powerpc qemu runs fail with the current upstream kernel.
Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
Unfortunately, that commit did not remove all instances of PPC_OF.
Practical impact is that the serial driver used by powerpc qemu
targets is no longer built into the test kernel.
Fixes: 52d996270032 ("powerpc: kill PPC_OF")
Cc: Kevin Hao <[email protected]>
Cc: Michael Ellerman <[email protected]>
Signed-off-by: Guenter Roeck <[email protected]>
---
drivers/tty/serial/Kconfig | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
index d2501f01cd03..77471d3db8d3 100644
--- a/drivers/tty/serial/Kconfig
+++ b/drivers/tty/serial/Kconfig
@@ -835,7 +835,7 @@ config SERIAL_MCF_CONSOLE
config SERIAL_PMACZILOG
tristate "Mac or PowerMac z85c30 ESCC support"
- depends on (M68K && MAC) || (PPC_OF && PPC_PMAC)
+ depends on (M68K && MAC) || PPC_PMAC
select SERIAL_CORE
help
This driver supports the Zilog z85C30 serial ports found on
@@ -1153,7 +1153,7 @@ config SERIAL_OMAP_CONSOLE
config SERIAL_OF_PLATFORM_NWPSERIAL
tristate "NWP serial port driver"
- depends on PPC_OF && PPC_DCR
+ depends on PPC_DCR
select SERIAL_OF_PLATFORM
select SERIAL_CORE_CONSOLE
select SERIAL_CORE
--
2.1.0
On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
> powerpc qemu runs fail with the current upstream kernel.
> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
> Unfortunately, that commit did not remove all instances of PPC_OF.
> Practical impact is that the serial driver used by powerpc qemu
> targets is no longer built into the test kernel.
Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
been merged into tty-next, but still not propagate to upstream yet. I failed
to reminder Michael of this when the pulling request is sent to Linus.
Thanks,
Kevin
>
> Fixes: 52d996270032 ("powerpc: kill PPC_OF")
> Cc: Kevin Hao <[email protected]>
> Cc: Michael Ellerman <[email protected]>
> Signed-off-by: Guenter Roeck <[email protected]>
> ---
> drivers/tty/serial/Kconfig | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
> index d2501f01cd03..77471d3db8d3 100644
> --- a/drivers/tty/serial/Kconfig
> +++ b/drivers/tty/serial/Kconfig
> @@ -835,7 +835,7 @@ config SERIAL_MCF_CONSOLE
>
> config SERIAL_PMACZILOG
> tristate "Mac or PowerMac z85c30 ESCC support"
> - depends on (M68K && MAC) || (PPC_OF && PPC_PMAC)
> + depends on (M68K && MAC) || PPC_PMAC
> select SERIAL_CORE
> help
> This driver supports the Zilog z85C30 serial ports found on
> @@ -1153,7 +1153,7 @@ config SERIAL_OMAP_CONSOLE
>
> config SERIAL_OF_PLATFORM_NWPSERIAL
> tristate "NWP serial port driver"
> - depends on PPC_OF && PPC_DCR
> + depends on PPC_DCR
> select SERIAL_OF_PLATFORM
> select SERIAL_CORE_CONSOLE
> select SERIAL_CORE
> --
> 2.1.0
>
On 04/16/2015 11:01 PM, Kevin Hao wrote:
> On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
>> powerpc qemu runs fail with the current upstream kernel.
>> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
>> Unfortunately, that commit did not remove all instances of PPC_OF.
>> Practical impact is that the serial driver used by powerpc qemu
>> targets is no longer built into the test kernel.
>
> Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
> ("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
> been merged into tty-next, but still not propagate to upstream yet. I failed
> to reminder Michael of this when the pulling request is sent to Linus.
>
Guess that explains why I don't see the breakage in linux-next.
This kind of problem seems to be happening a lot in this commit window.
Is there a new mechanism in place which requires splitting such series
into multiple parts ? Personally I preferred the "old" style, where
the entire series would have been handled by one maintainer, with Acks
from the others.
