Skip to contentSkip to conversation

KVM: SVM: Add Page Modification Logging (PML) support

PATCH Superseded Nikunj A Dadhania 2 participants 13 messages Sep 15, 2025 – Sep 24, 2025

Subsystems (3)

v2 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#0 lore raw

v2 cover KVM: SVM: Add Page Modification Logging (PML) support

This series implements Page Modification Logging (PML) for guests, bringing hardware-assisted dirty logging support. PML is designed to track guest modified memory pages. PML enables the hypervisor to identify which pages in a guest's memory have been modified since the last checkpoint or during live migration.

The PML feature uses two new VMCB fields (PML_ADDR and PML_INDEX) and generates a VMEXIT when the 4KB log buffer becomes full.

Patch breakdown:
1. Refactor existing VMX PML code to be shared between VMX and SVM
2. Prepare to share the PML page in VMX and SVM
3. Add AMD SVM PML CPUID
4. Implement SVM PML support using the shared infrastructure

The feature is enabled by default when hardware support is detected and can be disabled via the 'pml' module parameter.

Changelog:

v2:
* Rebased on latest kvm/next
* Added patch to move pml_pg field from struct vcpu_vmx to struct kvm_vcpu_arch to share the PML page. (Kai Huang)
* Dropped the SNP safe allocation optimization patch, will submit it separately.
* Update commit message adding explicit mention that AMD PML follows VMX behavior (Kai Huang)
* Updated SNP erratum comment to include PML buffer alongside VMCB, VMSA, and AVIC pages. (Kai Huang)

RFC: https://lore.kernel.org/kvm/[email protected]/

Nikunj A Dadhania (4):
KVM: x86: Carve out PML flush routine
KVM: x86: Move PML page to common vcpu arch structure x86/cpufeatures: Add Page modification logging
KVM: SVM: Add Page modification logging support

 arch/x86/include/asm/cpufeatures.h |  1 +
 arch/x86/include/asm/kvm_host.h    |  2 +
 arch/x86/include/asm/svm.h         |  6 +-
 arch/x86/include/uapi/asm/svm.h    |  2 +
 arch/x86/kernel/cpu/scattered.c    |  1 +
 arch/x86/kvm/svm/sev.c             |  2 +-
 arch/x86/kvm/svm/svm.c             | 99 +++++++++++++++++++++++++++++-
 arch/x86/kvm/svm/svm.h             |  4 ++
 arch/x86/kvm/vmx/vmx.c             | 48 ++++-----------
 arch/x86/kvm/vmx/vmx.h             |  7 ---
 arch/x86/kvm/x86.c                 | 31 ++++++++++
 arch/x86/kvm/x86.h                 |  7 +++
 12 files changed, 163 insertions(+), 47 deletions(-)

base-commit: a6ad54137af92535cfe32e19e5f3bc1bb7dbd383

Signature · 1 line
2.48.1
#1 lore raw

PATCH v2 1/4 KVM: x86: Carve out PML flush routine

Move the PML (Page Modification Logging) buffer flushing logic from VMX-specific code to common x86 KVM code to enable reuse by SVM and avoid code duplication.

The AMD SVM PML implementations share the same behavior as VMX PML:
1) The PML buffer is a 4K page with 512 entries
2) Hardware records dirty GPAs in reverse order (from index 511 to 0)
3) Hardware clears bits 11:0 when recording GPAs

The PML constants (PML_LOG_NR_ENTRIES and PML_HEAD_INDEX) are moved from vmx.h to x86.h to make them available to both VMX and SVM.

No functional change intended for VMX, except tone down the WARN_ON() to WARN_ON_ONCE() for the page alignment check. If hardware exhibits this behavior once, it's likely to occur repeatedly, so use WARN_ON_ONCE() to avoid log flooding while still capturing the unexpected condition.

The refactoring prepares for SVM to leverage the same PML flushing implementation.

---
 arch/x86/kvm/vmx/vmx.c | 26 ++------------------------
 arch/x86/kvm/vmx/vmx.h |  5 -----
 arch/x86/kvm/x86.c     | 31 +++++++++++++++++++++++++++++++
 arch/x86/kvm/x86.h     |  7 +++++++
 4 files changed, 40 insertions(+), 29 deletions(-)
arch/x86/kvm/vmx/vmx.c +2 -24
index aa157fe5b7b3..a0955155d7ca 100644
@@ -6107,37 +6107,15 @@ static void vmx_destroy_pml_buffer(struct vcpu_vmx *vmx)
 static void vmx_flush_pml_buffer(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
-	u16 pml_idx, pml_tail_index;
-	u64 *pml_buf;
-	int i;
+	u16 pml_idx;
 
 	pml_idx = vmcs_read16(GUEST_PML_INDEX);
 
 	/* Do nothing if PML buffer is empty */
 	if (pml_idx == PML_HEAD_INDEX)
 		return;
-	/*
-	 * PML index always points to the next available PML buffer entity
-	 * unless PML log has just overflowed.
-	 */
-	pml_tail_index = (pml_idx >= PML_LOG_NR_ENTRIES) ? 0 : pml_idx + 1;
 
