[PATCH v3 0/5] KVM: nVMX: Mark APIC page dirty on VM-Exit

[PATCH v3 0/5] KVM: nVMX: Mark APIC page dirty on VM-Exit

[Sean Christopherson]

Extended version of Fred's patch to mark the APIC access page dirty on
VM-Exit (KVM already marks it dirty when it's unmapped).

v3:
 - Fix a benign memslots bug in __kvm_vcpu_map().
 - Mark vmcs12 pages dirty if and only if they're mapped (out-of-band).
 - Don't mark the APIC access page dirty when deliver nested posted IRQ.

v2: https://lore.kernel.org/all/20250910085156.1419090-1-griffoul@gmail.com

Fred Griffoul (1):
  KVM: nVMX: Mark APIC access page dirty when syncing vmcs12 pages

Sean Christopherson (4):
  KVM: Use vCPU specific memslots in __kvm_vcpu_map()
  KVM: x86: Mark vmcs12 pages as dirty if and only if they're mapped
  KVM: nVMX: Precisely mark vAPIC and PID maps dirty when delivering
    nested PI
  KVM: VMX: Move nested_mark_vmcs12_pages_dirty() to vmx.c, and rename

 arch/x86/kvm/vmx/nested.c | 25 ++-----------------------
 arch/x86/kvm/vmx/vmx.c    | 11 ++++++++++-
 include/linux/kvm_host.h  |  9 ++++++++-
 virt/kvm/kvm_main.c       |  2 +-
 4 files changed, 21 insertions(+), 26 deletions(-)


base-commit: 115d5de2eef32ac5cd488404b44b38789362dbe6
-- 
2.52.0.rc2.455.g230fcf2819-goog

[PATCH v3 1/5] KVM: Use vCPU specific memslots in __kvm_vcpu_map()

[Sean Christopherson]

When establishing a "host access map", lookup the gfn in the vCPU specific
memslots, as the intent is that the mapping will be established for the
current vCPU context.  Specifically, using __kvm_vcpu_map() in x86's SMM
context should create mappings based on the SMM memslots, not the non-SMM
memslots.

Luckily, the bug is benign as x86 is the only architecture with multiple
memslot address spaces, and all of x86's usage is limited to non-SMM.  The
calls in (or reachable by) {svm,vmx}_enter_smm() are made before
enter_smm() sets HF_SMM_MASK, and the calls in {svm,vmx}_leave_smm() are
made after emulator_leave_smm() clears HF_SMM_MASK.

Note, kvm_vcpu_unmap() uses the vCPU specific memslots, only the map() side
of things is broken.

Fixes: 357a18ad230f ("KVM: Kill kvm_map_gfn() / kvm_unmap_gfn() and gfn_to_pfn_cache")
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 virt/kvm/kvm_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 9eca084bdcbe..afe13451ce7f 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -3118,7 +3118,7 @@ int __kvm_vcpu_map(struct kvm_vcpu *vcpu, gfn_t gfn, struct kvm_host_map *map,
 		   bool writable)
 {
 	struct kvm_follow_pfn kfp = {
-		.slot = gfn_to_memslot(vcpu->kvm, gfn),
+		.slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn),
 		.gfn = gfn,
 		.flags = writable ? FOLL_WRITE : 0,
 		.refcounted_page = &map->pinned_page,
-- 
2.52.0.rc2.455.g230fcf2819-goog

[PATCH v3 2/5] KVM: x86: Mark vmcs12 pages as dirty if and only if they're mapped

[Sean Christopherson]

Mark vmcs12 pages as dirty (in KVM's dirty log bitmap) if and only if the
page is mapped, i.e. if the page is actually "active" in vmcs02.  For some
pages, KVM simply disables the associated VMCS control if the vmcs12 page
is unreachable, i.e. it's possible for nested VM-Enter to succeed with a
"bad" vmcs12 page.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/vmx/nested.c | 15 +++------------
 include/linux/kvm_host.h  |  9 ++++++++-
 2 files changed, 11 insertions(+), 13 deletions(-)

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index 40777278eabb..d4ef33578747 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -3983,23 +3983,14 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
 
 void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
 {
-	struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
-	gfn_t gfn;
+	struct vcpu_vmx *vmx = to_vmx(vcpu);
 
 	/*
 	 * Don't need to mark the APIC access page dirty; it is never
 	 * written to by the CPU during APIC virtualization.
 	 */
-
-	if (nested_cpu_has(vmcs12, CPU_BASED_TPR_SHADOW)) {
-		gfn = vmcs12->virtual_apic_page_addr >> PAGE_SHIFT;
-		kvm_vcpu_mark_page_dirty(vcpu, gfn);
-	}
-
-	if (nested_cpu_has_posted_intr(vmcs12)) {
-		gfn = vmcs12->posted_intr_desc_addr >> PAGE_SHIFT;
-		kvm_vcpu_mark_page_dirty(vcpu, gfn);
-	}
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
 }
 
 static int vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index d93f75b05ae2..536d05e2726f 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -1381,6 +1381,7 @@ bool kvm_vcpu_is_visible_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
 unsigned long kvm_host_page_size(struct kvm_vcpu *vcpu, gfn_t gfn);
 void mark_page_dirty_in_slot(struct kvm *kvm, const struct kvm_memory_slot *memslot, gfn_t gfn);
 void mark_page_dirty(struct kvm *kvm, gfn_t gfn);
+void kvm_vcpu_mark_page_dirty(struct kvm_vcpu *vcpu, gfn_t gfn);
 
 int __kvm_vcpu_map(struct kvm_vcpu *vcpu, gpa_t gpa, struct kvm_host_map *map,
 		   bool writable);
@@ -1398,6 +1399,13 @@ static inline int kvm_vcpu_map_readonly(struct kvm_vcpu *vcpu, gpa_t gpa,
 	return __kvm_vcpu_map(vcpu, gpa, map, false);
 }
 
+static inline void kvm_vcpu_map_mark_dirty(struct kvm_vcpu *vcpu,
+					   struct kvm_host_map *map)
+{
+	if (kvm_vcpu_mapped(map))
+		kvm_vcpu_mark_page_dirty(vcpu, map->gfn);
+}
+
 unsigned long kvm_vcpu_gfn_to_hva(struct kvm_vcpu *vcpu, gfn_t gfn);
 unsigned long kvm_vcpu_gfn_to_hva_prot(struct kvm_vcpu *vcpu, gfn_t gfn, bool *writable);
 int kvm_vcpu_read_guest_page(struct kvm_vcpu *vcpu, gfn_t gfn, void *data, int offset,
@@ -1410,7 +1418,6 @@ int kvm_vcpu_write_guest_page(struct kvm_vcpu *vcpu, gfn_t gfn, const void *data
 			      int offset, int len);
 int kvm_vcpu_write_guest(struct kvm_vcpu *vcpu, gpa_t gpa, const void *data,
 			 unsigned long len);
-void kvm_vcpu_mark_page_dirty(struct kvm_vcpu *vcpu, gfn_t gfn);
 
 /**
  * kvm_gpc_init - initialize gfn_to_pfn_cache.
-- 
2.52.0.rc2.455.g230fcf2819-goog

[PATCH v3 3/5] KVM: nVMX: Precisely mark vAPIC and PID maps dirty when delivering nested PI

[Sean Christopherson]

Explicitly mark the vmcs12 vAPIC and PI descriptor pages as dirty when
delivering a nested posted interrupt instead of marking all vmcs12 pages
as dirty.  This will allow marking the APIC access page (and any future
 vmcs12 pages) as dirty in nested_mark_vmcs12_pages_dirty() without over-
dirtying in the nested PI case.  Manually marking the vAPIC and PID pages
as dirty also makes the flow a bit more self-documenting, e.g. it's not
obvious at first glance that vmx->nested.pi_desc is actually a host kernel
mapping of a vmcs12 page.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/vmx/nested.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index d4ef33578747..d0cf99903971 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -4027,7 +4027,8 @@ static int vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)
 		}
 	}
 
-	nested_mark_vmcs12_pages_dirty(vcpu);
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
 	return 0;
 
 mmio_needed:
-- 
2.52.0.rc2.455.g230fcf2819-goog

[PATCH v3 4/5] KVM: VMX: Move nested_mark_vmcs12_pages_dirty() to vmx.c, and rename

