public inbox for kvm@vger.kernel.org
 help / color / mirror / Atom feed
From: Avi Kivity <avi@redhat.com>
To: linux-kernel@vger.kernel.org
Cc: kvm@vger.kernel.org, Marcelo Tosatti <mtosatti@redhat.com>
Subject: [PATCH 32/40] KVM: PIT: fix injection logic and count
Date: Tue, 23 Sep 2008 16:46:46 +0300	[thread overview]
Message-ID: <1222177614-26669-33-git-send-email-avi@redhat.com> (raw)
In-Reply-To: <1222177614-26669-1-git-send-email-avi@redhat.com>

From: Marcelo Tosatti <mtosatti@redhat.com>

The PIT injection logic is problematic under the following cases:

1) If there is a higher priority vector to be delivered by the time
kvm_pit_timer_intr_post is invoked ps->inject_pending won't be set.
This opens the possibility for missing many PIT event injections (say if
guest executes hlt at this point).

2) ps->inject_pending is racy with more than two vcpus. Since there's no locking
around read/dec of pt->pending, two vcpu's can inject two interrupts for a single
pt->pending count.

Fix 1 by using an irq ack notifier: only reinject when the previous irq
has been acked. Fix 2 with appropriate locking around manipulation of
pending count and irq_ack by the injection / ack paths.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
---
 arch/x86/kvm/i8254.c |   70 +++++++++++++++++++++++++-------------------------
 arch/x86/kvm/i8254.h |    7 ++---
 arch/x86/kvm/irq.c   |    1 -
 3 files changed, 38 insertions(+), 40 deletions(-)

diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index c0f7872..7d04dd3 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -207,6 +207,8 @@ static int __pit_timer_fn(struct kvm_kpit_state *ps)
 
 	pt->timer.expires = ktime_add_ns(pt->timer.expires, pt->period);
 	pt->scheduled = ktime_to_ns(pt->timer.expires);
+	if (pt->period)
+		ps->channels[0].count_load_time = pt->timer.expires;
 
 	return (pt->period == 0 ? 0 : 1);
 }
@@ -215,12 +217,22 @@ int pit_has_pending_timer(struct kvm_vcpu *vcpu)
 {
 	struct kvm_pit *pit = vcpu->kvm->arch.vpit;
 
-	if (pit && vcpu->vcpu_id == 0 && pit->pit_state.inject_pending)
+	if (pit && vcpu->vcpu_id == 0 && pit->pit_state.irq_ack)
 		return atomic_read(&pit->pit_state.pit_timer.pending);
-
 	return 0;
 }
 
+void kvm_pit_ack_irq(struct kvm_irq_ack_notifier *kian)
+{
+	struct kvm_kpit_state *ps = container_of(kian, struct kvm_kpit_state,
+						 irq_ack_notifier);
+	spin_lock(&ps->inject_lock);
+	if (atomic_dec_return(&ps->pit_timer.pending) < 0)
+		WARN_ON(1);
+	ps->irq_ack = 1;
+	spin_unlock(&ps->inject_lock);
+}
+
 static enum hrtimer_restart pit_timer_fn(struct hrtimer *data)
 {
 	struct kvm_kpit_state *ps;
@@ -255,8 +267,9 @@ static void destroy_pit_timer(struct kvm_kpit_timer *pt)
 	hrtimer_cancel(&pt->timer);
 }
 
-static void create_pit_timer(struct kvm_kpit_timer *pt, u32 val, int is_period)
+static void create_pit_timer(struct kvm_kpit_state *ps, u32 val, int is_period)
 {
+	struct kvm_kpit_timer *pt = &ps->pit_timer;
 	s64 interval;
 
 	interval = muldiv64(val, NSEC_PER_SEC, KVM_PIT_FREQ);
@@ -268,6 +281,7 @@ static void create_pit_timer(struct kvm_kpit_timer *pt, u32 val, int is_period)
 	pt->period = (is_period == 0) ? 0 : interval;
 	pt->timer.function = pit_timer_fn;
 	atomic_set(&pt->pending, 0);
+	ps->irq_ack = 1;
 