-	/*
-	 * PML log is written backwards: the CPU first writes the entry 511
-	 * then the entry 510, and so on.
-	 *
-	 * Read the entries in the same order they were written, to ensure that
-	 * the dirty ring is filled in the same order the CPU wrote them.
-	 */
-	pml_buf = page_address(vmx->pml_pg);
-
-	for (i = PML_HEAD_INDEX; i >= pml_tail_index; i--) {
-		u64 gpa;
-
-		gpa = pml_buf[i];
-		WARN_ON(gpa & (PAGE_SIZE - 1));
-		kvm_vcpu_mark_page_dirty(vcpu, gpa >> PAGE_SHIFT);
-	}
+	kvm_flush_pml_buffer(vcpu, vmx->pml_pg, pml_idx);
 
 	/* reset PML index */
 	vmcs_write16(GUEST_PML_INDEX, PML_HEAD_INDEX);
arch/x86/kvm/vmx/vmx.h +0 -5
index d3389baf3ab3..4494c253727f 100644
@@ -269,11 +269,6 @@ struct vcpu_vmx {
 	unsigned int ple_window;
 	bool ple_window_dirty;
 
-	/* Support for PML */
-#define PML_LOG_NR_ENTRIES	512
-	/* PML is written backwards: this is the first entry written by the CPU */
-#define PML_HEAD_INDEX		(PML_LOG_NR_ENTRIES-1)
-
 	struct page *pml_pg;
 
 	/* apic deadline value in host tsc */
arch/x86/kvm/x86.c +31 -0
index 33fba801b205..123ebe7be184 100644
@@ -6417,6 +6417,37 @@ void kvm_arch_sync_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot)
 		kvm_vcpu_kick(vcpu);
 }
 
+void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, struct page *pml_page, u16 pml_idx)
+{
+	u16 pml_tail_index;
+	u64 *pml_buf;
+	int i;
+
+	/*
+	 * PML index always points to the next available PML buffer entity
+	 * unless PML log has just overflowed.
+	 */
+	pml_tail_index = (pml_idx >= PML_LOG_NR_ENTRIES) ? 0 : pml_idx + 1;
+
+	/*
+	 * PML log is written backwards: the CPU first writes the entry 511
+	 * then the entry 510, and so on.
+	 *
+	 * Read the entries in the same order they were written, to ensure that
+	 * the dirty ring is filled in the same order the CPU wrote them.
+	 */
+	pml_buf = page_address(pml_page);
+
+	for (i = PML_HEAD_INDEX; i >= pml_tail_index; i--) {
+		u64 gpa;
+
+		gpa = pml_buf[i];
+		WARN_ON_ONCE(gpa & (PAGE_SIZE - 1));
+		kvm_vcpu_mark_page_dirty(vcpu, gpa >> PAGE_SHIFT);
+	}
+}
+EXPORT_SYMBOL_GPL(kvm_flush_pml_buffer);
+
 int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
 			    struct kvm_enable_cap *cap)
 {
arch/x86/kvm/x86.h +7 -0
index bcfd9b719ada..23c188c0a24b 100644
@@ -699,4 +699,11 @@ int ____kvm_emulate_hypercall(struct kvm_vcpu *vcpu, int cpl,
 
 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
 
+/* Support for PML */
+#define PML_LOG_NR_ENTRIES	512
+/* PML is written backwards: this is the first entry written by the CPU */
+#define PML_HEAD_INDEX		(PML_LOG_NR_ENTRIES-1)
+
+void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, struct page *pml_pg, u16 pml_idx);
+
 #endif
Signature · 1 line
2.48.1
#2 lore raw

PATCH v2 2/4 KVM: x86: Move PML page to common vcpu arch structure

In reply to Nikunj A Dadhania

Move the PML page from VMX-specific vcpu_vmx structure to the common kvm_vcpu_arch structure to share it between VMX and SVM implementations.

Update all VMX references accordingly, and simplify the kvm_flush_pml_buffer() interface by removing the page parameter since it can now access the page directly from the vcpu structure.

No functional change, restructuring to prepare for SVM PML support.

---
 arch/x86/include/asm/kvm_host.h |  2 ++
 arch/x86/kvm/vmx/vmx.c          | 24 ++++++++++++------------
 arch/x86/kvm/vmx/vmx.h          |  2 --
 arch/x86/kvm/x86.c              |  4 ++--
 arch/x86/kvm/x86.h              |  2 +-
 5 files changed, 17 insertions(+), 17 deletions(-)
arch/x86/include/asm/kvm_host.h +2 -0
index c56cc54d682a..62a7d519fbaf 100644
@@ -857,6 +857,8 @@ struct kvm_vcpu_arch {
 	 */
 	struct kvm_mmu_memory_cache mmu_external_spt_cache;
 
