Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1195174 > unrolled thread
| Started by | Paolo Bonzini <pbonzini@redhat.com> |
|---|---|
| First post | 2015-07-29 15:30 +0200 |
| Last post | 2015-07-30 09:30 +0200 |
| Articles | 3 — 2 participants |
Back to article view | Back to linux.kernel
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
[PATCH 4/4] KVM: x86: clean/fix memory barriers in irqchip_in_kernel Paolo Bonzini <pbonzini@redhat.com> - 2015-07-29 15:30 +0200
Re: [PATCH 4/4] KVM: x86: clean/fix memory barriers in irqchip_in_kernel Steve Rutherford <srutherford@google.com> - 2015-07-30 05:40 +0200
Re: [PATCH 4/4] KVM: x86: clean/fix memory barriers in irqchip_in_kernel Paolo Bonzini <pbonzini@redhat.com> - 2015-07-30 09:30 +0200
| From | Paolo Bonzini <pbonzini@redhat.com> |
|---|---|
| Date | 2015-07-29 15:30 +0200 |
| Subject | [PATCH 4/4] KVM: x86: clean/fix memory barriers in irqchip_in_kernel |
| Message-ID | <pRzAK-Um-11@gated-at.bofh.it> |
The memory barriers are trying to protect against concurrent RCU-based
interrupt injection, but the IRQ routing table is not valid at the time
kvm->arch.vpic is written. Fix this by writing kvm->arch.vpic last.
kvm_destroy_pic then need not set kvm->arch.vpic to NULL; modify it
to take a struct kvm_pic* and reuse it if the IOAPIC creation fails.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
arch/x86/kvm/i8259.c | 15 +++++----------
arch/x86/kvm/irq.h | 8 ++++----
arch/x86/kvm/x86.c | 17 ++++++-----------
3 files changed, 15 insertions(+), 25 deletions(-)
diff --git a/arch/x86/kvm/i8259.c b/arch/x86/kvm/i8259.c
index fef922ff2635..7cc2360f1848 100644
--- a/arch/x86/kvm/i8259.c
+++ b/arch/x86/kvm/i8259.c
@@ -651,15 +651,10 @@ fail_unlock:
return NULL;
}
-void kvm_destroy_pic(struct kvm *kvm)
+void kvm_destroy_pic(struct kvm_pic *vpic)
{
- struct kvm_pic *vpic = kvm->arch.vpic;
-
- if (vpic) {
- kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS, &vpic->dev_master);
- kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS, &vpic->dev_slave);
- kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS, &vpic->dev_eclr);
- kvm->arch.vpic = NULL;
- kfree(vpic);
- }
+ kvm_io_bus_unregister_dev(vpic->kvm, KVM_PIO_BUS, &vpic->dev_master);
+ kvm_io_bus_unregister_dev(vpic->kvm, KVM_PIO_BUS, &vpic->dev_slave);
+ kvm_io_bus_unregister_dev(vpic->kvm, KVM_PIO_BUS, &vpic->dev_eclr);
+ kfree(vpic);
}
diff --git a/arch/x86/kvm/irq.h b/arch/x86/kvm/irq.h
index ad68c73008c5..3d782a2c336a 100644
--- a/arch/x86/kvm/irq.h
+++ b/arch/x86/kvm/irq.h
@@ -74,7 +74,7 @@ struct kvm_pic {
};
struct kvm_pic *kvm_create_pic(struct kvm *kvm);
-void kvm_destroy_pic(struct kvm *kvm);
+void kvm_destroy_pic(struct kvm_pic *vpic);
int kvm_pic_read_irq(struct kvm *kvm);
void kvm_pic_update_irq(struct kvm_pic *s);
@@ -85,11 +85,11 @@ static inline struct kvm_pic *pic_irqchip(struct kvm *kvm)
static inline int irqchip_in_kernel(struct kvm *kvm)
{
- int ret;
+ struct kvm_pic *vpic = pic_irqchip(kvm);
- ret = (pic_irqchip(kvm) != NULL);
+ /* Read vpic before kvm->irq_routing. */
smp_rmb();
- return ret;
+ return vpic != NULL;
}
void kvm_pic_reset(struct kvm_kpic_state *s);
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 2d62229aac26..23e47a0b054b 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3626,30 +3626,25 @@ long kvm_arch_vm_ioctl(struct file *filp,
