[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v4 00/11] virtio: virtio-blk data plane
From: |
Stefan Hajnoczi |
Subject: |
Re: [Qemu-devel] [PATCH v4 00/11] virtio: virtio-blk data plane |
Date: |
Thu, 29 Nov 2012 15:48:04 +0100 |
User-agent: |
Mutt/1.5.21 (2010-09-15) |
On Thu, Nov 29, 2012 at 04:09:28PM +0200, Michael S. Tsirkin wrote:
> On Thu, Nov 29, 2012 at 10:18:59AM +0100, Stefan Hajnoczi wrote:
> > Michael, Paolo: Are you happy with v4?
>
> Looks pretty clean by itself. I sent some comments but they can be
> addressed later. What worries me most is the code duplication with
> regular virtio.
>
> I see two ways to reduce the maintainance somewhat
> - split out ring handling code in virtio-blk
> to a separate file to make it more obvious which part
> is inactive when data plane runs.
> - share ring processing code with virtio/virtio-blk
> (e.g. use callbacks)
>
> Was any thought given to implementing one of these two
> approaches?
Yes, your option #2 is where I'd like to move once threaded memory
dispatch is working. I hope we can run virtio.c code in a thread
outside the global mutex soon. That way we can kill
hw/dataplane/vring.[ch].
Ping Fan Liu has been working on the memory API and device emulation
stuff that we need in order to eventually use virtio.c outside the
global mutex.
Stefan