svn commit: r333745 - in head/sys/contrib/ck: include src

[Pedro Giffuni]

On 17/05/2018 14:27, Emmanuel Vadot wrote:
> On Thu, 17 May 2018 14:20:05 -0500
> Pedro Giffuni <pfg at FreeBSD.org> wrote:
>
>> On 17/05/2018 14:12, Matthew Macy wrote:
>>> How do I avoid problems while allowing timely updates?
>>>
>>> -M
>>>
>>> On Thu, May 17, 2018 at 11:38 AM, Emmanuel Vadot <manu at bidouilliste.com> wrote:
>>>>    Hi Matt,
>>>>
>>>> On Thu, 17 May 2018 18:14:10 +0000 (UTC)
>>>> Matt Macy <mmacy at FreeBSD.org> wrote:
>>>>
>>>>> Author: mmacy
>>>>> Date: Thu May 17 18:14:10 2018
>>>>> New Revision: 333745
>>>>> URL: https://svnweb.freebsd.org/changeset/base/333745
>>>>>
>>>>> Log:
>>>>>     ck: add support for executing callbacks outside of main poll loop
>>>>>
>>>>>     Pull in change from upstream deca119d14bfffd440770eb67cbdbeaf7b57eb7b
>>>>>
>>>>>     |    ck_epoch: introduce ck_epoch_deferred
>>>>>     |
>>>>>     |    Allow for deferral to occur outside epoch poll critical loop (which may access per-CPU structures).
>>>>>     |
>>>>>
>>>>>     Approved by:        sbruno
>>>>>
>>>>> Modified:
>>>>>     head/sys/contrib/ck/include/ck_epoch.h
>>>>>     head/sys/contrib/ck/src/ck_epoch.c
>>>>>
>>>>    CK was imported in vendor-sys/ck, commiting directly into head will
>>>> cause some problems in the future.
>> Actually ... committing to head is fine: some things just have to be fixed.
>   This is not true, it can cause a lot of problems with futures updates
> if the change isn't in the vendor repo.
OK ... I meant generally you can. The catch here is that the change was 
already upstream so it should have been done in the vendor area and then 
merged. It's best to fix this properly now.

Sorry for misguiding.

Pedro.

>> We do ask you try to upstream the change and re-merge things when possible.
>   This is already upstream, Matt is just updating CK with a patch he did
> ( see
> https://github.com/concurrencykit/ck/commit/deca119d14bfffd440770eb67cbdbeaf7b57eb7b)
>
>> Pedro.