Patchwork [0/4] linux-yocto: consolidated update

login
register
mail settings
Submitter Bruce Ashfield
Date Aug. 15, 2012, 8:06 p.m.
Message ID <cover.1345059176.git.bruce.ashfield@windriver.com>
Download mbox
Permalink /patch/34663/
State New
Headers show

Pull-request

git://git.pokylinux.org/poky-contrib zedd/kernel-dev

Comments

Bruce Ashfield - Aug. 15, 2012, 8:06 p.m.
Richard/Saul,

Here's a consolidated update that I've had under test in one
way or another for 3 weeks now. Tom Z and others have also been
using parts of this for their work, so it has seen a reasonably
wide set of testers.

This series further streamlines working with linux-yocto* based
recipes/repositories by breaking the last branch based triggers
in the tooling. Branches were the most important piece of 
information when working with the tools and a linux-yocto repository,
but this has now been generalized, and BSP descriptions (among
other things) are now found by MACHINE and kernel type only. 

With this, KBRANCH now becomes what it was supposed to be, an 
human readible indication of the branch that is supposed to be
built, and one that the kern-tools will validate. If that branch
is not validated, an error is thrown. If the default branch is
left as-is, then the in-tree description leaves you on the 
proper branch and all is well.

I also have a collection of bug fixes to the tools, including
support for having a meta branch not called 'meta'.

I'd recommend that this go through some AB tests, and I'll be
around if there's a corner case that I managed to miss.

Cheers,

Bruce

The following changes since commit ee11e7ac55fa41cb69e9519f3e879a1746c4cf33:

  linux-yocto/meta-yocto/3.4: BSP, uprobes, config and cleanups (2012-08-15 13:52:00 -0400)

are available in the git repository at:
  git://git.pokylinux.org/poky-contrib zedd/kernel-dev
  http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=zedd/kernel-dev

Bruce Ashfield (4):
  linux-yocto: make KBRANCH the exception and not the rule
  linux-yocto/3.4: remove explicit KBRANCH designations
  linux-yocto: explicitly export KMETA to scripts
  kern-tools: fixes (branching,buildall) + cleanups (checkpoint,cleaner)

 meta/classes/kernel-yocto.bbclass                  |   93 ++++++++++++++------
 .../kern-tools/kern-tools-native_git.bb            |    2 +-
 meta/recipes-kernel/linux/linux-yocto_3.4.bb       |   11 +--
 3 files changed, 69 insertions(+), 37 deletions(-)
Richard Purdie - Aug. 18, 2012, 4:57 p.m.
On Wed, 2012-08-15 at 16:06 -0400, Bruce Ashfield wrote:
> Here's a consolidated update that I've had under test in one
> way or another for 3 weeks now. Tom Z and others have also been
> using parts of this for their work, so it has seen a reasonably
> wide set of testers.
> 
> This series further streamlines working with linux-yocto* based
> recipes/repositories by breaking the last branch based triggers
> in the tooling. Branches were the most important piece of 
> information when working with the tools and a linux-yocto repository,
> but this has now been generalized, and BSP descriptions (among
> other things) are now found by MACHINE and kernel type only. 
> 
> With this, KBRANCH now becomes what it was supposed to be, an 
> human readible indication of the branch that is supposed to be
> built, and one that the kern-tools will validate. If that branch
> is not validated, an error is thrown. If the default branch is
> left as-is, then the in-tree description leaves you on the 
> proper branch and all is well.
> 
> I also have a collection of bug fixes to the tools, including
> support for having a meta branch not called 'meta'.
> 
> I'd recommend that this go through some AB tests, and I'll be
> around if there's a corner case that I managed to miss.

I think 

http://autobuilder.yoctoproject.org:8010/builders/nightly-tiny/builds/199/steps/shell_34/logs/stdio

is due to something in this series...

