summaryrefslogtreecommitdiff
path: root/arch/x86/kvm
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86/kvm')
-rw-r--r--arch/x86/kvm/cpuid.c24
-rw-r--r--arch/x86/kvm/hyperv.c40
-rw-r--r--arch/x86/kvm/hyperv.h2
-rw-r--r--arch/x86/kvm/mmu.h24
-rw-r--r--arch/x86/kvm/mmu/mmu.c111
-rw-r--r--arch/x86/kvm/mmu/spte.c28
-rw-r--r--arch/x86/kvm/mmu/spte.h10
-rw-r--r--arch/x86/kvm/mmu/tdp_iter.h34
-rw-r--r--arch/x86/kvm/mmu/tdp_mmu.c99
-rw-r--r--arch/x86/kvm/pmu.h9
-rw-r--r--arch/x86/kvm/svm/avic.c3
-rw-r--r--arch/x86/kvm/svm/pmu.c29
-rw-r--r--arch/x86/kvm/svm/sev.c112
-rw-r--r--arch/x86/kvm/svm/svm.c1
-rw-r--r--arch/x86/kvm/svm/svm.h2
-rw-r--r--arch/x86/kvm/vmx/nested.c5
-rw-r--r--arch/x86/kvm/vmx/pmu_intel.c8
-rw-r--r--arch/x86/kvm/vmx/vmx.c7
-rw-r--r--arch/x86/kvm/vmx/vmx.h1
-rw-r--r--arch/x86/kvm/x86.c95
20 files changed, 440 insertions, 204 deletions
diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index b24ca7f4ed7c..0c1ba6aa0765 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -887,6 +887,11 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
union cpuid10_eax eax;
union cpuid10_edx edx;
+ if (!static_cpu_has(X86_FEATURE_ARCH_PERFMON)) {
+ entry->eax = entry->ebx = entry->ecx = entry->edx = 0;
+ break;
+ }
+
perf_get_x86_pmu_capability(&cap);
/*
@@ -1085,12 +1090,21 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
case 0x80000000:
entry->eax = min(entry->eax, 0x80000021);
/*
- * Serializing LFENCE is reported in a multitude of ways,
- * and NullSegClearsBase is not reported in CPUID on Zen2;
- * help userspace by providing the CPUID leaf ourselves.
+ * Serializing LFENCE is reported in a multitude of ways, and
+ * NullSegClearsBase is not reported in CPUID on Zen2; help
+ * userspace by providing the CPUID leaf ourselves.
+ *
+ * However, only do it if the host has CPUID leaf 0x8000001d.
+ * QEMU thinks that it can query the host blindly for that
+ * CPUID leaf if KVM reports that it supports 0x8000001d or
+ * above. The processor merrily returns values from the
+ * highest Intel leaf which QEMU tries to use as the guest's
+ * 0x8000001d. Even worse, this can result in an infinite
+ * loop if said highest leaf has no subleaves indexed by ECX.
*/
- if (static_cpu_has(X86_FEATURE_LFENCE_RDTSC)
- || !static_cpu_has_bug(X86_BUG_NULL_SEG))
+ if (entry->eax >= 0x8000001d &&
+ (static_cpu_has(X86_FEATURE_LFENCE_RDTSC)
+ || !static_cpu_has_bug(X86_BUG_NULL_SEG)))
entry->eax = max(entry->eax, 0x80000021);
break;
case 0x80000001:
diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
index 123b677111c5..46f9dfb60469 100644
--- a/arch/x86/kvm/hyperv.c
+++ b/arch/x86/kvm/hyperv.c
@@ -1135,11 +1135,13 @@ void kvm_hv_setup_tsc_page(struct kvm *kvm,
BUILD_BUG_ON(sizeof(tsc_seq) != sizeof(hv->tsc_ref.tsc_sequence));
BUILD_BUG_ON(offsetof(struct ms_hyperv_tsc_page, tsc_sequence) != 0);
+ mutex_lock(&hv->hv_lock);
+
if (hv->hv_tsc_page_status == HV_TSC_PAGE_BROKEN ||
+ hv->hv_tsc_page_status == HV_TSC_PAGE_SET ||
hv->hv_tsc_page_status == HV_TSC_PAGE_UNSET)
- return;
+ goto out_unlock;
- mutex_lock(&hv->hv_lock);
if (!(hv->hv_tsc_page & HV_X64_MSR_TSC_REFERENCE_ENABLE))
goto out_unlock;
@@ -1201,45 +1203,19 @@ out_unlock:
mutex_unlock(&hv->hv_lock);
}
-void kvm_hv_invalidate_tsc_page(struct kvm *kvm)
+void kvm_hv_request_tsc_page_update(struct kvm *kvm)
{
struct kvm_hv *hv = to_kvm_hv(kvm);
- u64 gfn;
- int idx;
-
- if (hv->hv_tsc_page_status == HV_TSC_PAGE_BROKEN ||
- hv->hv_tsc_page_status == HV_TSC_PAGE_UNSET ||
- tsc_page_update_unsafe(hv))
- return;
mutex_lock(&hv->hv_lock);
- if (!(hv->hv_tsc_page & HV_X64_MSR_TSC_REFERENCE_ENABLE))
- goto out_unlock;
-
- /* Preserve HV_TSC_PAGE_GUEST_CHANGED/HV_TSC_PAGE_HOST_CHANGED states */
- if (hv->hv_tsc_page_status == HV_TSC_PAGE_SET)
- hv->hv_tsc_page_status = HV_TSC_PAGE_UPDATING;
+ if (hv->hv_tsc_page_status == HV_TSC_PAGE_SET &&
+ !tsc_page_update_unsafe(hv))
+ hv->hv_tsc_page_status = HV_TSC_PAGE_HOST_CHANGED;
- gfn = hv->hv_tsc_page >> HV_X64_MSR_TSC_REFERENCE_ADDRESS_SHIFT;
-
- hv->tsc_ref.tsc_sequence = 0;
-
- /*
- * Take the srcu lock as memslots will be accessed to check the gfn
- * cache generation against the memslots generation.
- */
- idx = srcu_read_lock(&kvm->srcu);
- if (kvm_write_guest(kvm, gfn_to_gpa(gfn),
- &hv->tsc_ref, sizeof(hv->tsc_ref.tsc_sequence)))
- hv->hv_tsc_page_status = HV_TSC_PAGE_BROKEN;
- srcu_read_unlock(&kvm->srcu, idx);
-
-out_unlock:
mutex_unlock(&hv->hv_lock);
}
-
static bool hv_check_msr_access(struct kvm_vcpu_hv *hv_vcpu, u32 msr)
{
if (!hv_vcpu->enforce_cpuid)
diff --git a/arch/x86/kvm/hyperv.h b/arch/x86/kvm/hyperv.h
index e19c00ee9ab3..da2737f2a956 100644
--- a/arch/x86/kvm/hyperv.h
+++ b/arch/x86/kvm/hyperv.h
@@ -137,7 +137,7 @@ void kvm_hv_process_stimers(struct kvm_vcpu *vcpu);
void kvm_hv_setup_tsc_page(struct kvm *kvm,
struct pvclock_vcpu_time_info *hv_clock);
-void kvm_hv_invalidate_tsc_page(struct kvm *kvm);
+void kvm_hv_request_tsc_page_update(struct kvm *kvm);
void kvm_hv_init_vm(struct kvm *kvm);
void kvm_hv_destroy_vm(struct kvm *kvm);
diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
index e6cae6f22683..a335e7f1f69e 100644
--- a/arch/x86/kvm/mmu.h
+++ b/arch/x86/kvm/mmu.h
@@ -65,6 +65,30 @@ static __always_inline u64 rsvd_bits(int s, int e)
return ((2ULL << (e - s)) - 1) << s;
}
+/*
+ * The number of non-reserved physical address bits irrespective of features
+ * that repurpose legal bits, e.g. MKTME.
