qemu-devel
[Top][All Lists]
Advanced

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

[Qemu-devel] Re: [PATCH 0/5] virtio: cleanup layout assumptions


From: Juan Quintela
Subject: [Qemu-devel] Re: [PATCH 0/5] virtio: cleanup layout assumptions
Date: Thu, 18 Mar 2010 09:23:10 +0100
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)

"Michael S. Tsirkin" <address@hidden> wrote:
> This patchset helps remove the assumption that vdev must be the 1st
> member in any virtio device implementation. Other assumptions (made by
> qdev) can be removed in a similar way separately, this patchset does not
> do this yet.
> Juan, this is an alternative to your 'tell virtio about DO_UPCAST'
> patch, pls take a look.

Answered in the other series.  I don't see what removing that assumption
brings us.  qdev "requires" that 1st element is a qdev.  And that
requeriment is not going out anytime soon.

I think that this series are "over" engineering, at least until somebody
cames with a good use for not being the 1st element.

Why do you want this?  Notice that rest of qemu has this same assumption
for rest of devices.  Principle of Least Surprise means use my
"cleanup".

Furthermore, your VIRTIO_COMMON_INIT() macro is ugly as hell, again to
gain us nothing.

Without a good reason why we don't want VirtIODevice to be the 1st
element, I think that my series are more coherent (but on the other
hand, what would you expect from me :)

Later, Juan.




reply via email to

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