Cheers,

Richard
Bruce Ashfield - Aug. 18, 2012, 5:15 p.m.
On Sat, Aug 18, 2012 at 12:57 PM, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> On Wed, 2012-08-15 at 16:06 -0400, Bruce Ashfield wrote:
>> Here's a consolidated update that I've had under test in one
>> way or another for 3 weeks now. Tom Z and others have also been
>> using parts of this for their work, so it has seen a reasonably
>> wide set of testers.
>>
>> This series further streamlines working with linux-yocto* based
>> recipes/repositories by breaking the last branch based triggers
>> in the tooling. Branches were the most important piece of
>> information when working with the tools and a linux-yocto repository,
>> but this has now been generalized, and BSP descriptions (among
>> other things) are now found by MACHINE and kernel type only.
>>
>> With this, KBRANCH now becomes what it was supposed to be, an
>> human readible indication of the branch that is supposed to be
>> built, and one that the kern-tools will validate. If that branch
>> is not validated, an error is thrown. If the default branch is
>> left as-is, then the in-tree description leaves you on the
>> proper branch and all is well.
>>
>> I also have a collection of bug fixes to the tools, including
>> support for having a meta branch not called 'meta'.
>>
>> I'd recommend that this go through some AB tests, and I'll be
>> around if there's a corner case that I managed to miss.
>
> I think
>
> http://autobuilder.yoctoproject.org:8010/builders/nightly-tiny/builds/199/steps/shell_34/logs/stdio

Judging by the timestamps on your two email, is this not addressed by
the workaround
for git's behaviour ? On the surface, it looks to be the same thing,
since a branch that
should exist in the repository isn't there.

I'll fire of a poky-tiny build for qemu here, to see if I can
reproduce this one.

Cheers,

Bruce

>
> is due to something in this series...
>
> Cheers,
>
> Richard
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
Richard Purdie - Aug. 18, 2012, 10:18 p.m.
On Sat, 2012-08-18 at 13:15 -0400, Bruce Ashfield wrote:
> On Sat, Aug 18, 2012 at 12:57 PM, Richard Purdie
> <richard.purdie@linuxfoundation.org> wrote:
> > On Wed, 2012-08-15 at 16:06 -0400, Bruce Ashfield wrote:
> >> Here's a consolidated update that I've had under test in one
> >> way or another for 3 weeks now. Tom Z and others have also been
> >> using parts of this for their work, so it has seen a reasonably
> >> wide set of testers.
> >>
> >> This series further streamlines working with linux-yocto* based
> >> recipes/repositories by breaking the last branch based triggers
> >> in the tooling. Branches were the most important piece of
> >> information when working with the tools and a linux-yocto repository,
> >> but this has now been generalized, and BSP descriptions (among
> >> other things) are now found by MACHINE and kernel type only.
> >>
> >> With this, KBRANCH now becomes what it was supposed to be, an
> >> human readible indication of the branch that is supposed to be
> >> built, and one that the kern-tools will validate. If that branch
> >> is not validated, an error is thrown. If the default branch is
> >> left as-is, then the in-tree description leaves you on the
> >> proper branch and all is well.
> >>
> >> I also have a collection of bug fixes to the tools, including
> >> support for having a meta branch not called 'meta'.
> >>
> >> I'd recommend that this go through some AB tests, and I'll be
> >> around if there's a corner case that I managed to miss.
> >
> > I think
> >
> > http://autobuilder.yoctoproject.org:8010/builders/nightly-tiny/builds/199/steps/shell_34/logs/stdio
> 
> Judging by the timestamps on your two email, is this not addressed by
> the workaround
> for git's behaviour ?

Correct, it fails with the fix applied so I think its something
different.

>  On the surface, it looks to be the same thing,
> since a branch that
> should exist in the repository isn't there.
> 
> I'll fire of a poky-tiny build for qemu here, to see if I can
> reproduce this one.

We also have an atom-pc failure which may or may not be related:

http://autobuilder.yoctoproject.org:8010/builders/nightly-x86/builds/648/steps/shell_50/logs/stdio