Guenter
On Fri, Apr 17, 2015 at 7:01 AM, Kevin Hao <[email protected]> wrote:
> On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
>> powerpc qemu runs fail with the current upstream kernel.
>> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
>> Unfortunately, that commit did not remove all instances of PPC_OF.
>> Practical impact is that the serial driver used by powerpc qemu
>> targets is no longer built into the test kernel.
>
> Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
> ("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
> been merged into tty-next, but still not propagate to upstream yet. I failed
> to reminder Michael of this when the pulling request is sent to Linus.
This means you're doing it wrong. If there are dependencies between
patches, then they *must* all be in the same tree. You cannot depend
on the order that branches get merged because it breaks bisecting.
Or, to put it another way, every branch must compile on it's own
without needing to be merged into mainline or linux-next first.
There are two ways to handle this. The easiest is to commit all the
patches to a single branch. Then there is guaranteed to be no
problems. However, if the patches really need to be split up between
several branches, then the way to handle it is to put the prerequisite
patches into a new clean branch, and then merge that branch into each
tree that needs the prerequisite. That way every single tree includes
the required commit without committing it more than once or breaking
bisectability.
g.
>
> Thanks,
> Kevin
>>
>> Fixes: 52d996270032 ("powerpc: kill PPC_OF")
>> Cc: Kevin Hao <[email protected]>
>> Cc: Michael Ellerman <[email protected]>
>> Signed-off-by: Guenter Roeck <[email protected]>
>> ---
>> drivers/tty/serial/Kconfig | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
>> index d2501f01cd03..77471d3db8d3 100644
>> --- a/drivers/tty/serial/Kconfig
>> +++ b/drivers/tty/serial/Kconfig
>> @@ -835,7 +835,7 @@ config SERIAL_MCF_CONSOLE
>>
>> config SERIAL_PMACZILOG
>> tristate "Mac or PowerMac z85c30 ESCC support"
>> - depends on (M68K && MAC) || (PPC_OF && PPC_PMAC)
>> + depends on (M68K && MAC) || PPC_PMAC
>> select SERIAL_CORE
>> help
>> This driver supports the Zilog z85C30 serial ports found on
>> @@ -1153,7 +1153,7 @@ config SERIAL_OMAP_CONSOLE
>>
>> config SERIAL_OF_PLATFORM_NWPSERIAL
>> tristate "NWP serial port driver"
>> - depends on PPC_OF && PPC_DCR
>> + depends on PPC_DCR
>> select SERIAL_OF_PLATFORM
>> select SERIAL_CORE_CONSOLE
>> select SERIAL_CORE
>> --
>> 2.1.0
>>
On Fri, Apr 17, 2015 at 7:17 AM, Guenter Roeck <[email protected]> wrote:
> On 04/16/2015 11:01 PM, Kevin Hao wrote:
>>
>> On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
>>>
>>> powerpc qemu runs fail with the current upstream kernel.
>>> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
>>> Unfortunately, that commit did not remove all instances of PPC_OF.
>>> Practical impact is that the serial driver used by powerpc qemu
>>> targets is no longer built into the test kernel.
>>
>>
>> Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
>> ("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
>> been merged into tty-next, but still not propagate to upstream yet. I
>> failed
>> to reminder Michael of this when the pulling request is sent to Linus.
>>
>
> Guess that explains why I don't see the breakage in linux-next.
>
> This kind of problem seems to be happening a lot in this commit window.
>
> Is there a new mechanism in place which requires splitting such series
> into multiple parts ? Personally I preferred the "old" style, where
> the entire series would have been handled by one maintainer, with Acks
> from the others.
The rules haven't changed. Maintainers are doing the wrong thing. If a
series is split up into multiple parts, then maintainers *must*
coordinate to put the prerequisites into a single branch that can be
merged into each branch handling it. However, it is still almost
always better to just merge the entire series via a single tree.
