Re: [PATCH] seqlock: Use WRITE_ONCE() when updating sequence

[Waiman Long <llong@redhat.com>]

On 12/18/24 1:52 PM, Paul E. McKenney wrote:
>> If the compiler really uses the variable as a scratch storage, it will be a
>> problem if the variable can be accessed concurrently from multiple CPUs. It
>> is a new compiler optimization strategy that I am aware before. In that
>> case, we may really need a way to mark these variables as not suitable for
>> such advanced optimization.
> These markings already exist, namely, the "volatile" keyword, READ_ONCE(),
> WRITE_ONCE(), and hopefully soon INC_ONCE().  These last three use
> volatile accesses internally.
>
> The scratch-storage possibility exists only just before a normal
> C-language store, not before volatile accesses.  So a compiler is
> forbidden from doing that scratch-value-store trick before a volatile
> store.

I am aware of that in READ_ONCE() and WRITE_ONCE() and I am looking 
forward to have a INC_ONCE() and maybe DEC_ONCE() soon.

Cheers,
Longman