Re: [RFC PATCH v3 04/11] contrib/plugins: add plugin showcasing new dicontinuity related API

[Pierrick Bouvier]

On 1/9/25 06:04, Alex Bennée wrote:
> Julian Ganz <neither@nut.email> writes:
>
> > We recently introduced new plugin API for registration of discontinuity
> > related callbacks. This change introduces a minimal plugin showcasing
> > the new API. It simply counts the occurances of interrupts, exceptions
> > and host calls per CPU and reports the counts when exitting.
> > ---
> >   contrib/plugins/meson.build |  3 +-
> >   contrib/plugins/traps.c     | 96 +++++++++++++++++++++++++++++++++++++
> >   2 files changed, 98 insertions(+), 1 deletion(-)
> >   create mode 100644 contrib/plugins/traps.c
> >
> > diff --git a/contrib/plugins/meson.build b/contrib/plugins/meson.build
> > index 63a32c2b4f..9a3015e1c1 100644
> > --- a/contrib/plugins/meson.build
> > +++ b/contrib/plugins/meson.build
> > @@ -1,5 +1,6 @@
> >   contrib_plugins = ['bbv', 'cache', 'cflow', 'drcov', 'execlog', 'hotblocks',
> > -                   'hotpages', 'howvec', 'hwprofile', 'ips', 'stoptrigger']
> > +                   'hotpages', 'howvec', 'hwprofile', 'ips', 'stoptrigger',
> > +                   'traps']
>
> I wonder if this is better in tests/tcg/plugins? We need to do something
> to ensure it gets covered by CI although we might want to be smarter
> about running it together with a test binary that will actually pick up
> something.
>
> >   if host_os != 'windows'
> >     # lockstep uses socket.h
> >     contrib_plugins += 'lockstep'
> <snip>
> > +QEMU_PLUGIN_EXPORT
> > +int qemu_plugin_install(qemu_plugin_id_t id, const qemu_info_t *info,
> > +                        int argc, char **argv)
> > +{
> > +    if (!info->system_emulation) {
> > +        fputs("trap plugin can only be used in system emulation mode.\n",
> > +              stderr);
> > +        return -1;
> > +    }
> > +
> > +    max_vcpus = info->system.max_vcpus;
> > +    traps = qemu_plugin_scoreboard_new(sizeof(TrapCounters));
> > +    qemu_plugin_register_vcpu_init_cb(id, vcpu_init);
> > +    qemu_plugin_vcpu_for_each(id, vcpu_init);
>
> Hmm at first glances this seems redundant - however I guess this is
> covering the use case you load the plugin after the system is up and
> running.
>
> I wonder if you have unearthed a foot-gun in the API that is easy to
> fall into? Maybe we should expand qemu_plugin_register_vcpu_init_cb to
> call the call back immediately for existing vcpus?

I'm not sure to see how we can load a plugin after the cpus have been 
initialized?
In case someone wants to dynamically create a new vcpu_init callback 
(after some event or translation for instance), it should be the 
responsibility of plugin to call it for existing vcpus.

qemu_plugin_vcpu_for_each is still useful in case you want to only 
iterate currently active cpus (and not the one who exited already in 
user mode).

> > +
> > +    qemu_plugin_register_vcpu_discon_cb(id, QEMU_PLUGIN_DISCON_TRAPS,
> > +                                        vcpu_discon);
> > +
> > +    qemu_plugin_register_atexit_cb(id, plugin_exit, NULL);
> > +
> > +    return 0;
> > +}