[lttng-dev] [PATCH 02/11] urcu/uatomic: Use atomic builtins if configured

[Mathieu Desnoyers]

On 6/29/23 13:22, Olivier Dion wrote:
> On Thu, 22 Jun 2023, "Paul E. McKenney" <paulmck at kernel.org> wrote:
>> On Thu, Jun 22, 2023 at 11:55:55AM -0400, Mathieu Desnoyers wrote:
>>> On 6/21/23 19:19, Paul E. McKenney wrote:
>> I suggest C11 volatile atomic load/store.  Load/store fusing is permitted
>> for non-volatile atomic loads and stores, and such fusing can ruin your
>> code's entire day.  ;-)
> 
> After some testing, I got a wall of warnings:
> 
>    -Wignored-qualifiers:
> 
>      Warn if the return type of a function has a type qualifier such as
>      "const".  For ISO C such a type qualifier has no effect, since the
>      value returned by a function is not an lvalue.  For C++, the warning
>      is only emitted for scalar types or "void".  ISO C prohibits
>      qualified "void" return types on function definitions, so such
>      return types always receive a warning even without this option.
> 
> Since we are using atomic builtins, for example load:
> 
>    type __atomic_load_n (type *ptr, int memorder)
> 
> If we put the qualifier volatile to TYPE, we end up with the same
> qualifier on the return value, triggering a warning for each atomic
> operation.
> 
> This seems to be only a problem when compiling in C++ [0] while in C it
> seems the compiler is more relaxed on this [1].
> 
> Ideas to make the toolchains happy? :-)

Change:

(__typeof__(*ptr) *volatile)(ptr);

(which applies the volatile to the pointer, rather than what is pointed to)

to either:

(volatile __typeof__(*ptr) *)(ptr);

or:

(__typeof__(*ptr) volatile *)(ptr);

Thanks,

Mathieu

> 
>    [0] https://godbolt.org/z/3nW14M3v1
>    [1] https://godbolt.org/z/TcTeMeKbW
> 

-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com