Make noise whenever you see this kind of breakage because it means a
maintainer has done the wrong thing.
g.
On Fri, Apr 17, 2015 at 08:00:45AM +0100, Grant Likely wrote:
> On Fri, Apr 17, 2015 at 7:17 AM, Guenter Roeck <[email protected]> wrote:
> > On 04/16/2015 11:01 PM, Kevin Hao wrote:
> >>
> >> On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
> >>>
> >>> powerpc qemu runs fail with the current upstream kernel.
> >>> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
> >>> Unfortunately, that commit did not remove all instances of PPC_OF.
> >>> Practical impact is that the serial driver used by powerpc qemu
> >>> targets is no longer built into the test kernel.
> >>
> >>
> >> Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
> >> ("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
> >> been merged into tty-next, but still not propagate to upstream yet. I
> >> failed
> >> to reminder Michael of this when the pulling request is sent to Linus.
> >>
> >
> > Guess that explains why I don't see the breakage in linux-next.
> >
> > This kind of problem seems to be happening a lot in this commit window.
> >
> > Is there a new mechanism in place which requires splitting such series
> > into multiple parts ? Personally I preferred the "old" style, where
> > the entire series would have been handled by one maintainer, with Acks
> > from the others.
>
> The rules haven't changed. Maintainers are doing the wrong thing. If a
> series is split up into multiple parts, then maintainers *must*
> coordinate to put the prerequisites into a single branch that can be
> merged into each branch handling it. However, it is still almost
> always better to just merge the entire series via a single tree.
>
> Make noise whenever you see this kind of breakage because it means a
> maintainer has done the wrong thing.
Well, the maintainer needs to be _told_ that the patch that is being
sent to them shouldn't go through their tree and that it depends on
other patches, so that they can properly just ack them.
Which is what happened here, someone sent me a patch, and I applied it.
Nothing broke that I could determine, and I never got a report of
something breaking, so how am I, the maintainer, doing the wrong thing?
thanks,
greg k-h
On Fri, Apr 17, 2015 at 07:58:10AM +0100, Grant Likely wrote:
> On Fri, Apr 17, 2015 at 7:01 AM, Kevin Hao <[email protected]> wrote:
> > On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
> >> powerpc qemu runs fail with the current upstream kernel.
> >> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
> >> Unfortunately, that commit did not remove all instances of PPC_OF.
> >> Practical impact is that the serial driver used by powerpc qemu
> >> targets is no longer built into the test kernel.
> >
> > Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
> > ("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
> > been merged into tty-next, but still not propagate to upstream yet. I failed
> > to reminder Michael of this when the pulling request is sent to Linus.
>
> This means you're doing it wrong.
Definitely.
> If there are dependencies between
> patches, then they *must* all be in the same tree. You cannot depend
> on the order that branches get merged because it breaks bisecting.
We don't try to depend on the order that branches get merged. We just want to
merge the patch which kill the PPC_OF completely only after all the patches for
the specific drivers have been merged. I should took the responsibility to
make sure of this, but of course I missed something. Yes. putting all the
patches into the same tree should be a better plan. I will make sure of this
next time. :-)
>
> Or, to put it another way, every branch must compile on it's own
> without needing to be merged into mainline or linux-next first.
This issue doesn't break compile. That is another reason we didn't find it
earlier.
Thanks,
Kevin
>
> There are two ways to handle this. The easiest is to commit all the
> patches to a single branch. Then there is guaranteed to be no
> problems. However, if the patches really need to be split up between
> several branches, then the way to handle it is to put the prerequisite
> patches into a new clean branch, and then merge that branch into each
> tree that needs the prerequisite. That way every single tree includes
> the required commit without committing it more than once or breaking
> bisectability.
>
> g.
