Two new breakages since Thursday

Two new breakages since Thursday

[Russell King - ARM Linux]

Two new breakages have appeared since Thursday.  The first affects both
ARM and ARM64 builds (a missing file), the second only ARM builds.

arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'

ARM64 breaks on a defconfig build, ARM breaks with these errors on
an allmodconfig build.

The first two are due to the mediatek patches merged via the arm-soc
tree.  The second is due to:

commit 20567658b8f010eea287087bfdbeb25757801aed
Author: Wenyou Yang <wenyou.yang@atmel.com>
Date:   Mon Mar 9 11:53:46 2015 +0800

    ARM: at91/pm_suspend: add the WFI instruction support for ARMv7

which looks to me like completely the wrong solution:

+       .macro at91_cpu_idle
+
+#if defined(CONFIG_CPU_V7)
+       mov     tmp1, #AT91_PMC_PCK
+       str     tmp1, [pmc, #AT91_PMC_SCDR]
+
+       dsb
+
+       wfi             @ Wait For Interrupt
+#else
+       mcr     p15, 0, tmp1, c7, c0, 4
+#endif
+
+       .endm

So, if we build a kernel with ARMc6 and ARMv7 CPUs, AT91 ends up using
ARMv7-only instructions in this function, which will cause an illegal
instruction exception on ARMv6 CPUs.

-- 
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.

Two new breakages since Thursday

[Alexandre Belloni]

Hi,

On 05/04/2015 at 09:06:45 +0100, Russell King - ARM Linux wrote :
> So, if we build a kernel with ARMc6 and ARMv7 CPUs, AT91 ends up using
> ARMv7-only instructions in this function, which will cause an illegal
> instruction exception on ARMv6 CPUs.
> 

A fix has been pushed to the at91 tree but has not been pulled in
arm-soc yet.

-- 
Alexandre Belloni, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

Two new breakages since Thursday

[Nicolas Ferre]

Le 05/04/2015 12:12, Alexandre Belloni a ?crit :
> Hi,
> 
> On 05/04/2015 at 09:06:45 +0100, Russell King - ARM Linux wrote :
>> So, if we build a kernel with ARMc6 and ARMv7 CPUs, AT91 ends up using
>> ARMv7-only instructions in this function, which will cause an illegal
>> instruction exception on ARMv6 CPUs.
>>
> 
> A fix has been pushed to the at91 tree but has not been pulled in
> arm-soc yet.

Yes, I'll try to re-spin my pull requests soon so that it can be merged
quickly.

Thanks for the "heads up". Bye,
-- 
Nicolas Ferre

Two new breakages since Thursday

[Russell King - ARM Linux]

And almost a week later, and most of these are _still_ present.

Given that the merge window is potentially opening on Sunday, this
is _very_ poor, because it means that we've had a week of failing
builds in the week before the merge window - the week where we
_should_ be stabilising the code and getting it ready for submission
to Linus.

Right now, arm-soc is *NOT* ready for submission to Linus.

Please fix this as a top priority.

On Sun, Apr 05, 2015 at 09:06:45AM +0100, Russell King - ARM Linux wrote:
> Two new breakages have appeared since Thursday.  The first affects both
> ARM and ARM64 builds (a missing file), the second only ARM builds.
> 
> arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'
> 
> ARM64 breaks on a defconfig build, ARM breaks with these errors on
> an allmodconfig build.
> 
> The first two are due to the mediatek patches merged via the arm-soc
> tree.  The second is due to:
> 
> commit 20567658b8f010eea287087bfdbeb25757801aed
> Author: Wenyou Yang <wenyou.yang@atmel.com>
> Date:   Mon Mar 9 11:53:46 2015 +0800
> 
>     ARM: at91/pm_suspend: add the WFI instruction support for ARMv7
> 
> which looks to me like completely the wrong solution:
> 
> +       .macro at91_cpu_idle
> +
> +#if defined(CONFIG_CPU_V7)
> +       mov     tmp1, #AT91_PMC_PCK
> +       str     tmp1, [pmc, #AT91_PMC_SCDR]
> +
> +       dsb
> +
> +       wfi             @ Wait For Interrupt
> +#else
> +       mcr     p15, 0, tmp1, c7, c0, 4
> +#endif
> +
> +       .endm
> 
> So, if we build a kernel with ARMc6 and ARMv7 CPUs, AT91 ends up using
> ARMv7-only instructions in this function, which will cause an illegal
> instruction exception on ARMv6 CPUs.
> 
> -- 
> FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
> according to speedtest.net.

