Skip to content

Commit

Permalink
---
Browse files Browse the repository at this point in the history 
yaml
---
r: 80631
b: refs/heads/master
c: c7addb9
h: refs/heads/master
i:
  80629: 63eac34
  80627: 5a5882b
  80623: ac5f619
v: v3
  • Loading branch information
Avi Kivity committed Jan 30, 2008
1 parent 2214c85 commit f24cb69
Show file tree
Hide file tree
Showing 6 changed files with 123 additions and 38 deletions.
2 changes: 1 addition & 1 deletion [refs]
View file
Original file line number Diff line number Diff line change
@@ -1,2 +1,2 @@
---
refs/heads/master: 51c6cf662b4b361a09fbd324f4c67875d9bcfbea
refs/heads/master: c7addb902054195b995114df154e061c7d604f69
3 changes: 3 additions & 0 deletions trunk/drivers/kvm/kvm.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -150,6 +150,8 @@ struct kvm_mmu {
int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err);
void (*free)(struct kvm_vcpu *vcpu);
gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva);
void (*prefetch_page)(struct kvm_vcpu *vcpu,
struct kvm_mmu_page *page);
hpa_t root_hpa;
int root_level;
int shadow_root_level;
Expand Down Expand Up @@ -536,6 +538,7 @@ void kvm_mmu_module_exit(void);
void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
int kvm_mmu_create(struct kvm_vcpu *vcpu);
int kvm_mmu_setup(struct kvm_vcpu *vcpu);
void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte);

int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
Expand Down
4 changes: 3 additions & 1 deletion trunk/drivers/kvm/kvm_main.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3501,7 +3501,9 @@ int kvm_init_x86(struct kvm_x86_ops *ops, unsigned int vcpu_size,
kvm_preempt_ops.sched_in = kvm_sched_in;
kvm_preempt_ops.sched_out = kvm_sched_out;

return r;
kvm_mmu_set_nonpresent_ptes(0ull, 0ull);

return 0;

out_free:
kmem_cache_destroy(kvm_vcpu_cache);
Expand Down
89 changes: 68 additions & 21 deletions trunk/drivers/kvm/mmu.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -156,6 +156,16 @@ static struct kmem_cache *pte_chain_cache;
static struct kmem_cache *rmap_desc_cache;
static struct kmem_cache *mmu_page_header_cache;

static u64 __read_mostly shadow_trap_nonpresent_pte;
static u64 __read_mostly shadow_notrap_nonpresent_pte;

void kvm_mmu_set_nonpresent_ptes(u64 trap_pte, u64 notrap_pte)
{
shadow_trap_nonpresent_pte = trap_pte;
shadow_notrap_nonpresent_pte = notrap_pte;
}
EXPORT_SYMBOL_GPL(kvm_mmu_set_nonpresent_ptes);

static int is_write_protection(struct kvm_vcpu *vcpu)
{
return vcpu->cr0 & X86_CR0_WP;
Expand All @@ -176,6 +186,13 @@ static int is_present_pte(unsigned long pte)
return pte & PT_PRESENT_MASK;
}

static int is_shadow_present_pte(u64 pte)
{
pte &= ~PT_SHADOW_IO_MARK;
return pte != shadow_trap_nonpresent_pte
&& pte != shadow_notrap_nonpresent_pte;
}

static int is_writeble_pte(unsigned long pte)
{
return pte & PT_WRITABLE_MASK;
Expand Down Expand Up @@ -450,7 +467,7 @@ static int is_empty_shadow_page(u64 *spt)
u64 *end;

for (pos = spt, end = pos + PAGE_SIZE / sizeof(u64); pos != end; pos++)
if (*pos != 0) {
if ((*pos & ~PT_SHADOW_IO_MARK) != shadow_trap_nonpresent_pte) {
printk(KERN_ERR "%s: %p %llx\n", __FUNCTION__,
pos, *pos);
return 0;
Expand Down Expand Up @@ -632,6 +649,7 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
page->gfn = gfn;
page->role = role;
hlist_add_head(&page->hash_link, bucket);
vcpu->mmu.prefetch_page(vcpu, page);
if (!metaphysical)
rmap_write_protect(vcpu, gfn);
return page;
Expand All @@ -648,9 +666,9 @@ static void kvm_mmu_page_unlink_children(struct kvm *kvm,