 	hrtimer_start(&pt->timer, ktime_add_ns(ktime_get(), interval),
 		      HRTIMER_MODE_ABS);
@@ -302,11 +316,11 @@ static void pit_load_count(struct kvm *kvm, int channel, u32 val)
 	case 1:
         /* FIXME: enhance mode 4 precision */
 	case 4:
-		create_pit_timer(&ps->pit_timer, val, 0);
+		create_pit_timer(ps, val, 0);
 		break;
 	case 2:
 	case 3:
-		create_pit_timer(&ps->pit_timer, val, 1);
+		create_pit_timer(ps, val, 1);
 		break;
 	default:
 		destroy_pit_timer(&ps->pit_timer);
@@ -520,7 +534,7 @@ void kvm_pit_reset(struct kvm_pit *pit)
 	mutex_unlock(&pit->pit_state.lock);
 
 	atomic_set(&pit->pit_state.pit_timer.pending, 0);
-	pit->pit_state.inject_pending = 1;
+	pit->pit_state.irq_ack = 1;
 }
 
 struct kvm_pit *kvm_create_pit(struct kvm *kvm)
@@ -534,6 +548,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm)
 
 	mutex_init(&pit->pit_state.lock);
 	mutex_lock(&pit->pit_state.lock);
+	spin_lock_init(&pit->pit_state.inject_lock);
 
 	/* Initialize PIO device */
 	pit->dev.read = pit_ioport_read;
@@ -555,6 +570,9 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm)
 	pit_state->pit = pit;
 	hrtimer_init(&pit_state->pit_timer.timer,
 		     CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	pit_state->irq_ack_notifier.gsi = 0;
+	pit_state->irq_ack_notifier.irq_acked = kvm_pit_ack_irq;
+	kvm_register_irq_ack_notifier(kvm, &pit_state->irq_ack_notifier);
 	mutex_unlock(&pit->pit_state.lock);
 
 	kvm_pit_reset(pit);
@@ -592,37 +610,19 @@ void kvm_inject_pit_timer_irqs(struct kvm_vcpu *vcpu)
 	struct kvm_kpit_state *ps;
 
 	if (vcpu && pit) {
+		int inject = 0;
 		ps = &pit->pit_state;
 
-		/* Try to inject pending interrupts when:
-		 * 1. Pending exists
-		 * 2. Last interrupt was accepted or waited for too long time*/
-		if (atomic_read(&ps->pit_timer.pending) &&
-		    (ps->inject_pending ||
-		    (jiffies - ps->last_injected_time
-				>= KVM_MAX_PIT_INTR_INTERVAL))) {
-			ps->inject_pending = 0;
-			__inject_pit_timer_intr(kvm);
-			ps->last_injected_time = jiffies;
-		}
-	}
-}
-
-void kvm_pit_timer_intr_post(struct kvm_vcpu *vcpu, int vec)
-{
-	struct kvm_arch *arch = &vcpu->kvm->arch;
-	struct kvm_kpit_state *ps;
-
-	if (vcpu && arch->vpit) {
-		ps = &arch->vpit->pit_state;
-		if (atomic_read(&ps->pit_timer.pending) &&
-		(((arch->vpic->pics[0].imr & 1) == 0 &&
-		  arch->vpic->pics[0].irq_base == vec) ||
-		  (arch->vioapic->redirtbl[0].fields.vector == vec &&
-		  arch->vioapic->redirtbl[0].fields.mask != 1))) {
-			ps->inject_pending = 1;
-			atomic_dec(&ps->pit_timer.pending);
-			ps->channels[0].count_load_time = ktime_get();
+		/* Try to inject pending interrupts when
+		 * last one has been acked.
+		 */
+		spin_lock(&ps->inject_lock);
+		if (atomic_read(&ps->pit_timer.pending) && ps->irq_ack) {
+			ps->irq_ack = 0;
+			inject = 1;
 		}
+		spin_unlock(&ps->inject_lock);
+		if (inject)
+			__inject_pit_timer_intr(kvm);
 	}
 }
diff --git a/arch/x86/kvm/i8254.h b/arch/x86/kvm/i8254.h
index db25c2a..e436d49 100644
--- a/arch/x86/kvm/i8254.h
+++ b/arch/x86/kvm/i8254.h
@@ -8,7 +8,6 @@ struct kvm_kpit_timer {
 	int irq;
 	s64 period; /* unit: ns */
 	s64 scheduled;
-	ktime_t last_update;
 	atomic_t pending;
 };
 
