linux-kernel - Re: [PATCH 2/3] KVM: nVMX: Fix fail to get nested ack intr's vector during nested vmexit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <53C78E37.5090807@redhat.com>
Date:	Thu, 17 Jul 2014 10:49:59 +0200
From:	Paolo Bonzini <pbonzini@...hat.com>
To:	Wanpeng Li <wanpeng.li@...ux.intel.com>,
	Jan Kiszka <jan.kiszka@...mens.com>
CC:	Marcelo Tosatti <mtosatti@...hat.com>,
	Gleb Natapov <gleb@...nel.org>, Bandan Das <bsd@...hat.com>,
	Zhang Yang <yang.z.zhang@...el.com>,
	Hu Robert <robert.hu@...el.com>, kvm@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/3] KVM: nVMX: Fix fail to get nested ack intr's vector
 during nested vmexit

Il 17/07/2014 06:56, Wanpeng Li ha scritto:
>  	    && nested_exit_intr_ack_set(vcpu)) {
>  		int irq = kvm_cpu_get_interrupt(vcpu);
> +
> +		if (irq < 0 && kvm_apic_vid_enabled(vcpu->kvm))
> +			irq = kvm_get_apic_interrupt(vcpu);

There's something weird in this patch.  If you "inline" 
kvm_cpu_get_interrupt, what you get is this:

         int irq;

	/* Beginning of kvm_cpu_get_interrupt... */
         if (!irqchip_in_kernel(v->kvm))
                 irq = v->arch.interrupt.nr;
	else {
	        irq = kvm_cpu_get_extint(v); /* PIC */
	        if (!kvm_apic_vid_enabled(v->kvm) && irq == -1)
			irq = kvm_get_apic_interrupt(v); /* APIC */
	}

	/* kvm_cpu_get_interrupt done. */
	if (irq < 0 && kvm_apic_vid_enabled(vcpu->kvm))
		irq = kvm_get_apic_interrupt(vcpu);

There are just two callers of kvm_cpu_get_interrupt, and the other is 
protected by kvm_cpu_has_injectable_intr so it won't be executed if 
virtual interrupt delivery is enabled.  So you patch is effectively the 
same as this:

diff --git a/arch/x86/kvm/irq.c b/arch/x86/kvm/irq.c
index bd0da43..a1ec6a5 100644
--- a/arch/x86/kvm/irq.c
+++ b/arch/x86/kvm/irq.c
@@ -108,7 +108,7 @@ int kvm_cpu_get_interrupt(struct kvm_vcpu *v)

  	vector = kvm_cpu_get_extint(v);

-	if (kvm_apic_vid_enabled(v->kvm) || vector != -1)
+	if (vector != -1)
  		return vector;			/* PIC */

  	return kvm_get_apic_interrupt(v);	/* APIC */

But in kvm_get_apic_interrupt I have just added this comment:

         /* Note that we never get here with APIC virtualization
	 * enabled.  */

because kvm_get_apic_interrupt calls apic_set_isr, and apic_set_isr must 
never be called with APIC virtualization enabled either.  With APIC 
virtualization enabled, isr_count is always 1, and highest_isr_cache is 
always -1, and apic_set_isr breaks both of these invariants.

Paolo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/