>
> >
> > Thanks,
> > Kevin
> >>
> >> Fixes: 52d996270032 ("powerpc: kill PPC_OF")
> >> Cc: Kevin Hao <[email protected]>
> >> Cc: Michael Ellerman <[email protected]>
> >> Signed-off-by: Guenter Roeck <[email protected]>
> >> ---
> >> drivers/tty/serial/Kconfig | 4 ++--
> >> 1 file changed, 2 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/tty/serial/Kconfig b/drivers/tty/serial/Kconfig
> >> index d2501f01cd03..77471d3db8d3 100644
> >> --- a/drivers/tty/serial/Kconfig
> >> +++ b/drivers/tty/serial/Kconfig
> >> @@ -835,7 +835,7 @@ config SERIAL_MCF_CONSOLE
> >>
> >> config SERIAL_PMACZILOG
> >> tristate "Mac or PowerMac z85c30 ESCC support"
> >> - depends on (M68K && MAC) || (PPC_OF && PPC_PMAC)
> >> + depends on (M68K && MAC) || PPC_PMAC
> >> select SERIAL_CORE
> >> help
> >> This driver supports the Zilog z85C30 serial ports found on
> >> @@ -1153,7 +1153,7 @@ config SERIAL_OMAP_CONSOLE
> >>
> >> config SERIAL_OF_PLATFORM_NWPSERIAL
> >> tristate "NWP serial port driver"
> >> - depends on PPC_OF && PPC_DCR
> >> + depends on PPC_DCR
> >> select SERIAL_OF_PLATFORM
> >> select SERIAL_CORE_CONSOLE
> >> select SERIAL_CORE
> >> --
> >> 2.1.0
> >>
On Fri, Apr 17, 2015 at 8:43 AM, Greg Kroah-Hartman
<[email protected]> wrote:
> On Fri, Apr 17, 2015 at 08:00:45AM +0100, Grant Likely wrote:
>> On Fri, Apr 17, 2015 at 7:17 AM, Guenter Roeck <[email protected]> wrote:
>> > On 04/16/2015 11:01 PM, Kevin Hao wrote:
>> >>
>> >> On Thu, Apr 16, 2015 at 10:20:59PM -0700, Guenter Roeck wrote:
>> >>>
>> >>> powerpc qemu runs fail with the current upstream kernel.
>> >>> Bisect points to commit 52d996270032 ("powerpc: kill PPC_OF").
>> >>> Unfortunately, that commit did not remove all instances of PPC_OF.
>> >>> Practical impact is that the serial driver used by powerpc qemu
>> >>> targets is no longer built into the test kernel.
>> >>
>> >>
>> >> Sorry for the break. This is a dependency issue. The patch 213dce3c17a6
>> >> ("tty: kconfig: remove the superfluous dependency on PPC_OF") has already
>> >> been merged into tty-next, but still not propagate to upstream yet. I
>> >> failed
>> >> to reminder Michael of this when the pulling request is sent to Linus.
>> >>
>> >
>> > Guess that explains why I don't see the breakage in linux-next.
>> >
>> > This kind of problem seems to be happening a lot in this commit window.
>> >
>> > Is there a new mechanism in place which requires splitting such series
>> > into multiple parts ? Personally I preferred the "old" style, where
>> > the entire series would have been handled by one maintainer, with Acks
>> > from the others.
>>
>> The rules haven't changed. Maintainers are doing the wrong thing. If a
>> series is split up into multiple parts, then maintainers *must*
>> coordinate to put the prerequisites into a single branch that can be
>> merged into each branch handling it. However, it is still almost
>> always better to just merge the entire series via a single tree.
>>
>> Make noise whenever you see this kind of breakage because it means a
>> maintainer has done the wrong thing.
>
> Well, the maintainer needs to be _told_ that the patch that is being
> sent to them shouldn't go through their tree and that it depends on
> other patches, so that they can properly just ack them.
>
> Which is what happened here, someone sent me a patch, and I applied it.
> Nothing broke that I could determine, and I never got a report of
> something breaking, so how am I, the maintainer, doing the wrong thing?
My apologies, Yes of course. s/a maintainer/someone/
g.