[GIT PULL] GIC DT binding support

[GIT PULL] GIC DT binding support

[Rob Herring]

Arnd,

I'm not sure if you want a rebased branch to fix merge conflicts or not.
If that's okay, please pull. I've also fixed the msm compile failure.

The dependencies are v3.1-rc9 and rmk's devel-stable.

The following changes since commit 3a8254364277fabe01bc0e12b9691722939f5ef3:

  Merge remote-tracking branch 'rmk/devel-stable' into HEAD (2011-10-24
14:02:37 -0500)

are available in the git repository at:

  git://sources.calxeda.com/kernel/linux.git gic-dt

Rob Herring (5):
      of/irq: introduce of_irq_init
      irq: support domains with non-zero hwirq base
      ARM: gic: add irq_domain support
      ARM: gic: add OF based initialization
      ARM: gic: fix irq_alloc_descs handling for sparse irq

 Documentation/devicetree/bindings/arm/gic.txt |   55 +++++++
 arch/arm/common/Kconfig                       |    1 +
 arch/arm/common/gic.c                         |  187
++++++++++++++++---------
 arch/arm/include/asm/hardware/gic.h           |   10 +-
 drivers/of/irq.c                              |  107 ++++++++++++++
 include/linux/irqdomain.h                     |   16 ++-
 include/linux/of_irq.h                        |    3 +
 kernel/irq/irqdomain.c                        |   12 +-
 8 files changed, 319 insertions(+), 72 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/arm/gic.txt

Rob

[GIT PULL] GIC DT binding support

[Rob Herring]

Arnd,

On 10/24/2011 04:38 PM, Rob Herring wrote:
> Arnd,
> 
> I'm not sure if you want a rebased branch to fix merge conflicts or not.
> If that's okay, please pull. I've also fixed the msm compile failure.
> 
> The dependencies are v3.1-rc9 and rmk's devel-stable.
> 

Here's an updated pull request to fix missing export.h include. Now it's
also dependent on module.h clean-up branch:

git://openlinux.windriver.com/people/paulg/linux-next for-sfr


The following changes since commit fa1d6d372245a160eecf5f57d64d2b21ea248e08:

  Merge branch 'for-sfr' of
git://openlinux.windriver.com/people/paulg/linux-next into HEAD
(2011-10-28 15:51:07 -0500)

are available in the git repository at:

  git://sources.calxeda.com/kernel/linux.git gic-dt

Rob Herring (5):
      of/irq: introduce of_irq_init
      irq: support domains with non-zero hwirq base
      ARM: gic: add irq_domain support
      ARM: gic: add OF based initialization
      ARM: gic: fix irq_alloc_descs handling for sparse irq

 Documentation/devicetree/bindings/arm/gic.txt |   55 +++++++
 arch/arm/common/Kconfig                       |    1 +
 arch/arm/common/gic.c                         |  188
++++++++++++++++--------
 arch/arm/include/asm/hardware/gic.h           |   10 +-
 drivers/of/irq.c                              |  107 ++++++++++++++
 include/linux/irqdomain.h                     |   16 ++-
 include/linux/of_irq.h                        |    3 +
 kernel/irq/irqdomain.c                        |   12 +-
 8 files changed, 320 insertions(+), 72 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/arm/gic.txt

Rob

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Friday 28 October 2011, Rob Herring wrote:
> Here's an updated pull request to fix missing export.h include. Now it's
> also dependent on module.h clean-up branch:
> 
> git://openlinux.windriver.com/people/paulg/linux-next for-sfr
> 
> 
> The following changes since commit fa1d6d372245a160eecf5f57d64d2b21ea248e08:
> 
>   Merge branch 'for-sfr' of
> git://openlinux.windriver.com/people/paulg/linux-next into HEAD
> (2011-10-28 15:51:07 -0500)
> 
> are available in the git repository at:
> 
>   git://sources.calxeda.com/kernel/linux.git gic-dt

I've pulled it and am now trying to clean up the mess resulting from
having the previous version in a number of branches already.

Can you check that the highbank/soc branch is ok this time?

	Arnd

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Friday 28 October 2011, Rob Herring wrote:
>   Merge branch 'for-sfr' of
> git://openlinux.windriver.com/people/paulg/linux-next into HEAD
> (2011-10-28 15:51:07 -0500)
> 

Rob, I think there is no chance that this is working out.

I've just spent a few hours trying to resolve all the dependencies from this,
but I don't think it will work out, unless Linus pulls that tree very soon,
which I think is very unlikely.

Have you made sure that Paul even considers this branch stable?