@@ -34,8 +33,9 @@ struct kvm_kpit_state {
 	u32    speaker_data_on;
 	struct mutex lock;
 	struct kvm_pit *pit;
-	bool inject_pending; /* if inject pending interrupts */
-	unsigned long last_injected_time;
+	spinlock_t inject_lock;
+	unsigned long irq_ack;
+	struct kvm_irq_ack_notifier irq_ack_notifier;
 };
 
 struct kvm_pit {
@@ -54,7 +54,6 @@ struct kvm_pit {
 #define KVM_PIT_CHANNEL_MASK	    0x3
 
 void kvm_inject_pit_timer_irqs(struct kvm_vcpu *vcpu);
-void kvm_pit_timer_intr_post(struct kvm_vcpu *vcpu, int vec);
 void kvm_pit_load_count(struct kvm *kvm, int channel, u32 val);
 struct kvm_pit *kvm_create_pit(struct kvm *kvm);
 void kvm_free_pit(struct kvm *kvm);
diff --git a/arch/x86/kvm/irq.c b/arch/x86/kvm/irq.c
index 3c508af..8c1b9c5 100644
--- a/arch/x86/kvm/irq.c
+++ b/arch/x86/kvm/irq.c
@@ -90,7 +90,6 @@ EXPORT_SYMBOL_GPL(kvm_inject_pending_timer_irqs);
 void kvm_timer_intr_post(struct kvm_vcpu *vcpu, int vec)
 {
 	kvm_apic_timer_intr_post(vcpu, vec);
-	kvm_pit_timer_intr_post(vcpu, vec);
 	/* TODO: PIT, RTC etc. */
 }
 EXPORT_SYMBOL_GPL(kvm_timer_intr_post);
-- 
1.6.0.1


  parent reply	other threads:[~2008-09-23 13:47 UTC|newest]

