From: Keir Fraser <keir.xen@gmail.com>
To: Jan Beulich <JBeulich@suse.com>,
xen-devel <xen-devel@lists.xenproject.org>
Cc: George Dunlap <George.Dunlap@eu.citrix.com>
Subject: Re: [PATCH] fix locking in cpu_disable_scheduler()
Date: Mon, 28 Oct 2013 16:24:44 +0100 [thread overview]
Message-ID: <CE94404C.3AFFC%keir.xen@gmail.com> (raw)
In-Reply-To: <526E994E02000078000FD571@nat28.tlf.novell.com>
On 28/10/2013 17:05, "Jan Beulich" <JBeulich@suse.com> wrote:
> So commit eedd6039 ("scheduler: adjust internal locking interface")
> uncovered - by now using proper spin lock constructs - a bug after all:
> When bringing down a CPU, cpu_disable_scheduler() gets called with
> interrupts disabled, and hence the use of vcpu_schedule_lock_irq() was
> never really correct (i.e. the caller ended up with interrupts enabled
> despite having disabled them explicitly).
>
> Fixing this however surfaced another problem: The call path
> vcpu_migrate() -> evtchn_move_pirqs() wants to acquire the event lock,
> which however is a non-IRQ-safe once, and hence check_lock() doesn't
> like this lock to be acquired when interrupts are already off. As we're
> in stop-machine context here, getting things wrong wrt interrupt state
> management during lock acquire/release is out of question though, so
> the simple solution to this appears to be to just suppress spin lock
> debugging for the period of time while the stop machine callback gets
> run.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>
> --- a/xen/common/schedule.c
> +++ b/xen/common/schedule.c
> @@ -601,7 +601,8 @@ int cpu_disable_scheduler(unsigned int c
> {
> for_each_vcpu ( d, v )
> {
> - spinlock_t *lock = vcpu_schedule_lock_irq(v);
> + unsigned long flags;
> + spinlock_t *lock = vcpu_schedule_lock_irqsave(v, &flags);
>
> cpumask_and(&online_affinity, v->cpu_affinity, c->cpu_valid);
> if ( cpumask_empty(&online_affinity) &&
> @@ -622,14 +623,12 @@ int cpu_disable_scheduler(unsigned int c
> if ( v->processor == cpu )
> {
> set_bit(_VPF_migrating, &v->pause_flags);
> - vcpu_schedule_unlock_irq(lock, v);
> + vcpu_schedule_unlock_irqrestore(lock, flags, v);
> vcpu_sleep_nosync(v);
> vcpu_migrate(v);
> }
> else
> - {
> - vcpu_schedule_unlock_irq(lock, v);
> - }
> + vcpu_schedule_unlock_irqrestore(lock, flags, v);
>
> /*
> * A vcpu active in the hypervisor will not be migratable.
> --- a/xen/common/stop_machine.c
> +++ b/xen/common/stop_machine.c
> @@ -110,6 +110,7 @@ int stop_machine_run(int (*fn)(void *),
> local_irq_disable();
> stopmachine_set_state(STOPMACHINE_DISABLE_IRQ);
> stopmachine_wait_state();
> + spin_debug_disable();
>
> stopmachine_set_state(STOPMACHINE_INVOKE);
> if ( (cpu == smp_processor_id()) || (cpu == NR_CPUS) )
> @@ -117,6 +118,7 @@ int stop_machine_run(int (*fn)(void *),
> stopmachine_wait_state();
> ret = stopmachine_data.fn_result;
>
> + spin_debug_enable();
> stopmachine_set_state(STOPMACHINE_EXIT);
> stopmachine_wait_state();
> local_irq_enable();
>
>
>
next prev parent reply other threads:[~2013-10-28 16:25 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-10-28 16:05 [PATCH] fix locking in cpu_disable_scheduler() Jan Beulich
2013-10-28 15:24 ` Keir Fraser [this message]
2013-10-28 16:24 ` Andrew Cooper
2013-10-29 7:39 ` Jan Beulich
2013-10-29 11:30 ` George Dunlap
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=CE94404C.3AFFC%keir.xen@gmail.com \
--to=keir.xen@gmail.com \
--cc=George.Dunlap@eu.citrix.com \
--cc=JBeulich@suse.com \
--cc=xen-devel@lists.xenproject.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).