Normally this kind of change gets pushed *last* in the merge window, to make
it possible to fix up all the stuff that breaks. This means I cannot
wait for the branch to get merged upstream before send a pull request for
my next/dt and next/soc branches.

Do you even depend on that branch? It seems to me that you merely have a
conflict, not a dependency, so why do you even try fixing it up?

	Arnd

[GIT PULL] GIC DT binding support

[Rob Herring]

Arnd,

On 10/30/2011 07:31 PM, Arnd Bergmann wrote:
> On Friday 28 October 2011, Rob Herring wrote:
>>   Merge branch 'for-sfr' of
>> git://openlinux.windriver.com/people/paulg/linux-next into HEAD
>> (2011-10-28 15:51:07 -0500)
>>
> 
> Rob, I think there is no chance that this is working out.
> 
> I've just spent a few hours trying to resolve all the dependencies from this,
> but I don't think it will work out, unless Linus pulls that tree very soon,
> which I think is very unlikely.
> 
> Have you made sure that Paul even considers this branch stable?

No.

> Normally this kind of change gets pushed *last* in the merge window, to make
> it possible to fix up all the stuff that breaks. This means I cannot
> wait for the branch to get merged upstream before send a pull request for
> my next/dt and next/soc branches.
> 
> Do you even depend on that branch? It seems to me that you merely have a
> conflict, not a dependency, so why do you even try fixing it up?

Sorry about this. I thought I needed to fix the linux-next build
breakage. I only depend on it when export.h gets introduced, so I can
drop that out. All the other dependencies are in Linus' master now. I
pushed out a branch without Paul's branch:

I get to Connect around 2PM tomorrow, we can discuss further then if
necessary.

The following changes since commit ce949717b559709423c1ef716a9db16d1dcadaed:

  Merge git://github.com/rustyrussell/linux (2011-10-29 07:52:16 -0700)

are available in the git repository at:

  git://sources.calxeda.com/kernel/linux.git gic-dt-v2

Rob Herring (5):
      of/irq: introduce of_irq_init
      irq: support domains with non-zero hwirq base
      ARM: gic: add irq_domain support
      ARM: gic: add OF based initialization
      ARM: gic: fix irq_alloc_descs handling for sparse irq

 Documentation/devicetree/bindings/arm/gic.txt |   55 +++++++
 arch/arm/common/Kconfig                       |    1 +
 arch/arm/common/gic.c                         |  187
++++++++++++++++---------
 arch/arm/include/asm/hardware/gic.h           |   10 +-
 drivers/of/irq.c                              |  107 ++++++++++++++
 include/linux/irqdomain.h                     |   16 ++-
 include/linux/of_irq.h                        |    3 +
 kernel/irq/irqdomain.c                        |   12 +-
 8 files changed, 319 insertions(+), 72 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/arm/gic.txt

Rob

[GIT PULL] GIC DT binding support

[Rob Herring]

Arnd,

Please pull GIC device tree support. This is the first of 2 pull
requests. You can ignore this one if there are no other dependencies on
GIC DT support.

This is based on rmk's for-next branch and v3.1-rc9. rc9 was needed for
this dependency:

commit eef24afb28561a5a9f4be8f8da97735b7e6a826f
irq: Fix check for already initialized irq_domain in irq_domain_add

Regards,
Rob

The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:

  Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)

are available in the git repository at:

  http://sources.calxeda.com/git/kernel/linux.git gic-dt

Rob Herring (4):
      of/irq: introduce of_irq_init
      irq: support domains with non-zero hwirq base
      ARM: gic: add irq_domain support
      ARM: gic: add OF based initialization

 Documentation/devicetree/bindings/arm/gic.txt |   55 ++++++++++
 arch/arm/common/Kconfig                       |    1 +
 arch/arm/common/gic.c                         |  142
+++++++++++++++++++------
 arch/arm/include/asm/hardware/gic.h           |    5 +-
 drivers/of/irq.c                              |  107 +++++++++++++++++++
 include/linux/irqdomain.h                     |   16 +++-
 include/linux/of_irq.h                        |    3 +
 kernel/irq/irqdomain.c                        |   12 +-
 8 files changed, 299 insertions(+), 42 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/arm/gic.txt

[GIT PULL] GIC DT binding support

[Shawn Guo]

On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
> Arnd,
> 
> Please pull GIC device tree support. This is the first of 2 pull
> requests. You can ignore this one if there are no other dependencies on
> GIC DT support.
> 
My imx6q series depends on it.

