Re: [RFC][PATCH] x86, reboot: use NMI instead of REBOOT_VECTOR to stop cpus

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Thursday, October 06, 2011, Don Zickus wrote:
> On Wed, Sep 28, 2011 at 02:55:48PM -0400, Don Zickus wrote:
> > A recent discussion started talking about the locking on the pstore fs
> > and how it relates to the kmsg infrastructure.  We noticed it was possible
> > for userspace to r/w to the pstore fs (grabbing the locks in the process)
> > and block the panic path from r/w to the same fs.
> 
> Poke?  Anyone have an opinion on this?

I think Matthew Garrett should have a look at this (now CCed).

> > The reason was the cpu with the lock could be doing work while the crashing
> > cpu is panic'ing.  Busting those spinlocks might cause those cpus to step
> > on each other's data.  Fine, fair enough.
> > 
> > It was suggested it would be nice to serialize the panic path (ie stop
> > the other cpus) and have only one cpu running.  This would allow us to
> > bust the spinlocks and not worry about another cpu stepping on the data.
> > 
> > Of course, smp_send_stop() does this in the panic case.  kmsg_dump() would
> > have to be moved to be called after it.  Easy enough.
> > 
> > The only problem is on x86 the smp_send_stop() function calls the
> > REBOOT_VECTOR.  Any cpu with irqs disabled (which pstore and its backend
> > ERST would do), block this IPI and thus do not stop.  This makes it
> > difficult to reliably log data to the pstore fs.
> > 
> > The patch below switches from the REBOOT_VECTOR to NMI (and mimics what
> > kdump does).  Switching to NMI allows us to deliver the IPI when irqs are
> > disabled, increasing the reliability of this function.
> > 
> > However, Andi carefully noted that on some machines this approach does not
> > work because of broken BIOSes or whatever.
> > 
> > I was hoping to get feedback on how much of a problem this really is.  Are
> > there that many machines?  I assume most modern machines have a reliable NMI
> > IPI mechanism (well on x86).  Is this just a problem on 32-bit machines?
> > Early SMP machines?
> > 
> > One idea I had was to create a blacklist of machines and have those machines
> > fallback to the original native_stop_other_cpus() that Andi wrote originally.
> > The hope was that list was small.
> > 
> > Does anyone have any feedback whether this is a good idea or not?  Perhaps I am
> > missing something?  Perhaps I should approach this problem differently?
> > 
> > [note] this patch sits on top of another NMI infrastructure change I have
> > submitted, so the nmi registeration might not apply cleanly without that patch.
> > However, for discussion purposes, I don't think that change is relevant, it is
> > more the idea/philosophy of this patch that I am worried about.
> > 
> > Thanks,
> > Don
> > ---
> >  arch/x86/kernel/smp.c |   56 +++++++++++++++++++++++++++++++++++++++++++++++-
> >  1 files changed, 54 insertions(+), 2 deletions(-)
> > 
> > diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c
> > index 013e7eb..e98f0a1 100644
> > --- a/arch/x86/kernel/smp.c
> > +++ b/arch/x86/kernel/smp.c
> > @@ -28,6 +28,7 @@
> >  #include <asm/mmu_context.h>
> >  #include <asm/proto.h>
> >  #include <asm/apic.h>
> > +#include <asm/nmi.h>
> >  /*
> >   *	Some notes on x86 processor bugs affecting SMP operation:
> >   *
> > @@ -147,6 +148,57 @@ void native_send_call_func_ipi(const struct cpumask *mask)
> >  	free_cpumask_var(allbutself);
> >  }
> >  
> > +static int stopping_cpu;
> > +
> > +static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
> > +{
> > +	/* We are registerd on stopping cpu too, avoid spurious NMI */
> > +	if (raw_smp_processor_id() == stopping_cpu)
> > +		return NMI_HANDLED;
> > +
> > +	stop_this_cpu(NULL);
> > +
> > +	return NMI_HANDLED;
> > +}
> > +
> > +static void native_nmi_stop_other_cpus(int wait)
> > +{
> > +	unsigned long flags;
> > +	unsigned long timeout;
> > +
> > +	if (reboot_force)
> > +		return;
> > +
> > +	/*
> > +	 * Use an own vector here because smp_call_function
> > +	 * does lots of things not suitable in a panic situation.
> > +	 */
> > +	if (num_online_cpus() > 1) {
> > +		stopping_cpu = safe_smp_processor_id();
> > +
> > +		if (register_nmi_handler(NMI_LOCAL, smp_stop_nmi_callback,
> > +					 NMI_FLAG_FIRST, "smp_stop"))
> > +			return;		/* return what? */
> > +
> > +		/* sync above data before sending NMI */
> > +		wmb();
> > +
> > +		apic->send_IPI_allbutself(NMI_VECTOR);
> > +
> > +		/*
> > +		 * Don't wait longer than a second if the caller
> > +		 * didn't ask us to wait.
> > +		 */
> > +		timeout = USEC_PER_SEC;
> > +		while (num_online_cpus() > 1 && (wait || timeout--))
> > +			udelay(1);
> > +	}
> > +
> > +	local_irq_save(flags);
> > +	disable_local_APIC();
> > +	local_irq_restore(flags);
> > +}
> > +
> >  /*
> >   * this function calls the 'stop' function on all other CPUs in the system.
> >   */
> > @@ -159,7 +211,7 @@ asmlinkage void smp_reboot_interrupt(void)
> >  	irq_exit();
> >  }
> >  
> > -static void native_stop_other_cpus(int wait)
> > +static void native_irq_stop_other_cpus(int wait)
> >  {
> >  	unsigned long flags;
> >  	unsigned long timeout;
> > @@ -229,7 +281,7 @@ struct smp_ops smp_ops = {
> >  	.smp_prepare_cpus	= native_smp_prepare_cpus,
> >  	.smp_cpus_done		= native_smp_cpus_done,
> >  
> > -	.stop_other_cpus	= native_stop_other_cpus,
> > +	.stop_other_cpus	= native_nmi_stop_other_cpus,
> >  	.smp_send_reschedule	= native_smp_send_reschedule,
> >  
> >  	.cpu_up			= native_cpu_up,
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
>