KVM: MMU: Simplify calculation of pte access
authorAvi Kivity <avi@qumranet.com>
Sun, 9 Dec 2007 14:15:46 +0000 (16:15 +0200)
committerAvi Kivity <avi@qumranet.com>
Wed, 30 Jan 2008 15:53:20 +0000 (17:53 +0200)
The nx bit is awkwardly placed in the 63rd bit position; furthermore it
has a reversed meaning compared to the other bits, which means we can't use
a bitwise and to calculate compounded access masks.

So, we simplify things by creating a new 3-bit exec/write/user access word,
and doing all calculations in that.

Signed-off-by: Avi Kivity <avi@qumranet.com>
drivers/kvm/mmu.c
drivers/kvm/paging_tmpl.h

index 62a74151f84798f1c103a775c397ba32130a5eac..f8a2137c64a218d02a92f43dcbdc44690bbcaf43 100644 (file)
@@ -85,7 +85,8 @@ static int dbg = 1;
 #define PT_PAGE_SIZE_MASK (1ULL << 7)
 #define PT_PAT_MASK (1ULL << 7)
 #define PT_GLOBAL_MASK (1ULL << 8)
-#define PT64_NX_MASK (1ULL << 63)
+#define PT64_NX_SHIFT 63
+#define PT64_NX_MASK (1ULL << PT64_NX_SHIFT)
 
 #define PT_PAT_SHIFT 7
 #define PT_DIR_PAT_SHIFT 12
@@ -153,6 +154,11 @@ static int dbg = 1;
 
 #define RMAP_EXT 4
 
+#define ACC_EXEC_MASK    1
+#define ACC_WRITE_MASK   PT_WRITABLE_MASK
+#define ACC_USER_MASK    PT_USER_MASK
+#define ACC_ALL          (ACC_EXEC_MASK | ACC_WRITE_MASK | ACC_USER_MASK)
+
 struct kvm_rmap_desc {
        u64 *shadow_ptes[RMAP_EXT];
        struct kvm_rmap_desc *more;
@@ -921,7 +927,7 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, struct page *page)
                                >> PAGE_SHIFT;
                        new_table = kvm_mmu_get_page(vcpu, pseudo_gfn,
                                                     v, level - 1,
-                                                    1, 3, &table[index]);
+                                                    1, ACC_ALL, &table[index]);
                        if (!new_table) {
                                pgprintk("nonpaging_map: ENOMEM\n");
                                kvm_release_page_clean(page);
@@ -988,7 +994,7 @@ static void mmu_alloc_roots(struct kvm_vcpu *vcpu)
 
                ASSERT(!VALID_PAGE(root));
                sp = kvm_mmu_get_page(vcpu, root_gfn, 0,
-                                     PT64_ROOT_LEVEL, 0, 0, NULL);
+                                     PT64_ROOT_LEVEL, 0, ACC_ALL, NULL);
                root = __pa(sp->spt);
                ++sp->root_count;
                vcpu->mmu.root_hpa = root;
@@ -1009,7 +1015,7 @@ static void mmu_alloc_roots(struct kvm_vcpu *vcpu)
                        root_gfn = 0;
                sp = kvm_mmu_get_page(vcpu, root_gfn, i << 30,
                                      PT32_ROOT_LEVEL, !is_paging(vcpu),
-                                     0, NULL);
+                                     ACC_ALL, NULL);
                root = __pa(sp->spt);
                ++sp->root_count;
                vcpu->mmu.pae_root[i] = root | PT_PRESENT_MASK;
index 8086f827120fa0f6e585d92862609d8f2b352e12..7688cbf413c89cc17dddce2876ff1984abd0cd10 100644 (file)
@@ -66,7 +66,8 @@ struct guest_walker {
        int level;
        gfn_t table_gfn[PT_MAX_FULL_LEVELS];
        pt_element_t pte;
-       pt_element_t inherited_ar;
+       unsigned pt_access;
+       unsigned pte_access;
        gfn_t gfn;
        u32 error_code;
 };
@@ -110,7 +111,7 @@ static int FNAME(walk_addr)(struct guest_walker *walker,
 {
        pt_element_t pte;
        gfn_t table_gfn;
-       unsigned index;
+       unsigned index, pt_access, pte_access;
        gpa_t pte_gpa;
 