-- 
Regards,
Shawn

[GIT PULL] GIC DT binding support

[Domenico Andreoli]

On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
> 
> The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:
> 
>   Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)
> 
> are available in the git repository at:
> 
>   http://sources.calxeda.com/git/kernel/linux.git gic-dt

I got this when I tried to pick you patch.

$ git fetch calxeda
fatal: http://sources.calxeda.com/git/kernel/linux.git/info/refs not found: did you run git update-server-info on the server?
$

cheers,
Domenico

[GIT PULL] GIC DT binding support

[Rob Herring]

On 10/18/2011 04:53 AM, Domenico Andreoli wrote:
> On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
>>
>> The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:
>>
>>   Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)
>>
>> are available in the git repository at:
>>
>>   http://sources.calxeda.com/git/kernel/linux.git gic-dt
> 
> I got this when I tried to pick you patch.
> 
> $ git fetch calxeda
> fatal: http://sources.calxeda.com/git/kernel/linux.git/info/refs not found: did you run git update-server-info on the server?
> $

Strange. I had tested it out and it worked at the time.

The git protocol is setup now, so use it:

git://sources.calxeda.com/kernel/linux.git

Rob

[GIT PULL] GIC DT binding support

[Domenico Andreoli]

On Tue, Oct 18, 2011 at 09:50:22AM -0500, Rob Herring wrote:
> On 10/18/2011 04:53 AM, Domenico Andreoli wrote:
> > On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
> >>
> >> The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:
> >>
> >>   Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)
> >>
> >> are available in the git repository at:
> >>
> >>   http://sources.calxeda.com/git/kernel/linux.git gic-dt
> > 
> > I got this when I tried to pick you patch.
> > 
> > $ git fetch calxeda
> > fatal: http://sources.calxeda.com/git/kernel/linux.git/info/refs not found: did you run git update-server-info on the server?
> > $
> 
> Strange. I had tested it out and it worked at the time.
> 
> The git protocol is setup now, so use it:
> 
> git://sources.calxeda.com/kernel/linux.git

yes, this works. what about gitweb? ;)

thanks,
Domenico

[GIT PULL] GIC DT binding support

[Rob Herring]

On 10/18/2011 09:58 AM, Domenico Andreoli wrote:
> On Tue, Oct 18, 2011 at 09:50:22AM -0500, Rob Herring wrote:
>> On 10/18/2011 04:53 AM, Domenico Andreoli wrote:
>>> On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
>>>>
>>>> The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:
>>>>
>>>>   Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)
>>>>
>>>> are available in the git repository at:
>>>>
>>>>   http://sources.calxeda.com/git/kernel/linux.git gic-dt
>>>
>>> I got this when I tried to pick you patch.
>>>
>>> $ git fetch calxeda
>>> fatal: http://sources.calxeda.com/git/kernel/linux.git/info/refs not found: did you run git update-server-info on the server?
>>> $
>>
>> Strange. I had tested it out and it worked at the time.
>>
>> The git protocol is setup now, so use it:
>>
>> git://sources.calxeda.com/kernel/linux.git
> 
> yes, this works. what about gitweb? ;)
> 

sources.calxeda.com/gitweb

Rob

[GIT PULL] GIC DT binding support

[Shawn Guo]

On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
> Arnd,
> 
> Please pull GIC device tree support. This is the first of 2 pull
> requests. You can ignore this one if there are no other dependencies on
> GIC DT support.
> 
> This is based on rmk's for-next branch and v3.1-rc9. rc9 was needed for
> this dependency:
> 
> commit eef24afb28561a5a9f4be8f8da97735b7e6a826f
> irq: Fix check for already initialized irq_domain in irq_domain_add
> 
> Regards,
> Rob
> 
> The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:
> 
>   Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)
> 
> are available in the git repository at:
> 
>   http://sources.calxeda.com/git/kernel/linux.git gic-dt
> 
> Rob Herring (4):
>       of/irq: introduce of_irq_init
>       irq: support domains with non-zero hwirq base
>       ARM: gic: add irq_domain support
>       ARM: gic: add OF based initialization
> 
Hi Rob,

Russell is going to take Marc Zyngier's ppi series for v3.2 merge
window.  I just noticed your patch 'ARM: gic: add irq_domain support'
gets some conflict with Marc's series on function gic_dist_init(),
which is not so obvious to resolve.  You may want to take a look. 

