kernel build fails

Ian Lepore ian at freebsd.org
Tue Feb 24 01:14:50 UTC 2015


On Mon, 2015-02-23 at 19:58 -0500, Brett Wynkoop wrote:
> On Mon, 23 Feb 2015 15:21:45 -0800
> Michael Mitchell <mmitchel at gmail.com> wrote:
> 
> > i thought any arch except armv6 is currently problematic. (this is my
> > experience with 10.x so far).
> 
> > > /usr/src/sys/arm/compile/BEAGLEBONE/../../../cddl/compat/opensolaris/sys/debug_compat.h
> > > /usr/src/sys/arm/compile/BEAGLEBONE/../../../arm/arm/genassym.c
> > > cc: error: argument unused during compilation:
> > > '-march=armv7a' [-Werror,-Wunused-command-line-argument] *** Error
> > > code 1
> > >
> > > Stop.
> > > make[3]: stopped in /usr/src/sys/modules/dtrace/dtrace
> > > *** Error code 1
> > >
> > > Stop.
> > > make[2]: stopped in /usr/src/sys/modules/dtrace
> > > *** Error code 1
> > >
> > > Stop.
> > > make[1]: stopped in /usr/src/sys/modules
> > > *** Error code 1
> > >
> > > Stop.
> > > make: stopped in /usr/src/sys/arm/compile/BEAGLEBONE
> > > root at beaglebone:/usr/src/sys/arm/compile/BEAGLEBONE #
> > >
> 
> So the question is where did the -march-arm7a come from?  It seems to
> me it should not be in the kernel tree and someone introduced it in
> error.
> 
> Am I the only person who tries native kernel builds?
> 
> -Brett
> 

Why do you think -march=armv7 shouldn't be present on a build for an
armv7 kernel?  In fact it's now required to build the kernel correctly.
I can build wandboard kernels on the wandboard.  I haven't tried a bbb
(because it's really too slow to build on).

The real question is "why does -mcpu=arm1176jzf-s appear on your compile
command line?"  I wonder if the answer is that you've set the wrong
CPUTYPE in make.conf or src.conf.

-- Ian




More information about the freebsd-arm mailing list