2010-04-22 06:17:04

by Xiao Guangrong

[permalink] [raw]
Subject: [PATCH 9/10] KVM MMU: separate invlpg code form kvm_mmu_pte_write()

Let invlpg not depends on kvm_mmu_pte_write path, later patch will need
this feature

Signed-off-by: Xiao Guangrong <[email protected]>
---
arch/x86/kvm/mmu.c | 40 ++++++++++++++++++++++++----------------
1 files changed, 24 insertions(+), 16 deletions(-)

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index e0bb4d8..f092e71 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -2278,14 +2278,21 @@ static bool is_rsvd_bits_set(struct kvm_vcpu *vcpu, u64 gpte, int level)
return (gpte & vcpu->arch.mmu.rsvd_bits_mask[bit7][level-1]) != 0;
}

+static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
+ u64 gpte);
+static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu,
+ struct kvm_mmu_page *sp,
+ u64 *spte,
+ const void *new);
+
static void paging_invlpg(struct kvm_vcpu *vcpu, gva_t gva)
{
+ struct kvm_mmu_page *sp = NULL;
struct kvm_shadow_walk_iterator iterator;
- gpa_t pte_gpa = -1;
- int level;
- u64 *sptep;
- int need_flush = 0;
+ gfn_t gfn = -1;
+ u64 *sptep = NULL, gentry;
unsigned pte_size = 0;
+ int invlpg_counter, level, offset = 0, need_flush = 0;

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

@@ -2294,14 +2301,14 @@ static void paging_invlpg(struct kvm_vcpu *vcpu, gva_t gva)
sptep = iterator.sptep;

if (is_last_spte(*sptep, level)) {
- struct kvm_mmu_page *sp = page_header(__pa(sptep));
- int offset = 0;
+
+ sp = page_header(__pa(sptep));

if (!sp->role.cr4_pae)
offset = sp->role.quadrant << PT64_LEVEL_BITS;;
pte_size = sp->role.cr4_pae ? 8 : 4;
- pte_gpa = (sp->gfn << PAGE_SHIFT);
- pte_gpa += (sptep - sp->spt + offset) * pte_size;
+ gfn = (sp->gfn << PAGE_SHIFT);
+ offset = (sptep - sp->spt + offset) * pte_size;

if (is_shadow_present_pte(*sptep)) {
rmap_remove(vcpu->kvm, sptep);
@@ -2320,16 +2327,22 @@ static void paging_invlpg(struct kvm_vcpu *vcpu, gva_t gva)
if (need_flush)
kvm_flush_remote_tlbs(vcpu->kvm);

- atomic_inc(&vcpu->kvm->arch.invlpg_counter);
+ invlpg_counter = atomic_add_return(1, &vcpu->kvm->arch.invlpg_counter);

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

- if (pte_gpa == -1)
+ if (gfn == -1)
return;

if (mmu_topup_memory_caches(vcpu))
return;
- kvm_mmu_pte_write(vcpu, pte_gpa, NULL, pte_size, 0);
+
+ kvm_read_guest_page(vcpu->kvm, gfn, &gentry, offset, pte_size);
+ mmu_guess_page_from_pte_write(vcpu, gfn_to_gpa(gfn) + offset, gentry);
+ spin_lock(&vcpu->kvm->mmu_lock);
+ if (atomic_read(&vcpu->kvm->arch.invlpg_counter) == invlpg_counter)
+ mmu_pte_write_new_pte(vcpu, sp, sptep, &gentry);
+ spin_unlock(&vcpu->kvm->mmu_lock);
}

#define PTTYPE 64
@@ -2675,12 +2688,9 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
int flooded = 0;
int npte;
int r;
- int invlpg_counter;

pgprintk("%s: gpa %llx bytes %d\n", __func__, gpa, bytes);

- invlpg_counter = atomic_read(&vcpu->kvm->arch.invlpg_counter);
-
/*
* Assume that the pte write on a page table of the same type
* as the current vcpu paging mode. This is nearly always true
@@ -2713,8 +2723,6 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,

mmu_guess_page_from_pte_write(vcpu, gpa, gentry);
spin_lock(&vcpu->kvm->mmu_lock);
- if (atomic_read(&vcpu->kvm->arch.invlpg_counter) != invlpg_counter)
- gentry = 0;
kvm_mmu_access_page(vcpu, gfn);
kvm_mmu_free_some_pages(vcpu);
++vcpu->kvm->stat.mmu_pte_write;
--
1.6.1.2