+ */
+extern u8 __read_mostly shadow_phys_bits;
+
+static inline gfn_t kvm_mmu_max_gfn(void)
+{
+ /*
+ * Note that this uses the host MAXPHYADDR, not the guest's.
+ * EPT/NPT cannot support GPAs that would exceed host.MAXPHYADDR;
+ * assuming KVM is running on bare metal, guest accesses beyond
+ * host.MAXPHYADDR will hit a #PF(RSVD) and never cause a vmexit
+ * (either EPT Violation/Misconfig or #NPF), and so KVM will never
+ * install a SPTE for such addresses. If KVM is running as a VM
+ * itself, on the other hand, it might see a MAXPHYADDR that is less
+ * than hardware's real MAXPHYADDR. Using the host MAXPHYADDR
+ * disallows such SPTEs entirely and simplifies the TDP MMU.
+ */
+ int max_gpa_bits = likely(tdp_enabled) ? shadow_phys_bits : 52;
+
+ return (1ULL << (max_gpa_bits - PAGE_SHIFT)) - 1;
+}
+
void kvm_mmu_set_mmio_spte_mask(u64 mmio_value, u64 mmio_mask, u64 access_mask);
void kvm_mmu_set_ept_masks(bool has_ad_bits, bool has_exec_only);
diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 8f19ea752704..311e4e1d7870 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -473,30 +473,6 @@ retry:
}
#endif
-static bool spte_has_volatile_bits(u64 spte)
-{
- if (!is_shadow_present_pte(spte))
- return false;
-
- /*
- * Always atomically update spte if it can be updated
- * out of mmu-lock, it can ensure dirty bit is not lost,
- * also, it can help us to get a stable is_writable_pte()
- * to ensure tlb flush is not missed.
- */
- if (spte_can_locklessly_be_made_writable(spte) ||
- is_access_track_spte(spte))
- return true;
-
- if (spte_ad_enabled(spte)) {
- if ((spte & shadow_accessed_mask) == 0 ||
- (is_writable_pte(spte) && (spte & shadow_dirty_mask) == 0))
- return true;
- }
-
- return false;
-}
-
/* Rules for using mmu_spte_set:
* Set the sptep from nonpresent to present.
* Note: the sptep being assigned *must* be either not present
@@ -557,7 +533,7 @@ static bool mmu_spte_update(u64 *sptep, u64 new_spte)
* we always atomically update it, see the comments in
* spte_has_volatile_bits().
*/
- if (spte_can_locklessly_be_made_writable(old_spte) &&
+ if (is_mmu_writable_spte(old_spte) &&
!is_writable_pte(new_spte))
flush = true;
@@ -591,7 +567,8 @@ static int mmu_spte_clear_track_bits(struct kvm *kvm, u64 *sptep)
u64 old_spte = *sptep;
int level = sptep_to_sp(sptep)->role.level;
- if (!spte_has_volatile_bits(old_spte))
+ if (!is_shadow_present_pte(old_spte) ||
+ !spte_has_volatile_bits(old_spte))
__update_clear_spte_fast(sptep, 0ull);
else
old_spte = __update_clear_spte_slow(sptep, 0ull);
@@ -1187,7 +1164,7 @@ static bool spte_write_protect(u64 *sptep, bool pt_protect)
u64 spte = *sptep;
if (!is_writable_pte(spte) &&
- !(pt_protect && spte_can_locklessly_be_made_writable(spte)))
+ !(pt_protect && is_mmu_writable_spte(spte)))
return false;
rmap_printk("spte %p %llx\n", sptep, *sptep);
@@ -2804,8 +2781,12 @@ static int host_pfn_mapping_level(struct kvm *kvm, gfn_t gfn, kvm_pfn_t pfn,
const struct kvm_memory_slot *slot)
{
unsigned long hva;
- pte_t *pte;
- int level;
+ unsigned long flags;
+ int level = PG_LEVEL_4K;
+ pgd_t pgd;
+ p4d_t p4d;
+ pud_t pud;
+ pmd_t pmd;
if (!PageCompound(pfn_to_page(pfn)) && !kvm_is_zone_device_pfn(pfn))
return PG_LEVEL_4K;
@@ -2820,10 +2801,43 @@ static int host_pfn_mapping_level(struct kvm *kvm, gfn_t gfn, kvm_pfn_t pfn,
*/
hva = __gfn_to_hva_memslot(slot, gfn);
- pte = lookup_address_in_mm(kvm->mm, hva, &level);
- if (unlikely(!pte))
- return PG_LEVEL_4K;
+ /*
+ * Lookup the mapping level in the current mm. The information
+ * may become stale soon, but it is safe to use as long as
+ * 1) mmu_notifier_retry was checked after taking mmu_lock, and
+ * 2) mmu_lock is taken now.
+ *
+ * We still need to disable IRQs to prevent concurrent tear down
+ * of page tables.
+ */
+ local_irq_save(flags);
+
+ pgd = READ_ONCE(*pgd_offset(kvm->mm, hva));
+ if (pgd_none(pgd))
+ goto out;
+ p4d = READ_ONCE(*p4d_offset(&pgd, hva));
+ if (p4d_none(p4d) || !p4d_present(p4d))
+ goto out;
+
+ pud = READ_ONCE(*pud_offset(&p4d, hva));
+ if (pud_none(pud) || !pud_present(pud))
+ goto out;
+
+ if (pud_large(pud)) {
+ level = PG_LEVEL_1G;
+ goto out;
+ }
+
+ pmd = READ_ONCE(*pmd_offset(&pud, hva));
+ if (pmd_none(pmd) || !pmd_present(pmd))
+ goto out;
+
+ if (pmd_large(pmd))
+ level = PG_LEVEL_2M;
+
+out:
+ local_irq_restore(flags);
return level;
}
@@ -2992,9 +3006,15 @@ static bool handle_abnormal_pfn(struct kvm_vcpu *vcpu, struct kvm_page_fault *fa
/*
* If MMIO caching is disabled, emulate immediately without
* touching the shadow page tables as attempting to install an
- * MMIO SPTE will just be an expensive nop.
+ * MMIO SPTE will just be an expensive nop. Do not cache MMIO
+ * whose gfn is greater than host.MAXPHYADDR, any guest that
+ * generates such gfns is running nested and is being tricked
+ * by L0 userspace (you can observe gfn > L1.MAXPHYADDR if
+ * and only if L1's MAXPHYADDR is inaccurate with respect to
+ * the hardware's).
*/
- if (unlikely(!shadow_mmio_value)) {
+ if (unlikely(!shadow_mmio_value) ||
+ unlikely(fault->gfn > kvm_mmu_max_gfn())) {
*ret_val = RET_PF_EMULATE;
return true;
}
@@ -3153,8 +3173,7 @@ static int fast_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
* be removed in the fast path only if the SPTE was
* write-protected for dirty-logging or access tracking.
