2021-01-15 16:42:58

by Yu Zhang

Subject: [PATCH] KVM: x86/MMU: Do not check unsync status for root SP.

In shadow page table, only leaf SPs may be marked as unsync.
And for non-leaf SPs, we use unsync_children to keep the number
of the unsynced children. In kvm_mmu_sync_root(), sp->unsync
shall always be zero for the root SP, hence no need to check it.

Signed-off-by: Yu Zhang <[email protected]>
---
arch/x86/kvm/mmu/mmu.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 6d16481a..1a6bb03 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -3412,8 +3412,7 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
* mmu_need_write_protect() describe what could go wrong if this
* requirement isn't satisfied.
*/
- if (!smp_load_acquire(&sp->unsync) &&
- !smp_load_acquire(&sp->unsync_children))
+ if (!smp_load_acquire(&sp->unsync_children))
return;

spin_lock(&vcpu->kvm->mmu_lock);
--
1.9.1

2021-01-26 19:02:06

by Yu Zhang

Subject: Re: [PATCH] KVM: x86/MMU: Do not check unsync status for root SP.

Hi Paolo,

Any comments? Thanks!

B.R.
Yu

On Sat, Jan 16, 2021 at 08:21:00AM +0800, Yu Zhang wrote:
> In shadow page table, only leaf SPs may be marked as unsync.
> And for non-leaf SPs, we use unsync_children to keep the number
> of the unsynced children. In kvm_mmu_sync_root(), sp->unsync
> shall always be zero for the root SP, hence no need to check it.
>
> Signed-off-by: Yu Zhang <[email protected]>
> ---
> arch/x86/kvm/mmu/mmu.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index 6d16481a..1a6bb03 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -3412,8 +3412,7 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
> * mmu_need_write_protect() describe what could go wrong if this
> * requirement isn't satisfied.
> */
> - if (!smp_load_acquire(&sp->unsync) &&
> - !smp_load_acquire(&sp->unsync_children))
> + if (!smp_load_acquire(&sp->unsync_children))
> return;
>
> spin_lock(&vcpu->kvm->mmu_lock);
> --
> 1.9.1
>

2021-02-05 17:57:07

by Sean Christopherson

Subject: Re: [PATCH] KVM: x86/MMU: Do not check unsync status for root SP.

On Sat, Jan 16, 2021, Yu Zhang wrote:
> In shadow page table, only leaf SPs may be marked as unsync.
> And for non-leaf SPs, we use unsync_children to keep the number
> of the unsynced children. In kvm_mmu_sync_root(), sp->unsync
> shall always be zero for the root SP, hence no need to check it.
>
> Signed-off-by: Yu Zhang <[email protected]>
> ---
> arch/x86/kvm/mmu/mmu.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index 6d16481a..1a6bb03 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -3412,8 +3412,7 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
> * mmu_need_write_protect() describe what could go wrong if this
> * requirement isn't satisfied.
> */
> - if (!smp_load_acquire(&sp->unsync) &&
> - !smp_load_acquire(&sp->unsync_children))
> + if (!smp_load_acquire(&sp->unsync_children))
> return;
>
> spin_lock(&vcpu->kvm->mmu_lock);

Looks good. To make this less scary and more obviously correct, maybe move the
the WARN on !PG_LEVEL_4K into kvm_unsync_page() instead of having the WARN in
its sole caller, and add a WARN in mmu_sync_children()?

diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 86af58294272..bc8ee05bb3d3 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -1995,6 +1995,12 @@ static void mmu_sync_children(struct kvm_vcpu *vcpu,
LIST_HEAD(invalid_list);
bool flush = false;

+ /*
+ * Only 4k SPTEs can directly be made unsync, the root shadow page
+ * should never be unsyc'd.
+ */
+ WARN_ON_ONCE(sp->unsync);
+
while (mmu_unsync_walk(parent, &pages)) {
bool protected = false;

@@ -2502,6 +2508,8 @@ EXPORT_SYMBOL_GPL(kvm_mmu_unprotect_page);

static void kvm_unsync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
{
+ WARN_ON(sp->role.level != PG_LEVEL_4K);
+
trace_kvm_mmu_unsync_page(sp);
++vcpu->kvm->stat.mmu_unsync;
sp->unsync = 1;
@@ -2524,7 +2532,6 @@ bool mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn,
if (sp->unsync)
continue;

- WARN_ON(sp->role.level != PG_LEVEL_4K);
kvm_unsync_page(vcpu, sp);
}

@@ -3406,8 +3413,7 @@ void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
* mmu_need_write_protect() describe what could go wrong if this
* requirement isn't satisfied.
*/
- if (!smp_load_acquire(&sp->unsync) &&
- !smp_load_acquire(&sp->unsync_children))
+ if (!smp_load_acquire(&sp->unsync_children))
return;

write_lock(&vcpu->kvm->mmu_lock);