Building a kernel

[Michael Hill]

Since the last version of Fedlet came out, I've been building a kernel
package on my Baytrail tablet from the fc24 source rpm following these
instructions:

https://fedoraproject.org/wiki/Building_a_custom_kernel

This worked for a number of 4.2 kernels as well as a couple of 4.4 rc
kernels. On or about rc6 and continuing with 4.4.0-1, I've been
running this step:

rpmbuild -bb --with baseonly --without debuginfo --target=i686 kernel.spec

...and (after the patches are applied), failing like this:

Applying: i2c: designware: Convert to use unified device property API
Applying: drm/udl: Use unlocked gem unreferencing
+ chmod +x scripts/checkpatch.pl
+ touch .scmversion
+ mkdir configs
+ for i in '*.config'
+ mv kernel-4.4.0-aarch64.config .config
++ head -1 .config
++ cut -b 3-
+ Arch=arm64
+ grep -E '^CONFIG_'
+ make ARCH=arm64 listnewconfig
+ '[' -s .newoptions ']'
+ cat .newoptions
CONFIG_CM3218X
+ exit 1
error: Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)

RPM build errors:
    Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)

a. Is there something obvious I can fix?

b. Is there a better place to ask?

Thanks,

Mike

[Laura Abbott]

On 01/23/2016 04:57 PM, Michael Hill wrote:
> Since the last version of Fedlet came out, I've been building a kernel
> package on my Baytrail tablet from the fc24 source rpm following these
> instructions:
>
> https://fedoraproject.org/wiki/Building_a_custom_kernel
>
> This worked for a number of 4.2 kernels as well as a couple of 4.4 rc
> kernels. On or about rc6 and continuing with 4.4.0-1, I've been
> running this step:
>
> rpmbuild -bb --with baseonly --without debuginfo --target=i686 kernel.spec
>
> ...and (after the patches are applied), failing like this:
>
> Applying: i2c: designware: Convert to use unified device property API
> Applying: drm/udl: Use unlocked gem unreferencing
> chmod +x scripts/checkpatch.pl
> touch .scmversion
> mkdir configs
> for i in '*.config'
> mv kernel-4.4.0-aarch64.config .config
>
> ++ head -1 .config
> ++ cut -b 3-
> Arch=arm64
> grep -E '^CONFIG_'
> make ARCH=arm64 listnewconfig
> '[' -s .newoptions ']'
> cat .newoptions
>
> CONFIG_CM3218X
> exit 1
>
> error: Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)
>
> RPM build errors:
>      Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)
>
> a. Is there something obvious I can fix?
>
> b. Is there a better place to ask?

That (cryptic) error is indicating that CONFIG_CM3218X needs to be set in a config
file somewhere. Try adding CONFIG_CM3218X=<whatever> to config-local and
build again. You will need to figure out whether <whatever> should be y, n or m

Laura

[Michael Hill]

Thanks, Laura. CM3218X is under light sensors, and should be a module.
The line first gets removed when I run 'make oldconfig,' so I skipped
'make oldconfig.' I still get the same error and it seems to be
introduced in the line 'mv kernel-4.4.0-aarch64.config .config' under
'for i in *.config' in the spec file. I've built kernels before, but
in this case I want to wind up with a Fedora kernel package I can
install and remove cleanly, otherwise I wouldn't be trying to use
rpmbuild. I want an i686 kernel... does my command line look correct?

rpmbuild -bb --with baseonly --without debuginfo --target=i686 kernel.spec

Thanks again,

Mike

On Mon, Jan 25, 2016 at 11:39 AM, Laura Abbott  wrote:
> On 01/23/2016 04:57 PM, Michael Hill wrote:
> > Since the last version of Fedlet came out, I've been building a kernel
> > package on my Baytrail tablet from the fc24 source rpm following these
> > instructions:
> >
> > https://fedoraproject.org/wiki/Building_a_custom_kernel
> >
> > This worked for a number of 4.2 kernels as well as a couple of 4.4 rc
> > kernels. On or about rc6 and continuing with 4.4.0-1, I've been
> > running this step:
> >
> > rpmbuild -bb --with baseonly --without debuginfo --target=i686 kernel.spec
> >
> > ...and (after the patches are applied), failing like this:
> >
> > Applying: i2c: designware: Convert to use unified device property API
> > Applying: drm/udl: Use unlocked gem unreferencing
> > chmod +x scripts/checkpatch.pl
> > touch .scmversion
> > mkdir configs
> > for i in '*.config'
> > mv kernel-4.4.0-aarch64.config .config
> >
> > ++ head -1 .config
> > ++ cut -b 3-
> > Arch=arm64
> > grep -E '^CONFIG_'
> > make ARCH=arm64 listnewconfig
> > '[' -s .newoptions ']'
> > cat .newoptions
> >
> > CONFIG_CM3218X
> > exit 1
> >
> > error: Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)
> >
> > RPM build errors:
> >      Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)
> >
> > a. Is there something obvious I can fix?
> >
> > b. Is there a better place to ask?
>
> That (cryptic) error is indicating that CONFIG_CM3218X needs to be set in a
> config
> file somewhere. Try adding CONFIG_CM3218X=<whatever> to config-local and
> build again. You will need to figure out whether <whatever> should be y, n
> or m
>
> Laura

[Peter Robinson]

