Re: [PATCH v7 03/10] kthread: Add create_kthread_worker*()

[Andrew Morton]

On Mon, 30 May 2016 16:59:24 +0200 Petr Mladek <pmladek@xxxxxxxx> wrote:

> Kthread workers are currently created using the classic kthread API,
> namely kthread_run(). kthread_worker_fn() is passed as the @threadfn
> parameter.
> 
> This patch defines create_kthread_worker() and
> create_kthread_worker_on_cpu() functions that hide implementation details.

I hate to nick pits, but the naming isn't good.

A good, disciplined and pretty common naming scheme is to lead the
overall identifier with the name of the relevant subsystem.  kthread
has done that *fairly* well:


Things we got right:

kthread_create_on_node
kthread_create
kthread_create_on_cpu
kthread_run
kthread_bind
kthread_bind_mask
kthread_stop
kthread_should_stop
kthread_should_park
kthread_freezable_should_stop
kthread_data
kthread_park
kthread_unpark
kthread_parkme
kthreadd
kthread_work_func_t
KTHREAD_WORKER_INIT
KTHREAD_WORK_INIT
KTHREAD_WORKER_INIT_ONSTACK
kthread_worker_fn

Things we didn't:

probe_kthread_data
DEFINE_KTHREAD_WORKER
DEFINE_KTHREAD_WORK
DEFINE_KTHREAD_WORKER_ONSTACK
DEFINE_KTHREAD_WORKER_ONSTACK
__init_kthread_worker
init_kthread_worker
init_kthread_work
queue_kthread_work
flush_kthread_work
flush_kthread_worker


So I suggest kthread_create_worker() and
kthread_create_worker_on_cpu(), please.

And this might be a suitable time to regularize some of the "things we
didn't" identifiers, if you're feeling keen.