Regards,
Shawn

[GIT PULL] GIC DT binding support

[Russell King - ARM Linux]

On Thu, Oct 20, 2011 at 08:48:48PM +0800, Shawn Guo wrote:
> On Thu, Oct 13, 2011 at 10:01:21PM -0500, Rob Herring wrote:
> > Rob Herring (4):
> >       of/irq: introduce of_irq_init
> >       irq: support domains with non-zero hwirq base
> >       ARM: gic: add irq_domain support
> >       ARM: gic: add OF based initialization
> > 
> Hi Rob,
> 
> Russell is going to take Marc Zyngier's ppi series for v3.2 merge
> window.  I just noticed your patch 'ARM: gic: add irq_domain support'
> gets some conflict with Marc's series on function gic_dist_init(),
> which is not so obvious to resolve.  You may want to take a look. 

Don't be so sure - Marc has rebased commits which exist in Thomas' tree,
so I can't pull it as-is.  That needs to be resolving - but unfortunately
Thomas is not responding (I suspect he's busy elsewhere also because of
the proximity of the kernel summit.)

Given that this is my _last_ full working day before the kernel summit,
time is running very short to resolve these problems.  So, I think that
all this GIC stuff may have to wait until after the KS, and maybe for
the following merge window too.

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Friday 14 October 2011, Rob Herring wrote:
> Please pull GIC device tree support. This is the first of 2 pull
> requests. You can ignore this one if there are no other dependencies on
> GIC DT support.
> 
> This is based on rmk's for-next branch and v3.1-rc9. rc9 was needed for
> this dependency:

I don't see the for-next branch in the history. Do you mean it requires
the for-next branch as well in order to actually build?

Can you be more specific so I can watch for the dependencies to
get upstream first?

> The following changes since commit dd5e9eb0d051606ee3eda0c30222429c74e23f00:
> 
>   Merge commit 'v3.1-rc9' into upstream/base (2011-10-13 20:30:42 -0500)
> 
> are available in the git repository at:
> 
>   http://sources.calxeda.com/git/kernel/linux.git gic-dt

Pulled into my next/dt branch, thanks a lot!

	Arnd

[GIT PULL] GIC DT binding support

[Rob Herring]

Arnd,

On 10/20/2011 08:07 AM, Arnd Bergmann wrote:
> On Friday 14 October 2011, Rob Herring wrote:
>> Please pull GIC device tree support. This is the first of 2 pull
>> requests. You can ignore this one if there are no other dependencies on
>> GIC DT support.
>>
>> This is based on rmk's for-next branch and v3.1-rc9. rc9 was needed for
>> this dependency:
> 
> I don't see the for-next branch in the history. Do you mean it requires
> the for-next branch as well in order to actually build?

Probably because it is buried by rc9 commits:

git log v3.1-rc9..gic-dt

> 
> Can you be more specific so I can watch for the dependencies to
> get upstream first?
> 

I believe it conflicts with this commit in rmk/for-next:

commit b166bc3be08b744d2f4b14921a1efee14906b383
Author: Will Deacon <will.deacon@arm.com>
Date:   Tue Aug 23 22:20:03 2011 +0100

    ARM: 7061/1: gic: convert logical CPU numbers into physical numbers

And this one in rmk/devel-stable:

commit 254056f3b12563c11e6dbcfad2fbfce20a4f3302
Author: Colin Cross <ccross@android.com>
Date:   Thu Feb 10 12:54:10 2011 -0800

    ARM: gic: Use cpu pm notifiers to save gic state


BTW, Russell's for-next branch has been rebased. The conflict with the
1st commit is trivial, so I could rebase to merge of rmk/devel-stable
and v3.1-rc9.

I still need things from for-next for highbank. So perhaps I should send
a pull request after Russell's tree goes in?

Rob

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Thursday 20 October 2011, Rob Herring wrote:
> Arnd,
> 
> On 10/20/2011 08:07 AM, Arnd Bergmann wrote:
> > On Friday 14 October 2011, Rob Herring wrote:
> >> Please pull GIC device tree support. This is the first of 2 pull
> >> requests. You can ignore this one if there are no other dependencies on
> >> GIC DT support.
> >>
> >> This is based on rmk's for-next branch and v3.1-rc9. rc9 was needed for
> >> this dependency:
> > 
> > I don't see the for-next branch in the history. Do you mean it requires
> > the for-next branch as well in order to actually build?
> 
> Probably because it is buried by rc9 commits:
> 
> git log v3.1-rc9..gic-dt

Ah, I see them now. I should make sure I look more closely next time.
That definitely explains why I couldn't get this to merge into my
for-next branch cleanly.