-- 
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.

Two new breakages since Thursday

[Yingjoe Chen]

On Sat, 2015-04-11 at 10:00 +0100, Russell King - ARM Linux wrote:
> And almost a week later, and most of these are _still_ present.
> 
> Given that the merge window is potentially opening on Sunday, this
> is _very_ poor, because it means that we've had a week of failing
> builds in the week before the merge window - the week where we
> _should_ be stabilising the code and getting it ready for submission
> to Linus.
> 
> Right now, arm-soc is *NOT* ready for submission to Linus.
> 
> Please fix this as a top priority.
> 
> On Sun, Apr 05, 2015 at 09:06:45AM +0100, Russell King - ARM Linux wrote:
> > Two new breakages have appeared since Thursday.  The first affects both
> > ARM and ARM64 builds (a missing file), the second only ARM builds.
> > 
> > arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'

Hi Arnd/Olof,

The mediatek part is due to merge dependency in pinctrl driver. It is
discussed in
http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336258.html

I thought we already decided to revert DT node in arm-soc tree for 4.1.
What should we do next?

Joe.C

Two new breakages since Thursday

[Arnd Bergmann]

On Monday 13 April 2015, Yingjoe Chen wrote:
> 
> On Sat, 2015-04-11 at 10:00 +0100, Russell King - ARM Linux wrote:
> > And almost a week later, and most of these are still present.
> > 
> > Given that the merge window is potentially opening on Sunday, this
> > is very poor, because it means that we've had a week of failing
> > builds in the week before the merge window - the week where we
> > should be stabilising the code and getting it ready for submission
> > to Linus.
> > 
> > Right now, arm-soc is NOT ready for submission to Linus.
> > 
> > Please fix this as a top priority.
> > 
> > On Sun, Apr 05, 2015 at 09:06:45AM +0100, Russell King - ARM Linux wrote:
> > > Two new breakages have appeared since Thursday.  The first affects both
> > > ARM and ARM64 builds (a missing file), the second only ARM builds.
> > > 
> > > arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'
> 
> Hi Arnd/Olof,
> 
> The mediatek part is due to merge dependency in pinctrl driver. It is
> discussed in
> http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336258.html
> 
> I thought we already decided to revert DT node in arm-soc tree for 4.1.
> What should we do next?

I've finally managed to find my way through the email backlog from being
away for three weeks and got to this one now. Patch is reverted in next/dt now.

Sorry about the delay in processing this.

	Arnd

Two new breakages since Thursday

[Russell King - ARM Linux]

On Tue, Apr 14, 2015 at 12:41:34AM +0200, Arnd Bergmann wrote:
> On Monday 13 April 2015, Yingjoe Chen wrote:
> > 
> > On Sat, 2015-04-11 at 10:00 +0100, Russell King - ARM Linux wrote:
> > > And almost a week later, and most of these are still present.
> > > 
> > > Given that the merge window is potentially opening on Sunday, this
> > > is very poor, because it means that we've had a week of failing
> > > builds in the week before the merge window - the week where we
> > > should be stabilising the code and getting it ready for submission
> > > to Linus.
> > > 
> > > Right now, arm-soc is NOT ready for submission to Linus.
> > > 
> > > Please fix this as a top priority.
> > > 
> > > On Sun, Apr 05, 2015 at 09:06:45AM +0100, Russell King - ARM Linux wrote:
> > > > Two new breakages have appeared since Thursday.  The first affects both
> > > > ARM and ARM64 builds (a missing file), the second only ARM builds.
> > > > 
> > > > arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > > arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > > arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'
> > 
> > Hi Arnd/Olof,
> > 
> > The mediatek part is due to merge dependency in pinctrl driver. It is
> > discussed in
> > http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336258.html
> > 
> > I thought we already decided to revert DT node in arm-soc tree for 4.1.
> > What should we do next?
> 
> I've finally managed to find my way through the email backlog from being
> away for three weeks and got to this one now. Patch is reverted in next/dt now.
> 
> Sorry about the delay in processing this.

