Re: [Xen-devel] [PATCH v2 2/5] libxl: react correctly to POLLHUP

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

Ian Campbell writes ("Re: [Xen-devel] [PATCH v2 2/5] libxl: react correctly to 
POLLHUP"):
> On Fri, 2012-07-27 at 18:01 +0100, Ian Jackson wrote:

> > +static int datacopier_pollhup_handled(libxl__egc *egc,
> > +                                      libxl__datacopier_state *dc,
> > +                                      short revents, int onwrite)
> > +{
> > +    STATE_AO_GC(dc->ao);
> > +
> > +    if (dc->callback_pollhup && (revents & POLLHUP)) {
> > +        LOG(DEBUG, "received POLLHUP on %s during copy of %s",
> > +            onwrite ? dc->writewhat : dc->readwhat,
> > +            dc->copywhat);
> > +        libxl__datacopier_kill(dc);
> > +        dc->callback(egc, dc, onwrite, -1);
> 
> Shouldn't this be dc->callback_poolhup?

Yes.  I did say I hadn't executed it :-).

> > @@ -508,7 +508,10 @@ static void bootloader_gotptys(libxl__egc *egc, 
> > libxl__openpty_state *op)
> >      bl->keystrokes.maxsz = BOOTLOADER_BUF_OUT;
> >      bl->keystrokes.copywhat =
> >          GCSPRINTF("bootloader input for domain %"PRIu32, bl->domid);
> > -    bl->keystrokes.callback = bootloader_keystrokes_copyfail;
> > +    bl->keystrokes.callback =         bootloader_keystrokes_copyfail;
> > +    bl->keystrokes.callback_pollhup = bootloader_keystrokes_copyfail;
> > +        /* pollhup gets called with errnoval==-1 which is not otherwise
> > +         * possible since errnos are nonnegative, so it's unambiguous */
> 
> So the split into two callbacks is just for convenience of some future
> user, rather than something needed in this patch? Or have I missed
> somewhere which sets callback but not callback_pollhup?

Yes, you have.  datacopier is used for the qemu state file during
save/restore, and naturally wants to treat pollhup as fatal.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel