Gilles Chanteperdrix wrote:
> Jan Kiszka wrote:
>> Gilles Chanteperdrix wrote:
>>> Jan Kiszka wrote:
>>>> ...and also automatically fixes the missing LOCK prefix for
>>>> pthread_mutex_* services on x86_32 SMP.
>>> This looks to me as a half-way unification. Can we not totally get rid
>>> of atomic_32.h and atomic_64.h ? I mean since we are using unsigned long
>>> as atomic_t on both platforms, there should not be much difference
>>> (except maybe the inline asm).
>> I could merge all atomic_32/64.h hunks into atomic.h if that this
>> preferred, but I cannot help getting rid of the atomic_t vs. atomic64_t
>> differences, thus the sub-arch specific part cannot be reduced as far as
>> I see it ATM.
> We could use atomic_long_t on the two arches.

OK, but then it becomes wrapping business (2.4...) - on the long term a
vanishing issue, granted. Will look into this.


Attachment: signature.asc
Description: OpenPGP digital signature

Xenomai-core mailing list

Reply via email to