Re: [PATCH RFT v2] nfsd: provide locking for v4_end_grace

[Chuck Lever <chuck.lever@oracle.com>]

On 7/2/25 5:22 PM, NeilBrown wrote:
> On Wed, 02 Jul 2025, Chuck Lever wrote:
>> Hi Neil, handful of nits below.
>>
>>>
>>> [[ 
>>>  v2 - disable laundromat_work while _init is running as well as while
>>>  _exit is running.  Don't depend on ->nfsd_serv, test
>>>  ->client_tracking_ops instead.
>>> ]]
>>
>> Do you want the patch change history to appear in the commit log?
>> Asking because that is not usual practice.
> 
> Not really.  I'll send something new, likely tomorrow morning.
> What do you think is the best way to handle backporting?

The best practice is to have upstream patches that can apply cleanly to
LTS kernels.

- One patch that can be backported for the fix, one or more for
  clean-ups that apply only to master;

- Identify pre-requisites that need to be backported first to make
  your fix apply cleanly; or

- Handcraft the fix for each LTS kernel


> Should I
> submit the version that doesn't use disable_delayed_work(), then add a
> patch which changes to use that instead of a flag ?

Post an RFC here, let's see what you're thinking.


-- 
Chuck Lever