> On Nov 7, 2022, at 10:55 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
>
> On Mon, 2022-11-07 at 15:11 +0000, Chuck Lever III wrote:
>> Hi Jeff -
>>
>>> On Nov 7, 2022, at 10:01 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
>>>
>>> nfsd has grown a dedicated workqueue for the filecache, but this job is
>>> still queued to the system_wq. Change it to use the filecache_wq.
>>
>> Actually... there doesn't seem to be anything special about
>> nfsd_filecache_wq. 9542e6a643fc ("nfsd: Containerise filecache
>> laundrette") does not make clear why it was added.
>>
>> Playing devil's advocate: why not make the converse change and
>> replace it with system_wq?
>>
>
> Good question.
>
> At one time, allocating a workqueue gave you a dedicated pool of
> threads, but that's no longer the case.
>
> Documentation/core-api/workqueue.rst says: "A wq no longer manages
> execution resources but serves as a domain for forward progress
> guarantee, flush and work item attributes."
>
> Given that we're allocating this workqueue exactly the same way we
> allocate the system_wq, I don't think we need our own workqueue at all.
> I'd be fine with changing this to be the reverse if you prefer.
One less thing to fail on module initialization. Go for it.
>>> Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
>>> ---
>>> fs/nfsd/filecache.c | 2 +-
>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
>>> index 1e76b0d3b83a..018fd1565193 100644
>>> --- a/fs/nfsd/filecache.c
>>> +++ b/fs/nfsd/filecache.c
>>> @@ -212,7 +212,7 @@ static void
>>> nfsd_file_schedule_laundrette(void)
>>> {
>>> if (test_bit(NFSD_FILE_CACHE_UP, &nfsd_file_flags))
>>> - queue_delayed_work(system_wq, &nfsd_filecache_laundrette,
>>> + queue_delayed_work(nfsd_filecache_wq, &nfsd_filecache_laundrette,
>>> NFSD_LAUNDRETTE_DELAY);
>>> }
>>>
>>> --
>>> 2.38.1
>>>
>>
>> --
>> Chuck Lever
>>
>>
>>
>
> --
> Jeff Layton <jlayton@xxxxxxxxxx>
--
Chuck Lever