*/
- if (fault->write &&
- spte_can_locklessly_be_made_writable(spte)) {
+ if (fault->write && is_mmu_writable_spte(spte)) {
new_spte |= PT_WRITABLE_MASK;
/*
@@ -6237,12 +6256,24 @@ static int set_nx_huge_pages(const char *val, const struct kernel_param *kp)
return 0;
}
-int kvm_mmu_module_init(void)
+/*
+ * nx_huge_pages needs to be resolved to true/false when kvm.ko is loaded, as
+ * its default value of -1 is technically undefined behavior for a boolean.
+ */
+void kvm_mmu_x86_module_init(void)
{
- int ret = -ENOMEM;
-
if (nx_huge_pages == -1)
__set_nx_huge_pages(get_nx_auto_mode());
+}
+
+/*
+ * The bulk of the MMU initialization is deferred until the vendor module is
+ * loaded as many of the masks/values may be modified by VMX or SVM, i.e. need
+ * to be reset when a potentially different vendor module is loaded.
+ */
+int kvm_mmu_vendor_module_init(void)
+{
+ int ret = -ENOMEM;
/*
* MMU roles use union aliasing which is, generally speaking, an
@@ -6290,7 +6321,7 @@ void kvm_mmu_destroy(struct kvm_vcpu *vcpu)
mmu_free_memory_caches(vcpu);
}
-void kvm_mmu_module_exit(void)
+void kvm_mmu_vendor_module_exit(void)
{
mmu_destroy_caches();
percpu_counter_destroy(&kvm_total_used_mmu_pages);
diff --git a/arch/x86/kvm/mmu/spte.c b/arch/x86/kvm/mmu/spte.c
index 4739b53c9734..e5c0b6db6f2c 100644
--- a/arch/x86/kvm/mmu/spte.c
+++ b/arch/x86/kvm/mmu/spte.c
@@ -90,6 +90,34 @@ static bool kvm_is_mmio_pfn(kvm_pfn_t pfn)
E820_TYPE_RAM);
}
+/*
+ * Returns true if the SPTE has bits that may be set without holding mmu_lock.
+ * The caller is responsible for checking if the SPTE is shadow-present, and
+ * for determining whether or not the caller cares about non-leaf SPTEs.
+ */
+bool spte_has_volatile_bits(u64 spte)
+{
+ /*
+ * Always atomically update spte if it can be updated
+ * out of mmu-lock, it can ensure dirty bit is not lost,
+ * also, it can help us to get a stable is_writable_pte()
+ * to ensure tlb flush is not missed.
+ */
+ if (!is_writable_pte(spte) && is_mmu_writable_spte(spte))
+ return true;
+
+ if (is_access_track_spte(spte))
+ return true;
+
+ if (spte_ad_enabled(spte)) {
+ if (!(spte & shadow_accessed_mask) ||
+ (is_writable_pte(spte) && !(spte & shadow_dirty_mask)))
+ return true;
+ }
+
+ return false;
+}
+
bool make_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
const struct kvm_memory_slot *slot,
unsigned int pte_access, gfn_t gfn, kvm_pfn_t pfn,
diff --git a/arch/x86/kvm/mmu/spte.h b/arch/x86/kvm/mmu/spte.h
index 73f12615416f..80ab0f5cff01 100644
--- a/arch/x86/kvm/mmu/spte.h
+++ b/arch/x86/kvm/mmu/spte.h
@@ -201,12 +201,6 @@ static inline bool is_removed_spte(u64 spte)
*/
extern u64 __read_mostly shadow_nonpresent_or_rsvd_lower_gfn_mask;
-/*
- * The number of non-reserved physical address bits irrespective of features
- * that repurpose legal bits, e.g. MKTME.
- */
-extern u8 __read_mostly shadow_phys_bits;
-
static inline bool is_mmio_spte(u64 spte)
{
return (spte & shadow_mmio_mask) == shadow_mmio_value &&
@@ -396,7 +390,7 @@ static inline void check_spte_writable_invariants(u64 spte)
"kvm: Writable SPTE is not MMU-writable: %llx", spte);
}
-static inline bool spte_can_locklessly_be_made_writable(u64 spte)
+static inline bool is_mmu_writable_spte(u64 spte)
{
return spte & shadow_mmu_writable_mask;
}
@@ -410,6 +404,8 @@ static inline u64 get_mmio_spte_generation(u64 spte)
return gen;
}
+bool spte_has_volatile_bits(u64 spte);
+
bool make_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
const struct kvm_memory_slot *slot,
unsigned int pte_access, gfn_t gfn, kvm_pfn_t pfn,
diff --git a/arch/x86/kvm/mmu/tdp_iter.h b/arch/x86/kvm/mmu/tdp_iter.h
index b1eaf6ec0e0b..f0af385c56e0 100644
--- a/arch/x86/kvm/mmu/tdp_iter.h
+++ b/arch/x86/kvm/mmu/tdp_iter.h
@@ -6,6 +6,7 @@
#include <linux/kvm_host.h>
#include "mmu.h"
+#include "spte.h"
/*
* TDP MMU SPTEs are RCU protected to allow paging structures (non-leaf SPTEs)
@@ -17,9 +18,38 @@ static inline u64 kvm_tdp_mmu_read_spte(tdp_ptep_t sptep)
{
return READ_ONCE(*rcu_dereference(sptep));
}
-static inline void kvm_tdp_mmu_write_spte(tdp_ptep_t sptep, u64 val)
+
+static inline u64 kvm_tdp_mmu_write_spte_atomic(tdp_ptep_t sptep, u64 new_spte)
+{
+ return xchg(rcu_dereference(sptep), new_spte);
+}
+
+static inline void __kvm_tdp_mmu_write_spte(tdp_ptep_t sptep, u64 new_spte)
+{
+ WRITE_ONCE(*rcu_dereference(sptep), new_spte);
+}
+
+static inline u64 kvm_tdp_mmu_write_spte(tdp_ptep_t sptep, u64 old_spte,
+ u64 new_spte, int level)
{
- WRITE_ONCE(*rcu_dereference(sptep), val);
+ /*
+ * Atomically write the SPTE if it is a shadow-present, leaf SPTE with
+ * volatile bits, i.e. has bits that can be set outside of mmu_lock.
+ * The Writable bit can be set by KVM's fast page fault handler, and
+ * Accessed and Dirty bits can be set by the CPU.
+ *
+ * Note, non-leaf SPTEs do have Accessed bits and those bits are
+ * technically volatile, but KVM doesn't consume the Accessed bit of
+ * non-leaf SPTEs, i.e. KVM doesn't care if it clobbers the bit. This
+ * logic needs to be reassessed if KVM were to use non-leaf Accessed
+ * bits, e.g. to skip stepping down into child SPTEs when aging SPTEs.
+ */
+ if (is_shadow_present_pte(old_spte) && is_last_spte(old_spte, level) &&
+ spte_has_volatile_bits(old_spte))
+ return kvm_tdp_mmu_write_spte_atomic(sptep, new_spte);
+
+ __kvm_tdp_mmu_write_spte(sptep, new_spte);
+ return old_spte;
}
/*
diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
index d71d177ae6b8..922b06bf4b94 100644
--- a/arch/x86/kvm/mmu/tdp_mmu.c
+++ b/arch/x86/kvm/mmu/tdp_mmu.c
@@ -51,7 +51,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
if (!kvm->arch.tdp_mmu_enabled)
return;
- flush_workqueue(kvm->arch.tdp_mmu_zap_wq);
+ /* Also waits for any queued work items. */
destroy_workqueue(kvm->arch.tdp_mmu_zap_wq);
WARN_ON(!list_empty(&kvm->arch.tdp_mmu_pages));
@@ -426,9 +426,9 @@ static void handle_removed_pt(struct kvm *kvm, tdp_ptep_t pt, bool shared)
tdp_mmu_unlink_sp(kvm, sp, shared);
for (i = 0; i < PT64_ENT_PER_PAGE; i++) {
- u64 *sptep = rcu_dereference(pt) + i;
+ tdp_ptep_t sptep = pt + i;
gfn_t gfn = base_gfn + i * KVM_PAGES_PER_HPAGE(level);
- u64 old_child_spte;
+ u64 old_spte;
if (shared) {
/*
@@ -440,8 +440,8 @@ static void handle_removed_pt(struct kvm *kvm, tdp_ptep_t pt, bool shared)
* value to the removed SPTE value.