Thread overview: 41+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-09-23 13:46 [PATCH 00/40] KVM Updates for 2.6.28 merge window (part 1 of 3) Avi Kivity
2008-09-23 13:46 ` [PATCH 01/40] KVM: VMX: Rename misnamed msr bits Avi Kivity
2008-09-23 13:46 ` [PATCH 02/40] KVM: x86: accessors for guest registers Avi Kivity
2008-09-23 13:46 ` [PATCH 03/40] KVM: Move KVM TRACE DEFINITIONS to common header Avi Kivity
2008-09-23 13:46 ` [PATCH 04/40] KVM: Introduce kvm_set_irq to inject interrupts in guests Avi Kivity
2008-09-23 13:46 ` [PATCH 05/40] KVM: MMU: Separate the code for unlinking a shadow page from its parents Avi Kivity
2008-09-23 13:46 ` [PATCH 06/40] KVM: MMU: Simplify kvm_mmu_zap_page() Avi Kivity
2008-09-23 13:46 ` [PATCH 07/40] KVM: Move NMI IRET fault processing to new vmx_complete_interrupts() Avi Kivity
2008-09-23 13:46 ` [PATCH 08/40] KVM: VMX: Move nmi injection failure processing to vm exit path Avi Kivity
2008-09-23 13:46 ` [PATCH 09/40] KVM: Clear exception queue before emulating an instruction Avi Kivity
2008-09-23 13:46 ` [PATCH 10/40] KVM: VMX: Fix pending exception processing Avi Kivity
2008-09-23 13:46 ` [PATCH 11/40] KVM: Add a pending interrupt queue Avi Kivity
2008-09-23 13:46 ` [PATCH 12/40] KVM: VMX: Move interrupt post-processing to vmx_complete_interrupts() Avi Kivity
2008-09-23 13:46 ` [PATCH 13/40] KVM: VMX: Remove redundant check in handle_rmode_exception Avi Kivity
2008-09-23 13:46 ` [PATCH 14/40] KVM: Consolidate PIC isr clearing into a function Avi Kivity
2008-09-23 13:46 ` [PATCH 15/40] KVM: Consolidate XX_VECTOR defines Avi Kivity
2008-09-23 13:46 ` [PATCH 16/40] KVM: VMX: Reinject real mode exception Avi Kivity
2008-09-23 13:46 ` [PATCH 17/40] KVM: VMX: Unify register save/restore across 32 and 64 bit hosts Avi Kivity
2008-09-23 13:46 ` [PATCH 18/40] KVM: SVM: " Avi Kivity
2008-09-23 13:46 ` [PATCH 19/40] KVM: kvmtrace: Remove use of bit fields in kvm trace structure Avi Kivity
2008-09-23 13:46 ` [PATCH 20/40] KVM: kvmtrace: replace get_cycles with ktime_get v3 Avi Kivity
2008-09-23 13:46 ` [PATCH 21/40] KVM: ppc: enable KVM_TRACE building for powerpc Avi Kivity
2008-09-23 13:46 ` [PATCH 22/40] KVM: ppc: adds trace points for ppc tlb activity Avi Kivity
2008-09-23 13:46 ` [PATCH 23/40] KVM: ppc: trace powerpc instruction emulation Avi Kivity
2008-09-23 13:46 ` [PATCH 24/40] KVM: VMX: Avoid vmwrite(HOST_RSP) when possible Avi Kivity
2008-09-23 13:46 ` [PATCH 25/40] KVM: Ignore DEBUGCTL MSRs with no effect Avi Kivity
2008-09-23 13:46 ` [PATCH 26/40] KVM: ppc: guest breakpoint support Avi Kivity
2008-09-23 13:46 ` [PATCH 27/40] KVM: ppc: Stop saving host TLB state Avi Kivity
2008-09-23 13:46 ` [PATCH 28/40] KVM: ppc: Write only modified shadow entries into the TLB on exit Avi Kivity
2008-09-23 13:46 ` [PATCH 29/40] KVM: powerpc: Map guest userspace with TID=0 mappings Avi Kivity
2008-09-23 13:46 ` [PATCH 30/40] KVM: Add irq ack notifier list Avi Kivity
2008-09-23 13:46 ` [PATCH 31/40] KVM: irq ack notification Avi Kivity
2008-09-23 13:46 ` Avi Kivity [this message]
2008-09-23 13:46 ` [PATCH 33/40] x86: paravirt: factor out cpu_khz to common code Avi Kivity
2008-09-23 13:46 ` [PATCH 34/40] x86: KVM guest: use paravirt function to calculate cpu khz Avi Kivity
2008-09-23 13:46 ` [PATCH 35/40] KVM: direct mmio pfn check Avi Kivity
2008-09-23 13:46 ` [PATCH 36/40] KVM: pci device assignment Avi Kivity
2008-09-23 13:46 ` [PATCH 37/40] KVM: Reduce kvm stack usage in kvm_arch_vm_ioctl() Avi Kivity
2008-09-23 13:46 ` [PATCH 38/40] KVM: Reduce stack usage in kvm_vcpu_ioctl() Avi Kivity
2008-09-23 13:46 ` [PATCH 39/40] KVM: Reduce stack usage in kvm_arch_vcpu_ioctl() Avi Kivity
2008-09-23 13:46 ` [PATCH 40/40] KVM: Reduce stack usage in kvm_pv_mmu_op() Avi Kivity

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=1222177614-26669-33-git-send-email-avi@redhat.com \
    --to=avi@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mtosatti@redhat.com \
    /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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox