2017-07-28 00:35:52

by Paul E. McKenney

[permalink] [raw]
Subject: [RFC GIT PULL rcu/next] RCU/hotplug fix

Hello, Ingo,

This pull request is for an RCU change that permits waiting for grace
periods started by CPUs late in the process of going offline. Lack of
this capability is causing failures:

http://lkml.kernel.org/r/[email protected]

Although these failures predate v4.12, and thus are in some sense not
regressions, it would be good for them to be fixed sooner rather than
later, hence this RFC pull request.

This commit was posted to LKML:

http://lkml.kernel.org/r/[email protected]

This commit has been exposed to 0day test robot and -next testing
and is available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo

for you to fetch changes up to a58163d8ca2c8d288ee9f95989712f98473a5ac2:

rcu: Migrate callbacks earlier in the CPU-offline timeline (2017-07-25 13:03:43 -0700)

----------------------------------------------------------------
Paul E. McKenney (1):
rcu: Migrate callbacks earlier in the CPU-offline timeline

include/linux/rcupdate.h | 1 +
kernel/cpu.c | 1 +
kernel/rcu/tree.c | 209 +++++++++++++++++++++++++----------------------
3 files changed, 115 insertions(+), 96 deletions(-)


2017-08-15 08:43:30

by Ingo Molnar

[permalink] [raw]
Subject: Re: [RFC GIT PULL rcu/next] RCU/hotplug fix


* Paul E. McKenney <[email protected]> wrote:

> Hello, Ingo,
>
> This pull request is for an RCU change that permits waiting for grace
> periods started by CPUs late in the process of going offline. Lack of
> this capability is causing failures:
>
> http://lkml.kernel.org/r/[email protected]
>
> Although these failures predate v4.12, and thus are in some sense not
> regressions, it would be good for them to be fixed sooner rather than
> later, hence this RFC pull request.
>
> This commit was posted to LKML:
>
> http://lkml.kernel.org/r/[email protected]
>
> This commit has been exposed to 0day test robot and -next testing
> and is available in the git repository at:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo
>
> for you to fetch changes up to a58163d8ca2c8d288ee9f95989712f98473a5ac2:
>
> rcu: Migrate callbacks earlier in the CPU-offline timeline (2017-07-25 13:03:43 -0700)
>
> ----------------------------------------------------------------
> Paul E. McKenney (1):
> rcu: Migrate callbacks earlier in the CPU-offline timeline
>
> include/linux/rcupdate.h | 1 +
> kernel/cpu.c | 1 +
> kernel/rcu/tree.c | 209 +++++++++++++++++++++++++----------------------
> 3 files changed, 115 insertions(+), 96 deletions(-)

Pulled, thanks Paul!

Ingo

2017-08-23 16:24:46

by Jeffrey Hugo

[permalink] [raw]
Subject: Re: [RFC GIT PULL rcu/next] RCU/hotplug fix

On 8/15/2017 2:43 AM, Ingo Molnar wrote:
>
> * Paul E. McKenney <[email protected]> wrote:
>
>> Hello, Ingo,
>>
>> This pull request is for an RCU change that permits waiting for grace
>> periods started by CPUs late in the process of going offline. Lack of
>> this capability is causing failures:
>>
>> http://lkml.kernel.org/r/[email protected]
>>
>> Although these failures predate v4.12, and thus are in some sense not
>> regressions, it would be good for them to be fixed sooner rather than
>> later, hence this RFC pull request.
>>
>> This commit was posted to LKML:
>>
>> http://lkml.kernel.org/r/[email protected]
>>
>> This commit has been exposed to 0day test robot and -next testing
>> and is available in the git repository at:
>>
>> git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo
>>
>> for you to fetch changes up to a58163d8ca2c8d288ee9f95989712f98473a5ac2:
>>
>> rcu: Migrate callbacks earlier in the CPU-offline timeline (2017-07-25 13:03:43 -0700)
>>
>> ----------------------------------------------------------------
>> Paul E. McKenney (1):
>> rcu: Migrate callbacks earlier in the CPU-offline timeline
>>
>> include/linux/rcupdate.h | 1 +
>> kernel/cpu.c | 1 +
>> kernel/rcu/tree.c | 209 +++++++++++++++++++++++++----------------------
>> 3 files changed, 115 insertions(+), 96 deletions(-)
>
> Pulled, thanks Paul!
>
> Ingo
>

Ingo,

We've been keeping an eye on this fix, so I'm trying to ascertain it's
status. For my information, is this expected to hit Linus' tree in
4.13, or wait until the 4.14 merge window?

Thanks

--
Jeffrey Hugo
Qualcomm Datacenter Technologies as an affiliate of Qualcomm
Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

2017-08-24 07:53:25

by Ingo Molnar

[permalink] [raw]
Subject: Re: [RFC GIT PULL rcu/next] RCU/hotplug fix


* Jeffrey Hugo <[email protected]> wrote:

> > > This commit has been exposed to 0day test robot and -next testing
> > > and is available in the git repository at:
> > >
> > > git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo
> > >
> > > for you to fetch changes up to a58163d8ca2c8d288ee9f95989712f98473a5ac2:
> > >
> > > rcu: Migrate callbacks earlier in the CPU-offline timeline (2017-07-25 13:03:43 -0700)
> > >
> > > ----------------------------------------------------------------
> > > Paul E. McKenney (1):
> > > rcu: Migrate callbacks earlier in the CPU-offline timeline
> > >
> > > include/linux/rcupdate.h | 1 +
> > > kernel/cpu.c | 1 +
> > > kernel/rcu/tree.c | 209 +++++++++++++++++++++++++----------------------
> > > 3 files changed, 115 insertions(+), 96 deletions(-)
> >
> > Pulled, thanks Paul!
> >
> > Ingo
> >
>
> Ingo,
>
> We've been keeping an eye on this fix, so I'm trying to ascertain it's status.
> For my information, is this expected to hit Linus' tree in 4.13, or wait until
> the 4.14 merge window?

