Bug 1073

Summary: beagle-xm: random MAC address
Product: [Build System, Metadata & Runtime] Meta-yocto Reporter: Robert Berger <pokylinux>
Component: configurationAssignee: Bruce Ashfield <bruce.ashfield>
Status: RESOLVED FIXED QA Contact:
Severity: enhancement    
Priority: Low CC: poky.bs.watcher, poky.watcher, yang.shi
Version: 1.0   
Target Milestone: ---   
Hardware: Beagleboard   
OS: arm   
Whiteboard: RFC out to meta-ti (29/Jun/2011)
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: ---

Description Robert Berger 2011-05-15 02:30:17 UTC
The beagle-xm issues a random MAC address with each reboot, which makes it a bit "non standard" to work with DHCP.

I don't know if that's the right way to go, but one possible solution is here:
http://maxgalemin.blogspot.com/2010/11/patch-for-fixing-random-mac-address-on.html
Comment 1 Darren Hart 2011-05-19 13:27:22 UTC
Hi Robert, has this kernel patch been pushed upstream?
Comment 2 Robert Berger 2011-05-19 22:43:29 UTC
(In reply to comment #1)
> Hi Robert, has this kernel patch been pushed upstream?

Hi Darren,

Frankly I don't know. Let me ask around on the various mailing lists. It might be something you should not do in the kernel.
Comment 3 Robert Berger 2011-05-19 22:55:15 UTC
It looks like there is/was something upstream from the Linaro team:

https://bugs.launchpad.net/ubuntu/+source/linux-linaro/+bug/687396

There is also something on openembedded which uses the OMAP's die ID as a MAC (which I find a neat idea)

http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-October/026240.html
Comment 4 Darren Hart 2011-06-29 14:23:39 UTC
Sent a mail to oe-core and the meta-ti maintainers for their preference on a solution to this as we don't want to deviate from where they go with this.
Comment 5 Robert Berger 2011-06-29 14:57:44 UTC
Hi Darren,

Assuming the MAC address comes from u-boot it might be possible to get away without a kernel patch.

I did not try it out on the beagle-xm, but in general u-boot variables can be read from user space - see u-boot/tools/env.

Afterwards a script could be cooked, which patches this u-boot variable into the network config: hwaddress ether <something>

Just a thought at 1:00 am so I might also just be too tired for anything more creative.
Comment 6 Darren Hart 2011-06-29 16:12:06 UTC
There is some debate on how to best resolve this. We will wait for meta-ti to decide on an approach, and then follow that.
Comment 8 Darren Hart 2012-04-06 22:12:56 UTC
Bruce, can you (or a delegate) pick this one up - or close it as will not fix?
Comment 9 Yang Shi 2012-04-09 18:40:26 UTC
Checked https://patchwork.kernel.org/patch/263861/, it's commit f4e8ab7cc4e819011ca6325e54383b3da7a5d130 in mainline. It sounds this commit has been there since 2.6.x.
Comment 10 Bruce Ashfield 2012-04-10 05:02:30 UTC
This one is resolved in the 3.0 and 3.2 kernels, so we can mark this as fixed.