On Tue, Jan 26, 2016 at 12:04 AM, Michael Hill  wrote:
> Thanks, Laura. CM3218X is under light sensors, and should be a module.
> The line first gets removed when I run 'make oldconfig,' so I skipped
> 'make oldconfig.' I still get the same error and it seems to be
> introduced in the line 'mv kernel-4.4.0-aarch64.config .config' under
> 'for i in *.config' in the spec file. I've built kernels before, but
> in this case I want to wind up with a Fedora kernel package I can
> install and remove cleanly, otherwise I wouldn't be trying to use
> rpmbuild. I want an i686 kernel... does my command line look correct?
>
> rpmbuild -bb --with baseonly --without debuginfo --target=i686 kernel.spec

Add the following to config-generic:
# CONFIG_CM3218X is not set

Or:
CONFIG_CM3218X =m

It's not in the upstream kernel so it must be coming from a patch you apply.
> On Mon, Jan 25, 2016 at 11:39 AM, Laura Abbott  wrote:
> > On 01/23/2016 04:57 PM, Michael Hill wrote:
> > > Since the last version of Fedlet came out, I've been building a kernel
> > > package on my Baytrail tablet from the fc24 source rpm following these
> > > instructions:
> > >
> > > https://fedoraproject.org/wiki/Building_a_custom_kernel
> > >
> > > This worked for a number of 4.2 kernels as well as a couple of 4.4 rc
> > > kernels. On or about rc6 and continuing with 4.4.0-1, I've been
> > > running this step:
> > >
> > > rpmbuild -bb --with baseonly --without debuginfo --target=i686 kernel.spec
> > >
> > > ...and (after the patches are applied), failing like this:
> > >
> > > Applying: i2c: designware: Convert to use unified device property API
> > > Applying: drm/udl: Use unlocked gem unreferencing
> > > chmod +x scripts/checkpatch.pl
> > > touch .scmversion
> > > mkdir configs
> > > for i in '*.config'
> > > mv kernel-4.4.0-aarch64.config .config
> > >
> > > ++ head -1 .config
> > > ++ cut -b 3-
> > > Arch=arm64
> > > grep -E '^CONFIG_'
> > > make ARCH=arm64 listnewconfig
> > > '[' -s .newoptions ']'
> > > cat .newoptions
> > >
> > > CONFIG_CM3218X
> > > exit 1
> > >
> > > error: Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)
> > >
> > > RPM build errors:
> > >      Bad exit status from /var/tmp/rpm-tmp.niuQ3f (%prep)
> > >
> > > a. Is there something obvious I can fix?
> > >
> > > b. Is there a better place to ask?
> >
> > That (cryptic) error is indicating that CONFIG_CM3218X needs to be set in a
> > config
> > file somewhere. Try adding CONFIG_CM3218X=<whatever> to config-local and
> > build again. You will need to figure out whether <whatever> should be y, n
> > or m
> >
> > Laura
>
> kernel mailing list
> [e-mail]
> http://lists.fedoraproject.org/admin/lists/[e-mail]

[Michael Hill]

On Mon, Jan 25, 2016 at 8:17 PM, Peter Robinson  wrote:
> It's not in the upstream kernel so it must be coming from a patch you apply.

Thanks Peter, thanks Laura. I located and eliminated the offending
patch and the build is proceeding. The patchset is from the Ubuntu
crew so I'll ask them if it started failing for them.

Mike

[Josh Boyer]

On Mon, Jan 25, 2016 at 11:21:57PM -0500, Michael Hill wrote:
> On Mon, Jan 25, 2016 at 8:17 PM, Peter Robinson  wrote:
> > It's not in the upstream kernel so it must be coming from a patch you apply.
>
> Thanks Peter, thanks Laura. I located and eliminated the offending
> patch and the build is proceeding. The patchset is from the Ubuntu
> crew so I'll ask them if it started failing for them.

It is usually a good idea to explain fully what you're trying to do when
you ask for help.  I'm curious what patches you're applying to 4.4 and
why.

josh

[Michael Hill]

Hi Josh,

On Tue, Jan 26, 2016 at 7:47 AM, Josh Boyer  wrote:
> It is usually a good idea to explain fully what you're trying to do when
> you ask for help.  I'm curious what patches you're applying to 4.4 and
> why.

I installed Adam Williamson's Fedlet on my Baytrail tablet (a T100TA).
I run Rawhide elsewhere so I thought it would be nice to build a newer
kernel. The wiki instructions were pretty straightforward and I didn't
have any build issues until recently (although one of the early 4.4 rc
kernels I built and ran decided to eat the filesystem). There is an
active G+ Community that's Ubuntu-centric
(https://plus.google.com/communities/117853703024346186936) where I've
been following the patches (and config file) they use. For 4.4 they've
narrowed it down to 10 of the 11 in the 4.4 patches folder here:

https://drive.google.com/folderview?id=0B9C1WK1FQhjfcXNrbzN6djQzajg&usp=...

(For the 4.2 kernels, there were 22 patches.) The one that I can't get
to stay in .config lately is the Capella cm32181 device driver, but
I've been short-circuiting the instructions in the "Copy the Source
Tree and Generate a Patch" section and just applying them individually
so it's getting overwritten.

Mike

[Justin Forbes]

On Tue, Jan 26, 2016 at 7:27 AM, Michael Hill  wrote:
> (For the 4.2 kernels, there were 22 patches.) The one that I can't get
> to stay in .config lately is the Capella cm32181 device driver, but
> I've been short-circuiting the instructions in the "Copy the Source
> Tree and Generate a Patch" section and just applying them individually
> so it's getting overwritten.
>
> Mike

cm32181 is upstream, and has been for a while, the patch you are
adding is for a different Capella light sensor.

Justin

[Michael Hill]

On Tue, Jan 26, 2016 at 11:43 AM, Justin Forbes  wrote:
> cm32181 is upstream, and has been for a while, the patch you are
> adding is for a different Capella light sensor.

Good to know, thanks.

Mike