* Stefan Hajnoczi <stefa...@gmail.com> [2011-01-18 06:31:34]:
> On Tue, Jan 18, 2011 at 4:43 AM, Arun R Bharadwaj
> <a...@linux.vnet.ibm.com> wrote:
> > * Stefan Hajnoczi <stefa...@gmail.com> [2011-01-17 09:56:58]:
> >
> >> On Thu, Jan 13, 2011 at 12:14 PM, Arun R Bharadwaj
> >> <a...@linux.vnet.ibm.com> wrote:
> >> > +static void threadlet_io_completion_signal_handler(int signum)
> >> > +{
> >> > + qemu_service_io();
> >> > +}
> >> > +
> >> > +static void threadlet_register_signal_handler(void)
> >> > +{
> >> > + struct sigaction act;
> >> > + sigfillset(&act.sa_mask);
> >> > + act.sa_flags = 0; /* do not restart syscalls to interrupt select()
> >> > */
> >> > + act.sa_handler = threadlet_io_completion_signal_handler;
> >> > + sigaction(SIGUSR2, &act, NULL);
> >> > +}
> >> > +
> >> > +void threadlet_init(void)
> >> > +{
> >> > + threadlet_register_signal_handler();
> >> > +}
> >>
> >> This would be the right place to create qemu-threadlet.c, instead of
> >> adding the thread_init() prototype to qemu-thread.h and then including
> >> that in vl.c.
> >>
> >> Stefan
> >
> > I did not follow your comment here. How can we avoid including
> > threadler_init() in vl.c?
>
> Instead of adding threadlet_init() and related functions to
> posix-aio-compat.c and adding the prototype to qemu-thread.h, why not
> just create qemu-threadlet.c/qemu-threadlet.h and put these functions
> there instead?
>
> Stefan
Got it. So you mean I merge patch 8 and patch 10 into a single patch.
But wouldn't this mean we are moving code and adding new API in the
same patch? Anthony did not want this from what I recall. But I can do
it if you feel it makes things simple.
-arun