        pgprintk("%s: addr %lx\n", __FUNCTION__, addr);
@@ -128,7 +129,7 @@ walk:
        ASSERT((!is_long_mode(vcpu) && is_pae(vcpu)) ||
               (vcpu->cr3 & CR3_NONPAE_RESERVED_BITS) == 0);
 
-       walker->inherited_ar = PT_USER_MASK | PT_WRITABLE_MASK;
+       pt_access = ACC_ALL;
 
        for (;;) {
                index = PT_INDEX(addr, walker->level);
@@ -165,6 +166,14 @@ walk:
                        pte |= PT_ACCESSED_MASK;
                }
 
+               pte_access = pte & (PT_WRITABLE_MASK | PT_USER_MASK);
+               pte_access |= ACC_EXEC_MASK;
+#if PTTYPE == 64
+               if (is_nx(vcpu))
+                       pte_access &= ~(pte >> PT64_NX_SHIFT);
+#endif
+               pte_access &= pt_access;
+
                if (walker->level == PT_PAGE_TABLE_LEVEL) {
                        walker->gfn = gpte_to_gfn(pte);
                        break;
@@ -180,7 +189,7 @@ walk:
                        break;
                }
 
-               walker->inherited_ar &= pte;
+               pt_access = pte_access;
                --walker->level;
        }
 
@@ -197,7 +206,10 @@ walk:
        }
 
        walker->pte = pte;
-       pgprintk("%s: pte %llx\n", __FUNCTION__, (u64)pte);
+       walker->pt_access = pt_access;
+       walker->pte_access = pte_access;
+       pgprintk("%s: pte %llx pte_access %x pt_access %x\n",
+                __FUNCTION__, (u64)pte, pt_access, pte_access);
        return 1;
 
 not_present:
@@ -218,7 +230,8 @@ err:
 }
 
 static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
-                          u64 *shadow_pte, u64 access_bits,
+                          u64 *shadow_pte, unsigned pt_access,
+                          unsigned pte_access,
                           int user_fault, int write_fault,
                           int *ptwrite, struct guest_walker *walker,
                           gfn_t gfn)
@@ -228,12 +241,11 @@ static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
        int was_rmapped = is_rmap_pte(*shadow_pte);
        struct page *page;
 
-       pgprintk("%s: spte %llx gpte %llx access %llx write_fault %d"
+       pgprintk("%s: spte %llx gpte %llx access %x write_fault %d"
                 " user_fault %d gfn %lx\n",
-                __FUNCTION__, *shadow_pte, (u64)gpte, access_bits,
+                __FUNCTION__, *shadow_pte, (u64)gpte, pt_access,
                 write_fault, user_fault, gfn);
 
-       access_bits &= gpte;
        /*
         * We don't set the accessed bit, since we sometimes want to see
         * whether the guest actually used the pte (in order to detect
@@ -242,12 +254,12 @@ static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
        spte = PT_PRESENT_MASK | PT_DIRTY_MASK;
        spte |= gpte & PT64_NX_MASK;
        if (!dirty)
-               access_bits &= ~PT_WRITABLE_MASK;
+               pte_access &= ~ACC_WRITE_MASK;
 
        page = gfn_to_page(vcpu->kvm, gfn);
 
        spte |= PT_PRESENT_MASK;
-       if (access_bits & PT_USER_MASK)
+       if (pte_access & ACC_USER_MASK)
                spte |= PT_USER_MASK;
 
        if (is_error_page(page)) {
@@ -259,7 +271,7 @@ static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
 
        spte |= page_to_phys(page);
 
