On Mon, 2018-08-06 at 10:18 -0700, Tejun Heo wrote:

I'm worried that this is too inviting for misuses and subtle problems=

.

For example, your patch which uses this function uses
synchronize_rcu() to synchronize against per-cpu counts after the
first snoop; however, percpu_ref uses sched rcu, not the regu=

lar one,

so depending on the config, this will lead to *really* subtle
failures.  Even if that gets fixed, it's still leaking percpu-ref
internal details to its users - details which may change in the futur=

e

and will cause super subtle bugs.
=20
I'd go for something a lot more specific, like percpu_ref_is�

AF8-one(), so

that all the magics can be contained in percpu-ref implementation
proper.

Hello Tejun,

I will try to come up with an approach that does not require to modify the
per-cpu counter code.

Thanks,

Bart.