From: Igor Mammedov <imammedo@redhat.com>
To: Greg Kurz <groug@kaod.org>
Cc: qemu-devel@nongnu.org, Cornelia Huck <cohuck@redhat.com>,
"Dr. David Alan Gilbert" <dgilbert@redhat.com>,
Markus Armbruster <armbru@redhat.com>,
qemu-ppc@nongnu.org, David Gibson <david@gibson.dropbear.id.au>
Subject: Re: [Qemu-devel] [PATCH 2/2] monitor: add proper reference counting of the current CPU
Date: Mon, 16 Oct 2017 10:07:18 +0200 [thread overview]
Message-ID: <20171016100718.1ec98df6@nial.brq.redhat.com> (raw)
In-Reply-To: <20171013120546.7a89eeaf@bahia.lan>
On Fri, 13 Oct 2017 12:05:46 +0200
Greg Kurz <groug@kaod.org> wrote:
> On Fri, 13 Oct 2017 11:24:59 +0200
> Igor Mammedov <imammedo@redhat.com> wrote:
>
> > On Fri, 13 Oct 2017 10:35:31 +0200
> > Greg Kurz <groug@kaod.org> wrote:
> >
> > > If a CPU selected with the "cpu" command is hot-unplugged then "info cpus"
> > > causes QEMU to exit:
> > >
> > > (qemu) device_del cpu1
> > > (qemu) info cpus
> > > qemu:qemu_cpu_kick_thread: No such process
> > >
> > > This happens because "cpu" stores the pointer to the selected CPU into
> > > the monitor structure. When the CPU is hot-unplugged, we end up with a
> > > dangling pointer. The "info cpus" command then does:
> > >
> > > hmp_info_cpus()
> > > monitor_get_cpu_index()
> > > mon_get_cpu()
> > > cpu_synchronize_state() <--- called with dangling pointer
> > >
> > > This could cause a QEMU crash as well.
> > >
> > > This patch switches the monitor to use object_ref() to ensure the
> > > CPU object doesn't vanish unexpectedly. The reference is dropped
> > > either when "cpu" is used to switch to another CPU, or when the
> > > selected CPU is unrealized and cpu_list_remove() sets its cpu_index
> > > back to UNASSIGNED_CPU_INDEX.
> > I don't really like an idea of leaving dangling cpu around.
> > Is it possible to store QOM path on set_cpu in monitor and
> > resolving it each to instance each time it's needed?
> >
>
> It sounds workable. Also it would allow the fix to not depend on patch 1
> for ppc.
yep,
also if you'd touch these cpu_index based monitor commands,
it might make sense to convert them to use qmo path directly
instead if cpu_index (with command completion it would be
easy to use)
>
> Thanks for the suggestion!
>
> >
> > > Signed-off-by: Greg Kurz <groug@kaod.org>
> > > ---
> > > monitor.c | 12 ++++++++++++
> > > 1 file changed, 12 insertions(+)
> > >
> > > diff --git a/monitor.c b/monitor.c
> > > index fe0d1bdbb461..1c0b9a2c3ad3 100644
> > > --- a/monitor.c
> > > +++ b/monitor.c
> > > @@ -579,6 +579,9 @@ static void monitor_data_init(Monitor *mon)
> > >
> > > static void monitor_data_destroy(Monitor *mon)
> > > {
> > > + if (mon->mon_cpu) {
> > > + object_unref((Object *) mon->mon_cpu);
> > > + }
> > > qemu_chr_fe_deinit(&mon->chr, false);
> > > if (monitor_is_qmp(mon)) {
> > > json_message_parser_destroy(&mon->qmp.parser);
> > > @@ -1047,12 +1050,21 @@ int monitor_set_cpu(int cpu_index)
> > > if (cpu == NULL) {
> > > return -1;
> > > }
> > > + if (cur_mon->mon_cpu) {
> > > + object_unref((Object *) cur_mon->mon_cpu);
> > > + }
> > > cur_mon->mon_cpu = cpu;
> > > + object_ref((Object *) cpu);
> > > return 0;
> > > }
> > >
> > > CPUState *mon_get_cpu(void)
> > > {
> > > + if (cur_mon->mon_cpu &&
> > > + cur_mon->mon_cpu->cpu_index == UNASSIGNED_CPU_INDEX) {
> > > + object_unref((Object *) cur_mon->mon_cpu);
> > > + cur_mon->mon_cpu = NULL;
> > > + }
> > > if (!cur_mon->mon_cpu) {
> > > if (!first_cpu) {
> > > return NULL;
> > >
> > >
> >
>
next prev parent reply other threads:[~2017-10-16 8:07 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-10-13 8:35 [Qemu-devel] [PATCH 0/2] monitor: increase the refcount of the current CPU Greg Kurz
2017-10-13 8:35 ` [Qemu-devel] [PATCH 1/2] spapr_cpu_core: instantiate CPUs separately Greg Kurz
2017-10-13 9:21 ` Igor Mammedov
2017-10-13 9:43 ` Greg Kurz
2017-10-13 8:35 ` [Qemu-devel] [PATCH 2/2] monitor: add proper reference counting of the current CPU Greg Kurz
2017-10-13 9:24 ` Igor Mammedov
2017-10-13 10:05 ` Greg Kurz
2017-10-16 8:07 ` Igor Mammedov [this message]
2017-10-16 17:16 ` Greg Kurz
2017-10-13 8:46 ` [Qemu-devel] [PATCH 0/2] monitor: increase the refcount " Cornelia Huck
2017-10-13 9:14 ` Greg Kurz
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20171016100718.1ec98df6@nial.brq.redhat.com \
--to=imammedo@redhat.com \
--cc=armbru@redhat.com \
--cc=cohuck@redhat.com \
--cc=david@gibson.dropbear.id.au \
--cc=dgilbert@redhat.com \
--cc=groug@kaod.org \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).