2014-10-02 01:37:51

by Stephen Rothwell

[permalink] [raw]
Subject: linux-next: build failure after merge of the arm-soc tree

Hi all,

After merging the arm-soc tree, today's linux-next build (arm
multi_v7_defconfig) failed like this:

arch/arm/boot/dts/exynos5250-snow.dts:13:46: fatal error: dt-bindings/clock/maxim,max77686.h: No such file or directory
#include <dt-bindings/clock/maxim,max77686.h>
^

Caused by commit c04c92ed5614 ("ARM: dts: Add rtc_src clk for s3c-rtc
on exynos5250-snow").

Presumably there is some dependency on some tree that someone has
forgotten to merge :-(

I reverted that commit for today.

Which then got to:

/scratch/sfr/next/arch/arm/boot/dts/exynos5420-peach-pit.dts:15:46: fatal error: dt-bindings/clock/maxim,max77802.h: No such file or directory
#include <dt-bindings/clock/maxim,max77802.h>
^
compilation terminated.
/scratch/sfr/next/arch/arm/boot/dts/exynos5800-peach-pi.dts:15:46: fatal error: dt-bindings/clock/maxim,max77802.h: No such file or directory
#include <dt-bindings/clock/maxim,max77802.h>
^
compilation terminated.
Error: /scratch/sfr/next/arch/arm/boot/dts/hisi-x5hd2.dtsi:417.21-22 syntax error
FATAL ERROR: Unable to parse input tree

Caused by 8bd43b6e6df7 ("ARM: dts: Add rtc_src clk for s3c-rtc on
exynos Peach boards").

I give up ... please test your trees better ...

I have used the arm-soc tree form next-20141001 for today (which still
needs 5 reverts to get it to build :-().
--
Cheers,
Stephen Rothwell [email protected]


Attachments:
signature.asc (819.00 B)

2014-10-02 02:36:13

by Kukjin Kim

[permalink] [raw]
Subject: RE: linux-next: build failure after merge of the arm-soc tree

+ Mike

Stephen Rothwell wrote:
>
> Hi all,
>
Hi,

> After merging the arm-soc tree, today's linux-next build (arm
> multi_v7_defconfig) failed like this:
>
> arch/arm/boot/dts/exynos5250-snow.dts:13:46: fatal error: dt-bindings/clock/maxim,max77686.h: No such
> file or directory
> #include <dt-bindings/clock/maxim,max77686.h>
> ^
>
> Caused by commit c04c92ed5614 ("ARM: dts: Add rtc_src clk for s3c-rtc
> on exynos5250-snow").
>
> Presumably there is some dependency on some tree that someone has
> forgotten to merge :-(
>
Hmm...I thought it should be fine once Mike's clk tree is landing into the
-next tree. Firstly sorry about that and how should I do in this case?

> I reverted that commit for today.
>
> Which then got to:
>
> /scratch/sfr/next/arch/arm/boot/dts/exynos5420-peach-pit.dts:15:46: fatal error: dt-
> bindings/clock/maxim,max77802.h: No such file or directory
> #include <dt-bindings/clock/maxim,max77802.h>
> ^
> compilation terminated.
> /scratch/sfr/next/arch/arm/boot/dts/exynos5800-peach-pi.dts:15:46: fatal error: dt-
> bindings/clock/maxim,max77802.h: No such file or directory
> #include <dt-bindings/clock/maxim,max77802.h>
> ^
> compilation terminated.
> Error: /scratch/sfr/next/arch/arm/boot/dts/hisi-x5hd2.dtsi:417.21-22 syntax error
> FATAL ERROR: Unable to parse input tree
>
> Caused by 8bd43b6e6df7 ("ARM: dts: Add rtc_src clk for s3c-rtc on
> exynos Peach boards").
>
> I give up ... please test your trees better ...
>
So sorry, actually it's my fault, I missed the information to arm-soc when I
sent a pull-request, just I thought arm-soc were aware of that before happening
the error when samsung tree merged...

- Kukjin

> I have used the arm-soc tree form next-20141001 for today (which still
> needs 5 reverts to get it to build :-().

