On Sun, 1 Apr 2007, Avi Kivity wrote:

quoted text
> Davide Libenzi wrote:
> > On Sun, 1 Apr 2007, Avi Kivity wrote:
> > 
> >   
> > > What is the motivation for adding aio_resfd to an individual iocb instead
> > > of
> > > the entire io context?  It seems redundant, as you can already create
> > > multiple
> > > io contexts to wait on.
> > >     
> > 
> > To add it to the context, you need to either change the context create API
> > (I think no-go here), or add a new syscall just to handle that.
> > Doing it in the iocb gives finer grained setup, but can be more work for the
> > user that wants to use it for all the iocbs.
> >   
> 
> I think it's a bit too fine grained, and a new system call (io_bindfd()?)
> would be easier to use.  In addition, you would move the eventfd_fget() out of
> the submission path.

IMO the cost of the eventfd_fget() (have you seen it?) is not worth adding 
a new syscall.
Actually, the flags field that Linus suggested may be given an extra meaning of 
"bind to ctx", that'd solve the problem w/out new syscalls.



- Davide


-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/