r = kvm_ioapic_init(kvm);
if (r) {
mutex_lock(&kvm->slots_lock);
- kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS,
- &vpic->dev_master);
- kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS,
- &vpic->dev_slave);
- kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS,
- &vpic->dev_eclr);
+ kvm_destroy_pic(vpic);
mutex_unlock(&kvm->slots_lock);
- kfree(vpic);
goto create_irqchip_unlock;
}
} else
goto create_irqchip_unlock;
- smp_wmb();
- kvm->arch.vpic = vpic;
- smp_wmb();
r = kvm_setup_default_irq_routing(kvm);
if (r) {
mutex_lock(&kvm->slots_lock);
mutex_lock(&kvm->irq_lock);
kvm_ioapic_destroy(kvm);
- kvm_destroy_pic(kvm);
+ kvm_destroy_pic(vpic);
mutex_unlock(&kvm->irq_lock);
mutex_unlock(&kvm->slots_lock);
+ goto create_irqchip_unlock;
}
+ /* Write kvm->irq_routing before kvm->arch.vpic. */
+ smp_wmb();
+ kvm->arch.vpic = vpic;
create_irqchip_unlock:
mutex_unlock(&kvm->lock);
break;
--
1.8.3.1
--
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/
[toc] | [next] | [standalone]
| From | Steve Rutherford <srutherford@google.com> |
|---|---|
| Date | 2015-07-30 05:40 +0200 |
| Subject | Re: [PATCH 4/4] KVM: x86: clean/fix memory barriers in irqchip_in_kernel |
| Message-ID | <pRMRk-365-15@gated-at.bofh.it> |
| In reply to | #1195174 |
On Wed, Jul 29, 2015 at 03:28:58PM +0200, Paolo Bonzini wrote:
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 2d62229aac26..23e47a0b054b 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -3626,30 +3626,25 @@ long kvm_arch_vm_ioctl(struct file *filp,
> r = kvm_ioapic_init(kvm);
> if (r) {
> mutex_lock(&kvm->slots_lock);
> - kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS,
> - &vpic->dev_master);
> - kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS,
> - &vpic->dev_slave);
> - kvm_io_bus_unregister_dev(kvm, KVM_PIO_BUS,
> - &vpic->dev_eclr);
> + kvm_destroy_pic(vpic);
> mutex_unlock(&kvm->slots_lock);
> - kfree(vpic);
> goto create_irqchip_unlock;
> }
> } else
> goto create_irqchip_unlock;
> - smp_wmb();
> - kvm->arch.vpic = vpic;
> - smp_wmb();
> r = kvm_setup_default_irq_routing(kvm);
> if (r) {
> mutex_lock(&kvm->slots_lock);
> mutex_lock(&kvm->irq_lock);
> kvm_ioapic_destroy(kvm);
> - kvm_destroy_pic(kvm);
> + kvm_destroy_pic(vpic);
> mutex_unlock(&kvm->irq_lock);
> mutex_unlock(&kvm->slots_lock);
> + goto create_irqchip_unlock;
> }
> + /* Write kvm->irq_routing before kvm->arch.vpic. */
> + smp_wmb();
I assume this pairs with irqchip_in_kernel?
> + kvm->arch.vpic = vpic;
> create_irqchip_unlock:
> mutex_unlock(&kvm->lock);
> break;
> --
> 1.8.3.1
>
--
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/
[toc] | [prev] | [next] | [standalone]
| From | Paolo Bonzini <pbonzini@redhat.com> |
|---|---|
| Date | 2015-07-30 09:30 +0200 |
| Subject | Re: [PATCH 4/4] KVM: x86: clean/fix memory barriers in irqchip_in_kernel |
| Message-ID | <pRQrU-8ly-9@gated-at.bofh.it> |
| In reply to | #1195701 |
On 30/07/2015 05:32, Steve Rutherford wrote:
>> > + /* Write kvm->irq_routing before kvm->arch.vpic. */
>> > + smp_wmb();
> I assume this pairs with irqchip_in_kernel?
Yes, see the comment added there by this same patch ("read
kvm->arch.vpic before kvm->irq_routing").
Paolo
--
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/
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web