> > Can you be more specific so I can watch for the dependencies to
> > get upstream first?
> > 
> 
> I believe it conflicts with this commit in rmk/for-next:
> 
> commit b166bc3be08b744d2f4b14921a1efee14906b383
> Author: Will Deacon <will.deacon@arm.com>
> Date:   Tue Aug 23 22:20:03 2011 +0100
> 
>     ARM: 7061/1: gic: convert logical CPU numbers into physical numbers
> 
> And this one in rmk/devel-stable:
> 
> commit 254056f3b12563c11e6dbcfad2fbfce20a4f3302
> Author: Colin Cross <ccross@android.com>
> Date:   Thu Feb 10 12:54:10 2011 -0800
> 
>     ARM: gic: Use cpu pm notifiers to save gic state
> 

Ok. Conflicts are not the problem though, I can handle them and sfr can
handle them for linux-next, too. Real dependencies are the problem,
where you rely on a feature that is part of another tree.

> BTW, Russell's for-next branch has been rebased. The conflict with the
> 1st commit is trivial, so I could rebase to merge of rmk/devel-stable
> and v3.1-rc9.
> 
> I still need things from for-next for highbank. So perhaps I should send
> a pull request after Russell's tree goes in?

Yes and no. You simply cannot ask me to merge a branch that is based on
top of Russell's for-next branch, since that is getting rebased. It is
also bad if the stuff doesn't have any linux-next exposure, so we should
try to find another way out.

I've now rebased your tree on top of 3.1-rc9 plus the stable branches
from Russell's tree that I already have as dependencies in arm-soc/for-next
(devel-stable, smp, debug). This has caused no conflicts for me, but
that doesn't mean that it's correct. Please check that what I have
in arm-soc/dt/gic and arm-soc/highbank/soc actually works for you
and does not contain branches that you don't actually need.

	Arnd

[GIT PULL] GIC DT binding support

[Rob Herring]

On 10/20/2011 11:12 AM, Arnd Bergmann wrote:
> On Thursday 20 October 2011, Rob Herring wrote:
>> Arnd,
>>
>> On 10/20/2011 08:07 AM, Arnd Bergmann wrote:
>>> On Friday 14 October 2011, Rob Herring wrote:
>>>> Please pull GIC device tree support. This is the first of 2 pull
>>>> requests. You can ignore this one if there are no other dependencies on
>>>> GIC DT support.
>>>>
>>>> This is based on rmk's for-next branch and v3.1-rc9. rc9 was needed for
>>>> this dependency:
>>>
>>> I don't see the for-next branch in the history. Do you mean it requires
>>> the for-next branch as well in order to actually build?
>>
>> Probably because it is buried by rc9 commits:
>>
>> git log v3.1-rc9..gic-dt
> 
> Ah, I see them now. I should make sure I look more closely next time.
> That definitely explains why I couldn't get this to merge into my
> for-next branch cleanly.
> 
>>> Can you be more specific so I can watch for the dependencies to
>>> get upstream first?
>>>
>>
>> I believe it conflicts with this commit in rmk/for-next:
>>
>> commit b166bc3be08b744d2f4b14921a1efee14906b383
>> Author: Will Deacon <will.deacon@arm.com>
>> Date:   Tue Aug 23 22:20:03 2011 +0100
>>
>>     ARM: 7061/1: gic: convert logical CPU numbers into physical numbers
>>
>> And this one in rmk/devel-stable:
>>
>> commit 254056f3b12563c11e6dbcfad2fbfce20a4f3302
>> Author: Colin Cross <ccross@android.com>
>> Date:   Thu Feb 10 12:54:10 2011 -0800
>>
>>     ARM: gic: Use cpu pm notifiers to save gic state
>>
> 
> Ok. Conflicts are not the problem though, I can handle them and sfr can
> handle them for linux-next, too. Real dependencies are the problem,
> where you rely on a feature that is part of another tree.
> 
>> BTW, Russell's for-next branch has been rebased. The conflict with the
>> 1st commit is trivial, so I could rebase to merge of rmk/devel-stable
>> and v3.1-rc9.
>>
>> I still need things from for-next for highbank. So perhaps I should send
>> a pull request after Russell's tree goes in?
> 
> Yes and no. You simply cannot ask me to merge a branch that is based on
> top of Russell's for-next branch, since that is getting rebased. It is
> also bad if the stuff doesn't have any linux-next exposure, so we should
> try to find another way out.
> 
> I've now rebased your tree on top of 3.1-rc9 plus the stable branches
> from Russell's tree that I already have as dependencies in arm-soc/for-next
> (devel-stable, smp, debug). This has caused no conflicts for me, but
> that doesn't mean that it's correct. Please check that what I have
> in arm-soc/dt/gic and arm-soc/highbank/soc actually works for you
> and does not contain branches that you don't actually need.
> 

dt/gic doesn't need Russell's debug branch, but highbank does.

highbank/soc doesn't build. It needs Russell's l2x0 and io branches.
l2x0 has been rebased recently, but the io branch seems to be stable
(but not published externally).