Well, if this is now gone, there's still the issue of the AT91 breakage
that also needs to be fixed which is still there when I pulled arm-soc
last night.  What's happening there?

-- 
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.

Two new breakages since Thursday

[Arnd Bergmann]

On Tuesday 14 April 2015 10:26:12 Russell King - ARM Linux wrote:
> On Tue, Apr 14, 2015 at 12:41:34AM +0200, Arnd Bergmann wrote:
> > On Monday 13 April 2015, Yingjoe Chen wrote:
> > > 
> > > On Sat, 2015-04-11 at 10:00 +0100, Russell King - ARM Linux wrote:
> > > > And almost a week later, and most of these are still present.
> > > > 
> > > > Given that the merge window is potentially opening on Sunday, this
> > > > is very poor, because it means that we've had a week of failing
> > > > builds in the week before the merge window - the week where we
> > > > should be stabilising the code and getting it ready for submission
> > > > to Linus.
> > > > 
> > > > Right now, arm-soc is NOT ready for submission to Linus.
> > > > 
> > > > Please fix this as a top priority.
> > > > 
> > > > On Sun, Apr 05, 2015 at 09:06:45AM +0100, Russell King - ARM Linux wrote:
> > > > > Two new breakages have appeared since Thursday.  The first affects both
> > > > > ARM and ARM64 builds (a missing file), the second only ARM builds.
> > > > > 
> > > > > arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > > > arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > > > arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'
> > > 
> > > Hi Arnd/Olof,
> > > 
> > > The mediatek part is due to merge dependency in pinctrl driver. It is
> > > discussed in
> > > http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336258.html
> > > 
> > > I thought we already decided to revert DT node in arm-soc tree for 4.1.
> > > What should we do next?
> > 
> > I've finally managed to find my way through the email backlog from being
> > away for three weeks and got to this one now. Patch is reverted in next/dt now.
> > 
> > Sorry about the delay in processing this.
> 
> Well, if this is now gone, there's still the issue of the AT91 breakage
> that also needs to be fixed which is still there when I pulled arm-soc
> last night.  What's happening there?
> 

I managed to catch up with my email last night and pulled in these two fixes
as well as a number of others that we had neglected over the last two weeks.

I found one new regression with the overnight build tests in a branch I pulled
in as 'next/late' from Tony, and backed that out now. I'll wait for you and
Stephen Rothwell (who also reported these issues) to confirm that it's ok
now and we'll hopefully be ready to send pull requests by the end of the
week.

	Arnd

Two new breakages since Thursday

[Tony Lindgren]