Cheers,

Richard
Bruce Ashfield - Aug. 19, 2012, 4:44 a.m.
On Sat, Aug 18, 2012 at 6:18 PM, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> On Sat, 2012-08-18 at 13:15 -0400, Bruce Ashfield wrote:
>> On Sat, Aug 18, 2012 at 12:57 PM, Richard Purdie
>> <richard.purdie@linuxfoundation.org> wrote:
>> > On Wed, 2012-08-15 at 16:06 -0400, Bruce Ashfield wrote:
>> >> Here's a consolidated update that I've had under test in one
>> >> way or another for 3 weeks now. Tom Z and others have also been
>> >> using parts of this for their work, so it has seen a reasonably
>> >> wide set of testers.
>> >>
>> >> This series further streamlines working with linux-yocto* based
>> >> recipes/repositories by breaking the last branch based triggers
>> >> in the tooling. Branches were the most important piece of
>> >> information when working with the tools and a linux-yocto repository,
>> >> but this has now been generalized, and BSP descriptions (among
>> >> other things) are now found by MACHINE and kernel type only.
>> >>
>> >> With this, KBRANCH now becomes what it was supposed to be, an
>> >> human readible indication of the branch that is supposed to be
>> >> built, and one that the kern-tools will validate. If that branch
>> >> is not validated, an error is thrown. If the default branch is
>> >> left as-is, then the in-tree description leaves you on the
>> >> proper branch and all is well.
>> >>
>> >> I also have a collection of bug fixes to the tools, including
>> >> support for having a meta branch not called 'meta'.
>> >>
>> >> I'd recommend that this go through some AB tests, and I'll be
>> >> around if there's a corner case that I managed to miss.
>> >
>> > I think
>> >
>> > http://autobuilder.yoctoproject.org:8010/builders/nightly-tiny/builds/199/steps/shell_34/logs/stdio
>>
>> Judging by the timestamps on your two email, is this not addressed by
>> the workaround
>> for git's behaviour ?
>
> Correct, it fails with the fix applied so I think its something
> different.

poky-tiny and linux-yocto-tiny .. argh .. they are a bit out of date, and out
of site. Darren and I will have to drag them forward, since they are rotting
a bit.

Anyway, I got to trade 4 or 5 hours against the time that the fetcher stole
from you today. :(

I fixed the poky-tiny build here, and the patch is simple enough in the end:

http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=zedd/kernel-tiny

    linux-yocto-tiny: set default branch

    To streamline the creation of build time branches (branches that are
    not always present in the upstream kernel repository), linux-yocto-tiny
    should specify a default kernel branch. By setting the default branch
    (KBRANCH_DEFAULT) and also setting the build branch (KBRANCH) to that
    default, the tools will allow the board description to be processed
    and no branching forced.

    Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>


>
>>  On the surface, it looks to be the same thing,
>> since a branch that
>> should exist in the repository isn't there.
>>
>> I'll fire of a poky-tiny build for qemu here, to see if I can
>> reproduce this one.
>
> We also have an atom-pc failure which may or may not be related:
>
> http://autobuilder.yoctoproject.org:8010/builders/nightly-x86/builds/648/steps/shell_50/logs/stdio

Not related to the other fix, but I have a fix for the atom-pc build
against 3.0 here:

http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=zedd/kernel-3.0-fix

commit 569438febfdd6bffceb919e1050b020c7fcb8f89
Author: Bruce Ashfield <bruce.ashfield@windriver.com>
Date:   Sun Aug 19 00:28:41 2012 -0400

    linux-yocto/3.0: update meta SRCREV

    Bumping the meta SRCREV to pickup this fix:

        meta: rename virto.scc to virtio.scc

        The virtio configuration block is misnamed. BSPs that include it with
        the proper name, now throw an error (as they should). So fixing the
        name of the fragment fixes the build.

    Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

The tools did their job and flagged an invalid include in that BSP.
Before, those
options were tossed in the bin, now we error.

I've fixed the BSP configuration, and checked the other kernels, it builds here
now.

It's late, so I tossed these fixes out like this, if you want me to resend them
again tomorrow, I can, but I won't be around until the end of the day.

