Patchwork [1/5] KVM: Provide mmu notifier retry test based on struct kvm

login
register
mail settings
Submitter Paul Mackerras
Date Oct. 16, 2012, 3:59 a.m.
Message ID <20121016035933.GM1218@drongo>
Download mbox | patch
Permalink /patch/191715/
State New
Headers show

Comments

Paul Mackerras - Oct. 16, 2012, 3:59 a.m.
The mmu_notifier_retry() function, used to test whether any page
invalidations are in progress, currently takes a vcpu pointer, though
the code only needs the VM's struct kvm pointer.  Forthcoming patches
to the powerpc Book3S HV code will need to test for retry within a VM
ioctl, where a struct kvm pointer is available but a struct vcpu
pointer isn't.  Therefore this creates a variant of mmu_notifier_retry
called kvm_mmu_notifier_retry that takes a struct kvm pointer, and
implements mmu_notifier_retry in terms of it.

Signed-off-by: Paul Mackerras <paulus@samba.org>
---
 include/linux/kvm_host.h |   11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
Avi Kivity - Oct. 16, 2012, 9:44 a.m.
On 10/16/2012 05:59 AM, Paul Mackerras wrote:
> The mmu_notifier_retry() function, used to test whether any page
> invalidations are in progress, currently takes a vcpu pointer, though
> the code only needs the VM's struct kvm pointer.  Forthcoming patches
> to the powerpc Book3S HV code will need to test for retry within a VM
> ioctl, where a struct kvm pointer is available but a struct vcpu
> pointer isn't.  Therefore this creates a variant of mmu_notifier_retry
> called kvm_mmu_notifier_retry that takes a struct kvm pointer, and
> implements mmu_notifier_retry in terms of it.

Why not change mmu_notifier_retry() and all its callers?
Alexander Graf - Oct. 16, 2012, 10:06 a.m.
On 16.10.2012, at 11:44, Avi Kivity wrote:

> On 10/16/2012 05:59 AM, Paul Mackerras wrote:
>> The mmu_notifier_retry() function, used to test whether any page
>> invalidations are in progress, currently takes a vcpu pointer, though
>> the code only needs the VM's struct kvm pointer.  Forthcoming patches
>> to the powerpc Book3S HV code will need to test for retry within a VM
>> ioctl, where a struct kvm pointer is available but a struct vcpu
>> pointer isn't.  Therefore this creates a variant of mmu_notifier_retry
>> called kvm_mmu_notifier_retry that takes a struct kvm pointer, and
>> implements mmu_notifier_retry in terms of it.
> 
> Why not change mmu_notifier_retry() and all its callers?

Why not use Christoffer's patch? :)

Alex

--
To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 6afc5be..1cc1e1d 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -841,9 +841,9 @@  extern struct kvm_stats_debugfs_item debugfs_entries[];
 extern struct dentry *kvm_debugfs_dir;
 
 #if defined(CONFIG_MMU_NOTIFIER) && defined(KVM_ARCH_WANT_MMU_NOTIFIER)
-static inline int mmu_notifier_retry(struct kvm_vcpu *vcpu, unsigned long mmu_seq)
+static inline int kvm_mmu_notifier_retry(struct kvm *kvm, unsigned long mmu_seq)
 {
-	if (unlikely(vcpu->kvm->mmu_notifier_count))
+	if (unlikely(kvm->mmu_notifier_count))
 		return 1;
 	/*
 	 * Ensure the read of mmu_notifier_count happens before the read
@@ -856,10 +856,15 @@  static inline int mmu_notifier_retry(struct kvm_vcpu *vcpu, unsigned long mmu_se
 	 * can't rely on kvm->mmu_lock to keep things ordered.
 	 */
 	smp_rmb();
-	if (vcpu->kvm->mmu_notifier_seq != mmu_seq)
+	if (kvm->mmu_notifier_seq != mmu_seq)
 		return 1;
 	return 0;
 }
+
+static inline int mmu_notifier_retry(struct kvm_vcpu *vcpu, unsigned long mmu_seq)
+{
+	return kvm_mmu_notifier_retry(vcpu->kvm, mmu_seq);
+}
 #endif
 
 #ifdef KVM_CAP_IRQ_ROUTING