[Qemu-devel] Re: [patch 2/7] qemu: separate thread for io

[Anthony Liguori]

Marcelo Tosatti wrote:
> There was a significant (25% IIRC) reduction in iperf performance. This
> is sort of expected, since there are no optimizations at all (should
> collapse the signals sent to TCG context, for one). But my thinking is
> to merge the iothread (along the lines of this patchset), stabilize and
> then optimize.
>
> How about that?
>   

I don't mind that but I'll need to check out how things behave with TCG 
and with other guest architectures.  I'm willing to accept a temporary 
performance regression in something like iperf because I don't think 
people are relying on QEMU network performance that much today (outside 
of KVM/Xen) but if we have significant performance regressions with 
non-x86 boards with things like basic boot time, we'll have to fix those 
first.

We also need to not completely break the Windows build before merging.

> > Have you thought about how this is going to affect kvm-userspace?  
> >     
>
> Oops, no. But I can be held accountable for kvm-userspace iothread until
> it can be fully replaced by upstream.
>   

Before merging into QEMU, we should at least make sure that it's not 
going to create a nightmare for Avi :-)

> > Do you think we can eliminate the io threading code in kvm-userspace
> > after this goes in?
> >     
>
> Not immediately, need to generalize some of the changes introduced
> by the patchset and merge the remaining logic of kvm-userspace's
> qemu-kvm.c.
>   

Yeah, I thought it would take some work.  Great work Marcelo!

Regards,

Anthony Liguori