+	struct page *pml_page;
+
 	/*
 	 * QEMU userspace and the guest each have their own FPU state.
 	 * In vcpu_run, we switch between the user and guest FPU contexts.
arch/x86/kvm/vmx/vmx.c +12 -12
index a0955155d7ca..9520e11b08d0 100644
@@ -4612,7 +4612,8 @@ int vmx_vcpu_precreate(struct kvm *kvm)
 
 static void init_vmcs(struct vcpu_vmx *vmx)
 {
-	struct kvm *kvm = vmx->vcpu.kvm;
+	struct kvm_vcpu *vcpu = &vmx->vcpu;
+	struct kvm *kvm = vcpu->kvm;
 	struct kvm_vmx *kvm_vmx = to_kvm_vmx(kvm);
 
 	if (nested)
@@ -4703,7 +4704,7 @@ static void init_vmcs(struct vcpu_vmx *vmx)
 		vmcs_write64(XSS_EXIT_BITMAP, VMX_XSS_EXIT_BITMAP);
 
 	if (enable_pml) {
-		vmcs_write64(PML_ADDRESS, page_to_phys(vmx->pml_pg));
+		vmcs_write64(PML_ADDRESS, page_to_phys(vcpu->arch.pml_page));
 		vmcs_write16(GUEST_PML_INDEX, PML_HEAD_INDEX);
 	}
 
@@ -6096,17 +6097,16 @@ void vmx_get_entry_info(struct kvm_vcpu *vcpu, u32 *intr_info, u32 *error_code)
 		*error_code = 0;
 }
 
-static void vmx_destroy_pml_buffer(struct vcpu_vmx *vmx)
+static void vmx_destroy_pml_buffer(struct kvm_vcpu *vcpu)
 {
-	if (vmx->pml_pg) {
-		__free_page(vmx->pml_pg);
-		vmx->pml_pg = NULL;
+	if (vcpu->arch.pml_page) {
+		__free_page(vcpu->arch.pml_page);
+		vcpu->arch.pml_page = NULL;
 	}
 }
 
 static void vmx_flush_pml_buffer(struct kvm_vcpu *vcpu)
 {
-	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	u16 pml_idx;
 
 	pml_idx = vmcs_read16(GUEST_PML_INDEX);
@@ -6115,7 +6115,7 @@ static void vmx_flush_pml_buffer(struct kvm_vcpu *vcpu)
 	if (pml_idx == PML_HEAD_INDEX)
 		return;
 
-	kvm_flush_pml_buffer(vcpu, vmx->pml_pg, pml_idx);
+	kvm_flush_pml_buffer(vcpu, pml_idx);
 
 	/* reset PML index */
 	vmcs_write16(GUEST_PML_INDEX, PML_HEAD_INDEX);
@@ -7388,7 +7388,7 @@ void vmx_vcpu_free(struct kvm_vcpu *vcpu)
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 
 	if (enable_pml)
-		vmx_destroy_pml_buffer(vmx);
+		vmx_destroy_pml_buffer(vcpu);
 	free_vpid(vmx->vpid);
 	nested_vmx_free_vcpu(vcpu);
 	free_loaded_vmcs(vmx->loaded_vmcs);
@@ -7417,8 +7417,8 @@ int vmx_vcpu_create(struct kvm_vcpu *vcpu)
 	 * for the guest), etc.
 	 */
 	if (enable_pml) {
-		vmx->pml_pg = alloc_page(GFP_KERNEL_ACCOUNT | __GFP_ZERO);
-		if (!vmx->pml_pg)
+		vcpu->arch.pml_page = alloc_page(GFP_KERNEL_ACCOUNT | __GFP_ZERO);
+		if (!vcpu->arch.pml_page)
 			goto free_vpid;
 	}
 
@@ -7489,7 +7489,7 @@ int vmx_vcpu_create(struct kvm_vcpu *vcpu)
 free_vmcs:
 	free_loaded_vmcs(vmx->loaded_vmcs);
 free_pml:
-	vmx_destroy_pml_buffer(vmx);
+	vmx_destroy_pml_buffer(vcpu);
 free_vpid:
 	free_vpid(vmx->vpid);
 	return err;
arch/x86/kvm/vmx/vmx.h +0 -2
index 4494c253727f..6fafb6228c17 100644
@@ -269,8 +269,6 @@ struct vcpu_vmx {
 	unsigned int ple_window;
 	bool ple_window_dirty;
 
-	struct page *pml_pg;
-
 	/* apic deadline value in host tsc */
 	u64 hv_deadline_tsc;
 
arch/x86/kvm/x86.c +2 -2
index 123ebe7be184..afa7f8b46416 100644
@@ -6417,7 +6417,7 @@ void kvm_arch_sync_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot)
 		kvm_vcpu_kick(vcpu);
 }
 
