[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] Some thoughts about per-queue iothreads in VirtIOBlock
From: |
Paolo Bonzini |
Subject: |
Re: [Qemu-devel] Some thoughts about per-queue iothreads in VirtIOBlock |
Date: |
Tue, 20 Mar 2018 16:34:01 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.6.0 |
On 20/03/2018 15:45, Edgar Kaziakhmedov wrote:
> Hi!
>
> Recently, I have started working on the real multi-queue implementation
> for the VirtIOBlock device and came to conclusion
> that for the time being it may work only with RAW images. The reason is
> obviously lack of the true thread-safe multi-queue block
> layer, but recently I have seen the following message from Stefan
>
> https://lists.gnu.org/archive/html/qemu-devel/2018-02/msg04791.html
>
> It appears like Paolo has already started implementing this feature, so
> I wonder whether the work is abandoned or not, and
> could anyone point me out to any of those patches
I have almost finished getting rid of the AioContext lock
(aio_context_acquire and aio_context_release), which is the final step
of thread-safety and a prerequisite for multiqueue. Unfortunately it's
stalled a bit due to all the Meltdown and Spectre mess, which yes is
still going on after some months. :(
> As I understood from
> Stefan description, it is expected to change significantly
> the approach we use to interact with BlockDriverState.
I don't think the change is very large actually, at least for on-disk
devices. I haven't yet considered network devices such as libiscsi or ceph.
Paolo