*/
for (;;) {
- old_child_spte = xchg(sptep, REMOVED_SPTE);
- if (!is_removed_spte(old_child_spte))
+ old_spte = kvm_tdp_mmu_write_spte_atomic(sptep, REMOVED_SPTE);
+ if (!is_removed_spte(old_spte))
break;
cpu_relax();
}
@@ -455,23 +455,43 @@ static void handle_removed_pt(struct kvm *kvm, tdp_ptep_t pt, bool shared)
* are guarded by the memslots generation, not by being
* unreachable.
*/
- old_child_spte = READ_ONCE(*sptep);
- if (!is_shadow_present_pte(old_child_spte))
+ old_spte = kvm_tdp_mmu_read_spte(sptep);
+ if (!is_shadow_present_pte(old_spte))
continue;
/*
- * Marking the SPTE as a removed SPTE is not
- * strictly necessary here as the MMU lock will
- * stop other threads from concurrently modifying
- * this SPTE. Using the removed SPTE value keeps
- * the two branches consistent and simplifies
- * the function.
+ * Use the common helper instead of a raw WRITE_ONCE as
+ * the SPTE needs to be updated atomically if it can be
+ * modified by a different vCPU outside of mmu_lock.
+ * Even though the parent SPTE is !PRESENT, the TLB
+ * hasn't yet been flushed, and both Intel and AMD
+ * document that A/D assists can use upper-level PxE
+ * entries that are cached in the TLB, i.e. the CPU can
+ * still access the page and mark it dirty.
+ *
+ * No retry is needed in the atomic update path as the
+ * sole concern is dropping a Dirty bit, i.e. no other
+ * task can zap/remove the SPTE as mmu_lock is held for
+ * write. Marking the SPTE as a removed SPTE is not
+ * strictly necessary for the same reason, but using
+ * the remove SPTE value keeps the shared/exclusive
+ * paths consistent and allows the handle_changed_spte()
+ * call below to hardcode the new value to REMOVED_SPTE.
+ *
+ * Note, even though dropping a Dirty bit is the only
+ * scenario where a non-atomic update could result in a
+ * functional bug, simply checking the Dirty bit isn't
+ * sufficient as a fast page fault could read the upper
+ * level SPTE before it is zapped, and then make this
+ * target SPTE writable, resume the guest, and set the
+ * Dirty bit between reading the SPTE above and writing
+ * it here.
*/
- WRITE_ONCE(*sptep, REMOVED_SPTE);
+ old_spte = kvm_tdp_mmu_write_spte(sptep, old_spte,
+ REMOVED_SPTE, level);
}
handle_changed_spte(kvm, kvm_mmu_page_as_id(sp), gfn,
- old_child_spte, REMOVED_SPTE, level,
- shared);
+ old_spte, REMOVED_SPTE, level, shared);
}
call_rcu(&sp->rcu_head, tdp_mmu_free_sp_rcu_callback);
@@ -667,14 +687,13 @@ static inline int tdp_mmu_zap_spte_atomic(struct kvm *kvm,
KVM_PAGES_PER_HPAGE(iter->level));
/*
- * No other thread can overwrite the removed SPTE as they
- * must either wait on the MMU lock or use
- * tdp_mmu_set_spte_atomic which will not overwrite the
- * special removed SPTE value. No bookkeeping is needed
- * here since the SPTE is going from non-present
- * to non-present.
+ * No other thread can overwrite the removed SPTE as they must either
+ * wait on the MMU lock or use tdp_mmu_set_spte_atomic() which will not
+ * overwrite the special removed SPTE value. No bookkeeping is needed
+ * here since the SPTE is going from non-present to non-present. Use
+ * the raw write helper to avoid an unnecessary check on volatile bits.
*/
- kvm_tdp_mmu_write_spte(iter->sptep, 0);
+ __kvm_tdp_mmu_write_spte(iter->sptep, 0);
return 0;
}
@@ -699,10 +718,13 @@ static inline int tdp_mmu_zap_spte_atomic(struct kvm *kvm,
* unless performing certain dirty logging operations.
* Leaving record_dirty_log unset in that case prevents page
* writes from being double counted.
+ *
+ * Returns the old SPTE value, which _may_ be different than @old_spte if the
+ * SPTE had voldatile bits.
*/
-static void __tdp_mmu_set_spte(struct kvm *kvm, int as_id, tdp_ptep_t sptep,
- u64 old_spte, u64 new_spte, gfn_t gfn, int level,
- bool record_acc_track, bool record_dirty_log)
+static u64 __tdp_mmu_set_spte(struct kvm *kvm, int as_id, tdp_ptep_t sptep,
+ u64 old_spte, u64 new_spte, gfn_t gfn, int level,
+ bool record_acc_track, bool record_dirty_log)
{
lockdep_assert_held_write(&kvm->mmu_lock);
@@ -715,7 +737,7 @@ static void __tdp_mmu_set_spte(struct kvm *kvm, int as_id, tdp_ptep_t sptep,
*/
WARN_ON(is_removed_spte(old_spte) || is_removed_spte(new_spte));
- kvm_tdp_mmu_write_spte(sptep, new_spte);
+ old_spte = kvm_tdp_mmu_write_spte(sptep, old_spte, new_spte, level);
__handle_changed_spte(kvm, as_id, gfn, old_spte, new_spte, level, false);
@@ -724,6 +746,7 @@ static void __tdp_mmu_set_spte(struct kvm *kvm, int as_id, tdp_ptep_t sptep,
if (record_dirty_log)
handle_changed_spte_dirty_log(kvm, as_id, gfn, old_spte,
new_spte, level);
+ return old_spte;
}
static inline void _tdp_mmu_set_spte(struct kvm *kvm, struct tdp_iter *iter,
@@ -732,9 +755,10 @@ static inline void _tdp_mmu_set_spte(struct kvm *kvm, struct tdp_iter *iter,
{
WARN_ON_ONCE(iter->yielded);
- __tdp_mmu_set_spte(kvm, iter->as_id, iter->sptep, iter->old_spte,
- new_spte, iter->gfn, iter->level,
- record_acc_track, record_dirty_log);
+ iter->old_spte = __tdp_mmu_set_spte(kvm, iter->as_id, iter->sptep,
+ iter->old_spte, new_spte,
+ iter->gfn, iter->level,
+ record_acc_track, record_dirty_log);
}
static inline void tdp_mmu_set_spte(struct kvm *kvm, struct tdp_iter *iter,
@@ -815,14 +839,15 @@ static inline bool __must_check tdp_mmu_iter_cond_resched(struct kvm *kvm,
return iter->yielded;
}
-static inline gfn_t tdp_mmu_max_gfn_host(void)
+static inline gfn_t tdp_mmu_max_gfn_exclusive(void)
{
/*
- * Bound TDP MMU walks at host.MAXPHYADDR, guest accesses beyond that
- * will hit a #PF(RSVD) and never hit an EPT Violation/Misconfig / #NPF,
- * and so KVM will never install a SPTE for such addresses.
