Reverting ARCH_SUNXI arm64 support(?)

Reverting ARCH_SUNXI arm64 support(?)

[Andre Przywara]

Hi Will, Catalin,

Suzuki reported a build failure with certain (non-defconfig) .configs
due to the new ARCH_SUNXI support in 4.6-rc [1].
As unfortunately we couldn't agree on a solution and also the support
for the A64/Pine64 is partial in 4.6 anyway, can you please revert:
ce3dd55b99b151a90ac1701c6825f2ae2d49b54e ("arm64: Introduce Allwinner
SoC config option")
So with this revert we have the arm64 defconfig now selecting useless
symbols, it didn't complain, but I was wondering if we should revert the
defconfig patch as well then? But this looks like pain :-(

If all this reverting is too much due to conflicts, we might think about
going with Suzuki's original one-liner [1], which would be an easy fix -
though Olof opposed it back then (in favour of a more involved solution,
which is no option for 4.6 anymore).

Regardless of this the plan for the future would be:
1) Send a patch to properly express the dependencies for the sunxi IRQ
controllers - this is independent from the arm64 support. I will do this
ASAP in hope that it still reaches 4.7.
2) Investigate whether Maxime's new sunxi clock architecture[2] would be
a fit for the A64 support.
3) Use the next weeks to rework the support patches and aim for a 4.8-rc
merge - depending on the outcome and merge status of 2)

Thoughts?

Cheers,
Andre.

[1]
http://lists.infradead.org/pipermail/linux-arm-kernel/2016-March/418175.html
   with the thread continuing here:
http://lists.infradead.org/pipermail/linux-arm-kernel/2016-April/420326.html

[2]
http://lists.infradead.org/pipermail/linux-arm-kernel/2016-May/427839.html

Reverting ARCH_SUNXI arm64 support(?)

[Will Deacon]

On Mon, May 09, 2016 at 09:53:11AM +0100, Andre Przywara wrote:
> Hi Will, Catalin,

Hi Andre,

> Suzuki reported a build failure with certain (non-defconfig) .configs
> due to the new ARCH_SUNXI support in 4.6-rc [1].
> As unfortunately we couldn't agree on a solution and also the support
> for the A64/Pine64 is partial in 4.6 anyway, can you please revert:
> ce3dd55b99b151a90ac1701c6825f2ae2d49b54e ("arm64: Introduce Allwinner
> SoC config option")

Given that we (Catalin and I) didn't commit that patch, it seems weird
that we should be reverting it. Shouldn't this be handled in arm-soc or
via the sunxi subtree?

We're also a week from the merge window, so it doesn't seem like a good
time to be hacking at .config/kconfig-related issues.

Will

Reverting ARCH_SUNXI arm64 support(?)

[Andre Przywara]

Hi,

On 09/05/16 10:09, Will Deacon wrote:
> On Mon, May 09, 2016 at 09:53:11AM +0100, Andre Przywara wrote:
>> Hi Will, Catalin,
> 
> Hi Andre,
> 
>> Suzuki reported a build failure with certain (non-defconfig) .configs
>> due to the new ARCH_SUNXI support in 4.6-rc [1].
>> As unfortunately we couldn't agree on a solution and also the support
>> for the A64/Pine64 is partial in 4.6 anyway, can you please revert:
>> ce3dd55b99b151a90ac1701c6825f2ae2d49b54e ("arm64: Introduce Allwinner
>> SoC config option")
> 
> Given that we (Catalin and I) didn't commit that patch, it seems weird
> that we should be reverting it. Shouldn't this be handled in arm-soc or
> via the sunxi subtree?
> 
> We're also a week from the merge window, so it doesn't seem like a good
> time to be hacking at .config/kconfig-related issues.

I agree, that's why I was proposing taking Suzuki's first patch:

diff --git a/arch/arm64/Kconfig.platforms b/arch/arm64/Kconfig.platforms
index efa77c1..521b1ec 100644
--- a/arch/arm64/Kconfig.platforms
+++ b/arch/arm64/Kconfig.platforms
@@ -2,6 +2,7 @@ menu "Platform selection"

 config ARCH_SUNXI
 	bool "Allwinner sunxi 64-bit SoC Family"
+	select GENERIC_IRQ_CHIP
 	help
 	  This enables support for Allwinner sunxi based SoCs like the A64.


This is what defconfig selects anyway and also the change would be
confined to this (new) config option.

Olof, Arnd: would it be an option to take this patch still?

Cheers,
Andre.

Reverting ARCH_SUNXI arm64 support(?)

[Maxime Ripard]

On Mon, May 09, 2016 at 12:15:24PM +0100, Andre Przywara wrote:
> Hi,
> 
> On 09/05/16 10:09, Will Deacon wrote:
> > On Mon, May 09, 2016 at 09:53:11AM +0100, Andre Przywara wrote:
> >> Hi Will, Catalin,
> > 
> > Hi Andre,
> > 
> >> Suzuki reported a build failure with certain (non-defconfig) .configs
> >> due to the new ARCH_SUNXI support in 4.6-rc [1].
> >> As unfortunately we couldn't agree on a solution and also the support
> >> for the A64/Pine64 is partial in 4.6 anyway, can you please revert:
> >> ce3dd55b99b151a90ac1701c6825f2ae2d49b54e ("arm64: Introduce Allwinner
> >> SoC config option")
> > 
> > Given that we (Catalin and I) didn't commit that patch, it seems weird
> > that we should be reverting it. Shouldn't this be handled in arm-soc or
> > via the sunxi subtree?
> > 
> > We're also a week from the merge window, so it doesn't seem like a good
> > time to be hacking at .config/kconfig-related issues.
> 
> I agree, that's why I was proposing taking Suzuki's first patch:
> 
> diff --git a/arch/arm64/Kconfig.platforms b/arch/arm64/Kconfig.platforms
> index efa77c1..521b1ec 100644
> --- a/arch/arm64/Kconfig.platforms
> +++ b/arch/arm64/Kconfig.platforms
> @@ -2,6 +2,7 @@ menu "Platform selection"
> 
>  config ARCH_SUNXI
>  	bool "Allwinner sunxi 64-bit SoC Family"
> +	select GENERIC_IRQ_CHIP
>  	help
>  	  This enables support for Allwinner sunxi based SoCs like the A64.
> 
> 
> This is what defconfig selects anyway and also the change would be
> confined to this (new) config option.
> 
> Olof, Arnd: would it be an option to take this patch still?

FWIW, I'd be in favour for such patch as a late patch for 4.6.

We can always fix things up properly in a later patch for 4.7 or 4.8.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20160509/6d3f1d45/attachment.sig>

Reverting ARCH_SUNXI arm64 support(?)

[Olof Johansson]

On Mon, May 9, 2016 at 1:57 PM, Maxime Ripard
<maxime.ripard@free-electrons.com> wrote:
> On Mon, May 09, 2016 at 12:15:24PM +0100, Andre Przywara wrote:
>> Hi,
>>
>> On 09/05/16 10:09, Will Deacon wrote:
>> > On Mon, May 09, 2016 at 09:53:11AM +0100, Andre Przywara wrote:
>> >> Hi Will, Catalin,
>> >
>> > Hi Andre,
>> >
>> >> Suzuki reported a build failure with certain (non-defconfig) .configs
>> >> due to the new ARCH_SUNXI support in 4.6-rc [1].
>> >> As unfortunately we couldn't agree on a solution and also the support
>> >> for the A64/Pine64 is partial in 4.6 anyway, can you please revert:
>> >> ce3dd55b99b151a90ac1701c6825f2ae2d49b54e ("arm64: Introduce Allwinner
>> >> SoC config option")
>> >
>> > Given that we (Catalin and I) didn't commit that patch, it seems weird
>> > that we should be reverting it. Shouldn't this be handled in arm-soc or
>> > via the sunxi subtree?
>> >
>> > We're also a week from the merge window, so it doesn't seem like a good
>> > time to be hacking at .config/kconfig-related issues.
>>
>> I agree, that's why I was proposing taking Suzuki's first patch:
>>
>> diff --git a/arch/arm64/Kconfig.platforms b/arch/arm64/Kconfig.platforms
>> index efa77c1..521b1ec 100644
>> --- a/arch/arm64/Kconfig.platforms
>> +++ b/arch/arm64/Kconfig.platforms
>> @@ -2,6 +2,7 @@ menu "Platform selection"
>>
>>  config ARCH_SUNXI
>>       bool "Allwinner sunxi 64-bit SoC Family"
>> +     select GENERIC_IRQ_CHIP
>>       help
>>         This enables support for Allwinner sunxi based SoCs like the A64.
>>
>>
>> This is what defconfig selects anyway and also the change would be
>> confined to this (new) config option.
>>
>> Olof, Arnd: would it be an option to take this patch still?
>
> FWIW, I'd be in favour for such patch as a late patch for 4.6.
>
> We can always fix things up properly in a later patch for 4.7 or 4.8.

+1. Please send said patch to arm at kernel.org and we can apply it (cc
Maxime so he can cc) and send up.


-Olof