Rob

[GIT PULL] GIC DT binding support

[Russell King - ARM Linux]

On Thu, Oct 20, 2011 at 12:09:03PM -0500, Rob Herring wrote:
> highbank/soc doesn't build. It needs Russell's l2x0 and io branches.
> l2x0 has been rebased recently, but the io branch seems to be stable
> (but not published externally).

Actually, l2x0 _hasn't_.  Let's be clear about what happened.  A load
of patches were all merged into the 'misc' branch over time, containing
multiple different subjects.  At the time they were merged there was
no clear separate line of development of each sub-system, so they
just ended up in a common branch without any particular ordering or
grouping.

People wanted to base work on a select few of those commits.  Rather
than having the entire set of commits merged into various peoples trees,
many of which have nothing to do with what they were working on, I
split various commits out of the 'misc' branch and created a set of
more fine-grained topic branches.

That is arguably not rebasing - it is re-committing, in such a way that
those parts can then be frozen, stabilized and published without
unintended dependencies.  That's got to be a good thing.

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Thursday 20 October 2011 19:56:12 Russell King - ARM Linux wrote:
> On Thu, Oct 20, 2011 at 12:09:03PM -0500, Rob Herring wrote:
> > highbank/soc doesn't build. It needs Russell's l2x0 and io branches.
> > l2x0 has been rebased recently, but the io branch seems to be stable
> > (but not published externally).

I've kept the dt/gic branch the way I had it now, I guess the extra
branch in it doesn't hurt and removing it would cause extra work.
I'll just have to wait until it's merged then.

The highbank/soc branch is now rebased on top of
dt/gic+depends/rmk/io+depends/rmk/l2x0+depends/rmk/gpio.

I ended up having to add the gpio branch to the ones you mentioned to
get it all to compile.

> Actually, l2x0 _hasn't_.  Let's be clear about what happened.  A load
> of patches were all merged into the 'misc' branch over time, containing
> multiple different subjects.  At the time they were merged there was
> no clear separate line of development of each sub-system, so they
> just ended up in a common branch without any particular ordering or
> grouping.
> 
> People wanted to base work on a select few of those commits.  Rather
> than having the entire set of commits merged into various peoples trees,
> many of which have nothing to do with what they were working on, I
> split various commits out of the 'misc' branch and created a set of
> more fine-grained topic branches.
> 
> That is arguably not rebasing - it is re-committing, in such a way that
> those parts can then be frozen, stabilized and published without
> unintended dependencies.  That's got to be a good thing.

Yes, this certainly sounds good to me. One can still call it rebasing,
since you rebased the misc branch to create the l2x0 branch, but that
was of course specifically done to make it possible to create a stable
highbank tree, which is what Rob needs.

I'm not sure if I used the intended process to get at those branches.
Since they are not published on ftp.arm.linux.org.uk as separate branches,
I looked at the big merge changeset in the for-next branch and recreated
the branches I needed, which feels like I'm doing something that wasn't
your intention. Is there a better way?

	Arnd

[GIT PULL] GIC DT binding support

[Russell King - ARM Linux]

On Thu, Oct 20, 2011 at 11:04:35PM +0200, Arnd Bergmann wrote:
> I'm not sure if I used the intended process to get at those branches.
> Since they are not published on ftp.arm.linux.org.uk as separate branches,
> I looked at the big merge changeset in the for-next branch and recreated
> the branches I needed, which feels like I'm doing something that wasn't
> your intention. Is there a better way?

The best way is to say "Hey, my work depends on X, Y, Z" and I'll see
what can be done to make them stable _and_ say whether they are ripe
for that yet (or, eg, whether they're waiting for peoples acks.)

It's actually what I documented a while ago at
http://www.arm.linux.org.uk/developer/git-arm.php

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Friday 21 October 2011 19:33:52 Russell King - ARM Linux wrote:
> On Thu, Oct 20, 2011 at 11:04:35PM +0200, Arnd Bergmann wrote:
> > I'm not sure if I used the intended process to get at those branches.
> > Since they are not published on ftp.arm.linux.org.uk as separate branches,
> > I looked at the big merge changeset in the for-next branch and recreated
> > the branches I needed, which feels like I'm doing something that wasn't
> > your intention. Is there a better way?
> 
> The best way is to say "Hey, my work depends on X, Y, Z" and I'll see
> what can be done to make them stable and say whether they are ripe
> for that yet (or, eg, whether they're waiting for peoples acks.)
> 
> It's actually what I documented a while ago at
> http://www.arm.linux.org.uk/developer/git-arm.php

Ok, that is a very useful documentation, but the part I still don't
understand is how I should get the commit ID of the stable branches.
I noticed that you are publishing an l2x0 branch now, in addition
to the pm and smp branches that were already there the last time
I looked, but you don't publish the gpio, io and debug branches
that I have pulled into the arm-soc/next-soc branch because imx6 and
highbank depend on them.

