DragonFly BSD
DragonFly kernel List (threaded) for 2011-04
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

gsoc: porting virtio device drivers from NetBSD to DragonFly BSD

From: Stéphanie Ouillon <stephanie@xxxxxxxxx>
Date: Wed, 27 Apr 2011 19:54:28 +0200


My project is to port virtio device drivers from NetBSD to DragonFly BSD, so that when DragonFly is a KVM guest, it gets better performances. Some work was already done by Tim Bisson and Pratyush Kshirsagar.

  • Three viritio devices are concerned: the block device, the network device and the memory ballooning device. They are set up through a common virtio driver.
                1) The block device driver has been already ported and it has been tested. I need to clear some portions of the code because                     the driver was written to work independently.
                2) The network device driver exists at the state of a skeleton of driver. It needs to be completed, and tested.
                3) The memory ballooning driver has to be completely ported.

  • Here is what I plan to do first (the sooner the better):
                While looking at the NetBSD code:
                1) Read documentation about some notions (dma, mbuf, I/O routines, ...) to understand the NetBSD code. Read about device                     drivers implementation in FreeBSD.
                2) Get used with virtio specifications.
                3) The block device driver has already been ported to DragonFly and it has been tested: it may be a good starting point to get                     used with this code in order to see how to deal with virtio specifications and some differences between NetBSD and                                 DragonFly BSD.

                After this phase of documentation, I think I will have a better undertanding of some key points of the virtio drivers.

  • After that, I'll test again by myself the block device driver (so I'll be able to check and compare the results).
  • Then, I'll code the network device driver, which would have to be finished and tested by the mid-term.
  • The memory ballooning device will be coded during the second period.

Stéphanie Ouillon

Le 27/04/11 17:58, Pratyush Kshirsagar a écrit :
Congrats, your proposal for the Virtio Drivers is being accepted.
Better you mail brief description of the working status on the mailing list and any questions.
I think Justin mailed about steps needed to be taken.
Also, we would like you to start as soon as possible.

Thank You,
Pratyush Kshirsagar. 

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