Re: [Qemu-devel] [RFC v3 12/19] tcg: add kick timer for single-threaded vCPU emulation

[Richard Henderson]

On 06/27/2016 02:20 PM, Sergey Fedorov wrote:
> On 03/06/16 23:40, Alex Bennée wrote:
> > diff --git a/cpus.c b/cpus.c
> > index 1694ce9..12e04c9 100644
> > --- a/cpus.c
> > +++ b/cpus.c
> > @@ -1208,9 +1208,29 @@ static int tcg_cpu_exec(CPUState *cpu)
> >      return ret;
> >  }
> >
> > +/* Single-threaded TCG
> > + *
> > + * In the single-threaded case each vCPU is simulated in turn. If
> > + * there is more than a single vCPU we create a simple timer to kick
> > + * the vCPU and ensure we don't get stuck in a tight loop in one vCPU.
> > + * This is done explicitly rather than relying on side-effects
> > + * elsewhere.
> > + */
> > +static void qemu_cpu_kick_no_halt(void);
> > +#define TCG_KICK_FREQ (qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + \
> > +                       NANOSECONDS_PER_SECOND / 10)
>
> Hmm, it doesn't look nice to wrap calculation of the next timeout in a
> macro and name it '*_FREQ'. I think we'd better do like this:
>
> #define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
>
> static inline int64_t qemu_tcg_next_kick(void)
> {
>     return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
> }
>
> and use it like this:
>
> timer_mod(kick_timer, qemu_tcg_next_kick());

Agreed.

As an aside, surely a period of 10ns is too small.
That's on the order of 20-50 host instructions.


r~