-void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, struct page *pml_page, u16 pml_idx)
+void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, u16 pml_idx)
 {
 	u16 pml_tail_index;
 	u64 *pml_buf;
@@ -6436,7 +6436,7 @@ void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, struct page *pml_page, u16 pml_
 	 * Read the entries in the same order they were written, to ensure that
 	 * the dirty ring is filled in the same order the CPU wrote them.
 	 */
-	pml_buf = page_address(pml_page);
+	pml_buf = page_address(vcpu->arch.pml_page);
 
 	for (i = PML_HEAD_INDEX; i >= pml_tail_index; i--) {
 		u64 gpa;
arch/x86/kvm/x86.h +1 -1
index 23c188c0a24b..92016081a7e7 100644
@@ -704,6 +704,6 @@ int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
 /* PML is written backwards: this is the first entry written by the CPU */
 #define PML_HEAD_INDEX		(PML_LOG_NR_ENTRIES-1)
 
-void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, struct page *pml_pg, u16 pml_idx);
+void kvm_flush_pml_buffer(struct kvm_vcpu *vcpu, u16 pml_idx);
 
 #endif
Signature · 1 line
2.48.1
#3 lore raw

Re: [PATCH v2 2/4] KVM: x86: Move PML page to common vcpu arch structure

Quoted 11 lines · Nikunj A Dadhania

Move the PML page from VMX-specific vcpu_vmx structure to the common kvm_vcpu_arch structure to share it between VMX and SVM implementations.

Update all VMX references accordingly, and simplify the kvm_flush_pml_buffer() interface by removing the page parameter since it can now access the page directly from the vcpu structure.

No functional change, restructuring to prepare for SVM PML support.

Nit: IMHO it's also better to explain why we only moved the PML buffer pointer but not the code which allocates/frees the PML buffer:

Move the PML page to x86 common code only without moving the PML page allocation code, since for AMD the PML buffer must be allocated using snp_safe_alloc_page().

#4 lore raw

Re: [PATCH v2 2/4] KVM: x86: Move PML page to common vcpu arch structure

Quoted 13 lines · Kai
Quoted 11 lines · Nikunj A Dadhania

Move the PML page from VMX-specific vcpu_vmx structure to the common kvm_vcpu_arch structure to share it between VMX and SVM implementations.

Update all VMX references accordingly, and simplify the kvm_flush_pml_buffer() interface by removing the page parameter since it can now access the page directly from the vcpu structure.

No functional change, restructuring to prepare for SVM PML support.

Quoted 6 lines

Nit: IMHO it's also better to explain why we only moved the PML buffer pointer but not the code which allocates/frees the PML buffer:

Move the PML page to x86 common code only without moving the PML page allocation code, since for AMD the PML buffer must be allocated using snp_safe_alloc_page().

Ack

Regards
Nikunj

#5 lore raw

Re: [PATCH v2 2/4] KVM: x86: Move PML page to common vcpu arch structure

Quoted 15 lines · Nikunj A. Dadhania
Quoted 13 lines · Kai
Quoted 11 lines · Nikunj A Dadhania

Move the PML page from VMX-specific vcpu_vmx structure to the common kvm_vcpu_arch structure to share it between VMX and SVM implementations.

Update all VMX references accordingly, and simplify the kvm_flush_pml_buffer() interface by removing the page parameter since it can now access the page directly from the vcpu structure.

No functional change, restructuring to prepare for SVM PML support.

Quoted 9 lines
Quoted 6 lines

Nit: IMHO it's also better to explain why we only moved the PML buffer pointer but not the code which allocates/frees the PML buffer:

Move the PML page to x86 common code only without moving the PML page allocation code, since for AMD the PML buffer must be allocated using snp_safe_alloc_page().

Ack

Btw, just asking: why not just merging this patch to the first patch?

I don't have strong preference but seems patch 1/2 are connected (they both move VMX specific code to x86 common for share) and could be one.

#6 lore raw

Re: [PATCH v2 2/4] KVM: x86: Move PML page to common vcpu arch structure

Quoted 19 lines · Kai
Quoted 18 lines · Nikunj A. Dadhania
Quoted 16 lines · Kai
Quoted 11 lines · Nikunj A Dadhania

Move the PML page from VMX-specific vcpu_vmx structure to the common kvm_vcpu_arch structure to share it between VMX and SVM implementations.

Update all VMX references accordingly, and simplify the kvm_flush_pml_buffer() interface by removing the page parameter since it can now access the page directly from the vcpu structure.

No functional change, restructuring to prepare for SVM PML support.

Nit: IMHO it's also better to explain why we only moved the PML buffer pointer but not the code which allocates/frees the PML buffer:

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

Move the PML page to x86 common code only without moving the PML page allocation code, since for AMD the PML buffer must be allocated using snp_safe_alloc_page().

Ack

Btw, just asking: why not just merging this patch to the first patch?

Just kept it separate as VMX and SVM is not sharing the page allocation and if we do not want to have this churn of moving the pml_page to vcpu structure. We can drop this patch.

Quoted 2 lines

I don't have strong preference but seems patch 1/2 are connected (they both move VMX specific code to x86 common for share) and could be one.

Same here, I can squash this with patch 1.

