Re: [Qemu-devel] [PATCH] main: force enabling of I/O thread

[Andreas Färber]

Am 29.08.2011 um 20:03 schrieb Anthony Liguori:

> On 08/22/2011 08:35 AM, Jan Kiszka wrote:
> > On 2011-08-22 15:24, Anthony Liguori wrote:
> > > Enabling the I/O thread by default seems like an important part of  
> > > declaring
> > > 1.0.  Besides allowing true SMP support with KVM, the I/O thread  
> > > means that the
> > > TCG VCPU doesn't have to multiplex itself with the I/O dispatch  
> > > routines which
> > > currently requires a (racey) signal based alarm system.
> > >
> > > I know there have been concerns about performance.  I think so far  
> > > the ones that
> > > have come up (virtio-net) are most likely due to secondary reasons  
> > > like
> > > decreased batching.
> >
> > iothread enabled without [1] gives unacceptable performance for ARM
> > emulation (Musicpal board) here. With that series applied, the host  
> > CPU
> > load is measurably higher, but no longer excessively.
>
> Given that [1] has been applied now, is there any objections to this  
> patch?

There had been additional patches to fix I/O thread issues on Darwin,  
have any such fixes been applied the last few weeks?

The patches that I tested some time ago only fixed the non-iothread  
case reliably.

Andreas