+ * Bound TDP MMU walks at host.MAXPHYADDR. KVM disallows memslots with
+ * a gpa range that would exceed the max gfn, and KVM does not create
+ * MMIO SPTEs for "impossible" gfns, instead sending such accesses down
+ * the slow emulation path every time.
*/
- return 1ULL << (shadow_phys_bits - PAGE_SHIFT);
+ return kvm_mmu_max_gfn() + 1;
}
static void __tdp_mmu_zap_root(struct kvm *kvm, struct kvm_mmu_page *root,
@@ -830,7 +855,7 @@ static void __tdp_mmu_zap_root(struct kvm *kvm, struct kvm_mmu_page *root,
{
struct tdp_iter iter;
- gfn_t end = tdp_mmu_max_gfn_host();
+ gfn_t end = tdp_mmu_max_gfn_exclusive();
gfn_t start = 0;
for_each_tdp_pte_min_level(iter, root, zap_level, start, end) {
@@ -923,7 +948,7 @@ static bool tdp_mmu_zap_leafs(struct kvm *kvm, struct kvm_mmu_page *root,
{
struct tdp_iter iter;
- end = min(end, tdp_mmu_max_gfn_host());
+ end = min(end, tdp_mmu_max_gfn_exclusive());
lockdep_assert_held_write(&kvm->mmu_lock);
diff --git a/arch/x86/kvm/pmu.h b/arch/x86/kvm/pmu.h
index 9e66fba1d6a3..22992b049d38 100644
--- a/arch/x86/kvm/pmu.h
+++ b/arch/x86/kvm/pmu.h
@@ -138,6 +138,15 @@ static inline u64 get_sample_period(struct kvm_pmc *pmc, u64 counter_value)
return sample_period;
}
+static inline void pmc_update_sample_period(struct kvm_pmc *pmc)
+{
+ if (!pmc->perf_event || pmc->is_paused)
+ return;
+
+ perf_event_period(pmc->perf_event,
+ get_sample_period(pmc, pmc->counter));
+}
+
void reprogram_gp_counter(struct kvm_pmc *pmc, u64 eventsel);
void reprogram_fixed_counter(struct kvm_pmc *pmc, u8 ctrl, int fixed_idx);
void reprogram_counter(struct kvm_pmu *pmu, int pmc_idx);
diff --git a/arch/x86/kvm/svm/avic.c b/arch/x86/kvm/svm/avic.c
index a1cf9c31273b..421619540ff9 100644
--- a/arch/x86/kvm/svm/avic.c
+++ b/arch/x86/kvm/svm/avic.c
@@ -837,7 +837,8 @@ bool avic_check_apicv_inhibit_reasons(enum kvm_apicv_inhibit reason)
BIT(APICV_INHIBIT_REASON_IRQWIN) |
BIT(APICV_INHIBIT_REASON_PIT_REINJ) |
BIT(APICV_INHIBIT_REASON_X2APIC) |
- BIT(APICV_INHIBIT_REASON_BLOCKIRQ);
+ BIT(APICV_INHIBIT_REASON_BLOCKIRQ) |
+ BIT(APICV_INHIBIT_REASON_SEV);
return supported & BIT(reason);
}
diff --git a/arch/x86/kvm/svm/pmu.c b/arch/x86/kvm/svm/pmu.c
index 24eb935b6f85..16a5ebb420cf 100644
--- a/arch/x86/kvm/svm/pmu.c
+++ b/arch/x86/kvm/svm/pmu.c
@@ -45,6 +45,22 @@ static struct kvm_event_hw_type_mapping amd_event_mapping[] = {
[7] = { 0xd1, 0x00, PERF_COUNT_HW_STALLED_CYCLES_BACKEND },
};
+/* duplicated from amd_f17h_perfmon_event_map. */
+static struct kvm_event_hw_type_mapping amd_f17h_event_mapping[] = {
+ [0] = { 0x76, 0x00, PERF_COUNT_HW_CPU_CYCLES },
+ [1] = { 0xc0, 0x00, PERF_COUNT_HW_INSTRUCTIONS },
+ [2] = { 0x60, 0xff, PERF_COUNT_HW_CACHE_REFERENCES },
+ [3] = { 0x64, 0x09, PERF_COUNT_HW_CACHE_MISSES },
+ [4] = { 0xc2, 0x00, PERF_COUNT_HW_BRANCH_INSTRUCTIONS },
+ [5] = { 0xc3, 0x00, PERF_COUNT_HW_BRANCH_MISSES },
+ [6] = { 0x87, 0x02, PERF_COUNT_HW_STALLED_CYCLES_FRONTEND },
+ [7] = { 0x87, 0x01, PERF_COUNT_HW_STALLED_CYCLES_BACKEND },
+};
+
+/* amd_pmc_perf_hw_id depends on these being the same size */
+static_assert(ARRAY_SIZE(amd_event_mapping) ==
+ ARRAY_SIZE(amd_f17h_event_mapping));
+
static unsigned int get_msr_base(struct kvm_pmu *pmu, enum pmu_type type)
{
struct kvm_vcpu *vcpu = pmu_to_vcpu(pmu);
@@ -140,6 +156,7 @@ static inline struct kvm_pmc *get_gp_pmc_amd(struct kvm_pmu *pmu, u32 msr,
static unsigned int amd_pmc_perf_hw_id(struct kvm_pmc *pmc)
{
+ struct kvm_event_hw_type_mapping *event_mapping;
u8 event_select = pmc->eventsel & ARCH_PERFMON_EVENTSEL_EVENT;
u8 unit_mask = (pmc->eventsel & ARCH_PERFMON_EVENTSEL_UMASK) >> 8;
int i;
@@ -148,15 +165,20 @@ static unsigned int amd_pmc_perf_hw_id(struct kvm_pmc *pmc)
if (WARN_ON(pmc_is_fixed(pmc)))
return PERF_COUNT_HW_MAX;
+ if (guest_cpuid_family(pmc->vcpu) >= 0x17)
+ event_mapping = amd_f17h_event_mapping;
+ else
+ event_mapping = amd_event_mapping;
+
for (i = 0; i < ARRAY_SIZE(amd_event_mapping); i++)
- if (amd_event_mapping[i].eventsel == event_select
- && amd_event_mapping[i].unit_mask == unit_mask)
+ if (event_mapping[i].eventsel == event_select
+ && event_mapping[i].unit_mask == unit_mask)
break;
if (i == ARRAY_SIZE(amd_event_mapping))
return PERF_COUNT_HW_MAX;
- return amd_event_mapping[i].event_type;
+ return event_mapping[i].event_type;
}
/* check if a PMC is enabled by comparing it against global_ctrl bits. Because
@@ -257,6 +279,7 @@ static int amd_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
pmc = get_gp_pmc_amd(pmu, msr, PMU_TYPE_COUNTER);
if (pmc) {
pmc->counter += data - pmc_read_counter(pmc);
+ pmc_update_sample_period(pmc);
return 0;
}
/* MSR_EVNTSELn */
diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index 75fa6dd268f0..7c392873626f 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -260,6 +260,8 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
INIT_LIST_HEAD(&sev->regions_list);
INIT_LIST_HEAD(&sev->mirror_vms);
+ kvm_set_apicv_inhibit(kvm, APICV_INHIBIT_REASON_SEV);
+
return 0;
e_free:
@@ -465,6 +467,7 @@ static void sev_clflush_pages(struct page *pages[], unsigned long npages)
page_virtual = kmap_atomic(pages[i]);
clflush_cache_range(page_virtual, PAGE_SIZE);
kunmap_atomic(page_virtual);
+ cond_resched();
}
}
@@ -1591,24 +1594,51 @@ static void sev_unlock_two_vms(struct kvm *dst_kvm, struct kvm *src_kvm)
atomic_set_release(&src_sev->migration_in_progress, 0);
}
+/* vCPU mutex subclasses. */
+enum sev_migration_role {
+ SEV_MIGRATION_SOURCE = 0,
+ SEV_MIGRATION_TARGET,
+ SEV_NR_MIGRATION_ROLES,
+};
-static int sev_lock_vcpus_for_migration(struct kvm *kvm)
+static int sev_lock_vcpus_for_migration(struct kvm *kvm,
+ enum sev_migration_role role)
{
struct kvm_vcpu *vcpu;
unsigned long i, j;
+ bool first = true;
kvm_for_each_vcpu(i, vcpu, kvm) {
- if (mutex_lock_killable(&vcpu->mutex))
+ if (mutex_lock_killable_nested(&vcpu->mutex, role))
goto out_unlock;
+
+ if (first) {
+ /*
+ * Reset the role to one that avoids colliding with
+ * the role used for the first vcpu mutex.