[Sean Christopherson]

Move nested_mark_vmcs12_pages_dirty() to vmx.c now that it's only used in
the VM-Exit path, and add "all" to its name to document that its purpose
is to mark all (mapped-out-of-band) vmcs12 pages as dirty.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/vmx/nested.c | 13 -------------
 arch/x86/kvm/vmx/vmx.c    | 14 +++++++++++++-
 2 files changed, 13 insertions(+), 14 deletions(-)

diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index d0cf99903971..97554eda440c 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -3980,19 +3980,6 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
 	}
 }
 
-
-void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
-{
-	struct vcpu_vmx *vmx = to_vmx(vcpu);
-
-	/*
-	 * Don't need to mark the APIC access page dirty; it is never
-	 * written to by the CPU during APIC virtualization.
-	 */
-	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
-	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
-}
-
 static int vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index 4cbe8c84b636..cc38d08935e8 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -6378,6 +6378,18 @@ static void vmx_flush_pml_buffer(struct kvm_vcpu *vcpu)
 	vmcs_write16(GUEST_PML_INDEX, PML_HEAD_INDEX);
 }
 
+static void nested_vmx_mark_all_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_vmx *vmx = to_vmx(vcpu);
+
+	/*
+	 * Don't need to mark the APIC access page dirty; it is never
+	 * written to by the CPU during APIC virtualization.
+	 */
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
+}
+
 static void vmx_dump_sel(char *name, uint32_t sel)
 {
 	pr_err("%s sel=0x%04x, attr=0x%05x, limit=0x%08x, base=0x%016lx\n",
@@ -6655,7 +6667,7 @@ static int __vmx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t exit_fastpath)
 		 * Mark them dirty on every exit from L2 to prevent them from
 		 * getting out of sync with dirty tracking.
 		 */
-		nested_mark_vmcs12_pages_dirty(vcpu);
+		nested_vmx_mark_all_vmcs12_pages_dirty(vcpu);
 
 		/*
 		 * Synthesize a triple fault if L2 state is invalid.  In normal
-- 
2.52.0.rc2.455.g230fcf2819-goog

[PATCH v3 5/5] KVM: nVMX: Mark APIC access page dirty when syncing vmcs12 pages

[Sean Christopherson]

From: Fred Griffoul <fgriffo@amazon.co.uk>

For consistency with commit 7afe79f5734a ("KVM: nVMX: Mark vmcs12's APIC
access page dirty when unmapping"), which marks the page dirty during
unmap operations, also mark it dirty during vmcs12 page synchronization.

Signed-off-by: Fred Griffoul <fgriffo@amazon.co.uk>
[sean: use kvm_vcpu_map_mark_dirty()]
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/vmx/vmx.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index cc38d08935e8..72ac42fdf3b1 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -6382,10 +6382,7 @@ static void nested_vmx_mark_all_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 
-	/*
-	 * Don't need to mark the APIC access page dirty; it is never
-	 * written to by the CPU during APIC virtualization.
-	 */
+	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.apic_access_page_map);
 	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
 	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
 }
-- 
2.52.0.rc2.455.g230fcf2819-goog

Re: [PATCH v3 0/5] KVM: nVMX: Mark APIC page dirty on VM-Exit

[Sean Christopherson]

On Fri, 21 Nov 2025 14:34:39 -0800, Sean Christopherson wrote:
> Extended version of Fred's patch to mark the APIC access page dirty on
> VM-Exit (KVM already marks it dirty when it's unmapped).
> 
> v3:
>  - Fix a benign memslots bug in __kvm_vcpu_map().
>  - Mark vmcs12 pages dirty if and only if they're mapped (out-of-band).
>  - Don't mark the APIC access page dirty when deliver nested posted IRQ.
> 
> [...]

Applied to kvm-x86 apic, thanks!