* Arnd Bergmann <arnd@arndb.de> [150414 02:37]:
> On Tuesday 14 April 2015 10:26:12 Russell King - ARM Linux wrote:
> > On Tue, Apr 14, 2015 at 12:41:34AM +0200, Arnd Bergmann wrote:
> > > On Monday 13 April 2015, Yingjoe Chen wrote:
> > > > 
> > > > On Sat, 2015-04-11 at 10:00 +0100, Russell King - ARM Linux wrote:
> > > > > And almost a week later, and most of these are still present.
> > > > > 
> > > > > Given that the merge window is potentially opening on Sunday, this
> > > > > is very poor, because it means that we've had a week of failing
> > > > > builds in the week before the merge window - the week where we
> > > > > should be stabilising the code and getting it ready for submission
> > > > > to Linus.
> > > > > 
> > > > > Right now, arm-soc is NOT ready for submission to Linus.
> > > > > 
> > > > > Please fix this as a top priority.
> > > > > 
> > > > > On Sun, Apr 05, 2015 at 09:06:45AM +0100, Russell King - ARM Linux wrote:
> > > > > > Two new breakages have appeared since Thursday.  The first affects both
> > > > > > ARM and ARM64 builds (a missing file), the second only ARM builds.
> > > > > > 
> > > > > > arch/arm64/boot/dts/mediatek/mt8173-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > > > > arch/arm/boot/dts/mt8135-pinfunc.h:18:40: fatal error: dt-bindings/pinctrl/mt65xx.h: No such file or directory
> > > > > > arch/arm/mach-at91/pm_suspend.S:143: Error: selected processor does not support ARM mode `dsb'
> > > > 
> > > > Hi Arnd/Olof,
> > > > 
> > > > The mediatek part is due to merge dependency in pinctrl driver. It is
> > > > discussed in
> > > > http://lists.infradead.org/pipermail/linux-arm-kernel/2015-April/336258.html
> > > > 
> > > > I thought we already decided to revert DT node in arm-soc tree for 4.1.
> > > > What should we do next?
> > > 
> > > I've finally managed to find my way through the email backlog from being
> > > away for three weeks and got to this one now. Patch is reverted in next/dt now.
> > > 
> > > Sorry about the delay in processing this.
> > 
> > Well, if this is now gone, there's still the issue of the AT91 breakage
> > that also needs to be fixed which is still there when I pulled arm-soc
> > last night.  What's happening there?
> > 
> 
> I managed to catch up with my email last night and pulled in these two fixes
> as well as a number of others that we had neglected over the last two weeks.
> 
> I found one new regression with the overnight build tests in a branch I pulled
> in as 'next/late' from Tony, and backed that out now. I'll wait for you and
> Stephen Rothwell (who also reported these issues) to confirm that it's ok
> now and we'll hopefully be ready to send pull requests by the end of the
> week.

Hmm what's this issue you're mentioning here for the next/late? Is it the
merge conflict with camera related changes fixed up by Stephen Rothwell
in next?

Regards,

Tony

Two new breakages since Thursday

[Arnd Bergmann]

On Tuesday 14 April 2015 07:28:47 Tony Lindgren wrote:
> > I managed to catch up with my email last night and pulled in these two fixes
> > as well as a number of others that we had neglected over the last two weeks.
> > 
> > I found one new regression with the overnight build tests in a branch I pulled
> > in as 'next/late' from Tony, and backed that out now. I'll wait for you and
> > Stephen Rothwell (who also reported these issues) to confirm that it's ok
> > now and we'll hopefully be ready to send pull requests by the end of the
> > week.
> 
> Hmm what's this issue you're mentioning here for the next/late? Is it the
> merge conflict with camera related changes fixed up by Stephen Rothwell
> in next?
> 
> 

This is the error I get during 'make dtbs':

  DTC     arch/arm/boot/dts/am3517_mt_ventoux.dtb
ERROR (phandle_references): Reference to non-existent node or label "omap3_scm_general"

The problem is evidently commit b8845074cfbbd ("ARM: dts: omap3: add
minimal l4 bus layout with control module support"), which for some reason
removes the tisyscon at 48002270 node that is still referenced through
the &omap3_scm_general label from omap34xx.dtsi and omap36xx.dtsi.

	Arnd

Two new breakages since Thursday

[Tony Lindgren]

* Arnd Bergmann <arnd@arndb.de> [150414 08:00]:
> On Tuesday 14 April 2015 07:28:47 Tony Lindgren wrote:
> > > I managed to catch up with my email last night and pulled in these two fixes
> > > as well as a number of others that we had neglected over the last two weeks.
> > > 
> > > I found one new regression with the overnight build tests in a branch I pulled
> > > in as 'next/late' from Tony, and backed that out now. I'll wait for you and
> > > Stephen Rothwell (who also reported these issues) to confirm that it's ok
> > > now and we'll hopefully be ready to send pull requests by the end of the
> > > week.
> > 
> > Hmm what's this issue you're mentioning here for the next/late? Is it the
> > merge conflict with camera related changes fixed up by Stephen Rothwell
> > in next?
> > 
> > 
> 
> This is the error I get during 'make dtbs':
> 
>   DTC     arch/arm/boot/dts/am3517_mt_ventoux.dtb
> ERROR (phandle_references): Reference to non-existent node or label "omap3_scm_general"
> 
> The problem is evidently commit b8845074cfbbd ("ARM: dts: omap3: add
> minimal l4 bus layout with control module support"), which for some reason
> removes the tisyscon at 48002270 node that is still referenced through
> the &omap3_scm_general label from omap34xx.dtsi and omap36xx.dtsi.