-       if ((access_bits & PT_WRITABLE_MASK)
+       if ((pte_access & ACC_WRITE_MASK)
            || (write_fault && !is_write_protection(vcpu) && !user_fault)) {
                struct kvm_mmu_page *shadow;
 
@@ -273,7 +285,7 @@ static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
                if (shadow) {
                        pgprintk("%s: found shadow page for %lx, marking ro\n",
                                 __FUNCTION__, gfn);
-                       access_bits &= ~PT_WRITABLE_MASK;
+                       pte_access &= ~ACC_WRITE_MASK;
                        if (is_writeble_pte(spte)) {
                                spte &= ~PT_WRITABLE_MASK;
                                kvm_x86_ops->tlb_flush(vcpu);
@@ -285,7 +297,7 @@ static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
 
 unshadowed:
 
-       if (access_bits & PT_WRITABLE_MASK)
+       if (pte_access & ACC_WRITE_MASK)
                mark_page_dirty(vcpu->kvm, gfn);
 
        pgprintk("%s: setting spte %llx\n", __FUNCTION__, spte);
@@ -317,8 +329,8 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page,
        if (bytes < sizeof(pt_element_t))
                return;
        pgprintk("%s: gpte %llx spte %p\n", __FUNCTION__, (u64)gpte, spte);
-       FNAME(set_pte)(vcpu, gpte, spte, PT_USER_MASK | PT_WRITABLE_MASK, 0,
-                      0, NULL, NULL, gpte_to_gfn(gpte));
+       FNAME(set_pte)(vcpu, gpte, spte, ACC_ALL, ACC_ALL,
+                      0, 0, NULL, NULL, gpte_to_gfn(gpte));
 }
 
 /*
@@ -331,6 +343,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
        hpa_t shadow_addr;
        int level;
        u64 *shadow_ent;
+       unsigned access = walker->pt_access;
 
        if (!is_present_pte(walker->pte))
                return NULL;
@@ -349,7 +362,6 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
                u64 shadow_pte;
                int metaphysical;
                gfn_t table_gfn;
-               unsigned hugepage_access = 0;
 
                shadow_ent = ((u64 *)__va(shadow_addr)) + index;
                if (is_shadow_present_pte(*shadow_ent)) {
@@ -365,20 +377,15 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
                if (level - 1 == PT_PAGE_TABLE_LEVEL
                    && walker->level == PT_DIRECTORY_LEVEL) {
                        metaphysical = 1;
-                       hugepage_access = walker->pte;
-                       hugepage_access &= PT_USER_MASK | PT_WRITABLE_MASK;
                        if (!is_dirty_pte(walker->pte))
-                               hugepage_access &= ~PT_WRITABLE_MASK;
-                       hugepage_access >>= PT_WRITABLE_SHIFT;
-                       if (walker->pte & PT64_NX_MASK)
-                               hugepage_access |= (1 << 2);
+                               access &= ~ACC_WRITE_MASK;
                        table_gfn = gpte_to_gfn(walker->pte);
                } else {
                        metaphysical = 0;
                        table_gfn = walker->table_gfn[level - 2];
                }
                shadow_page = kvm_mmu_get_page(vcpu, table_gfn, addr, level-1,
-                                              metaphysical, hugepage_access,
+                                              metaphysical, access,
                                               shadow_ent);
                shadow_addr = __pa(shadow_page->spt);
                shadow_pte = shadow_addr | PT_PRESENT_MASK | PT_ACCESSED_MASK
@@ -387,7 +394,8 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
        }
 
        FNAME(set_pte)(vcpu, walker->pte, shadow_ent,
-                      walker->inherited_ar, user_fault, write_fault,
+                      access, walker->pte_access & access,
+                      user_fault, write_fault,
                       ptwrite, walker, walker->gfn);
 
        return shadow_ent;