All of lore.kernel.org
 help / color / mirror / Atom feed
From: Marcelo Tosatti <mtosatti@redhat.com>
To: "Yang, Sheng" <sheng.yang@intel.com>
Cc: "kvm@vger.kernel.org" <kvm@vger.kernel.org>,
	Avi Kivity <avi@redhat.com>, Chris Wright <chrisw@redhat.com>
Subject: Re: KVM: protect assigned dev workqueue, int handler and irq acker
Date: Fri, 27 Feb 2009 14:54:31 -0300	[thread overview]
Message-ID: <20090227175431.GA10839@amt.cnet> (raw)
In-Reply-To: <200902271217.01918.sheng.yang@intel.com>

On Fri, Feb 27, 2009 at 12:17:01PM +0800, Yang, Sheng wrote:
> On Friday 27 February 2009 07:50:54 Marcelo Tosatti wrote:
> > Can someone with HW test this please?
> 
> Good catch! The patch works fine on my side.
> 
> Can it be a per-device lock? One big lock for all assigned device seems 
> restrict scalability. 

Since all state is per-device, yes.

Can you please review, test and ack the patch below?

Thanks.

> > -----
> >
> > kvm_assigned_dev_ack_irq is vulnerable to a race condition with the
> > interrupt handler function. It does:
> >
> >         if (dev->host_irq_disabled) {
> >                 enable_irq(dev->host_irq);
> >                 dev->host_irq_disabled = false;
> > 	}
> >
> > If an interrupt triggers before the host->dev_irq_disabled assignment,
> > it will disable the interrupt and set dev->host_irq_disabled to true.
> >
> > On return to kvm_assigned_dev_ack_irq, dev->host_irq_disabled is set to
> > false, and the next kvm_assigned_dev_ack_irq call will fail to reenable
> > it.
> >
> > Other than that, having the interrupt handler and work handlers run in
> > parallel sounds like asking for trouble (could not spot any obvious
> > problem, but better not have to, its fragile).
> 
> Well, my original purpose is a FIFO between interrupt handler and work(for 
> MSI-X), but seems too complex... And I also don't see any problem for now...
> 
> -- 
> regards
> Yang, Sheng


diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 3832243..16bf60b 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -349,6 +349,7 @@ struct kvm_assigned_dev_kernel {
 	int flags;
 	struct pci_dev *dev;
 	struct kvm *kvm;
+	spinlock_t assigned_dev_lock;
 };
 
 struct kvm_irq_mask_notifier {
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 4d2be16..b7d060f 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -41,6 +41,7 @@
 #include <linux/pagemap.h>
 #include <linux/mman.h>
 #include <linux/swap.h>
+#include <linux/spinlock.h>
 
 #include <asm/processor.h>
 #include <asm/io.h>
@@ -132,6 +133,7 @@ static void kvm_assigned_dev_interrupt_work_handler(struct work_struct *work)
 	 * finer-grained lock, update this
 	 */
 	mutex_lock(&kvm->lock);
+	spin_lock_irq(&assigned_dev->assigned_dev_lock);
 	if (assigned_dev->irq_requested_type & KVM_ASSIGNED_DEV_MSIX) {
 		struct kvm_guest_msix_entry *guest_entries =
 			assigned_dev->guest_msix_entries;
@@ -158,18 +160,21 @@ static void kvm_assigned_dev_interrupt_work_handler(struct work_struct *work)
 		}
 	}
 
+	spin_unlock_irq(&assigned_dev->assigned_dev_lock);
 	mutex_unlock(&assigned_dev->kvm->lock);
 }
 
 static irqreturn_t kvm_assigned_dev_intr(int irq, void *dev_id)
 {
+	unsigned long flags;
 	struct kvm_assigned_dev_kernel *assigned_dev =
 		(struct kvm_assigned_dev_kernel *) dev_id;
 
+	spin_lock_irqsave(&assigned_dev->assigned_dev_lock, flags);
 	if (assigned_dev->irq_requested_type == KVM_ASSIGNED_DEV_MSIX) {
 		int index = find_index_from_host_irq(assigned_dev, irq);
 		if (index < 0)
-			return IRQ_HANDLED;
+			goto out;
 		assigned_dev->guest_msix_entries[index].flags |=
 			KVM_ASSIGNED_MSIX_PENDING;
 	}
@@ -179,6 +184,8 @@ static irqreturn_t kvm_assigned_dev_intr(int irq, void *dev_id)
 	disable_irq_nosync(irq);
 	assigned_dev->host_irq_disabled = true;
 
+out:
+	spin_unlock_irqrestore(&assigned_dev->assigned_dev_lock, flags);
 	return IRQ_HANDLED;
 }
 
@@ -186,6 +193,7 @@ static irqreturn_t kvm_assigned_dev_intr(int irq, void *dev_id)
 static void kvm_assigned_dev_ack_irq(struct kvm_irq_ack_notifier *kian)
 {
 	struct kvm_assigned_dev_kernel *dev;
+	unsigned long flags;
 
 	if (kian->gsi == -1)
 		return;
@@ -198,10 +206,12 @@ static void kvm_assigned_dev_ack_irq(struct kvm_irq_ack_notifier *kian)
 	/* The guest irq may be shared so this ack may be
 	 * from another device.
 	 */
+	spin_lock_irqsave(&dev->assigned_dev_lock, flags);
 	if (dev->host_irq_disabled) {
 		enable_irq(dev->host_irq);
 		dev->host_irq_disabled = false;
 	}
+	spin_unlock_irqrestore(&dev->assigned_dev_lock, flags);
 }
 
 /* The function implicit hold kvm->lock mutex due to cancel_work_sync() */
@@ -604,6 +614,7 @@ static int kvm_vm_ioctl_assign_device(struct kvm *kvm,
 	match->dev = dev;
 	match->irq_source_id = -1;
 	match->kvm = kvm;
+	spin_lock_init(&match->assigned_dev_lock);
 
 	list_add(&match->list, &kvm->arch.assigned_dev_head);
 

  reply	other threads:[~2009-02-27 17:55 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-02-26 23:50 Marcelo Tosatti
2009-02-27  4:17 ` Yang, Sheng
2009-02-27 17:54   ` Marcelo Tosatti [this message]
2009-03-02  7:01     ` Yang, Sheng

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=20090227175431.GA10839@amt.cnet \
    --to=mtosatti@redhat.com \
    --cc=avi@redhat.com \
    --cc=chrisw@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=sheng.yang@intel.com \
    --subject='Re: KVM: protect assigned dev workqueue, int handler and irq acker' \
    /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

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.