@@ -484,6 +484,11 @@ static inline bool kvm_s2_trans_writable(struct kvm_s2_trans *trans)
BUG();
}
+static inline u64 kvm_encode_nested_level(struct kvm_s2_trans *trans)
+{
+ BUG();
+}
+
static inline void kvm_nested_s2_flush(struct kvm *kvm) {}
static inline void kvm_nested_s2_wp(struct kvm *kvm) {}
static inline void kvm_nested_s2_clear(struct kvm *kvm) {}
@@ -2,6 +2,7 @@
#ifndef __ARM64_KVM_NESTED_H
#define __ARM64_KVM_NESTED_H
+#include <linux/bitfield.h>
#include <linux/kvm_host.h>
static inline bool nested_virt_in_use(const struct kvm_vcpu *vcpu)
@@ -79,4 +80,9 @@ void access_nested_id_reg(struct kvm_vcpu *v, struct sys_reg_params *p,
#define KVM_NV_GUEST_MAP_SZ GENMASK_ULL(56, 55)
+static inline u64 kvm_encode_nested_level(struct kvm_s2_trans *trans)
+{
+ return FIELD_PREP(KVM_NV_GUEST_MAP_SZ, trans->level);
+}
+
#endif /* __ARM64_KVM_NESTED_H */
@@ -1715,6 +1715,11 @@ static bool fault_supports_stage2_huge_mapping(struct kvm_memory_slot *memslot,
(hva & ~(map_size - 1)) + map_size <= uaddr_end;
}
+#define set_desc_bits(which, desc, val) \
+ do { \
+ desc = __ ## which(which ## _val(desc) | val); \
+ } while(0)
+
static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
struct kvm_s2_trans *nested,
struct kvm_memory_slot *memslot,
@@ -1736,6 +1741,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
unsigned long vma_pagesize, flags = 0;
struct kvm_s2_mmu *mmu = vcpu->arch.hw_mmu;
unsigned long max_map_size = PUD_SIZE;
+ u64 l1_s2_level;
write_fault = kvm_is_write_fault(vcpu);
exec_fault = kvm_vcpu_trap_is_iabt(vcpu);
@@ -1845,10 +1851,18 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
* Potentially reduce shadow S2 permissions to match the guest's own
* S2. For exec faults, we'd only reach this point if the guest
* actually allowed it (see kvm_s2_handle_perm_fault).
+ *
+ * Also encode the level of the nested translation in the SW bits of
+ * the PTE/PMD/PUD. This will be retrived on TLB invalidation from
+ * the guest.
*/
if (kvm_is_shadow_s2_fault(vcpu)) {
writable &= kvm_s2_trans_writable(nested);
readable &= kvm_s2_trans_readable(nested);
+
+ l1_s2_level = kvm_encode_nested_level(nested);
+ } else {
+ l1_s2_level = 0;
}
spin_lock(&kvm->mmu_lock);
@@ -1902,6 +1916,8 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
if (needs_exec)
new_pud = kvm_s2pud_mkexec(new_pud);
+ set_desc_bits(pud, new_pud, l1_s2_level);
+
ret = stage2_set_pud_huge(mmu, memcache, fault_ipa, &new_pud);
} else if (vma_pagesize == PMD_SIZE) {
pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
@@ -1917,6 +1933,8 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
if (needs_exec)
new_pmd = kvm_s2pmd_mkexec(new_pmd);
+ set_desc_bits(pmd, new_pmd, l1_s2_level);
+
ret = stage2_set_pmd_huge(mmu, memcache, fault_ipa, &new_pmd);
} else {
pte_t new_pte = kvm_pfn_pte(pfn, mem_type);
@@ -1932,6 +1950,8 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
if (needs_exec)
new_pte = kvm_s2pte_mkexec(new_pte);
+ set_desc_bits(pte, new_pte, l1_s2_level);
+
ret = stage2_set_pte(mmu, memcache, fault_ipa, &new_pte, flags);
}
Populate bits [56:55] of the leaf entry with the level provided by the guest's S2 translation. Signed-off-by: Marc Zyngier <maz@kernel.org> --- arch/arm/include/asm/kvm_mmu.h | 5 +++++ arch/arm64/include/asm/kvm_nested.h | 6 ++++++ virt/kvm/arm/mmu.c | 20 ++++++++++++++++++++ 3 files changed, 31 insertions(+)