KVM: MMU: Don't allocate gfns page for direct mmu pages

When sp->role.direct is set, sp->gfns does not contain any essential
information, leaf sptes reachable from this sp are for a continuous
guest physical memory range (a linear range).
So sp->gfns[i] (if it was set) equals to sp->gfn + i. (PT_PAGE_TABLE_LEVEL)
Obviously, it is not essential information, we can calculate it when need.

It means we don't need sp->gfns when sp->role.direct=1,
Thus we can save one page usage for every kvm_mmu_page.

Note:
  Access to sp->gfns must be wrapped by kvm_mmu_page_get_gfn()
  or kvm_mmu_page_set_gfn().
  It is only exposed in FNAME(sync_page).

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>

3 files changed, 35 insertions, 10 deletions

diff --git a/Documentation/kvm/mmu.txt b/Documentation/kvm/mmu.txt
index 0e872ae30914..2201dcba92a1 100644
--- a/Documentation/kvm/mmu.txt
+++ b/Documentation/kvm/mmu.txt
@@ -180,7 +180,9 @@ Shadow pages contain the following information:
     guest pages as leaves.
   gfns:
     An array of 512 guest frame numbers, one for each present pte.  Used to
-    perform a reverse map from a pte to a gfn.
+    perform a reverse map from a pte to a gfn. When role.direct is set, any
+    element of this array can be calculated from the gfn field when used, in
+    this case, the array of gfns is not allocated. See role.direct and gfn.
   slot_bitmap:
     A bitmap containing one bit per memory slot.  If the page contains a pte
     mapping a page from memory slot n, then bit n of slot_bitmap will be set
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 07673487fd5d..f46b6c9aff27 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -397,6 +397,22 @@ static void mmu_free_rmap_desc(struct kvm_rmap_desc *rd)
         kmem_cache_free(rmap_desc_cache, rd);
 }
 
+static gfn_t kvm_mmu_page_get_gfn(struct kvm_mmu_page *sp, int index)
+{
+        if (!sp->role.direct)
+                return sp->gfns[index];
+
+        return sp->gfn + (index << ((sp->role.level - 1) * PT64_LEVEL_BITS));
+}
+
+static void kvm_mmu_page_set_gfn(struct kvm_mmu_page *sp, int index, gfn_t gfn)
+{
+        if (sp->role.direct)
+                BUG_ON(gfn != kvm_mmu_page_get_gfn(sp, index));
+        else
+                sp->gfns[index] = gfn;
+}
+
 /*
  * Return the pointer to the largepage write count for a given
  * gfn, handling slots that are not large page aligned.
@@ -547,7 +563,7 @@ static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn)
                 return count;
         gfn = unalias_gfn(vcpu->kvm, gfn);
         sp = page_header(__pa(spte));
-        sp->gfns[spte - sp->spt] = gfn;
+        kvm_mmu_page_set_gfn(sp, spte - sp->spt, gfn);
         rmapp = gfn_to_rmap(vcpu->kvm, gfn, sp->role.level);
         if (!*rmapp) {
                 rmap_printk("rmap_add: %p %llx 0->1\n", spte, *spte);
@@ -605,6 +621,7 @@ static void rmap_remove(struct kvm *kvm, u64 *spte)
         struct kvm_rmap_desc *prev_desc;
         struct kvm_mmu_page *sp;
         pfn_t pfn;
+        gfn_t gfn;
         unsigned long *rmapp;
         int i;
 
@@ -616,7 +633,8 @@ static void rmap_remove(struct kvm *kvm, u64 *spte)
                 kvm_set_pfn_accessed(pfn);
         if (is_writable_pte(*spte))
                 kvm_set_pfn_dirty(pfn);
-        rmapp = gfn_to_rmap(kvm, sp->gfns[spte - sp->spt], sp->role.level);
+        gfn = kvm_mmu_page_get_gfn(sp, spte - sp->spt);
+        rmapp = gfn_to_rmap(kvm, gfn, sp->role.level);
         if (!*rmapp) {
                 printk(KERN_ERR "rmap_remove: %p %llx 0->BUG\n", spte, *spte);
                 BUG();
@@ -900,7 +918,8 @@ static void kvm_mmu_free_page(struct kvm *kvm, struct kvm_mmu_page *sp)
         ASSERT(is_empty_shadow_page(sp->spt));
         list_del(&sp->link);
         __free_page(virt_to_page(sp->spt));
-        __free_page(virt_to_page(sp->gfns));
+        if (!sp->role.direct)
+                __free_page(virt_to_page(sp->gfns));
         kmem_cache_free(mmu_page_header_cache, sp);
         ++kvm->arch.n_free_mmu_pages;
 }
@@ -911,13 +930,15 @@ static unsigned kvm_page_table_hashfn(gfn_t gfn)
 }
 
 static struct kvm_mmu_page *kvm_mmu_alloc_page(struct kvm_vcpu *vcpu,
-                                               u64 *parent_pte)
+                                               u64 *parent_pte, int direct)
 {
         struct kvm_mmu_page *sp;
 
         sp = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_header_cache, sizeof *sp);
         sp->spt = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache, PAGE_SIZE);
-        sp->gfns = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache, PAGE_SIZE);
+        if (!direct)
+                sp->gfns = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache,
+                                                  PAGE_SIZE);
         set_page_private(virt_to_page(sp->spt), (unsigned long)sp);
         list_add(&sp->link, &vcpu->kvm->arch.active_mmu_pages);
         bitmap_zero(sp->slot_bitmap, KVM_MEMORY_SLOTS + KVM_PRIVATE_MEM_SLOTS);
@@ -1386,7 +1407,7 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
                         return sp;
                 }
         ++vcpu->kvm->stat.mmu_cache_miss;
-        sp = kvm_mmu_alloc_page(vcpu, parent_pte);
+        sp = kvm_mmu_alloc_page(vcpu, parent_pte, direct);
         if (!sp)
                 return sp;
         sp->gfn = gfn;
@@ -3403,7 +3424,7 @@ void inspect_spte_has_rmap(struct kvm *kvm, u64 *sptep)
 
         if (*sptep & PT_WRITABLE_MASK) {
                 rev_sp = page_header(__pa(sptep));
-                gfn = rev_sp->gfns[sptep - rev_sp->spt];
+                gfn = kvm_mmu_page_get_gfn(rev_sp, sptep - rev_sp->spt);
 
                 if (!gfn_to_memslot(kvm, gfn)) {
                         if (!printk_ratelimit())
@@ -3417,8 +3438,7 @@ void inspect_spte_has_rmap(struct kvm *kvm, u64 *sptep)
                         return;
                 }
 
-                rmapp = gfn_to_rmap(kvm, rev_sp->gfns[sptep - rev_sp->spt],
-                                    rev_sp->role.level);
+                rmapp = gfn_to_rmap(kvm, gfn, rev_sp->role.level);
                 if (!*rmapp) {
                         if (!printk_ratelimit())
                                 return;
diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h
index 167f53357eef..2ee7060a80a5 100644
--- a/arch/x86/kvm/paging_tmpl.h
+++ b/arch/x86/kvm/paging_tmpl.h
@@ -582,6 +582,9 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
 
         offset = nr_present = 0;
 
+        /* direct kvm_mmu_page can not be unsync. */
+        BUG_ON(sp->role.direct);
+
         if (PTTYPE == 32)
                 offset = sp->role.quadrant << PT64_LEVEL_BITS;