It's for v4.14, it's a pretty large (and thus risky) commit that came in
relatively late. If it's problem-free it can be sent to -stable.

If Paul insists I can try and send it to Linus, but it's getting fairly late
in the cycle.

Thanks,

Ingo

2017-08-24 14:35:39

by Paul E. McKenney

[permalink] [raw]
Subject: Re: [RFC GIT PULL rcu/next] RCU/hotplug fix

On Thu, Aug 24, 2017 at 09:53:20AM +0200, Ingo Molnar wrote:
>
> * Jeffrey Hugo <[email protected]> wrote:
>
> > > > This commit has been exposed to 0day test robot and -next testing
> > > > and is available in the git repository at:
> > > >
> > > > git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo
> > > >
> > > > for you to fetch changes up to a58163d8ca2c8d288ee9f95989712f98473a5ac2:
> > > >
> > > > rcu: Migrate callbacks earlier in the CPU-offline timeline (2017-07-25 13:03:43 -0700)
> > > >
> > > > ----------------------------------------------------------------
> > > > Paul E. McKenney (1):
> > > > rcu: Migrate callbacks earlier in the CPU-offline timeline
> > > >
> > > > include/linux/rcupdate.h | 1 +
> > > > kernel/cpu.c | 1 +
> > > > kernel/rcu/tree.c | 209 +++++++++++++++++++++++++----------------------
> > > > 3 files changed, 115 insertions(+), 96 deletions(-)
> > >
> > > Pulled, thanks Paul!
> > >
> > > Ingo
> > >
> >
> > Ingo,
> >
> > We've been keeping an eye on this fix, so I'm trying to ascertain it's status.
> > For my information, is this expected to hit Linus' tree in 4.13, or wait until
> > the 4.14 merge window?
>
> It's for v4.14, it's a pretty large (and thus risky) commit that came in
> relatively late. If it's problem-free it can be sent to -stable.
>
> If Paul insists I can try and send it to Linus, but it's getting fairly late
> in the cycle.

Hugo, I was interpreting your email of August 22nd as meaning that
the problem no longer existed in mainline:

https://marc.info/?l=linux-kernel&m=150343524911301&w=2

If my interpretation is correct, then this commit can wait until the
upcoming merge window. As Ingo says, I can send patches to -stable
(and backporting is required, BTW).

If my interpretation is incorrect, please let me know what the situation
really is.

Thanx, Paul

2017-08-24 14:49:52

by Jeffrey Hugo

[permalink] [raw]
Subject: Re: [RFC GIT PULL rcu/next] RCU/hotplug fix

On 8/24/2017 8:35 AM, Paul E. McKenney wrote:
> On Thu, Aug 24, 2017 at 09:53:20AM +0200, Ingo Molnar wrote:
>>
>> * Jeffrey Hugo <[email protected]> wrote:
>>
>>>>> This commit has been exposed to 0day test robot and -next testing
>>>>> and is available in the git repository at:
>>>>>
>>>>> git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo
>>>>>
>>>>> for you to fetch changes up to a58163d8ca2c8d288ee9f95989712f98473a5ac2:
>>>>>
>>>>> rcu: Migrate callbacks earlier in the CPU-offline timeline (2017-07-25 13:03:43 -0700)
>>>>>
>>>>> ----------------------------------------------------------------
>>>>> Paul E. McKenney (1):
>>>>> rcu: Migrate callbacks earlier in the CPU-offline timeline
>>>>>
>>>>> include/linux/rcupdate.h | 1 +
>>>>> kernel/cpu.c | 1 +
>>>>> kernel/rcu/tree.c | 209 +++++++++++++++++++++++++----------------------
>>>>> 3 files changed, 115 insertions(+), 96 deletions(-)
>>>>
>>>> Pulled, thanks Paul!
>>>>
>>>> Ingo
>>>>
>>>
>>> Ingo,
>>>
>>> We've been keeping an eye on this fix, so I'm trying to ascertain it's status.
>>> For my information, is this expected to hit Linus' tree in 4.13, or wait until
>>> the 4.14 merge window?
>>
>> It's for v4.14, it's a pretty large (and thus risky) commit that came in
>> relatively late. If it's problem-free it can be sent to -stable.
>>
>> If Paul insists I can try and send it to Linus, but it's getting fairly late
>> in the cycle.
>
> Hugo, I was interpreting your email of August 22nd as meaning that
> the problem no longer existed in mainline:
>
> https://marc.info/?l=linux-kernel&m=150343524911301&w=2
>
> If my interpretation is correct, then this commit can wait until the
> upcoming merge window. As Ingo says, I can send patches to -stable
> (and backporting is required, BTW).
>
> If my interpretation is incorrect, please let me know what the situation
> really is.
>
> Thanx, Paul
>

Sorry for any confusion. Yes, that interpretation is correct. I agree,
it is late in the cycle, and I am not advocating for inclusion into
4.13. We have internal processes concerning upstream fixes, and it is
helpful to identify exactly when a fix is expected to hit Linus' tree -
none of which is particularly important or relevant to the community,
I'm just trying to bridge a few gaps.

4.14 merge window is perfectly fine. I will plan on that. Thanks for
the clarification, and a huge thanks for looking at the issue and
solving it.

--
Jeffrey Hugo
Qualcomm Datacenter Technologies as an affiliate of Qualcomm
Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.