KVM: introduce kvm->srcu and convert kvm_set_memory_region to SRCU update
Use two steps for memslot deletion: mark the slot invalid (which stops
instantiation of new shadow pages for that slot, but allows destruction),
then instantiate the new empty slot.
Also simplifies kvm_handle_hva locking.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 87d296d..2bb24a8 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -44,6 +44,7 @@
#include <linux/bitops.h>
#include <linux/spinlock.h>
#include <linux/compat.h>
+#include <linux/srcu.h>
#include <asm/processor.h>
#include <asm/io.h>
@@ -213,7 +214,7 @@
unsigned long address)
{
struct kvm *kvm = mmu_notifier_to_kvm(mn);
- int need_tlb_flush;
+ int need_tlb_flush, idx;
/*
* When ->invalidate_page runs, the linux pte has been zapped
@@ -233,10 +234,12 @@
* pte after kvm_unmap_hva returned, without noticing the page
* is going to be freed.
*/
+ idx = srcu_read_lock(&kvm->srcu);
spin_lock(&kvm->mmu_lock);
kvm->mmu_notifier_seq++;
need_tlb_flush = kvm_unmap_hva(kvm, address);
spin_unlock(&kvm->mmu_lock);
+ srcu_read_unlock(&kvm->srcu, idx);
/* we've to flush the tlb before the pages can be freed */
if (need_tlb_flush)
@@ -250,11 +253,14 @@
pte_t pte)
{
struct kvm *kvm = mmu_notifier_to_kvm(mn);
+ int idx;
+ idx = srcu_read_lock(&kvm->srcu);
spin_lock(&kvm->mmu_lock);
kvm->mmu_notifier_seq++;
kvm_set_spte_hva(kvm, address, pte);
spin_unlock(&kvm->mmu_lock);
+ srcu_read_unlock(&kvm->srcu, idx);
}
static void kvm_mmu_notifier_invalidate_range_start(struct mmu_notifier *mn,
@@ -263,8 +269,9 @@
unsigned long end)
{
struct kvm *kvm = mmu_notifier_to_kvm(mn);
- int need_tlb_flush = 0;
+ int need_tlb_flush = 0, idx;
+ idx = srcu_read_lock(&kvm->srcu);
spin_lock(&kvm->mmu_lock);
/*
* The count increase must become visible at unlock time as no
@@ -275,6 +282,7 @@
for (; start < end; start += PAGE_SIZE)
need_tlb_flush |= kvm_unmap_hva(kvm, start);
spin_unlock(&kvm->mmu_lock);
+ srcu_read_unlock(&kvm->srcu, idx);
/* we've to flush the tlb before the pages can be freed */
if (need_tlb_flush)
@@ -312,11 +320,13 @@
unsigned long address)
{
struct kvm *kvm = mmu_notifier_to_kvm(mn);
- int young;
+ int young, idx;
+ idx = srcu_read_lock(&kvm->srcu);
spin_lock(&kvm->mmu_lock);
young = kvm_age_hva(kvm, address);
spin_unlock(&kvm->mmu_lock);
+ srcu_read_unlock(&kvm->srcu, idx);
if (young)
kvm_flush_remote_tlbs(kvm);
@@ -379,11 +389,15 @@
kvm->memslots = kzalloc(sizeof(struct kvm_memslots), GFP_KERNEL);
if (!kvm->memslots)
goto out_err;
+ if (init_srcu_struct(&kvm->srcu))
+ goto out_err;
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
page = alloc_page(GFP_KERNEL | __GFP_ZERO);
- if (!page)
+ if (!page) {
+ cleanup_srcu_struct(&kvm->srcu);
goto out_err;
+ }
kvm->coalesced_mmio_ring =
(struct kvm_coalesced_mmio_ring *)page_address(page);
@@ -391,6 +405,7 @@
r = kvm_init_mmu_notifier(kvm);
if (r) {
+ cleanup_srcu_struct(&kvm->srcu);
#ifdef KVM_COALESCED_MMIO_PAGE_OFFSET
put_page(page);
#endif
@@ -480,6 +495,7 @@
#else
kvm_arch_flush_shadow(kvm);
#endif
+ cleanup_srcu_struct(&kvm->srcu);
kvm_arch_destroy_vm(kvm);
hardware_disable_all();
mmdrop(mm);
@@ -521,12 +537,13 @@
struct kvm_userspace_memory_region *mem,
int user_alloc)
{
- int r;
+ int r, flush_shadow = 0;
gfn_t base_gfn;
unsigned long npages;
unsigned long i;
struct kvm_memory_slot *memslot;
struct kvm_memory_slot old, new;
+ struct kvm_memslots *slots, *old_memslots;
r = -EINVAL;
/* General sanity checks */
@@ -588,15 +605,7 @@
memset(new.rmap, 0, npages * sizeof(*new.rmap));
new.user_alloc = user_alloc;
- /*
- * hva_to_rmmap() serialzies with the mmu_lock and to be
- * safe it has to ignore memslots with !user_alloc &&
- * !userspace_addr.