Regards
Nikunj

#7 lore raw

Re: [PATCH v2 2/4] KVM: x86: Move PML page to common vcpu arch structure

Quoted 30 lines · Nikunj A. Dadhania
Quoted 19 lines · Kai
Quoted 18 lines · Nikunj A. Dadhania
Quoted 16 lines · Kai
Quoted 11 lines · Nikunj A Dadhania

Move the PML page from VMX-specific vcpu_vmx structure to the common kvm_vcpu_arch structure to share it between VMX and SVM implementations.

Update all VMX references accordingly, and simplify the kvm_flush_pml_buffer() interface by removing the page parameter since it can now access the page directly from the vcpu structure.

No functional change, restructuring to prepare for SVM PML support.

Nit: IMHO it's also better to explain why we only moved the PML buffer pointer but not the code which allocates/frees the PML buffer:

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

Move the PML page to x86 common code only without moving the PML page allocation code, since for AMD the PML buffer must be allocated using snp_safe_alloc_page().

Ack

Btw, just asking: why not just merging this patch to the first patch?

Quoted 8 lines

Just kept it separate as VMX and SVM is not sharing the page allocation and if we do not want to have this churn of moving the pml_page to vcpu structure. We can drop this patch.

Quoted 2 lines

I don't have strong preference but seems patch 1/2 are connected (they both move VMX specific code to x86 common for share) and could be one.

Same here, I can squash this with patch 1.

All right, up to you (and Sean/Paolo) :-)

#8 lore raw

PATCH v2 3/4 x86/cpufeatures: Add Page modification logging

In reply to Nikunj A Dadhania

Page modification logging(PML) is a hardware feature designed to track guest modified memory pages. PML enables the hypervisor to identify which pages in a guest's memory have been changed since the last checkpoint or during live migration.

The PML feature is advertised via CPUID leaf 0x8000000A ECX[4] bit.

---
 arch/x86/include/asm/cpufeatures.h | 1 +
 arch/x86/kernel/cpu/scattered.c    | 1 +
 2 files changed, 2 insertions(+)
arch/x86/include/asm/cpufeatures.h +1 -0
index 06fc0479a23f..60e95f2bea06 100644
@@ -228,6 +228,7 @@
 #define X86_FEATURE_PVUNLOCK		( 8*32+20) /* PV unlock function */
 #define X86_FEATURE_VCPUPREEMPT		( 8*32+21) /* PV vcpu_is_preempted function */
 #define X86_FEATURE_TDX_GUEST		( 8*32+22) /* "tdx_guest" Intel Trust Domain Extensions Guest */
+#define X86_FEATURE_PML			( 8*32+23) /* AMD Page Modification logging */
 
 /* Intel-defined CPU features, CPUID level 0x00000007:0 (EBX), word 9 */
 #define X86_FEATURE_FSGSBASE		( 9*32+ 0) /* "fsgsbase" RDFSBASE, WRFSBASE, RDGSBASE, WRGSBASE instructions*/
arch/x86/kernel/cpu/scattered.c +1 -0
index 6b868afb26c3..aeca6f55d2fa 100644
@@ -48,6 +48,7 @@ static const struct cpuid_bit cpuid_bits[] = {
 	{ X86_FEATURE_PROC_FEEDBACK,		CPUID_EDX, 11, 0x80000007, 0 },
 	{ X86_FEATURE_AMD_FAST_CPPC,		CPUID_EDX, 15, 0x80000007, 0 },
 	{ X86_FEATURE_MBA,			CPUID_EBX,  6, 0x80000008, 0 },
+	{ X86_FEATURE_PML,			CPUID_ECX,  4, 0x8000000A, 0 },
 	{ X86_FEATURE_COHERENCY_SFW_NO,		CPUID_EBX, 31, 0x8000001f, 0 },
 	{ X86_FEATURE_SMBA,			CPUID_EBX,  2, 0x80000020, 0 },
 	{ X86_FEATURE_BMEC,			CPUID_EBX,  3, 0x80000020, 0 },
Signature · 1 line
2.48.1
#9 lore raw

PATCH v2 4/4 KVM: SVM: Add Page modification logging support

In reply to Nikunj A Dadhania

Currently, dirty logging relies on write protecting guest memory and marking dirty GFNs during subsequent write faults. This method works but incurs overhead due to additional write faults for each dirty GFN.

Implement support for the Page Modification Logging (PML) feature, a hardware-assisted method for efficient dirty logging. PML automatically logs dirty GPA[51:12] to a 4K buffer when the CPU sets NPT D-bits. Two new VMCB fields are utilized: PML_ADDR and PML_INDEX. The PML_INDEX is initialized to 511 (8 bytes per GPA entry), and the CPU decreases the PML_INDEX after logging each GPA. When the PML buffer is full, a VMEXIT(PML_FULL) with exit code 0x407 is generated.

PML is enabled by default when supported and can be disabled via the 'pml' module parameter.

