Building a custom kernel via cross compiling

[Andrew Wing]

Hi all,

I am a newbie with kernel matters so I apologize in advance for what is a
newbie type setup question. My feeling is that if I send this query to the
more general Arm Fedora forum, the mere mention of kernel will provoke
glazed eyes :-)

The background is that I have an OEM board based on a TI AM3352 OMAP family
SoC looking somewhat like a Beaglebone black. Recently I've been exploring
getting this running under Fedora. In order to get  it working I've cobbled
together the F22 minimal image along with a device tree blob created by a
colleague for this board, and a customised MLO/u-boot.

My next task is to look through and understand the arm kernel sources,
customize various aspects of the kernel config for my board and re-build.

It seems that instructions are here:

https://fedoraproject.org/wiki/Building_a_custom_kernel

However, I want to cross compile on a host system. To ease this process I
have set up a  Fedora x86 box, also running f22, which is going to be the
host build machine.

I decided I should build from source rpm and was following the wiki
instructions(rpmdev-setuptree etc.) up to the point here...

*cd ~/rpmbuild/SPECS*
*rpmbuild -bp --target=$(uname -m) kernel.spec*

At this point I'd like to actually have a target of arm7l for my target but
entering this gets me an unknown target error (possibly not surprising
given the build machine/target architecture difference!).

My build machine does have a suitable linaro Arm compiler in place. It
occurs to me that I need a different kernel spec and to indicate to my
build system to use my compiler. Hopefully this can be achieved via some
slightly modified step from that given in the wiki page? Or should I be
tackling this in a completely different way?

Thanks in advance for any clarification.

regards,

Andrew