- */
- if (user_alloc)
- new.userspace_addr = mem->userspace_addr;
- else
- new.userspace_addr = 0;
+ new.userspace_addr = mem->userspace_addr;
}
if (!npages)
goto skip_lpage;
@@ -651,8 +660,9 @@
if (!new.dirty_bitmap)
goto out_free;
memset(new.dirty_bitmap, 0, dirty_bytes);
+ /* destroy any largepage mappings for dirty tracking */
if (old.npages)
- kvm_arch_flush_shadow(kvm);
+ flush_shadow = 1;
}
#else /* not defined CONFIG_S390 */
new.user_alloc = user_alloc;
@@ -660,34 +670,72 @@
new.userspace_addr = mem->userspace_addr;
#endif /* not defined CONFIG_S390 */
- if (!npages)
+ if (!npages) {
+ r = -ENOMEM;
+ slots = kzalloc(sizeof(struct kvm_memslots), GFP_KERNEL);
+ if (!slots)
+ goto out_free;
+ memcpy(slots, kvm->memslots, sizeof(struct kvm_memslots));
+ if (mem->slot >= slots->nmemslots)
+ slots->nmemslots = mem->slot + 1;
+ slots->memslots[mem->slot].flags |= KVM_MEMSLOT_INVALID;
+
+ old_memslots = kvm->memslots;
+ rcu_assign_pointer(kvm->memslots, slots);
+ synchronize_srcu_expedited(&kvm->srcu);
+ /* From this point no new shadow pages pointing to a deleted
+ * memslot will be created.
+ *
+ * validation of sp->gfn happens in:
+ * - gfn_to_hva (kvm_read_guest, gfn_to_pfn)
+ * - kvm_is_visible_gfn (mmu_check_roots)
+ */
kvm_arch_flush_shadow(kvm);
+ kfree(old_memslots);
+ }
r = kvm_arch_prepare_memory_region(kvm, &new, old, mem, user_alloc);
if (r)
goto out_free;
- spin_lock(&kvm->mmu_lock);
- if (mem->slot >= kvm->memslots->nmemslots)
- kvm->memslots->nmemslots = mem->slot + 1;
+#ifdef CONFIG_DMAR
+ /* map the pages in iommu page table */
+ if (npages) {
+ r = kvm_iommu_map_pages(kvm, &new);
+ if (r)
+ goto out_free;
+ }
+#endif
- *memslot = new;
- spin_unlock(&kvm->mmu_lock);
+ r = -ENOMEM;
+ slots = kzalloc(sizeof(struct kvm_memslots), GFP_KERNEL);
+ if (!slots)
+ goto out_free;
+ memcpy(slots, kvm->memslots, sizeof(struct kvm_memslots));
+ if (mem->slot >= slots->nmemslots)
+ slots->nmemslots = mem->slot + 1;
+
+ /* actual memory is freed via old in kvm_free_physmem_slot below */
+ if (!npages) {
+ new.rmap = NULL;
+ new.dirty_bitmap = NULL;
+ for (i = 0; i < KVM_NR_PAGE_SIZES - 1; ++i)
+ new.lpage_info[i] = NULL;
+ }
+
+ slots->memslots[mem->slot] = new;
+ old_memslots = kvm->memslots;
+ rcu_assign_pointer(kvm->memslots, slots);
+ synchronize_srcu_expedited(&kvm->srcu);
kvm_arch_commit_memory_region(kvm, mem, old, user_alloc);
- kvm_free_physmem_slot(&old, npages ? &new : NULL);
- /* Slot deletion case: we have to update the current slot */
- spin_lock(&kvm->mmu_lock);
- if (!npages)
- *memslot = old;
- spin_unlock(&kvm->mmu_lock);
-#ifdef CONFIG_DMAR
- /* map the pages in iommu page table */
- r = kvm_iommu_map_pages(kvm, memslot);
- if (r)
- goto out;
-#endif
+ kvm_free_physmem_slot(&old, &new);
+ kfree(old_memslots);
+
+ if (flush_shadow)
+ kvm_arch_flush_shadow(kvm);
+
return 0;
out_free:
@@ -787,7 +835,7 @@
struct kvm_memory_slot *gfn_to_memslot_unaliased(struct kvm *kvm, gfn_t gfn)
{
int i;
- struct kvm_memslots *slots = kvm->memslots;
+ struct kvm_memslots *slots = rcu_dereference(kvm->memslots);
for (i = 0; i < slots->nmemslots; ++i) {
struct kvm_memory_slot *memslot = &slots->memslots[i];
@@ -809,12 +857,15 @@
int kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
{
int i;
- struct kvm_memslots *slots = kvm->memslots;
+ struct kvm_memslots *slots = rcu_dereference(kvm->memslots);
gfn = unalias_gfn(kvm, gfn);
for (i = 0; i < KVM_MEMORY_SLOTS; ++i) {
struct kvm_memory_slot *memslot = &slots->memslots[i];
+ if (memslot->flags & KVM_MEMSLOT_INVALID)
+ continue;
+
if (gfn >= memslot->base_gfn
&& gfn < memslot->base_gfn + memslot->npages)
return 1;
@@ -823,13 +874,31 @@
}
EXPORT_SYMBOL_GPL(kvm_is_visible_gfn);
+int memslot_id(struct kvm *kvm, gfn_t gfn)
+{
+ int i;
+ struct kvm_memslots *slots = rcu_dereference(kvm->memslots);
+ struct kvm_memory_slot *memslot = NULL;
+
+ gfn = unalias_gfn(kvm, gfn);
+ for (i = 0; i < slots->nmemslots; ++i) {
+ memslot = &slots->memslots[i];
+
+ if (gfn >= memslot->base_gfn
+ && gfn < memslot->base_gfn + memslot->npages)
+ break;
+ }
+
+ return memslot - slots->memslots;
+}
+
unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
{
struct kvm_memory_slot *slot;
gfn = unalias_gfn(kvm, gfn);
slot = gfn_to_memslot_unaliased(kvm, gfn);
- if (!slot)
+ if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
return bad_hva();
return (slot->userspace_addr + (gfn - slot->base_gfn) * PAGE_SIZE);
}