+ */
+ role = SEV_NR_MIGRATION_ROLES;
+ first = false;
+ } else {
+ mutex_release(&vcpu->mutex.dep_map, _THIS_IP_);
+ }
}
return 0;
out_unlock:
+
+ first = true;
kvm_for_each_vcpu(j, vcpu, kvm) {
if (i == j)
break;
+ if (first)
+ first = false;
+ else
+ mutex_acquire(&vcpu->mutex.dep_map, role, 0, _THIS_IP_);
+
+
mutex_unlock(&vcpu->mutex);
}
return -EINTR;
@@ -1618,8 +1648,15 @@ static void sev_unlock_vcpus_for_migration(struct kvm *kvm)
{
struct kvm_vcpu *vcpu;
unsigned long i;
+ bool first = true;
kvm_for_each_vcpu(i, vcpu, kvm) {
+ if (first)
+ first = false;
+ else
+ mutex_acquire(&vcpu->mutex.dep_map,
+ SEV_NR_MIGRATION_ROLES, 0, _THIS_IP_);
+
mutex_unlock(&vcpu->mutex);
}
}
@@ -1745,10 +1782,10 @@ int sev_vm_move_enc_context_from(struct kvm *kvm, unsigned int source_fd)
charged = true;
}
- ret = sev_lock_vcpus_for_migration(kvm);
+ ret = sev_lock_vcpus_for_migration(kvm, SEV_MIGRATION_SOURCE);
if (ret)
goto out_dst_cgroup;
- ret = sev_lock_vcpus_for_migration(source_kvm);
+ ret = sev_lock_vcpus_for_migration(source_kvm, SEV_MIGRATION_TARGET);
if (ret)
goto out_dst_vcpu;
@@ -2223,51 +2260,47 @@ int sev_cpu_init(struct svm_cpu_data *sd)
* Pages used by hardware to hold guest encrypted state must be flushed before
* returning them to the system.
*/
-static void sev_flush_guest_memory(struct vcpu_svm *svm, void *va,
- unsigned long len)
+static void sev_flush_encrypted_page(struct kvm_vcpu *vcpu, void *va)
{
+ int asid = to_kvm_svm(vcpu->kvm)->sev_info.asid;
+
/*
- * If hardware enforced cache coherency for encrypted mappings of the
- * same physical page is supported, nothing to do.
+ * Note! The address must be a kernel address, as regular page walk
+ * checks are performed by VM_PAGE_FLUSH, i.e. operating on a user
+ * address is non-deterministic and unsafe. This function deliberately
+ * takes a pointer to deter passing in a user address.
*/
- if (boot_cpu_has(X86_FEATURE_SME_COHERENT))
- return;
+ unsigned long addr = (unsigned long)va;
/*
- * If the VM Page Flush MSR is supported, use it to flush the page
- * (using the page virtual address and the guest ASID).
+ * If CPU enforced cache coherency for encrypted mappings of the
+ * same physical page is supported, use CLFLUSHOPT instead. NOTE: cache
+ * flush is still needed in order to work properly with DMA devices.
*/
- if (boot_cpu_has(X86_FEATURE_VM_PAGE_FLUSH)) {
- struct kvm_sev_info *sev;
- unsigned long va_start;
- u64 start, stop;
-
- /* Align start and stop to page boundaries. */
- va_start = (unsigned long)va;
- start = (u64)va_start & PAGE_MASK;
- stop = PAGE_ALIGN((u64)va_start + len);
-
- if (start < stop) {
- sev = &to_kvm_svm(svm->vcpu.kvm)->sev_info;
+ if (boot_cpu_has(X86_FEATURE_SME_COHERENT)) {
+ clflush_cache_range(va, PAGE_SIZE);
+ return;
+ }
- while (start < stop) {
- wrmsrl(MSR_AMD64_VM_PAGE_FLUSH,
- start | sev->asid);
+ /*
+ * VM Page Flush takes a host virtual address and a guest ASID. Fall
+ * back to WBINVD if this faults so as not to make any problems worse
+ * by leaving stale encrypted data in the cache.
+ */
+ if (WARN_ON_ONCE(wrmsrl_safe(MSR_AMD64_VM_PAGE_FLUSH, addr | asid)))
+ goto do_wbinvd;
- start += PAGE_SIZE;
- }
+ return;
- return;
- }
+do_wbinvd:
+ wbinvd_on_all_cpus();
+}
- WARN(1, "Address overflow, using WBINVD\n");
- }
+void sev_guest_memory_reclaimed(struct kvm *kvm)
+{
+ if (!sev_guest(kvm))
+ return;
- /*
- * Hardware should always have one of the above features,
- * but if not, use WBINVD and issue a warning.