---
 arch/x86/include/asm/svm.h      |  6 +-
 arch/x86/include/uapi/asm/svm.h |  2 +
 arch/x86/kvm/svm/sev.c          |  2 +-
 arch/x86/kvm/svm/svm.c          | 99 ++++++++++++++++++++++++++++++++-
 arch/x86/kvm/svm/svm.h          |  4 ++
 5 files changed, 108 insertions(+), 5 deletions(-)
arch/x86/include/asm/svm.h +5 -1
index ffc27f676243..9fbada95afd5 100644
@@ -165,7 +165,10 @@ struct __attribute__ ((__packed__)) vmcb_control_area {
 	u8 reserved_9[22];
 	u64 allowed_sev_features;	/* Offset 0x138 */
 	u64 guest_sev_features;		/* Offset 0x140 */
-	u8 reserved_10[664];
+	u8 reserved_10[128];
+	u64 pml_addr;			/* Offset 0x1c8 */
+	u16 pml_index;			/* Offset 0x1d0 */
+	u8 reserved_11[526];
 	/*
 	 * Offset 0x3e0, 32 bytes reserved
 	 * for use by hypervisor/software.
@@ -239,6 +242,7 @@ struct __attribute__ ((__packed__)) vmcb_control_area {
 #define SVM_NESTED_CTL_NP_ENABLE	BIT(0)
 #define SVM_NESTED_CTL_SEV_ENABLE	BIT(1)
 #define SVM_NESTED_CTL_SEV_ES_ENABLE	BIT(2)
+#define SVM_NESTED_CTL_PML_ENABLE	BIT(11)
 
 
 #define SVM_TSC_RATIO_RSVD	0xffffff0000000000ULL
arch/x86/include/uapi/asm/svm.h +2 -0
index 9c640a521a67..f329dca167de 100644
@@ -101,6 +101,7 @@
 #define SVM_EXIT_AVIC_INCOMPLETE_IPI		0x401
 #define SVM_EXIT_AVIC_UNACCELERATED_ACCESS	0x402
 #define SVM_EXIT_VMGEXIT       0x403
+#define SVM_EXIT_PML_FULL	0x407
 
 /* SEV-ES software-defined VMGEXIT events */
 #define SVM_VMGEXIT_MMIO_READ			0x80000001
@@ -232,6 +233,7 @@
 	{ SVM_EXIT_AVIC_INCOMPLETE_IPI,		"avic_incomplete_ipi" }, \
 	{ SVM_EXIT_AVIC_UNACCELERATED_ACCESS,   "avic_unaccelerated_access" }, \
 	{ SVM_EXIT_VMGEXIT,		"vmgexit" }, \
+	{ SVM_EXIT_PML_FULL,		"pml_full" }, \
 	{ SVM_VMGEXIT_MMIO_READ,	"vmgexit_mmio_read" }, \
 	{ SVM_VMGEXIT_MMIO_WRITE,	"vmgexit_mmio_write" }, \
 	{ SVM_VMGEXIT_NMI_COMPLETE,	"vmgexit_nmi_complete" }, \
arch/x86/kvm/svm/sev.c +1 -1
index 5bac4d20aec0..b179a0a2581a 100644
@@ -4669,7 +4669,7 @@ struct page *snp_safe_alloc_page_node(int node, gfp_t gfp)
 	 * Allocate an SNP-safe page to workaround the SNP erratum where
 	 * the CPU will incorrectly signal an RMP violation #PF if a
 	 * hugepage (2MB or 1GB) collides with the RMP entry of a
-	 * 2MB-aligned VMCB, VMSA, or AVIC backing page.
+	 * 2MB-aligned VMCB, VMSA, PML or AVIC backing page.
 	 *
 	 * Allocate one extra page, choose a page which is not
 	 * 2MB-aligned, and free the other.
arch/x86/kvm/svm/svm.c +96 -3

178 lines not on this page · raw message

arch/x86/kvm/svm/svm.h +4 -0
index 70df7c6413cf..7e6ee4e80021 100644
@@ -335,6 +335,8 @@ struct vcpu_svm {
 
 	/* Guest GIF value, used when vGIF is not enabled */
 	bool guest_gif;
+
+	struct page *pml_page;
 };
 
 struct svm_cpu_data {
@@ -717,6 +719,8 @@ static inline void svm_enable_intercept_for_msr(struct kvm_vcpu *vcpu,
 	svm_set_intercept_for_msr(vcpu, msr, type, true);
 }
 
+void svm_update_cpu_dirty_logging(struct kvm_vcpu *vcpu);
+
 /* nested.c */
 
 #define NESTED_EXIT_HOST	0	/* Exit handled on host level */
Signature · 1 line
2.48.1
#10 lore raw

Re: [PATCH v2 4/4] KVM: SVM: Add Page modification logging support

Quoted 3 lines · Nikunj A Dadhania

Currently, dirty logging relies on write protecting guest memory and marking dirty GFNs during subsequent write faults.

Better to point out "On AMD platforms only".

Quoted 11 lines

This method works but
incurs overhead due to additional write faults for each dirty GFN.