I believe this issue got introduced on the 7th by the camera patches
I acked earlier. Stephen has been carrying this fix in next:

https://lkml.org/lkml/2015/4/6/436

The problem was we had two different SCM dts entries.. So with Tero's
changes we've now standardized on scm_conf.

Reagrds,

Tony

Two new breakages since Thursday

[Tony Lindgren]

* Tony Lindgren <tony@atomide.com> [150414 08:13]:
> * Arnd Bergmann <arnd@arndb.de> [150414 08:00]:
> > On Tuesday 14 April 2015 07:28:47 Tony Lindgren wrote:
> > > > I managed to catch up with my email last night and pulled in these two fixes
> > > > as well as a number of others that we had neglected over the last two weeks.
> > > > 
> > > > I found one new regression with the overnight build tests in a branch I pulled
> > > > in as 'next/late' from Tony, and backed that out now. I'll wait for you and
> > > > Stephen Rothwell (who also reported these issues) to confirm that it's ok
> > > > now and we'll hopefully be ready to send pull requests by the end of the
> > > > week.
> > > 
> > > Hmm what's this issue you're mentioning here for the next/late? Is it the
> > > merge conflict with camera related changes fixed up by Stephen Rothwell
> > > in next?
> > > 
> > > 
> > 
> > This is the error I get during 'make dtbs':
> > 
> >   DTC     arch/arm/boot/dts/am3517_mt_ventoux.dtb
> > ERROR (phandle_references): Reference to non-existent node or label "omap3_scm_general"
> > 
> > The problem is evidently commit b8845074cfbbd ("ARM: dts: omap3: add
> > minimal l4 bus layout with control module support"), which for some reason
> > removes the tisyscon at 48002270 node that is still referenced through
> > the &omap3_scm_general label from omap34xx.dtsi and omap36xx.dtsi.
> 
> I believe this issue got introduced on the 7th by the camera patches
> I acked earlier.

Actually, investigating it further, looks like my for-next branch is
missing my omap-for-v4.1/dt-v2 branch already merged into arm-soc
for-next.. Sorry about that, this merge conflict could have been
avoided if I had not missed that branch when rebuilding my for-next.

Anyways, Stephen's resolve below is correct. Arnd, I suggest you
remerge it with --no-commit and apply Stephen's resolution below
manually.

Regards,

Tony

> Stephen has been carrying this fix in next:
> 
> https://lkml.org/lkml/2015/4/6/436
> 
> The problem was we had two different SCM dts entries.. So with Tero's
> changes we've now standardized on scm_conf.

Two new breakages since Thursday

[Arnd Bergmann]