Cheers,

Bruce

>
> Cheers,
>
> Richard
>
>
Richard Purdie - Aug. 19, 2012, 9:47 a.m.
On Sun, 2012-08-19 at 00:44 -0400, Bruce Ashfield wrote:
> On Sat, Aug 18, 2012 at 6:18 PM, Richard Purdie
> <richard.purdie@linuxfoundation.org> wrote:
> > On Sat, 2012-08-18 at 13:15 -0400, Bruce Ashfield wrote:
> >> On Sat, Aug 18, 2012 at 12:57 PM, Richard Purdie
> >> <richard.purdie@linuxfoundation.org> wrote:
> >> > On Wed, 2012-08-15 at 16:06 -0400, Bruce Ashfield wrote:
> >> >> Here's a consolidated update that I've had under test in one
> >> >> way or another for 3 weeks now. Tom Z and others have also been
> >> >> using parts of this for their work, so it has seen a reasonably
> >> >> wide set of testers.
> >> >>
> >> >> This series further streamlines working with linux-yocto* based
> >> >> recipes/repositories by breaking the last branch based triggers
> >> >> in the tooling. Branches were the most important piece of
> >> >> information when working with the tools and a linux-yocto repository,
> >> >> but this has now been generalized, and BSP descriptions (among
> >> >> other things) are now found by MACHINE and kernel type only.
> >> >>
> >> >> With this, KBRANCH now becomes what it was supposed to be, an
> >> >> human readible indication of the branch that is supposed to be
> >> >> built, and one that the kern-tools will validate. If that branch
> >> >> is not validated, an error is thrown. If the default branch is
> >> >> left as-is, then the in-tree description leaves you on the
> >> >> proper branch and all is well.
> >> >>
> >> >> I also have a collection of bug fixes to the tools, including
> >> >> support for having a meta branch not called 'meta'.
> >> >>
> >> >> I'd recommend that this go through some AB tests, and I'll be
> >> >> around if there's a corner case that I managed to miss.
> >> >
> >> > I think
> >> >
> >> > http://autobuilder.yoctoproject.org:8010/builders/nightly-tiny/builds/199/steps/shell_34/logs/stdio
> >>
> >> Judging by the timestamps on your two email, is this not addressed by
> >> the workaround
> >> for git's behaviour ?
> >
> > Correct, it fails with the fix applied so I think its something
> > different.
> 
> poky-tiny and linux-yocto-tiny .. argh .. they are a bit out of date, and out
> of site. Darren and I will have to drag them forward, since they are rotting
> a bit.
> 
> Anyway, I got to trade 4 or 5 hours against the time that the fetcher stole
> from you today. :(
> 
> I fixed the poky-tiny build here, and the patch is simple enough in the end:
> 
> http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=zedd/kernel-tiny
> 
>     linux-yocto-tiny: set default branch
> 
>     To streamline the creation of build time branches (branches that are
>     not always present in the upstream kernel repository), linux-yocto-tiny
>     should specify a default kernel branch. By setting the default branch
>     (KBRANCH_DEFAULT) and also setting the build branch (KBRANCH) to that
>     default, the tools will allow the board description to be processed
>     and no branching forced.
> 
>     Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
> 
> 
> >
> >>  On the surface, it looks to be the same thing,
> >> since a branch that
> >> should exist in the repository isn't there.
> >>
> >> I'll fire of a poky-tiny build for qemu here, to see if I can
> >> reproduce this one.
> >
> > We also have an atom-pc failure which may or may not be related:
> >
> > http://autobuilder.yoctoproject.org:8010/builders/nightly-x86/builds/648/steps/shell_50/logs/stdio
> 
> Not related to the other fix, but I have a fix for the atom-pc build
> against 3.0 here:
> 
> http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=zedd/kernel-3.0-fix
> 
> commit 569438febfdd6bffceb919e1050b020c7fcb8f89
> Author: Bruce Ashfield <bruce.ashfield@windriver.com>
> Date:   Sun Aug 19 00:28:41 2012 -0400
> 
>     linux-yocto/3.0: update meta SRCREV
> 
>     Bumping the meta SRCREV to pickup this fix:
> 
>         meta: rename virto.scc to virtio.scc
> 
>         The virtio configuration block is misnamed. BSPs that include it with
>         the proper name, now throw an error (as they should). So fixing the
>         name of the fragment fixes the build.
> 
>     Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
> 
> The tools did their job and flagged an invalid include in that BSP.
> Before, those
> options were tossed in the bin, now we error.
> 
> I've fixed the BSP configuration, and checked the other kernels, it builds here
> now.
> 
> It's late, so I tossed these fixes out like this, if you want me to resend them
> again tomorrow, I can, but I won't be around until the end of the day.

Thanks for the fixes. I've pulled various things into master now and
will queue up another build to test everything.

Cheers,

Richard