2014-10-02 14:33:19

by Arnd Bergmann

[permalink] [raw]
Subject: Re: linux-next: build failure after merge of the arm-soc tree

On Thursday 02 October 2014 11:37:42 Stephen Rothwell wrote:
> After merging the arm-soc tree, today's linux-next build (arm
> multi_v7_defconfig) failed like this:
>
> arch/arm/boot/dts/exynos5250-snow.dts:13:46: fatal error: dt-bindings/clock/maxim,max77686.h: No such file or directory
> #include <dt-bindings/clock/maxim,max77686.h>
> ^
>
> Caused by commit c04c92ed5614 ("ARM: dts: Add rtc_src clk for s3c-rtc
> on exynos5250-snow").
>
> Presumably there is some dependency on some tree that someone has
> forgotten to merge
>
> I reverted that commit for today.
>
> Which then got to:
>
> /scratch/sfr/next/arch/arm/boot/dts/exynos5420-peach-pit.dts:15:46: fatal error: dt-bindings/clock/maxim,max77802.h: No such file or directory
> #include <dt-bindings/clock/maxim,max77802.h>
> ^
> compilation terminated.
> /scratch/sfr/next/arch/arm/boot/dts/exynos5800-peach-pi.dts:15:46: fatal error: dt-bindings/clock/maxim,max77802.h: No such file or directory
> #include <dt-bindings/clock/maxim,max77802.h>
> ^
> compilation terminated.
> Error: /scratch/sfr/next/arch/arm/boot/dts/hisi-x5hd2.dtsi:417.21-22 syntax error
> FATAL ERROR: Unable to parse input tree
>
> Caused by 8bd43b6e6df7 ("ARM: dts: Add rtc_src clk for s3c-rtc on
> exynos Peach boards").
>
> I give up ... please test your trees better ...

I'm deeply sorry about that. I actually saw the problem here but was too
tired to fix it up again by redoing all the merges in our next/late
branch.

I've changed the for-next branch now to drop all of next/late, I think
we're better off not sending any of it for 3.18, especially as I found
some other problems as well.

> I have used the arm-soc tree form next-20141001 for today (which still
> needs 5 reverts to get it to build :-().

D'oh. I actually fixed these already last friday in the next/dt branch
but for some reason forgot to include it in for-next.
I've made sure that everything is included now.

Arnd

2014-10-03 06:50:20

by Javier Martinez Canillas

[permalink] [raw]
Subject: Re: linux-next: build failure after merge of the arm-soc tree

Hello Arnd, Stephen,

On 10/02/2014 04:32 PM, Arnd Bergmann wrote:
> On Thursday 02 October 2014 11:37:42 Stephen Rothwell wrote:
>> After merging the arm-soc tree, today's linux-next build (arm
>> multi_v7_defconfig) failed like this:
>>
>> arch/arm/boot/dts/exynos5250-snow.dts:13:46: fatal error: dt-bindings/clock/maxim,max77686.h: No such file or directory
>> #include <dt-bindings/clock/maxim,max77686.h>
>> ^
>>
>> Caused by commit c04c92ed5614 ("ARM: dts: Add rtc_src clk for s3c-rtc
>> on exynos5250-snow").
>>
>> Presumably there is some dependency on some tree that someone has
>> forgotten to merge
>>
>> I reverted that commit for today.
>>
>> Which then got to:
>>
>> /scratch/sfr/next/arch/arm/boot/dts/exynos5420-peach-pit.dts:15:46: fatal error: dt-bindings/clock/maxim,max77802.h: No such file or directory
>> #include <dt-bindings/clock/maxim,max77802.h>
>> ^
>> compilation terminated.
>> /scratch/sfr/next/arch/arm/boot/dts/exynos5800-peach-pi.dts:15:46: fatal error: dt-bindings/clock/maxim,max77802.h: No such file or directory
>> #include <dt-bindings/clock/maxim,max77802.h>
>> ^
>> compilation terminated.
>> Error: /scratch/sfr/next/arch/arm/boot/dts/hisi-x5hd2.dtsi:417.21-22 syntax error
>> FATAL ERROR: Unable to parse input tree
>>
>> Caused by 8bd43b6e6df7 ("ARM: dts: Add rtc_src clk for s3c-rtc on
>> exynos Peach boards").
>>
>> I give up ... please test your trees better ...
>
> I'm deeply sorry about that. I actually saw the problem here but was too
> tired to fix it up again by redoing all the merges in our next/late
> branch.
>