On Tuesday 14 April 2015 09:21:55 Tony Lindgren wrote:
> * Tony Lindgren <tony@atomide.com> [150414 08:13]:
> > * Arnd Bergmann <arnd@arndb.de> [150414 08:00]:
> > > On Tuesday 14 April 2015 07:28:47 Tony Lindgren wrote:
> > > > > I managed to catch up with my email last night and pulled in these two fixes
> > > > > as well as a number of others that we had neglected over the last two weeks.
> > > > > 
> > > > > I found one new regression with the overnight build tests in a branch I pulled
> > > > > in as 'next/late' from Tony, and backed that out now. I'll wait for you and
> > > > > Stephen Rothwell (who also reported these issues) to confirm that it's ok
> > > > > now and we'll hopefully be ready to send pull requests by the end of the
> > > > > week.
> > > > 
> > > > Hmm what's this issue you're mentioning here for the next/late? Is it the
> > > > merge conflict with camera related changes fixed up by Stephen Rothwell
> > > > in next?
> > > > 
> > > > 
> > > 
> > > This is the error I get during 'make dtbs':
> > > 
> > >   DTC     arch/arm/boot/dts/am3517_mt_ventoux.dtb
> > > ERROR (phandle_references): Reference to non-existent node or label "omap3_scm_general"
> > > 
> > > The problem is evidently commit b8845074cfbbd ("ARM: dts: omap3: add
> > > minimal l4 bus layout with control module support"), which for some reason
> > > removes the tisyscon at 48002270 node that is still referenced through
> > > the &omap3_scm_general label from omap34xx.dtsi and omap36xx.dtsi.
> > 
> > I believe this issue got introduced on the 7th by the camera patches
> > I acked earlier.
> 
> Actually, investigating it further, looks like my for-next branch is
> missing my omap-for-v4.1/dt-v2 branch already merged into arm-soc
> for-next.. Sorry about that, this merge conflict could have been
> avoided if I had not missed that branch when rebuilding my for-next.
> 
> Anyways, Stephen's resolve below is correct. Arnd, I suggest you
> remerge it with --no-commit and apply Stephen's resolution below
> manually.
> 

I've pulled in the omap/dt branch into next/late now, and fixed the
merge up in that way. I'm putting it back into for-next now.

We're running a bit late with the normal pull request now, so
no promises that next/late will still make it in time.

	Arnd

Two new breakages since Thursday

[Tony Lindgren]

* Arnd Bergmann <arnd@arndb.de> [150415 12:52]:
> On Tuesday 14 April 2015 09:21:55 Tony Lindgren wrote:
> > * Tony Lindgren <tony@atomide.com> [150414 08:13]:
> > > * Arnd Bergmann <arnd@arndb.de> [150414 08:00]:
> > > > On Tuesday 14 April 2015 07:28:47 Tony Lindgren wrote:
> > > > > > I managed to catch up with my email last night and pulled in these two fixes
> > > > > > as well as a number of others that we had neglected over the last two weeks.
> > > > > > 
> > > > > > I found one new regression with the overnight build tests in a branch I pulled
> > > > > > in as 'next/late' from Tony, and backed that out now. I'll wait for you and
> > > > > > Stephen Rothwell (who also reported these issues) to confirm that it's ok
> > > > > > now and we'll hopefully be ready to send pull requests by the end of the
> > > > > > week.
> > > > > 
> > > > > Hmm what's this issue you're mentioning here for the next/late? Is it the
> > > > > merge conflict with camera related changes fixed up by Stephen Rothwell
> > > > > in next?
> > > > > 
> > > > > 
> > > > 
> > > > This is the error I get during 'make dtbs':
> > > > 
> > > >   DTC     arch/arm/boot/dts/am3517_mt_ventoux.dtb
> > > > ERROR (phandle_references): Reference to non-existent node or label "omap3_scm_general"
> > > > 
> > > > The problem is evidently commit b8845074cfbbd ("ARM: dts: omap3: add
> > > > minimal l4 bus layout with control module support"), which for some reason
> > > > removes the tisyscon at 48002270 node that is still referenced through
> > > > the &omap3_scm_general label from omap34xx.dtsi and omap36xx.dtsi.
> > > 
> > > I believe this issue got introduced on the 7th by the camera patches
> > > I acked earlier.
> > 
> > Actually, investigating it further, looks like my for-next branch is
> > missing my omap-for-v4.1/dt-v2 branch already merged into arm-soc
> > for-next.. Sorry about that, this merge conflict could have been
> > avoided if I had not missed that branch when rebuilding my for-next.
> > 
> > Anyways, Stephen's resolve below is correct. Arnd, I suggest you
> > remerge it with --no-commit and apply Stephen's resolution below
> > manually.
> > 
> 
> I've pulled in the omap/dt branch into next/late now, and fixed the
> merge up in that way. I'm putting it back into for-next now.

OK thanks!
 
> We're running a bit late with the normal pull request now, so
> no promises that next/late will still make it in time.

Yes sure totally understood.

Regards,

Tony