Re: [Qemu-devel] [PATCH v3 3/4] qdev: Simplify the SysBusDeviceClass::init path

[Eduardo Habkost <ehabkost@redhat.com>]

On Fri, Apr 20, 2018 at 09:22:55AM +0200, Markus Armbruster wrote:
> Philippe Mathieu-Daudé <f4bug@amsat.org> writes:
> 
> > On 04/19/2018 06:27 PM, Philippe Mathieu-Daudé wrote:
> >> The SysBusDevice is the last DeviceClass::init user.
> >> 
> >> Instead of using
> >>   SysBusDeviceClass::realize
> >>    -> DeviceClass::realize
> >>        -> DeviceClass::init
> >>            -> sysbus_device_init
> >>               -> SysBusDeviceClass::init
> >> 
> >> Simplify the path by directly calling SysBusDeviceClass::init
> >> in SysBusDeviceClass::realize:
> >> 
> >>   SysBusDeviceClass::realize
> >>    -> SysBusDeviceClass::init
> >> 
> >> Finally, remove the DeviceClass::init, there are no more users.
> >> 
> >> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> >> ---
> >>  include/hw/qdev-core.h |  2 --
> >>  hw/core/qdev.c         | 14 --------------
> >>  hw/core/sysbus.c       | 15 ++++++++++-----
> >>  3 files changed, 10 insertions(+), 21 deletions(-)
> >> 
> >> diff --git a/include/hw/qdev-core.h b/include/hw/qdev-core.h
> >> index 9453588160..6f60748043 100644
> >> --- a/include/hw/qdev-core.h
> >> +++ b/include/hw/qdev-core.h
> >> @@ -29,7 +29,6 @@ typedef enum DeviceCategory {
> >>      DEVICE_CATEGORY_MAX
> >>  } DeviceCategory;
> >>  
> >> -typedef int (*qdev_initfn)(DeviceState *dev);
> >>  typedef int (*qdev_event)(DeviceState *dev);
> >>  typedef void (*DeviceRealize)(DeviceState *dev, Error **errp);
> >>  typedef void (*DeviceUnrealize)(DeviceState *dev, Error **errp);
> >> @@ -124,7 +123,6 @@ typedef struct DeviceClass {
> >>      const struct VMStateDescription *vmsd;
> >>  
> >>      /* Private to qdev / bus.  */
> >> -    qdev_initfn init; /* TODO remove, once users are converted to realize */
> >
> > The DeviceClass documentation (top of this file) is now out of date :(
> >
> > If the maintainer taking this series prefer a respin, I plan to only
> > remove the 'init' references in the big comment.
> 
> I'd move the actual removal into the next and final patch.  Lets us fix
> up the comment in one go.

I'd prefer a respin, so we can ensure the documentation is
updated and the removal of DeviceClass::init and
DeviceClass::exit are done in the same patch.

(Sorry for taking so long to reply.)

-- 
Eduardo