I'm so sorry about the breakage, when I posted this patch I mentioned that
it was meant for the 3.18-rc cycle once 3.18-rc1 was released [0]. Because
it fixes a regression caused by a patch queued in the clock tree for 3.18.
I should had been more explicit about the dependencies to avoid confusions.

> I've changed the for-next branch now to drop all of next/late, I think
> we're better off not sending any of it for 3.18, especially as I found
> some other problems as well.
>

Do you think that it may be possible to send this for the 3.18-rc cycle
or should it have to wait for 3.19? Without this the real time clock does
not work on the Peach Chromebooks and your next/late branch contains other
fixes as well (e.g: the Exynos USB dr_mode DT patches).

>> I have used the arm-soc tree form next-20141001 for today (which still
>> needs 5 reverts to get it to build :-().
>
> D'oh. I actually fixed these already last friday in the next/dt branch
> but for some reason forgot to include it in for-next.
> I've made sure that everything is included now.
>
> Arnd
>

Thanks a lot and best regards,
Javier

[0]: https://www.mail-archive.com/[email protected]/msg36702.html

2014-10-03 07:56:51

by Russell King - ARM Linux

[permalink] [raw]
Subject: Re: linux-next: build failure after merge of the arm-soc tree

On Fri, Oct 03, 2014 at 08:50:09AM +0200, Javier Martinez Canillas wrote:
> Hello Arnd, Stephen,
>
> I'm so sorry about the breakage, when I posted this patch I mentioned that
> it was meant for the 3.18-rc cycle once 3.18-rc1 was released [0]. Because
> it fixes a regression caused by a patch queued in the clock tree for 3.18.
> I should had been more explicit about the dependencies to avoid confusions.

If a patch is known to cause a regression, it shouldn't be going in during
the merge window _or_ the fix patch should also go in during the merge
window via the same route to resolve the regression before it's noticable.

Given the number of changes which go in during the merge window, waiting
for -rc to fix a known bug *before* the merge window is not on - it breaks
the ability to bisect during the period where bisect really is needed.

--
FTTC broadband for 0.8mile line: currently at 9.5Mbps down 400kbps up
according to speedtest.net.

2014-10-03 08:07:25

by Javier Martinez Canillas

[permalink] [raw]
Subject: Re: linux-next: build failure after merge of the arm-soc tree

Hello Russell,

On 10/03/2014 09:56 AM, Russell King - ARM Linux wrote:
> On Fri, Oct 03, 2014 at 08:50:09AM +0200, Javier Martinez Canillas wrote:
>> Hello Arnd, Stephen,
>>
>> I'm so sorry about the breakage, when I posted this patch I mentioned that
>> it was meant for the 3.18-rc cycle once 3.18-rc1 was released [0]. Because
>> it fixes a regression caused by a patch queued in the clock tree for 3.18.
>> I should had been more explicit about the dependencies to avoid confusions.
>
> If a patch is known to cause a regression, it shouldn't be going in during
> the merge window _or_ the fix patch should also go in during the merge
> window via the same route to resolve the regression before it's noticable.
>
> Given the number of changes which go in during the merge window, waiting
> for -rc to fix a known bug *before* the merge window is not on - it breaks
> the ability to bisect during the period where bisect really is needed.
>

I'm not the author of the patch that caused the regression, I just was bitten
by it and proposed a fix. So please don't kill the messenger :-)

Sorry, I didn't know about this rule, otherwise I would had asked Mike to pick
the fix instead. Or at least ask Kukjin to provide an immutable branch so Mike
could merge it on his clock tree.

Best regards,
Javier