if (page->role.level == PT_PAGE_TABLE_LEVEL) {
for (i = 0; i < PT64_ENT_PER_PAGE; ++i) {
if (pt[i] & PT_PRESENT_MASK)
if (is_shadow_present_pte(pt[i]))
rmap_remove(&pt[i]);
pt[i] = 0;
pt[i] = shadow_trap_nonpresent_pte;
}
kvm_flush_remote_tlbs(kvm);
return;
Expand All @@ -659,8 +677,8 @@ static void kvm_mmu_page_unlink_children(struct kvm *kvm,
for (i = 0; i < PT64_ENT_PER_PAGE; ++i) {
ent = pt[i];

pt[i] = 0;
if (!(ent & PT_PRESENT_MASK))
pt[i] = shadow_trap_nonpresent_pte;
if (!is_shadow_present_pte(ent))
continue;
ent &= PT64_BASE_ADDR_MASK;
mmu_page_remove_parent_pte(page_header(ent), &pt[i]);
Expand Down Expand Up @@ -691,7 +709,7 @@ static void kvm_mmu_zap_page(struct kvm *kvm,
}
BUG_ON(!parent_pte);
kvm_mmu_put_page(page, parent_pte);
set_shadow_pte(parent_pte, 0);
set_shadow_pte(parent_pte, shadow_trap_nonpresent_pte);
}
kvm_mmu_page_unlink_children(kvm, page);
if (!page->root_count) {
Expand Down Expand Up @@ -798,7 +816,7 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, hpa_t p)

if (level == 1) {
pte = table[index];
if (is_present_pte(pte) && is_writeble_pte(pte))
if (is_shadow_present_pte(pte) && is_writeble_pte(pte))
return 0;
mark_page_dirty(vcpu->kvm, v >> PAGE_SHIFT);
page_header_update_slot(vcpu->kvm, table, v);
Expand All @@ -808,7 +826,7 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, hpa_t p)
return 0;
}

if (table[index] == 0) {
if (table[index] == shadow_trap_nonpresent_pte) {
struct kvm_mmu_page *new_table;
gfn_t pseudo_gfn;

Expand All @@ -829,6 +847,15 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, hpa_t p)
}
}

static void nonpaging_prefetch_page(struct kvm_vcpu *vcpu,
struct kvm_mmu_page *sp)
{
int i;

for (i = 0; i < PT64_ENT_PER_PAGE; ++i)
sp->spt[i] = shadow_trap_nonpresent_pte;
}

static void mmu_free_roots(struct kvm_vcpu *vcpu)
{
int i;
Expand Down Expand Up @@ -943,6 +970,7 @@ static int nonpaging_init_context(struct kvm_vcpu *vcpu)
context->page_fault = nonpaging_page_fault;
context->gva_to_gpa = nonpaging_gva_to_gpa;
context->free = nonpaging_free;
context->prefetch_page = nonpaging_prefetch_page;
context->root_level = 0;
context->shadow_root_level = PT32E_ROOT_LEVEL;
context->root_hpa = INVALID_PAGE;
Expand Down Expand Up @@ -989,6 +1017,7 @@ static int paging64_init_context_common(struct kvm_vcpu *vcpu, int level)
context->new_cr3 = paging_new_cr3;
context->page_fault = paging64_page_fault;
context->gva_to_gpa = paging64_gva_to_gpa;
context->prefetch_page = paging64_prefetch_page;
context->free = paging_free;
context->root_level = level;
context->shadow_root_level = level;
Expand All @@ -1009,6 +1038,7 @@ static int paging32_init_context(struct kvm_vcpu *vcpu)
context->page_fault = paging32_page_fault;
context->gva_to_gpa = paging32_gva_to_gpa;
context->free = paging_free;
context->prefetch_page = paging32_prefetch_page;
context->root_level = PT32_ROOT_LEVEL;
context->shadow_root_level = PT32E_ROOT_LEVEL;
context->root_hpa = INVALID_PAGE;
Expand Down Expand Up @@ -1081,30 +1111,33 @@ static void mmu_pte_write_zap_pte(struct kvm_vcpu *vcpu,
struct kvm_mmu_page *child;

pte = *spte;
if (is_present_pte(pte)) {
if (is_shadow_present_pte(pte)) {
if (page->role.level == PT_PAGE_TABLE_LEVEL)
rmap_remove(spte);
else {
child = page_header(pte & PT64_BASE_ADDR_MASK);
mmu_page_remove_parent_pte(child, spte);
}
}
set_shadow_pte(spte, 0);
set_shadow_pte(spte, shadow_trap_nonpresent_pte);
kvm_flush_remote_tlbs(vcpu->kvm);
}