Implement support for the Page Modification Logging (PML) feature, a hardware-assisted method for efficient dirty logging. PML automatically logs dirty GPA[51:12] to a 4K buffer when the CPU sets NPT D-bits. Two new VMCB fields are utilized: PML_ADDR and PML_INDEX. The PML_INDEX is initialized to 511 (8 bytes per GPA entry), and the CPU decreases the PML_INDEX after logging each GPA. When the PML buffer is full, a VMEXIT(PML_FULL) with exit code 0x407 is generated.

Quoted 2 lines

PML is enabled by default when supported and can be disabled via the 'pml' module parameter.

This changelog mentions nothing about interaction between PML vs nested.

On VMX, PML is emulated for L2 (for nested EPT) but is never enabled in hardware when CPU runs in L2, so:

1) PML is exposed to L1 (for nested EPT).
2) PML needs to be turned off when CPU runs in L2 otherwise L2's GPA could be logged, and turned on again after CPU leaves L2 (and restore PML buffer/index of VMCS01).

It doesn't seem this series supports emulating PML for L2 (for nested NPT), because AMD's PML is also enumerated via a CPUID bit (while VMX doesn't) and it's not exposed to guest, so we don't need to handle nested PML_FULL VMEXIT etc.

This is fine I think, and we can support this in the future if needed.

But 2) is also needed anyway for AMD's PML AFAICT, regardless of whether
1) is supported or not ?

If so, could we add some text to clarify all of these in the changelog?

[...]

arch/x86/kvm/svm/svm.c static int bus_lock_exit(struct kvm_vcpu *vcpu) · line 3242

Quoted 10 lines
arch/x86/kvm/svm/svm.c +9 -0
+void svm_update_cpu_dirty_logging(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	if (WARN_ON_ONCE(!pml))
+		return;
+
+	if (is_guest_mode(vcpu))
+		return;

VMX has a vmx->nested.update_vmcs01_cpu_dirty_logging boolean. It's set here to indicate PML enabling is not updated for L2 here, but later when switching to run in L1, the PML enabling needs to updated.

Shouldn't SVM have similar handling?

Quoted 13 lines
arch/x86/kvm/svm/svm.c +12 -0
+
+	/*
+	 * Note, nr_memslots_dirty_logging can be changed concurrently with this
+	 * code, but in that case another update request will be made and so the
+	 * guest will never run with a stale PML value.
+	 */
+	if (atomic_read(&vcpu->kvm->nr_memslots_dirty_logging))
+		svm->vmcb->control.nested_ctl |= SVM_NESTED_CTL_PML_ENABLE;
+	else
+		svm->vmcb->control.nested_ctl &= ~SVM_NESTED_CTL_PML_ENABLE;
+}
+

[...]

arch/x86/kvm/svm/svm.h struct vcpu_svm { · line 340

Quoted 9 lines
arch/x86/kvm/svm/svm.h +2 -0
@@ -335,6 +335,8 @@ struct vcpu_svm {
 
 	/* Guest GIF value, used when vGIF is not enabled */
 	bool guest_gif;
+
+	struct page *pml_page;
 };

This seems to be a leftover.

#11 lore raw

Re: [PATCH v2 4/4] KVM: SVM: Add Page modification logging support

Quoted 6 lines · Kai
Quoted 3 lines · Nikunj A Dadhania

Currently, dirty logging relies on write protecting guest memory and marking dirty GFNs during subsequent write faults.

Better to point out "On AMD platforms only".

Subject being "KVM: SVM: ", this seems redundant to me.

Quoted 17 lines
Quoted 12 lines

This method works but
incurs overhead due to additional write faults for each dirty GFN.

Implement support for the Page Modification Logging (PML) feature, a hardware-assisted method for efficient dirty logging. PML automatically logs dirty GPA[51:12] to a 4K buffer when the CPU sets NPT D-bits. Two new VMCB fields are utilized: PML_ADDR and PML_INDEX. The PML_INDEX is initialized to 511 (8 bytes per GPA entry), and the CPU decreases the PML_INDEX after logging each GPA. When the PML buffer is full, a VMEXIT(PML_FULL) with exit code 0x407 is generated.

Quoted 2 lines

PML is enabled by default when supported and can be disabled via the 'pml' module parameter.

This changelog mentions nothing about interaction between PML vs nested.

Quoted 7 lines

On VMX, PML is emulated for L2 (for nested EPT) but is never enabled in hardware when CPU runs in L2, so:

1) PML is exposed to L1 (for nested EPT).
2) PML needs to be turned off when CPU runs in L2 otherwise L2's GPA could be logged, and turned on again after CPU leaves L2 (and restore PML buffer/index of VMCS01).

I get your point and I see that when nested VM entry, PML is set in the nested_ctl for L2.

I am trying to create this scenario, and couldnt get the L2 GPA's.

Quoted 10 lines

It doesn't seem this series supports emulating PML for L2 (for nested NPT), because AMD's PML is also enumerated via a CPUID bit (while VMX doesn't) and it's not exposed to guest, so we don't need to handle nested PML_FULL VMEXIT etc.

