Memory Barrier

[osv at javad.com (Sergei Organov)]

Pavel Pisa <ppisa4lists at pikron.com> writes:
[...]
> As for the non-inlined version of _Thread_Enable_dispatch,
> there should not be problem. Calling function without static
> or inline attributes is considered as full compiler memory ordering
> barrier point.

No, it is not, at least not according to the C standard. Therefore,
while it's OK with current version of GCC, it may still break in the
future. I'd vote for including required barriers in both variants.

-- Sergei.