static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu,
struct kvm_mmu_page *page,
u64 *spte,
const void *new, int bytes)
const void *new, int bytes,
int offset_in_pte)
{
if (page->role.level != PT_PAGE_TABLE_LEVEL)
return;

if (page->role.glevels == PT32_ROOT_LEVEL)
paging32_update_pte(vcpu, page, spte, new, bytes);
paging32_update_pte(vcpu, page, spte, new, bytes,
offset_in_pte);
else
paging64_update_pte(vcpu, page, spte, new, bytes);
paging64_update_pte(vcpu, page, spte, new, bytes,
offset_in_pte);
}

void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
Expand All @@ -1126,6 +1159,7 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
int npte;

pgprintk("%s: gpa %llx bytes %d\n", __FUNCTION__, gpa, bytes);
kvm_mmu_audit(vcpu, "pre pte write");
if (gfn == vcpu->last_pt_write_gfn) {
++vcpu->last_pt_write_count;
if (vcpu->last_pt_write_count >= 3)
Expand Down Expand Up @@ -1181,10 +1215,12 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
spte = &page->spt[page_offset / sizeof(*spte)];
while (npte--) {
mmu_pte_write_zap_pte(vcpu, page, spte);
mmu_pte_write_new_pte(vcpu, page, spte, new, bytes);
mmu_pte_write_new_pte(vcpu, page, spte, new, bytes,
page_offset & (pte_size - 1));
++spte;
}
}
kvm_mmu_audit(vcpu, "post pte write");
}

int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva)
Expand Down Expand Up @@ -1359,22 +1395,33 @@ static void audit_mappings_page(struct kvm_vcpu *vcpu, u64 page_pte,
for (i = 0; i < PT64_ENT_PER_PAGE; ++i, va += va_delta) {
u64 ent = pt[i];

if (!(ent & PT_PRESENT_MASK))
if (ent == shadow_trap_nonpresent_pte)
continue;

va = canonicalize(va);
if (level > 1)
if (level > 1) {
if (ent == shadow_notrap_nonpresent_pte)
printk(KERN_ERR "audit: (%s) nontrapping pte"
" in nonleaf level: levels %d gva %lx"
" level %d pte %llx\n", audit_msg,
vcpu->mmu.root_level, va, level, ent);

audit_mappings_page(vcpu, ent, va, level - 1);
else {
} else {
gpa_t gpa = vcpu->mmu.gva_to_gpa(vcpu, va);
hpa_t hpa = gpa_to_hpa(vcpu, gpa);

if ((ent & PT_PRESENT_MASK)
if (is_shadow_present_pte(ent)
&& (ent & PT64_BASE_ADDR_MASK) != hpa)
printk(KERN_ERR "audit error: (%s) levels %d"
" gva %lx gpa %llx hpa %llx ent %llx\n",
printk(KERN_ERR "xx audit error: (%s) levels %d"
" gva %lx gpa %llx hpa %llx ent %llx %d\n",
audit_msg, vcpu->mmu.root_level,
va, gpa, hpa, ent);
va, gpa, hpa, ent, is_shadow_present_pte(ent));
else if (ent == shadow_notrap_nonpresent_pte
&& !is_error_hpa(hpa))
printk(KERN_ERR "audit: (%s) notrap shadow,"
" valid guest gva %lx\n", audit_msg, va);

}
}
}
Expand Down
52 changes: 39 additions & 13 deletions trunk/drivers/kvm/paging_tmpl.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
#define PT_INDEX(addr, level) PT64_INDEX(addr, level)
#define SHADOW_PT_INDEX(addr, level) PT64_INDEX(addr, level)
#define PT_LEVEL_MASK(level) PT64_LEVEL_MASK(level)
#define PT_LEVEL_BITS PT64_LEVEL_BITS
#ifdef CONFIG_X86_64
#define PT_MAX_FULL_LEVELS 4
#else
Expand All @@ -45,6 +46,7 @@
#define PT_INDEX(addr, level) PT32_INDEX(addr, level)
#define SHADOW_PT_INDEX(addr, level) PT64_INDEX(addr, level)
#define PT_LEVEL_MASK(level) PT32_LEVEL_MASK(level)
#define PT_LEVEL_BITS PT32_LEVEL_BITS
#define PT_MAX_FULL_LEVELS 2
#else
#error Invalid PTTYPE value
Expand Down Expand Up @@ -211,12 +213,12 @@ static void FNAME(set_pte_common)(struct kvm_vcpu *vcpu,
{
hpa_t paddr;
int dirty = gpte & PT_DIRTY_MASK;
u64 spte = *shadow_pte;
int was_rmapped = is_rmap_pte(spte);
u64 spte;
int was_rmapped = is_rmap_pte(*shadow_pte);