Does that mean you don't consider the latter three branches stable
yet, or does it just mean they are stable but you had no reason
to publish them?

	Arnd

[GIT PULL] GIC DT binding support

[Russell King - ARM Linux]

On Fri, Oct 21, 2011 at 09:31:46PM +0200, Arnd Bergmann wrote:
> Ok, that is a very useful documentation, but the part I still don't
> understand is how I should get the commit ID of the stable branches.

I'll publish the branch itself.

> I noticed that you are publishing an l2x0 branch now, in addition
> to the pm and smp branches that were already there the last time
> I looked, but you don't publish the gpio, io and debug branches
> that I have pulled into the arm-soc/next-soc branch because imx6 and
> highbank depend on them.

debug has been published for a while but not the other two.  gpio can
be done trivially, but (as I've already said) io has an additional
commit on top which I'm not publishing yet.  So I'd need to fudge stuff
in some way to publish that one.

Given the time now, I'm not going to change my tree at this point,
with very little time available before having to leave.

[GIT PULL] GIC DT binding support

[Rob Herring]

Russell,

On 10/20/2011 01:56 PM, Russell King - ARM Linux wrote:
> On Thu, Oct 20, 2011 at 12:09:03PM -0500, Rob Herring wrote:
>> highbank/soc doesn't build. It needs Russell's l2x0 and io branches.
>> l2x0 has been rebased recently, but the io branch seems to be stable
>> (but not published externally).
> 
> Actually, l2x0 _hasn't_.  Let's be clear about what happened.  A load
> of patches were all merged into the 'misc' branch over time, containing
> multiple different subjects.  At the time they were merged there was
> no clear separate line of development of each sub-system, so they
> just ended up in a common branch without any particular ordering or
> grouping.
> 
> People wanted to base work on a select few of those commits.  Rather
> than having the entire set of commits merged into various peoples trees,
> many of which have nothing to do with what they were working on, I
> split various commits out of the 'misc' branch and created a set of
> more fine-grained topic branches.
> 
> That is arguably not rebasing - it is re-committing, in such a way that
> those parts can then be frozen, stabilized and published without
> unintended dependencies.  That's got to be a good thing.

Understood, but from a git perspective the commit ids have changed.
While git can deal with trivial re-commits like this, my understanding
is that should be avoided for upstream merges. Otherwise, I could just
cherry-pick them and re-commit them in my tree.

Ultimately, the people working off of those branches do need them to be
stable to merge their branches. So are you saying your l2x0 branch
(91c2ebb90b1890abc648ba9dec5608cbc97e1cb9) is now stable? Can you
publish it as a branch. iMX6Q is also dependent on it.

Rob

[GIT PULL] GIC DT binding support

[Russell King - ARM Linux]

On Thu, Oct 20, 2011 at 04:08:32PM -0500, Rob Herring wrote:
> Understood, but from a git perspective the commit ids have changed.
> While git can deal with trivial re-commits like this, my understanding
> is that should be avoided for upstream merges. Otherwise, I could just
> cherry-pick them and re-commit them in my tree.

It's precisely that need people feel to cherry-pick stuff which is
the need to keep my tree 'unstable' so that I _can_ reorganize it
to permit people to base work off it.

> Ultimately, the people working off of those branches do need them to be
> stable to merge their branches. So are you saying your l2x0 branch
> (91c2ebb90b1890abc648ba9dec5608cbc97e1cb9) is now stable? Can you
> publish it as a branch. iMX6Q is also dependent on it.

Indeed I can - all it takes is someone to tell me what they require
and I can sort it out - and this is especially so as I'm not planning
to do anything further with my git tree after this evening (on the
assumption that the merge window will open immediately after the
kernel summit.)

I already have several 'stable' branches: pm, smp, debug, and now
l2x0.

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Thursday 20 October 2011 16:08:32 Rob Herring wrote:
> > unintended dependencies.  That's got to be a good thing.
> 
> Understood, but from a git perspective the commit ids have changed.
> While git can deal with trivial re-commits like this, my understanding
> is that should be avoided for upstream merges. Otherwise, I could just
> cherry-pick them and re-commit them in my tree.
> 
> Ultimately, the people working off of those branches do need them to be
> stable to merge their branches. So are you saying your l2x0 branch
> (91c2ebb90b1890abc648ba9dec5608cbc97e1cb9) is now stable? Can you
> publish it as a branch. iMX6Q is also dependent on it.

