This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: 2.1.127pre2


Linus Torvalds writes:
> 
> 
> On Mon, 26 Oct 1998, David S. Miller wrote:
> > 
> > Linus, can we have a "rwlock_init()" just like we do for spinlocks so
> > functions can initialize rwlock's in this manner without the compile
> > bombing on non-SMP because it will in turn from:
> > 
> > 	tmp->foo_rwlock = RWLOCK_INIT;
> > 
> > end up seeing:
> > 
> > 	tmp->foo_rwlock = { };
> > 
> > which last time I checked is not legal C :-)
> 
> No, we should do it as
> 
> 	tmp->foo_rwlock = (rw_lock_t) RWLOCK_INIT;
> 
> which is legal gcc. 

But gross. Having to do that when you use RWLOCK_INIT is ugly. What
about putting the cast into the macro itself?

				Regards,

					Richard....


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]