Re: [Qemu-devel] [PATCH 12/15] qtest: add support for target-i386 -M pc

[Anthony Liguori]

On 01/10/2012 01:56 PM, Paolo Bonzini wrote:
> On 01/10/2012 08:10 PM, Anthony Liguori wrote:
> > This involves forcing the CPU into the halted state if qtest is enabled and
> > replacing the local APIC with the qtest interrupt controller.
> >
> > It should be pretty straight forward to do the same for other machine types on
> > other architectures.
> >
> > Signed-off-by: Anthony Liguori<address@hidden>
> > ---
> > hw/pc.c | 7 ++++++-
> > hw/pc_piix.c | 9 ++++++---
> > 2 files changed, 12 insertions(+), 4 deletions(-)
> >
> > diff --git a/hw/pc.c b/hw/pc.c
> > index 85304cf..fac5098 100644
> > --- a/hw/pc.c
> > +++ b/hw/pc.c
> > @@ -43,6 +43,7 @@
> > #include "ui/qemu-spice.h"
> > #include "memory.h"
> > #include "exec-memory.h"
> > +#include "qtest.h"
> >
> > /* output Bochs bios info messages */
> > //#define DEBUG_BIOS
> > @@ -926,7 +927,11 @@ static void pc_cpu_reset(void *opaque)
> > CPUState *env = opaque;
> >
> > cpu_reset(env);
> > - env->halted = !cpu_is_bsp(env);
> > + if (qtest_enabled()) {
> > + env->halted = 1;
> > + } else {
> > + env->halted = !cpu_is_bsp(env);
> > + }
>
> This is wrong. qtest and Xen should simply not create the CPU threads at all.

This is easier said than done.  I started down this road and there's a huge 
amount of code that assumes that first_cpu != NULL.

I agree it's where we want to go though.

Regards,

Anthony Liguori

> Paolo