Re: [PATCH] x86/irq: Plug vector setup race

Re: [PATCH] x86/irq: Plug vector setup race

[Thomas Gleixner]

On Thu, Jul 24 2025 at 12:49, Thomas Gleixner wrote:

Hogan!

> Hogan reported a vector setup race, which overwrites the interrupt
> descriptor in the per CPU vector array resulting in a disfunctional device.
>
> CPU0				CPU1
> 				interrupt is raised in APIC IRR
> 				but not handled
>   free_irq()
>     per_cpu(vector_irq, CPU1)[vector] = VECTOR_SHUTDOWN;
>
>   request_irq()			common_interrupt()
>   				  d = this_cpu_read(vector_irq[vector]);
>
>     per_cpu(vector_irq, CPU1)[vector] = desc;
>
>     				  if (d == VECTOR_SHUTDOWN)
> 				    this_cpu_write(vector_irq[vector], VECTOR_UNUSED);
>
> free_irq() cannot observe the pending vector in the CPU1 APIC as there is
> no way to query the remote CPUs APIC IRR.
>
> This requires that request_irq() uses the same vector/CPU as the one which
> was freed, but this also can be triggered by a spurious interrupt.
>
> Prevent this by reevaluating vector_irq under the vector lock, which is
> held by the interrupt activation code when vector_irq is updated.

Does this fix your problem?

Thanks,

        tglx

Re: [PATCH] x86/irq: Plug vector setup race

[Hogan Wang]

> On Thu, Jul 24 2025 at 12:49, Thomas Gleixner wrote:
> 

Thank you very much for your professional, friendly, and detailed response.
Based on the clear modification suggestions you provided, I conducted
retesting and validation using the following methods:

1) Start a virtual machine with 192U384G specification, and configure
   one VirtioNet network card and one VirtioSCSI disk.
2) After the virtual machine starts successfully, execute the following
   script inside the virtual machine. The interrupt number 30 is the
   VirtioNet MSI-x interrupt.

for((;;))
    do (for((i=0;i<192;i++))
	    do echo $i > /proc/irq/30/smp_affinity_list
		sleep 0.1
	done)
done

After a 7x24-hour test, no error logs of the type "No irq handler for
vector" were found, I believe this issue should have already been
resolved. 

As you said, this fix cannot solve the problem of lost interrupts.

I believe an effective solution to the issue of lost interrupts
might be to modify the vifo module to avoid un-plug/plug irq,
and instead use a more lightweight method to switch interrupt
modes. Just like:

vfio_irq_handler()
	if kvm_mode
		vfio_send_eventfd(kvm_irq_fd);
	else
		vfio_send_eventfd(qemu_irq_fd);

However, this will bring about some troubles:
1) The kvm_mode variable should be protected, leading to performance loss. 
2) The VFIO interface requires the passing of two eventfds. 
3) Add another interface to implement mode switching. 

Do you have a better solution to fix this interrupt loss issue? 
	
There is a question that has been troubling me: Why are interrupts
still reported after they have been masked and the interrupt remapping
table entries have been disabled? Is this interrupt cached somewhere?

> Hogan!
> 
> > Hogan reported a vector setup race, which overwrites the interrupt 
> > descriptor in the per CPU vector array resulting in a disfunctional device.
> >
> > CPU0				CPU1
> > 				interrupt is raised in APIC IRR
> > 				but not handled
> >   free_irq()
> >     per_cpu(vector_irq, CPU1)[vector] = VECTOR_SHUTDOWN;
> >
> >   request_irq()			common_interrupt()
> >   				  d = this_cpu_read(vector_irq[vector]);
> >
> >     per_cpu(vector_irq, CPU1)[vector] = desc;
> >
> >     				  if (d == VECTOR_SHUTDOWN)
> > 				    this_cpu_write(vector_irq[vector], VECTOR_UNUSED);
> >
> > free_irq() cannot observe the pending vector in the CPU1 APIC as there 
> > is no way to query the remote CPUs APIC IRR.
> >
> > This requires that request_irq() uses the same vector/CPU as the one 
> > which was freed, but this also can be triggered by a spurious interrupt.
> >
> > Prevent this by reevaluating vector_irq under the vector lock, which 
> > is held by the interrupt activation code when vector_irq is updated.
> 
> Does this fix your problem?