- */
- WARN_ONCE(1, "Using WBINVD to flush guest memory\n");
wbinvd_on_all_cpus();
}
@@ -2281,7 +2314,8 @@ void sev_free_vcpu(struct kvm_vcpu *vcpu)
svm = to_svm(vcpu);
if (vcpu->arch.guest_state_protected)
- sev_flush_guest_memory(svm, svm->sev_es.vmsa, PAGE_SIZE);
+ sev_flush_encrypted_page(vcpu, svm->sev_es.vmsa);
+
__free_page(virt_to_page(svm->sev_es.vmsa));
if (svm->sev_es.ghcb_sa_free)
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index bd4c64b362d2..7e45d03cd018 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -4620,6 +4620,7 @@ static struct kvm_x86_ops svm_x86_ops __initdata = {
.mem_enc_ioctl = sev_mem_enc_ioctl,
.mem_enc_register_region = sev_mem_enc_register_region,
.mem_enc_unregister_region = sev_mem_enc_unregister_region,
+ .guest_memory_reclaimed = sev_guest_memory_reclaimed,
.vm_copy_enc_context_from = sev_vm_copy_enc_context_from,
.vm_move_enc_context_from = sev_vm_move_enc_context_from,
diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
index f77a7d2d39dd..f76deff71002 100644
--- a/arch/x86/kvm/svm/svm.h
+++ b/arch/x86/kvm/svm/svm.h
@@ -609,6 +609,8 @@ int sev_mem_enc_unregister_region(struct kvm *kvm,
struct kvm_enc_region *range);
int sev_vm_copy_enc_context_from(struct kvm *kvm, unsigned int source_fd);
int sev_vm_move_enc_context_from(struct kvm *kvm, unsigned int source_fd);
+void sev_guest_memory_reclaimed(struct kvm *kvm);
+
void pre_sev_run(struct vcpu_svm *svm, int cpu);
void __init sev_set_cpu_caps(void);
void __init sev_hardware_setup(void);
diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index f18744f7ff82..856c87563883 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -4618,6 +4618,11 @@ void nested_vmx_vmexit(struct kvm_vcpu *vcpu, u32 vm_exit_reason,
kvm_make_request(KVM_REQ_APIC_PAGE_RELOAD, vcpu);
}
+ if (vmx->nested.update_vmcs01_apicv_status) {
+ vmx->nested.update_vmcs01_apicv_status = false;
+ kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu);
+ }
+
if ((vm_exit_reason != -1) &&
(enable_shadow_vmcs || evmptr_is_valid(vmx->nested.hv_evmcs_vmptr)))
vmx->nested.need_vmcs12_to_shadow_sync = true;
diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c
index bc3f8512bb64..b82b6709d7a8 100644
--- a/arch/x86/kvm/vmx/pmu_intel.c
+++ b/arch/x86/kvm/vmx/pmu_intel.c
@@ -431,15 +431,11 @@ static int intel_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
!(msr & MSR_PMC_FULL_WIDTH_BIT))
data = (s64)(s32)data;
pmc->counter += data - pmc_read_counter(pmc);
- if (pmc->perf_event && !pmc->is_paused)
- perf_event_period(pmc->perf_event,
- get_sample_period(pmc, data));
+ pmc_update_sample_period(pmc);
return 0;
} else if ((pmc = get_fixed_pmc(pmu, msr))) {
pmc->counter += data - pmc_read_counter(pmc);
- if (pmc->perf_event && !pmc->is_paused)
- perf_event_period(pmc->perf_event,
- get_sample_period(pmc, data));
+ pmc_update_sample_period(pmc);
return 0;
} else if ((pmc = get_gp_pmc(pmu, msr, MSR_P6_EVNTSEL0))) {
if (data == pmc->eventsel)
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index 04d170c4b61e..610355b9ccce 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -4174,6 +4174,11 @@ static void vmx_refresh_apicv_exec_ctrl(struct kvm_vcpu *vcpu)
{
struct vcpu_vmx *vmx = to_vmx(vcpu);
+ if (is_guest_mode(vcpu)) {
+ vmx->nested.update_vmcs01_apicv_status = true;
+ return;
+ }
+
pin_controls_set(vmx, vmx_pin_based_exec_ctrl(vmx));
if (cpu_has_secondary_exec_ctrls()) {
if (kvm_vcpu_apicv_active(vcpu))
@@ -5467,7 +5472,7 @@ static bool vmx_emulation_required_with_pending_exception(struct kvm_vcpu *vcpu)
struct vcpu_vmx *vmx = to_vmx(vcpu);
return vmx->emulation_required && !vmx->rmode.vm86_active &&
- vcpu->arch.exception.pending;
+ (vcpu->arch.exception.pending || vcpu->arch.exception.injected);
}
static int handle_invalid_guest_state(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/vmx/vmx.h b/arch/x86/kvm/vmx/vmx.h
index 9c6bfcd84008..b98c7e96697a 100644
--- a/arch/x86/kvm/vmx/vmx.h
+++ b/arch/x86/kvm/vmx/vmx.h
@@ -183,6 +183,7 @@ struct nested_vmx {
bool change_vmcs01_virtual_apic_mode;
bool reload_vmcs01_apic_access_page;
bool update_vmcs01_cpu_dirty_logging;
+ bool update_vmcs01_apicv_status;
/*
* Enlightened VMCS has been enabled. It does not mean that L1 has to
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 0c0ca599a353..4790f0d7d40b 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2901,7 +2901,7 @@ static void kvm_end_pvclock_update(struct kvm *kvm)
static void kvm_update_masterclock(struct kvm *kvm)
{
- kvm_hv_invalidate_tsc_page(kvm);
+ kvm_hv_request_tsc_page_update(kvm);
kvm_start_pvclock_update(kvm);
pvclock_update_vm_gtod_copy(kvm);
kvm_end_pvclock_update(kvm);
@@ -3113,8 +3113,7 @@ static int kvm_guest_time_update(struct kvm_vcpu *v)
offsetof(struct compat_vcpu_info, time));
if (vcpu->xen.vcpu_time_info_set)
kvm_setup_pvclock_page(v, &vcpu->xen.vcpu_time_info_cache, 0);
- if (!v->vcpu_idx)
- kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
+ kvm_hv_setup_tsc_page(v->kvm, &vcpu->hv_clock);
return 0;
}
@@ -6241,7 +6240,7 @@ static int kvm_vm_ioctl_set_clock(struct kvm *kvm, void __user *argp)
if (data.flags & ~KVM_CLOCK_VALID_FLAGS)
return -EINVAL;
- kvm_hv_invalidate_tsc_page(kvm);
+ kvm_hv_request_tsc_page_update(kvm);
kvm_start_pvclock_update(kvm);
pvclock_update_vm_gtod_copy(kvm);
@@ -8926,7 +8925,7 @@ int kvm_arch_init(void *opaque)
}
kvm_nr_uret_msrs = 0;
- r = kvm_mmu_module_init();
+ r = kvm_mmu_vendor_module_init();
if (r)
goto out_free_percpu;
@@ -8974,7 +8973,7 @@ void kvm_arch_exit(void)
cancel_work_sync(&pvclock_gtod_work);
#endif
kvm_x86_ops.hardware_enable = NULL;
- kvm_mmu_module_exit();
+ kvm_mmu_vendor_module_exit();
free_percpu(user_return_msrs);
kmem_cache_destroy(x86_emulator_cache);
#ifdef CONFIG_KVM_XEN
@@ -9112,7 +9111,7 @@ static void kvm_apicv_init(struct kvm *kvm)
if (!enable_apicv)
set_or_clear_apicv_inhibit(inhibits,
- APICV_INHIBIT_REASON_ABSENT, true);
+ APICV_INHIBIT_REASON_DISABLE, true);
}
static void kvm_sched_yield(struct kvm_vcpu *vcpu, unsigned long dest_id)
@@ -9890,6 +9889,11 @@ void kvm_arch_mmu_notifier_invalidate_range(struct kvm *kvm,
kvm_make_all_cpus_request(kvm, KVM_REQ_APIC_PAGE_RELOAD);
}
+void kvm_arch_guest_memory_reclaimed(struct kvm *kvm)
+{
+ static_call_cond(kvm_x86_guest_memory_reclaimed)(kvm);
+}
+
static void kvm_vcpu_reload_apic_access_page(struct kvm_vcpu *vcpu)
{
if (!lapic_in_kernel(vcpu))
@@ -10016,12 +10020,14 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
if (kvm_check_request(KVM_REQ_HV_CRASH, vcpu)) {
vcpu->run->exit_reason = KVM_EXIT_SYSTEM_EVENT;
vcpu->run->system_event.type = KVM_SYSTEM_EVENT_CRASH;
+ vcpu->run->system_event.ndata = 0;
r = 0;
goto out;
}
if (kvm_check_request(KVM_REQ_HV_RESET, vcpu)) {
vcpu->run->exit_reason = KVM_EXIT_SYSTEM_EVENT;
vcpu->run->system_event.type = KVM_SYSTEM_EVENT_RESET;
+ vcpu->run->system_event.ndata = 0;
r = 0;
goto out;
}
@@ -10098,7 +10104,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
/* Store vcpu->apicv_active before vcpu->mode. */
smp_store_release(&vcpu->mode, IN_GUEST_MODE);
- srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
+ kvm_vcpu_srcu_read_unlock(vcpu);
/*
* 1) We should set ->mode before checking ->requests. Please see
@@ -10129,7 +10135,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
smp_wmb();
local_irq_enable();
preempt_enable();
- vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
+ kvm_vcpu_srcu_read_lock(vcpu);
r = 1;
goto cancel_injection;
}
@@ -10255,7 +10261,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
local_irq_enable();
preempt_enable();
- vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
+ kvm_vcpu_srcu_read_lock(vcpu);
/*
* Profile KVM exit RIPs:
@@ -10285,7 +10291,7 @@ out:
}
/* Called within kvm->srcu read side. */
-static inline int vcpu_block(struct kvm *kvm, struct kvm_vcpu *vcpu)
+static inline int vcpu_block(struct kvm_vcpu *vcpu)
{
bool hv_timer;
@@ -10301,12 +10307,12 @@ static inline int vcpu_block(struct kvm *kvm, struct kvm_vcpu *vcpu)
if (hv_timer)
kvm_lapic_switch_to_sw_timer(vcpu);
- srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
+ kvm_vcpu_srcu_read_unlock(vcpu);
if (vcpu->arch.mp_state == KVM_MP_STATE_HALTED)
kvm_vcpu_halt(vcpu);
else
kvm_vcpu_block(vcpu);
- vcpu->srcu_idx = srcu_read_lock(&kvm->srcu);
+ kvm_vcpu_srcu_read_lock(vcpu);
if (hv_timer)
kvm_lapic_switch_to_hv_timer(vcpu);
@@ -10348,7 +10354,6 @@ static inline bool kvm_vcpu_running(struct kvm_vcpu *vcpu)
static int vcpu_run(struct kvm_vcpu *vcpu)
{
int r;
- struct kvm *kvm = vcpu->kvm;
vcpu->arch.l1tf_flush_l1d = true;
@@ -10356,7 +10361,7 @@ static int vcpu_run(struct kvm_vcpu *vcpu)
if (kvm_vcpu_running(vcpu)) {
r = vcpu_enter_guest(vcpu);
} else {
- r = vcpu_block(kvm, vcpu);
+ r = vcpu_block(vcpu);
}
if (r <= 0)
@@ -10375,9 +10380,9 @@ static int vcpu_run(struct kvm_vcpu *vcpu)
}
if (__xfer_to_guest_mode_work_pending()) {
- srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
+ kvm_vcpu_srcu_read_unlock(vcpu);
r = xfer_to_guest_mode_handle_work(vcpu);
- vcpu->srcu_idx = srcu_read_lock(&kvm->srcu);
+ kvm_vcpu_srcu_read_lock(vcpu);
if (r)
return r;
}
@@ -10388,12 +10393,7 @@ static int vcpu_run(struct kvm_vcpu *vcpu)
static inline int complete_emulated_io(struct kvm_vcpu *vcpu)
{
- int r;
-
- vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
- r = kvm_emulate_instruction(vcpu, EMULTYPE_NO_DECODE);
- srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
- return r;
+ return kvm_emulate_instruction(vcpu, EMULTYPE_NO_DECODE);
}
static int complete_emulated_pio(struct kvm_vcpu *vcpu)
@@ -10485,7 +10485,6 @@ static void kvm_put_guest_fpu(struct kvm_vcpu *vcpu)
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
{
struct kvm_run *kvm_run = vcpu->run;
- struct kvm *kvm = vcpu->kvm;
int r;
vcpu_load(vcpu);
@@ -10493,7 +10492,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
kvm_run->flags = 0;
kvm_load_guest_fpu(vcpu);
- vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
+ kvm_vcpu_srcu_read_lock(vcpu);
if (unlikely(vcpu->arch.mp_state == KVM_MP_STATE_UNINITIALIZED)) {
if (kvm_run->immediate_exit) {
r = -EINTR;
@@ -10505,9 +10504,9 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
*/
WARN_ON_ONCE(kvm_lapic_hv_timer_in_use(vcpu));
- srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
+ kvm_vcpu_srcu_read_unlock(vcpu);
kvm_vcpu_block(vcpu);
- vcpu->srcu_idx = srcu_read_lock(&kvm->srcu);
+ kvm_vcpu_srcu_read_lock(vcpu);
if (kvm_apic_accept_events(vcpu) < 0) {
r = 0;
@@ -10568,7 +10567,7 @@ out:
if (kvm_run->kvm_valid_regs)
store_regs(vcpu);
post_kvm_run_save(vcpu);
- srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
+ kvm_vcpu_srcu_read_unlock(vcpu);
kvm_sigset_deactivate(vcpu);
vcpu_put(vcpu);
@@ -10986,6 +10985,9 @@ static void kvm_arch_vcpu_guestdbg_update_apicv_inhibit(struct kvm *kvm)
struct kvm_vcpu *vcpu;
unsigned long i;
+ if (!enable_apicv)
+ return;
+
down_write(&kvm->arch.apicv_update_lock);
kvm_for_each_vcpu(i, vcpu, kvm) {
@@ -11197,8 +11199,21 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
r = kvm_create_lapic(vcpu, lapic_timer_advance_ns);
if (r < 0)
goto fail_mmu_destroy;
- if (kvm_apicv_activated(vcpu->kvm))
+
+ /*
+ * Defer evaluating inhibits until the vCPU is first run, as
+ * this vCPU will not get notified of any changes until this
+ * vCPU is visible to other vCPUs (marked online and added to
+ * the set of vCPUs). Opportunistically mark APICv active as
+ * VMX in particularly is highly unlikely to have inhibits.
+ * Ignore the current per-VM APICv state so that vCPU creation
+ * is guaranteed to run with a deterministic value, the request
+ * will ensure the vCPU gets the correct state before VM-Entry.
+ */
+ if (enable_apicv) {
vcpu->arch.apicv_active = true;
+ kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu);
+ }
} else
static_branch_inc(&kvm_has_noapic_vcpu);
@@ -11996,8 +12011,12 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
struct kvm_memory_slot *new,
enum kvm_mr_change change)
{
- if (change == KVM_MR_CREATE || change == KVM_MR_MOVE)
+ if (change == KVM_MR_CREATE || change == KVM_MR_MOVE) {
+ if ((new->base_gfn + new->npages - 1) > kvm_mmu_max_gfn())
+ return -EINVAL;
+
return kvm_alloc_memslot_metadata(kvm, new);
+ }
if (change == KVM_MR_FLAGS_ONLY)
memcpy(&new->arch, &old->arch, sizeof(old->arch));
@@ -12986,3 +13005,19 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_enter);
EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_exit);
EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_msr_protocol_enter);
EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_msr_protocol_exit);
+
+static int __init kvm_x86_init(void)
+{
+ kvm_mmu_x86_module_init();
+ return 0;
+}
+module_init(kvm_x86_init);
+
+static void __exit kvm_x86_exit(void)
+{
+ /*
+ * If module_init() is implemented, module_exit() must also be
+ * implemented to allow module unload.
+ */
+}
+module_exit(kvm_x86_exit);
generated by cgit (git 2.34.1) at 2024-05-25 22:30:59 +0000