| Message ID | 4DCEF5B1.3050706@cn.fujitsu.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 05/15/2011 12:35 AM, Xiao Guangrong wrote: > Simply return from kvm_mmu_pte_write path if no shadow page is > write-protected, then we can avoid to walk all shadow pages and hold > mmu-lock > diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c > index 2841805..971e2d2 100644 > --- a/arch/x86/kvm/mmu.c > +++ b/arch/x86/kvm/mmu.c > @@ -498,6 +498,7 @@ static void account_shadowed(struct kvm *kvm, gfn_t gfn) > linfo = lpage_info_slot(gfn, slot, i); > linfo->write_count += 1; > } > + atomic_inc(&kvm->arch.indirect_shadow_pages); > } > > static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) > @@ -513,6 +514,7 @@ static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) > linfo->write_count -= 1; > WARN_ON(linfo->write_count< 0); > } > + atomic_dec(&kvm->arch.indirect_shadow_pages); > } These atomic ops are always called from within the spinlock, so we don't need an atomic_t here. Sorry, I should have noticed this on the first version.
On 05/15/2011 04:20 PM, Avi Kivity wrote: > On 05/15/2011 12:35 AM, Xiao Guangrong wrote: >> Simply return from kvm_mmu_pte_write path if no shadow page is >> write-protected, then we can avoid to walk all shadow pages and hold >> mmu-lock >> diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c >> index 2841805..971e2d2 100644 >> --- a/arch/x86/kvm/mmu.c >> +++ b/arch/x86/kvm/mmu.c >> @@ -498,6 +498,7 @@ static void account_shadowed(struct kvm *kvm, gfn_t gfn) >> linfo = lpage_info_slot(gfn, slot, i); >> linfo->write_count += 1; >> } >> + atomic_inc(&kvm->arch.indirect_shadow_pages); >> } >> >> static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) >> @@ -513,6 +514,7 @@ static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) >> linfo->write_count -= 1; >> WARN_ON(linfo->write_count< 0); >> } >> + atomic_dec(&kvm->arch.indirect_shadow_pages); >> } > > These atomic ops are always called from within the spinlock, so we don't need an atomic_t here. > > Sorry, I should have noticed this on the first version. We read indirect_shadow_pages atomically on pte write path, that is allowed out of mmu_lock -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 05/15/2011 11:33 AM, Xiao Guangrong wrote: > > > > These atomic ops are always called from within the spinlock, so we don't need an atomic_t here. > > > > Sorry, I should have noticed this on the first version. > > We read indirect_shadow_pages atomically on pte write path, that is allowed out of mmu_lock Reading is fine: #define atomic_read(v) (*(volatile int *)&(v)->counter)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index d2ac8e2..d2e5fb8 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -442,6 +442,7 @@ struct kvm_arch { unsigned int n_requested_mmu_pages; unsigned int n_max_mmu_pages; atomic_t invlpg_counter; + atomic_t indirect_shadow_pages; struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES]; /* * Hash table of struct kvm_mmu_page. diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index 2841805..971e2d2 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -498,6 +498,7 @@ static void account_shadowed(struct kvm *kvm, gfn_t gfn) linfo = lpage_info_slot(gfn, slot, i); linfo->write_count += 1; } + atomic_inc(&kvm->arch.indirect_shadow_pages); } static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) @@ -513,6 +514,7 @@ static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) linfo->write_count -= 1; WARN_ON(linfo->write_count < 0); } + atomic_dec(&kvm->arch.indirect_shadow_pages); } static int has_wrprotected_page(struct kvm *kvm, @@ -3233,6 +3235,13 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, int level, npte, invlpg_counter, r, flooded = 0; bool remote_flush, local_flush, zap_page; + /* + * If we don't have indirect shadow pages, it means no page is + * write-protected, so we can exit simply. + */ + if (!atomic_read(&vcpu->kvm->arch.indirect_shadow_pages)) + return; + zap_page = remote_flush = local_flush = false; offset = offset_in_page(gpa);
Simply return from kvm_mmu_pte_write path if no shadow page is write-protected, then we can avoid to walk all shadow pages and hold mmu-lock Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> --- arch/x86/include/asm/kvm_host.h | 1 + arch/x86/kvm/mmu.c | 9 +++++++++ 2 files changed, 10 insertions(+), 0 deletions(-)