[1/5] KVM: Use vCPU specific memslots in __kvm_vcpu_map()
      https://github.com/kvm-x86/linux/commit/44da6629d282
[2/5] KVM: x86: Mark vmcs12 pages as dirty if and only if they're mapped
      https://github.com/kvm-x86/linux/commit/70b02809ded9
[3/5] KVM: nVMX: Precisely mark vAPIC and PID maps dirty when delivering nested PI
      https://github.com/kvm-x86/linux/commit/f74bb1d2eda1
[4/5] KVM: VMX: Move nested_mark_vmcs12_pages_dirty() to vmx.c, and rename
      https://github.com/kvm-x86/linux/commit/57dfa61f6248
[5/5] KVM: nVMX: Mark APIC access page dirty when syncing vmcs12 pages
      https://github.com/kvm-x86/linux/commit/c9d7134679eb

--
https://github.com/kvm-x86/linux/tree/next

Re: [PATCH v3 4/5] KVM: VMX: Move nested_mark_vmcs12_pages_dirty() to vmx.c, and rename

[Binbin Wu]

On 11/22/2025 6:34 AM, Sean Christopherson wrote:
> Move nested_mark_vmcs12_pages_dirty() to vmx.c now that it's only used in
> the VM-Exit path, and add "all" to its name to document that its purpose
> is to mark all (mapped-out-of-band) vmcs12 pages as dirty.
> 
> No functional change intended.
> 
> Signed-off-by: Sean Christopherson <seanjc@google.com>
> ---
>  arch/x86/kvm/vmx/nested.c | 13 -------------
>  arch/x86/kvm/vmx/vmx.c    | 14 +++++++++++++-
>  2 files changed, 13 insertions(+), 14 deletions(-)
> 
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index d0cf99903971..97554eda440c 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -3980,19 +3980,6 @@ static void vmcs12_save_pending_event(struct kvm_vcpu *vcpu,
>  	}
>  }
>  
> -
> -void nested_mark_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)

The declaration in the header file is left over.
Since the patch series has been merged in the kvm-x86 next branch, I sent a
patch to remove it.

https://lore.kernel.org/kvm/20260113084748.1714633-1-binbin.wu@linux.intel.com/

> -{
> -	struct vcpu_vmx *vmx = to_vmx(vcpu);
> -
> -	/*
> -	 * Don't need to mark the APIC access page dirty; it is never
> -	 * written to by the CPU during APIC virtualization.
> -	 */
> -	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
> -	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
> -}
> -
>  static int vmx_complete_nested_posted_interrupt(struct kvm_vcpu *vcpu)
>  {
>  	struct vcpu_vmx *vmx = to_vmx(vcpu);
> diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> index 4cbe8c84b636..cc38d08935e8 100644
> --- a/arch/x86/kvm/vmx/vmx.c
> +++ b/arch/x86/kvm/vmx/vmx.c
> @@ -6378,6 +6378,18 @@ static void vmx_flush_pml_buffer(struct kvm_vcpu *vcpu)
>  	vmcs_write16(GUEST_PML_INDEX, PML_HEAD_INDEX);
>  }
>  
> +static void nested_vmx_mark_all_vmcs12_pages_dirty(struct kvm_vcpu *vcpu)
> +{
> +	struct vcpu_vmx *vmx = to_vmx(vcpu);
> +
> +	/*
> +	 * Don't need to mark the APIC access page dirty; it is never
> +	 * written to by the CPU during APIC virtualization.
> +	 */
> +	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.virtual_apic_map);
> +	kvm_vcpu_map_mark_dirty(vcpu, &vmx->nested.pi_desc_map);
> +}
> +
>  static void vmx_dump_sel(char *name, uint32_t sel)
>  {
>  	pr_err("%s sel=0x%04x, attr=0x%05x, limit=0x%08x, base=0x%016lx\n",
> @@ -6655,7 +6667,7 @@ static int __vmx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t exit_fastpath)
>  		 * Mark them dirty on every exit from L2 to prevent them from
>  		 * getting out of sync with dirty tracking.
>  		 */
> -		nested_mark_vmcs12_pages_dirty(vcpu);
> +		nested_vmx_mark_all_vmcs12_pages_dirty(vcpu);
>  
>  		/*
>  		 * Synthesize a triple fault if L2 state is invalid.  In normal