This is fine I think, and we can support this in the future if needed.

But 2) is also needed anyway for AMD's PML AFAICT, regardless of whether
1) is supported or not ?

I see your point, we will need to disable PML for L2.

Quoted 21 lines

If so, could we add some text to clarify all of these in the changelog?

[...]

arch/x86/kvm/svm/svm.c static int bus_lock_exit(struct kvm_vcpu *vcpu) · line 3242

Quoted 10 lines
arch/x86/kvm/svm/svm.c +9 -0
+void svm_update_cpu_dirty_logging(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	if (WARN_ON_ONCE(!pml))
+		return;
+
+	if (is_guest_mode(vcpu))
+		return;

VMX has a vmx->nested.update_vmcs01_cpu_dirty_logging boolean. It's set here to indicate PML enabling is not updated for L2 here, but later when switching to run in L1, the PML enabling needs to updated.

Quoted 1 line

Shouldn't SVM have similar handling?

Sure, will get back to you on this.

Quoted 20 lines
Quoted 14 lines
arch/x86/kvm/svm/svm.c +12 -0
+
+	/*
+	 * Note, nr_memslots_dirty_logging can be changed concurrently with this
+	 * code, but in that case another update request will be made and so the
+	 * guest will never run with a stale PML value.
+	 */
+	if (atomic_read(&vcpu->kvm->nr_memslots_dirty_logging))
+		svm->vmcb->control.nested_ctl |= SVM_NESTED_CTL_PML_ENABLE;
+	else
+		svm->vmcb->control.nested_ctl &= ~SVM_NESTED_CTL_PML_ENABLE;
+}
+

[...]

Quoted 4 lines
arch/x86/kvm/svm/svm.h +0 -0
@@ -335,6 +335,8 @@ struct vcpu_svm {
 
Quoted 7 lines

arch/x86/kvm/svm/svm.h struct vcpu_svm { · line 340

Quoted 5 lines
arch/x86/kvm/svm/svm.h +2 -0
 	/* Guest GIF value, used when vGIF is not enabled */
 	bool guest_gif;
+
+	struct page *pml_page;
 };

This seems to be a leftover.

Sure.

Regards
Nikunj

#12 lore raw

Re: [PATCH v2 4/4] KVM: SVM: Add Page modification logging support

Quoted 12 lines · Nikunj A Dadhania
Quoted 10 lines · Kai
Quoted 3 lines · Nikunj A Dadhania

PML is enabled by default when supported and can be disabled via the 'pml' module parameter.

This changelog mentions nothing about interaction between PML vs nested.

On VMX, PML is emulated for L2 (for nested EPT) but is never enabled in hardware when CPU runs in L2, so:

Quoted 7 lines
Quoted 4 lines

1) PML is exposed to L1 (for nested EPT).
2) PML needs to be turned off when CPU runs in L2 otherwise L2's GPA could be logged, and turned on again after CPU leaves L2 (and restore PML buffer/index of VMCS01).

I get your point and I see that when nested VM entry, PML is set in the nested_ctl for L2.

As nested_ctl from L1 was getting copied into L2, PML flag was also set for L2. I have disabled this now.

Quoted 2 lines

I am trying to create this scenario, and couldnt get the L2 GPA's.

L2 GPA's were not logged because PML_ADDR and PML_INDEX of the VMCB02 was not populated.

Quoted 35 lines
Quoted 11 lines

It doesn't seem this series supports emulating PML for L2 (for nested NPT), because AMD's PML is also enumerated via a CPUID bit (while VMX doesn't) and it's not exposed to guest, so we don't need to handle nested PML_FULL VMEXIT etc.

This is fine I think, and we can support this in the future if needed.

But 2) is also needed anyway for AMD's PML AFAICT, regardless of whether
1) is supported or not ?

I see your point, we will need to disable PML for L2.

Quoted 21 lines

If so, could we add some text to clarify all of these in the changelog?

[...]

arch/x86/kvm/svm/svm.c static int bus_lock_exit(struct kvm_vcpu *vcpu) · line 3242

Quoted 10 lines
arch/x86/kvm/svm/svm.c +9 -0
+void svm_update_cpu_dirty_logging(struct kvm_vcpu *vcpu)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	if (WARN_ON_ONCE(!pml))
+		return;
+
+	if (is_guest_mode(vcpu))
+		return;

VMX has a vmx->nested.update_vmcs01_cpu_dirty_logging boolean. It's set here to indicate PML enabling is not updated for L2 here, but later when switching to run in L1, the PML enabling needs to updated.

Quoted 3 lines
Quoted 1 line

Shouldn't SVM have similar handling?

Sure, will get back to you on this.

Yes, this will be required as it is handling a case when there is a dirty logging change for L1 while L2 is running. So this variable is used to record such change and when L2 exits, the pending change needs to be made in L1's VMCB.

Thanks for the detailed review.

Regards
Nikunj