. However, 'init 0' and 'init 6' are work as expected.
I have the file '/etc/inittab' , but its content is also ignored.
I can start the X by executing '/etc/init.d/gdm start'. However, its link in /etc/rc5.d/ is also ignored
by init.

How can i build working init?

Thanx in advance
Gyorgy Kovesdi

Re: USB OTG not working for linux-mainline-3.2 (was Problem with usbhid kernel module)

Frank Agius — 2012-05-23T13:02:01

This relates to usb code changes committed July 2011.  As a result of 
these chages a gadget driver must be loaded in order for host mode to be 
activated.  If you load a gadget driver, then host mode should be activated.

frank

Re: USB OTG not working for linux-mainline-3.2(was Problem with usbhid kernel module)

Gyorgy Kovesdi — 2012-05-23T13:06:50

I forgot about an important detail: my USB is working, devices other than mouse and keyboard
are functional, it is only a module insertion problem. I don't know where the module 'usbhid'
should be inserted, so IMHO it is an Angstrom-related problem.

On 2012 May 23 13:05:02 Andreas Müller wrote:

Re: compiling a kernel module [solved]

chrbruno — 2012-05-23T11:09:18

Hello,

I just tested your kernel workflow
I have modified the em28xx-core.c file (in 
kernel....drivers/media/video/em28xx/ )
after the -c compile and -c build, I see that the em28xx-core.o has not 
been updated

finally, i ran the command    bitbake -b linux-mainline_3.2.bb
compilation is not finished now, but as far as I see, the module (.ko) 
is up to date

This solution seems to be working for what I need (module-only 
recompilation)
Now, I will test your suggestions to reconfigure the kernel

Thanks to you all for your help !
Best Regards,

Chris


Le 22/05/2012 21:17, Bob Feretich a écrit :

Problem with usbhid kernel module

Gyorgy Kovesdi — 2012-05-23T08:16:22

Hi,

I have a Beagleboard C4, using it with USB keyboard and mouse.
Building the current version of Angstrom results a non-functional keyboard and mouse at the first boot.
After some digging, i found that the kernel module "usbhid" is not compiled in. It is built as a module (m), but
is not loaded during startup. Some weeks ago, it was compiled in (y) in the default kernel config of Beagle.

This way the only possibility to get it working is to log in using the serial port, and set up manually.
I think it is related to other Beagle boards (e.g. xM), however, other people reported that it is working. Do they
use it other way, or do i made something wrong?

Regards
Gyorgy Kovesdi

Re: compiling a kernel module

Gyorgy Kovesdi — 2012-05-23T07:03:51

$> bitbake virtual/kernel -c menuconfig

Re: compiling a kernel module

chrbruno — 2012-05-22T19:38:50

Le 22/05/2012 19:49, Don deJuan a écrit :

You are right, bitbake -f -c compile virtual/kernel leaves the sources 
in place (with my modifications)
But I cannot find the module (em28xx.ko) file with a recent date&time. 
It seems it is not rebuilt.
I think another approach is to get a pure kernel from kernel.org, apply 
the patches, overwrite the .config and "make uImage modules"
I will try this !
thanks anyway for your help !
Best Regards,
Chris

Re: compiling a kernel module

Bob Feretich — 2012-05-22T19:17:08

I generally do my kernel builds in 3 steps (steps 3-5). (5 steps for the 
first time)
The kernel is built when Angstrom builds, but sources are deleted. I 
like the source available for examination and want to be able to easily 
reconfigure the kernel.

 1. bitbake -c clean virtual/kernel          // ready OE to rebuild the
    kernel
 2. bitbake -f -c compile virtual/kernel   // execute the OE build up to
    the compile step
 3. cd ....../git
    make menuconfig                             // edit the kernel
    configuration
 4. bitbake -f -c compile virtual/kernel   // force re-do of the compile
    step with the new config settings
 5. bitbake  -c build virtual/kernel          // execute the OE steps
    after the compile, but don't erase
                                                            //     
    intermediate results (kernel sources)

With the intermediate results left intact, I can easily build driver 
modules independently for testing.
bitbake  -c build -b <my driver recipe .bb>
(Generate

Re: compiling a kernel module

Don deJuan — 2012-05-22T17:49:59

When you are done modifying your code could you not just run
bitbake -f -c compile virtual/kernel ?
I have not messed with kernel modules other than changing the config. By 
doing the -f -c compile , if I am correct should only force a compile of 
virtual/kernel again leaving the modifications in place.
Hopefully a bitbake/OE master will chime in the actual proper commands 
to do exactly what you wish.
HTH

Re: compiling a kernel module

chrbruno — 2012-05-22T17:41:29

Thanks for your reply,

in fact, I use the virtual/kernel as a target
I don't know what name to provide for "YourModule"
The module I work on is driver/media/video/em28xx but "em28xx" is not 
recognized as a target by bitbake

when I run a bitbake -c clean virtual/kernel, it removes all the kernel 
and module sources (the source code I want to modify)

I think I will dig into the OE documentation !
or another approach would be to get the .config/patches  from angstrom 
tree, and compile a pure kernel using the classical "linux" method

Best Regards,
Chris


Le 22/05/2012 18:10, Don deJuan a écrit :