Re: [Xen-devel] [PATCH v2 05/15] libxl: change libxl__ao_device_remove to libxl__ao_device

[Roger Pau Monne <roger.pau@xxxxxxxxxx>]

Ian Jackson wrote:
> Roger Pau Monne writes ("[PATCH v2 05/15] libxl: change libxl__ao_device_remove to 
> libxl__ao_device"):
> > Introduce a new structure to track state of device backends, that will
> > be used in following patches on this series.
> >
> > This structure if used for both device creation and device
> > destruction and removes libxl__ao_device_remove.
> > diff --git a/tools/libxl/libxl_internal.h b/tools/libxl/libxl_internal.h
> > index ae5527b..b62110a 100644
> > --- a/tools/libxl/libxl_internal.h
> > +++ b/tools/libxl/libxl_internal.h
> > @@ -1802,6 +1795,44 @@ _hidden void 
> > libxl__bootloader_init(libxl__bootloader_state *bl);
> ...
> > +/* Device AO operations */
> >
> > -typedef struct {
> > -    libxl__ao *ao;
> > -    libxl__ev_devstate ds;
> > -} libxl__ao_device_remove;
> > +void libxl__init_ao_device(libxl__ao_device *aorm, libxl__ao *ao,
> > +                           libxl__ao_device **base)
> > +{
> > +    aorm->ao = ao;
> > +    aorm->active = 1;
> > +    aorm->rc = 0;
> > +    aorm->base = base;
> > +}
> I think a function "init" should set "active" to 0, surely ?
> Since the operation has not in fact been started.  So perhaps this
> should just be a memset.
Forget to answer this, we cannot use active = 0 at init because in the 
hypothetical case that a device add/remove finished before queuing all 
the others (so the others would be active == 0) it would mark the ao as 
complete, since it would see all other operations as finished.
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel