In some cases, the NX hugepage mitigation for iTLB multihit is not
needed for all guests on a host. Allow disabling the mitigation on a
per-VM basis to avoid the performance hit of NX hugepages on trusted
workloads.
In order to disable NX hugepages on a VM, ensure that the userspace
actor has permission to reboot the system. Since disabling NX hugepages
would allow a guest to crash the system, it is similar to reboot
permissions.
Ideally, KVM would require userspace to prove it has access to KVM's
nx_huge_pages module param, e.g. so that userspace can opt out without
needing full reboot permissions. But getting access to the module param
file info is difficult because it is buried in layers of sysfs and module
glue. Requiring CAP_SYS_BOOT is sufficient for all known use cases.
Suggested-by: Jim Mattson <[email protected]>
Signed-off-by: Ben Gardon <[email protected]>
---
Documentation/virt/kvm/api.rst | 13 +++++++++++++
arch/x86/include/asm/kvm_host.h | 2 ++
arch/x86/kvm/mmu.h | 9 +++++----
arch/x86/kvm/mmu/spte.c | 7 ++++---
arch/x86/kvm/mmu/spte.h | 3 ++-
arch/x86/kvm/mmu/tdp_mmu.c | 3 ++-
arch/x86/kvm/x86.c | 23 +++++++++++++++++++++++
include/uapi/linux/kvm.h | 1 +
8 files changed, 52 insertions(+), 9 deletions(-)
diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
index 72183ae628f7..021452a9fa91 100644
--- a/Documentation/virt/kvm/api.rst
+++ b/Documentation/virt/kvm/api.rst
@@ -7855,6 +7855,19 @@ At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting
this capability will disable PMU virtualization for that VM. Usermode
should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
+8.36 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
+---------------------------
+
+:Capability KVM_CAP_PMU_CAPABILITY
+:Architectures: x86
+:Type: vm
+:Returns 0 on success, -EPERM if the userspace process does not
+ have CAP_SYS_BOOT
+
+This capability disables the NX huge pages mitigation for iTLB MULTIHIT.
+
+The capability has no effect if the nx_huge_pages module parameter is not set.
+
9. Known KVM API problems
=========================
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 2c20f715f009..b8ab4fa7d4b2 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1240,6 +1240,8 @@ struct kvm_arch {
hpa_t hv_root_tdp;
spinlock_t hv_root_tdp_lock;
#endif
+
+ bool disable_nx_huge_pages;
};
struct kvm_vm_stat {
diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
index 671cfeccf04e..148f630af78a 100644
--- a/arch/x86/kvm/mmu.h
+++ b/arch/x86/kvm/mmu.h
@@ -173,9 +173,10 @@ struct kvm_page_fault {
int kvm_tdp_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
extern int nx_huge_pages;
-static inline bool is_nx_huge_page_enabled(void)
+static inline bool is_nx_huge_page_enabled(struct kvm *kvm)
{
- return READ_ONCE(nx_huge_pages);
+ return READ_ONCE(nx_huge_pages) &&
+ !kvm->arch.disable_nx_huge_pages;
}
static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
@@ -191,8 +192,8 @@ static inline int kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa,
.user = err & PFERR_USER_MASK,
.prefetch = prefetch,
.is_tdp = likely(vcpu->arch.mmu->page_fault == kvm_tdp_page_fault),
- .nx_huge_page_workaround_enabled = is_nx_huge_page_enabled(),
-
+ .nx_huge_page_workaround_enabled =
+ is_nx_huge_page_enabled(vcpu->kvm),
.max_level = KVM_MAX_HUGEPAGE_LEVEL,
.req_level = PG_LEVEL_4K,
.goal_level = PG_LEVEL_4K,
diff --git a/arch/x86/kvm/mmu/spte.c b/arch/x86/kvm/mmu/spte.c
index 4739b53c9734..877ad30bc7ad 100644
--- a/arch/x86/kvm/mmu/spte.c
+++ b/arch/x86/kvm/mmu/spte.c
@@ -116,7 +116,7 @@ bool make_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
spte |= spte_shadow_accessed_mask(spte);
if (level > PG_LEVEL_4K && (pte_access & ACC_EXEC_MASK) &&
- is_nx_huge_page_enabled()) {
+ is_nx_huge_page_enabled(vcpu->kvm)) {
pte_access &= ~ACC_EXEC_MASK;
}
@@ -215,7 +215,8 @@ static u64 make_spte_executable(u64 spte)
* This is used during huge page splitting to build the SPTEs that make up the
* new page table.
*/
-u64 make_huge_page_split_spte(u64 huge_spte, int huge_level, int index)
+u64 make_huge_page_split_spte(struct kvm *kvm, u64 huge_spte, int huge_level,
+ int index)
{
u64 child_spte;
int child_level;
@@ -243,7 +244,7 @@ u64 make_huge_page_split_spte(u64 huge_spte, int huge_level, int index)
* When splitting to a 4K page, mark the page executable as the
* NX hugepage mitigation no longer applies.
*/
- if (is_nx_huge_page_enabled())
+ if (is_nx_huge_page_enabled(kvm))
child_spte = make_spte_executable(child_spte);
}
diff --git a/arch/x86/kvm/mmu/spte.h b/arch/x86/kvm/mmu/spte.h
index 73f12615416f..e4142caff4b1 100644
--- a/arch/x86/kvm/mmu/spte.h
+++ b/arch/x86/kvm/mmu/spte.h
@@ -415,7 +415,8 @@ bool make_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
unsigned int pte_access, gfn_t gfn, kvm_pfn_t pfn,
u64 old_spte, bool prefetch, bool can_unsync,
bool host_writable, u64 *new_spte);
-u64 make_huge_page_split_spte(u64 huge_spte, int huge_level, int index);
+u64 make_huge_page_split_spte(struct kvm *kvm, u64 huge_spte, int huge_level,
+ int index);
u64 make_nonleaf_spte(u64 *child_pt, bool ad_disabled);
u64 make_mmio_spte(struct kvm_vcpu *vcpu, u64 gfn, unsigned int access);
u64 mark_spte_for_access_track(u64 spte);
diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
index 566548a3efa7..03aa1e0f60e2 100644
--- a/arch/x86/kvm/mmu/tdp_mmu.c
+++ b/arch/x86/kvm/mmu/tdp_mmu.c
@@ -1469,7 +1469,8 @@ static int tdp_mmu_split_huge_page(struct kvm *kvm, struct tdp_iter *iter,
* not been linked in yet and thus is not reachable from any other CPU.
*/
for (i = 0; i < PT64_ENT_PER_PAGE; i++)
- sp->spt[i] = make_huge_page_split_spte(huge_spte, level, i);
+ sp->spt[i] = make_huge_page_split_spte(kvm, huge_spte,
+ level, i);
/*
* Replace the huge spte with a pointer to the populated lower level
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 665c1fa8bb57..27631c3b53c2 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -4286,6 +4286,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
case KVM_CAP_SYS_ATTRIBUTES:
case KVM_CAP_VAPIC:
case KVM_CAP_ENABLE_CAP:
+ case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES:
r = 1;
break;
case KVM_CAP_EXIT_HYPERCALL:
@@ -6079,6 +6080,28 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
}
mutex_unlock(&kvm->lock);
break;
+ case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES:
+ r = -EINVAL;
+ if (cap->args[0])
+ break;
+
+ /*
+ * Since the risk of disabling NX hugepages is a guest crashing
+ * the system, ensure the userspace process has permission to
+ * reboot the system.
+ */
+ if (!capable(CAP_SYS_BOOT)) {
+ r = -EPERM;
+ break;
+ }
+
+ mutex_lock(&kvm->lock);
+ if (!kvm->created_vcpus) {
+ kvm->arch.disable_nx_huge_pages = true;
+ r = 0;
+ }
+ mutex_unlock(&kvm->lock);
+ break;
default:
r = -EINVAL;
break;
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index dd1d8167e71f..7155488164bd 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -1148,6 +1148,7 @@ struct kvm_ppc_resize_hpt {
#define KVM_CAP_PMU_CAPABILITY 212
#define KVM_CAP_DISABLE_QUIRKS2 213
#define KVM_CAP_VM_TSC_CONTROL 214
+#define KVM_CAP_VM_DISABLE_NX_HUGE_PAGES 215
#ifdef KVM_CAP_IRQ_ROUTING
--
2.35.1.1178.g4f1659d476-goog
On Wed, Apr 13, 2022, Ben Gardon wrote:
> diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
> index 72183ae628f7..021452a9fa91 100644
> --- a/Documentation/virt/kvm/api.rst
> +++ b/Documentation/virt/kvm/api.rst
> @@ -7855,6 +7855,19 @@ At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting
> this capability will disable PMU virtualization for that VM. Usermode
> should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
>
> +8.36 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
> +---------------------------
> +
> +:Capability KVM_CAP_PMU_CAPABILITY
> +:Architectures: x86
> +:Type: vm
> +:Returns 0 on success, -EPERM if the userspace process does not
> + have CAP_SYS_BOOT
Needs to document the -EINVAL cases, especially the requirement that this be
called before VMs are created. The
> +This capability disables the NX huge pages mitigation for iTLB MULTIHIT.
> +
> +The capability has no effect if the nx_huge_pages module parameter is not set.
> +
> 9. Known KVM API problems
> =========================
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 2c20f715f009..b8ab4fa7d4b2 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -1240,6 +1240,8 @@ struct kvm_arch {
> hpa_t hv_root_tdp;
> spinlock_t hv_root_tdp_lock;
> #endif
> +
> + bool disable_nx_huge_pages;
> };
>
> struct kvm_vm_stat {
> diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h
> index 671cfeccf04e..148f630af78a 100644
> --- a/arch/x86/kvm/mmu.h
> +++ b/arch/x86/kvm/mmu.h
> @@ -173,9 +173,10 @@ struct kvm_page_fault {
> int kvm_tdp_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
>
> extern int nx_huge_pages;
> -static inline bool is_nx_huge_page_enabled(void)
> +static inline bool is_nx_huge_page_enabled(struct kvm *kvm)
> {
> - return READ_ONCE(nx_huge_pages);
> + return READ_ONCE(nx_huge_pages) &&
> + !kvm->arch.disable_nx_huge_pages;
No need for a newline, that fits on a single line.
> diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
> index 566548a3efa7..03aa1e0f60e2 100644
> --- a/arch/x86/kvm/mmu/tdp_mmu.c
> +++ b/arch/x86/kvm/mmu/tdp_mmu.c
> @@ -1469,7 +1469,8 @@ static int tdp_mmu_split_huge_page(struct kvm *kvm, struct tdp_iter *iter,
> * not been linked in yet and thus is not reachable from any other CPU.
> */
> for (i = 0; i < PT64_ENT_PER_PAGE; i++)
> - sp->spt[i] = make_huge_page_split_spte(huge_spte, level, i);
> + sp->spt[i] = make_huge_page_split_spte(kvm, huge_spte,
> + level, i);
Just let this poke past 80 chars.
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 665c1fa8bb57..27631c3b53c2 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -4286,6 +4286,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
> case KVM_CAP_SYS_ATTRIBUTES:
> case KVM_CAP_VAPIC:
> case KVM_CAP_ENABLE_CAP:
> + case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES:
> r = 1;
> break;
> case KVM_CAP_EXIT_HYPERCALL:
> @@ -6079,6 +6080,28 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
> }
> mutex_unlock(&kvm->lock);
> break;
> + case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES:
> + r = -EINVAL;
> + if (cap->args[0])
> + break;
> +
> + /*
> + * Since the risk of disabling NX hugepages is a guest crashing
> + * the system, ensure the userspace process has permission to
> + * reboot the system.
Since I'm nitpicking already and there's also a comment...
Can you call out that, unlike the actual reboot() syscall, the process needs the
capability in the init? namespace (I don't actual know the terminology) because
exposing /dev/kvm into a container doesn't magically limit the iTLB multihit bug
to that container. I.e. that this _must_ use capable(), not ns_capable().
Amusingly, someone could subvert the selftest's SYS_reboot heuristic by running
the test in a container :-)