The point is that Russell's 'misc' branch is not stable, which helps a lot
because it lets him commit stuff early for testing and then take it out
again when it turns out to be buggy. The for-next branch in Russell's
tree is not stable and gets recreated by pulling in the other branches.
This is similar to what I do with the arm-soc for-next branch, which 
is the sum of the other next/* branches in arm-soc.

My understanding is that the other branches in Russell's tree are stable in
the sense that they are never rebased, though it might happen that the entire
branch does not get submitted if a problem gets found with the series. That of
course implies that all other dependent trees cannot get submitted.

	Arnd

[GIT PULL] GIC DT binding support

[Arnd Bergmann]

On Thursday 20 October 2011 18:12:28 Arnd Bergmann wrote:
> I've now rebased your tree on top of 3.1-rc9 plus the stable branches
> from Russell's tree that I already have as dependencies in arm-soc/for-next
> (devel-stable, smp, debug). This has caused no conflicts for me, but
> that doesn't mean that it's correct. Please check that what I have
> in arm-soc/dt/gic and arm-soc/highbank/soc actually works for you
> and does not contain branches that you don't actually need.

I needed this patch in the end:

8<-----
ARM: gic: fix build error with CONFIG_IRQ_DOMAIN disabled

In file included from arch/arm/mach-msm/timer.c:26:0:
asm/hardware/gic.h:42:50: warning: 'struct device_node' declared inside parameter list [enabled by default]
asm/hardware/gic.h:42:50: warning: its scope is only this definition or declaration, which is probably not what you want [enabled by default]
asm/hardware/gic.h:58:20: error: field 'domain' has incomplete type

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
Ok to apply on top?

diff --git a/arch/arm/include/asm/hardware/gic.h b/arch/arm/include/asm/hardware/gic.h
index 1a776a1..a06c3ac 100644
--- a/arch/arm/include/asm/hardware/gic.h
+++ b/arch/arm/include/asm/hardware/gic.h
@@ -35,6 +35,8 @@
 #ifndef __ASSEMBLY__
 #include <linux/irqdomain.h>
 
+struct device_node;
+
 extern void __iomem *gic_cpu_base_addr;
 extern struct irq_chip gic_arch_extn;
 
@@ -55,7 +57,9 @@ struct gic_chip_data {
 	u32 __percpu *saved_ppi_enable;
 	u32 __percpu *saved_ppi_conf;
 #endif
+#ifdef CONFIG_IRQ_DOMAIN
 	struct irq_domain domain;
+#endif
 	unsigned int gic_irqs;
 };
 #endif

[GIT PULL] GIC DT binding support

[Rob Herring]

On 10/21/2011 03:51 AM, Arnd Bergmann wrote:
> On Thursday 20 October 2011 18:12:28 Arnd Bergmann wrote:
>> I've now rebased your tree on top of 3.1-rc9 plus the stable branches
>> from Russell's tree that I already have as dependencies in arm-soc/for-next
>> (devel-stable, smp, debug). This has caused no conflicts for me, but
>> that doesn't mean that it's correct. Please check that what I have
>> in arm-soc/dt/gic and arm-soc/highbank/soc actually works for you
>> and does not contain branches that you don't actually need.
> 
> I needed this patch in the end:
> 
> 8<-----
> ARM: gic: fix build error with CONFIG_IRQ_DOMAIN disabled
> 
> In file included from arch/arm/mach-msm/timer.c:26:0:
> asm/hardware/gic.h:42:50: warning: 'struct device_node' declared inside parameter list [enabled by default]
> asm/hardware/gic.h:42:50: warning: its scope is only this definition or declaration, which is probably not what you want [enabled by default]
> asm/hardware/gic.h:58:20: error: field 'domain' has incomplete type
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
> Ok to apply on top?

Okay.

> 
> diff --git a/arch/arm/include/asm/hardware/gic.h b/arch/arm/include/asm/hardware/gic.h
> index 1a776a1..a06c3ac 100644
> --- a/arch/arm/include/asm/hardware/gic.h
> +++ b/arch/arm/include/asm/hardware/gic.h
> @@ -35,6 +35,8 @@
>  #ifndef __ASSEMBLY__
>  #include <linux/irqdomain.h>
>  
> +struct device_node;
> +

You don't actually need this line.

Thanks,
Rob