pgprintk("%s: spte %llx gpte %llx access %llx write_fault %d"
" user_fault %d gfn %lx\n",
__FUNCTION__, spte, (u64)gpte, access_bits,
__FUNCTION__, *shadow_pte, (u64)gpte, access_bits,
write_fault, user_fault, gfn);

if (write_fault && !dirty) {
Expand All @@ -236,7 +238,7 @@ static void FNAME(set_pte_common)(struct kvm_vcpu *vcpu,
FNAME(mark_pagetable_dirty)(vcpu->kvm, walker);
}

spte |= PT_PRESENT_MASK | PT_ACCESSED_MASK | PT_DIRTY_MASK;
spte = PT_PRESENT_MASK | PT_ACCESSED_MASK | PT_DIRTY_MASK;
spte |= gpte & PT64_NX_MASK;
if (!dirty)
access_bits &= ~PT_WRITABLE_MASK;
Expand All @@ -248,10 +250,8 @@ static void FNAME(set_pte_common)(struct kvm_vcpu *vcpu,
spte |= PT_USER_MASK;

if (is_error_hpa(paddr)) {
spte |= gaddr;
spte |= PT_SHADOW_IO_MARK;
spte &= ~PT_PRESENT_MASK;
set_shadow_pte(shadow_pte, spte);
set_shadow_pte(shadow_pte,
shadow_trap_nonpresent_pte | PT_SHADOW_IO_MARK);
return;
}

Expand Down Expand Up @@ -286,6 +286,7 @@ static void FNAME(set_pte_common)(struct kvm_vcpu *vcpu,
if (access_bits & PT_WRITABLE_MASK)
mark_page_dirty(vcpu->kvm, gaddr >> PAGE_SHIFT);

pgprintk("%s: setting spte %llx\n", __FUNCTION__, spte);
set_shadow_pte(shadow_pte, spte);
page_header_update_slot(vcpu->kvm, shadow_pte, gaddr);
if (!was_rmapped)
Expand All @@ -304,14 +305,18 @@ static void FNAME(set_pte)(struct kvm_vcpu *vcpu, pt_element_t gpte,
}

static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page,
u64 *spte, const void *pte, int bytes)
u64 *spte, const void *pte, int bytes,
int offset_in_pte)
{
pt_element_t gpte;

if (bytes < sizeof(pt_element_t))
return;
gpte = *(const pt_element_t *)pte;
if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK))
if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK)) {
if (!offset_in_pte && !is_present_pte(gpte))
set_shadow_pte(spte, shadow_notrap_nonpresent_pte);
return;
}
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,
Expand Down Expand Up @@ -368,7 +373,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
unsigned hugepage_access = 0;

shadow_ent = ((u64 *)__va(shadow_addr)) + index;
if (is_present_pte(*shadow_ent) || is_io_pte(*shadow_ent)) {
if (is_shadow_present_pte(*shadow_ent)) {
if (level == PT_PAGE_TABLE_LEVEL)
break;
shadow_addr = *shadow_ent & PT64_BASE_ADDR_MASK;
Expand Down Expand Up @@ -500,6 +505,26 @@ static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t vaddr)
return gpa;
}

static void FNAME(prefetch_page)(struct kvm_vcpu *vcpu,
struct kvm_mmu_page *sp)
{
int i;
pt_element_t *gpt;

if (sp->role.metaphysical || PTTYPE == 32) {
nonpaging_prefetch_page(vcpu, sp);
return;
}

gpt = kmap_atomic(gfn_to_page(vcpu->kvm, sp->gfn), KM_USER0);
for (i = 0; i < PT64_ENT_PER_PAGE; ++i)
if (is_present_pte(gpt[i]))
sp->spt[i] = shadow_trap_nonpresent_pte;
else
sp->spt[i] = shadow_notrap_nonpresent_pte;
kunmap_atomic(gpt, KM_USER0);
}

#undef pt_element_t
#undef guest_walker
#undef FNAME
Expand All @@ -508,4 +533,5 @@ static gpa_t FNAME(gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t vaddr)
#undef SHADOW_PT_INDEX
#undef PT_LEVEL_MASK
#undef PT_DIR_BASE_ADDR_MASK
#undef PT_LEVEL_BITS
#undef PT_MAX_FULL_LEVELS
Loading

0 comments on commit f24cb69

Please sign in to comment.