[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH v2 00/18] qom: dynamic properties and compositio

From: Anthony Liguori
Subject: Re: [Qemu-devel] [PATCH v2 00/18] qom: dynamic properties and composition tree (v2)
Date: Sun, 04 Dec 2011 15:01:59 -0600
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv: Gecko/20110831 Lightning/1.0b2 Thunderbird/3.1.13

On 12/03/2011 03:34 PM, Anthony Liguori wrote:
On 12/03/2011 08:24 AM, Paolo Bonzini wrote:
On 12/03/2011 03:40 AM, Anthony Liguori wrote:
That is still true. The next step, inheritance, will pull the properties
into a base class. That base class can be used elsewhere outside of the
device model.

But this is already a 20 patch series. If you want all of that in one
series, it's going to be 100 patches that are not terribly easy to
review at once.

Without a design document and a roadmap, however, it's impossible to try to
understand how the pieces will be together. 100 patches may require some time to
digest, but 20 patches require a crystal ball to figure out what's ahead.

You can see a bit further by looking at:


That fills out the composition tree pretty well for the pc. The next step is
aggressive refactoring such that the qdev objects reflect the composition. IOW,
we should create the rtc from within the piix3 initialization function.

I've begun the work of introducing proper inheritance. There's a lot going on but the basic idea is:

1) introduce QOM base type (Object), make qdev inherit from it

2) create a dynamic typeinfo based DeviceInfo, make device class point to 

3) model qdev hierarchy in QOM

4) starting from the bottom of the hierarchy, remove DeviceInfo subclass and push that functionality into QOM classes

5) once (4) is complete, remove DeviceInfo

6) refactor any use of multiple child busses into separate devices with one bus

7) refactor busstate as an interface

8) refactor device model to make more aggressive use of composition

9) refactor life cycle events into virtual methods

The tree I've posted is on step (4).


Anthony Liguori

reply via email to

[Prev in Thread] Current Thread [Next in Thread]