Thanks,

		Hogan
-- 
2.45.1

Re: [PATCH] x86/irq: Plug vector setup race

[Thomas Gleixner]

On Fri, Aug 01 2025 at 22:56, Hogan Wang wrote:
> I believe an effective solution to the issue of lost interrupts
> might be to modify the vifo module to avoid un-plug/plug irq,
> and instead use a more lightweight method to switch interrupt
> modes. Just like:
>
> vfio_irq_handler()
> 	if kvm_mode
> 		vfio_send_eventfd(kvm_irq_fd);
> 	else
> 		vfio_send_eventfd(qemu_irq_fd);
>
> However, this will bring about some troubles:
> 1) The kvm_mode variable should be protected, leading to performance loss. 
> 2) The VFIO interface requires the passing of two eventfds. 
> 3) Add another interface to implement mode switching. 
>
> Do you have a better solution to fix this interrupt loss issue?

Interesting. I looked at vfio_irq_handler(), which is in the platform/
part of VFIO. The corresponding vfio_set_trigger(), which switches eventfds
does the right thing:

     disable_irq();
     update(trigger);
     enable_irq();

disable_irq() ensures that there is no interrupt handler in progress, so
it becomes safe to switch the trigger in the data structure which is has
been handed to request_irq() as @dev_id argument. For edge type
interupts this ensures that a interrupt which arrives while disabled is
retriggered on enable, so that no interrupt can get lost.

The PCI variant is using the trigger itself as the @dev_id argument and
therefore has to do the free_irq()/request_irq() dance. It shouldn't be
hard to convert the PCI implementation over to the disable/enable scheme.

> There is a question that has been troubling me: Why are interrupts
> still reported after they have been masked and the interrupt remapping
> table entries have been disabled? Is this interrupt cached somewhere?

Let me bring back the picture I used before:

	 CPU0				CPU1
	 vmenter(vCPU0)
	 ....                           local_irq_disable()
	  msi_set_affinity()
 #1	    mask(MSI-X)
	      vmexit()                  
 #2      ...                             interrupt is raised in APIC
                                         but not handled

 #3      really_mask(MSI-X)
         free_irq()
 	   mask();        

 #4	   __synchronize_irq()

	   msi_domain_deactivate()
	     write_msg(0);
	   x86_vector_deactivate()
 #5          per_cpu(vector_irq, cpu)[vector] = VECTOR_SHUTDOWN;

 #6                                     local_irq_enable()
                                         interrupt is handled and
					 observes VECTOR_SHUTDOWN
					 writes VECTOR_UNUSED
	request_irq()
	  x86_vector_activate()
	     per_cpu(vector_irq, cpu)[vector] = desc;

	   msi_domain_deactivate()
	     write_msg(msg);
	   unmask();

#1 is the mask operation in the VM, which is trapped, i.e. the interrupt
   is not yet masked at the MSIX level.

#2 The device raises the interupt _before_ the host can mask the
   interrupt at the PCI-MSIX level (#3).

   The interrupt is sent to the APIC of the target CPU 1, which sets the
   corresponding IRR bit in the APIC if the CPU cannot handle it at that
   point, because it has interrupts disabled.

#4 cannot observe the pending IRR bit on CPU1's APIC and therefore
   concludes that there is no interrupt in flight.

If the host side VMM manages to shut down the interrupt completely (#5)
_before_ CPU1 reenables interrupts (#6), then CPU1 will observe
VECTOR_SHUTDOWN and treats it as a spurious interrupt.

The same problem exists on bare metal, when a driver leaves the device
interrupts enabled and then does a free/request dance:

	 CPU0				CPU1
	 ....                           local_irq_disable()
 #1	 free_irq()
 #2      ...                             interrupt is raised in APIC
                                         but not handled

 #3       really_mask(MSI-X)

 #4	   __synchronize_irq()

	   msi_domain_deactivate()
	     write_msg(0);
	   x86_vector_deactivate()
 #5          per_cpu(vector_irq, cpu)[vector] = VECTOR_SHUTDOWN;

 #6                                     local_irq_enable()
                                         interrupt is handled and
					 observes VECTOR_SHUTDOWN
					 writes VECTOR_UNUSED
	request_irq()
	  x86_vector_activate()
	     per_cpu(vector_irq, cpu)[vector] = desc;

	   msi_domain_deactivate()
	     write_msg(msg);
	   unmask();

See?

Thanks,

        tglx