Skip to contentSkip to conversation

TDX MMU prep series part 1

PATCH Superseded Rick Edgecombe 4 participants 47 messages Jun 19, 2024 – Jul 18, 2024

Subsystems (4)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover TDX MMU prep series part 1

Hi,

This is v3 of the TDX MMU prep series, split out of the giant 130 patch TDX base enabling series [0]. It is focusing on the changes to the x86 MMU to support TDX’s separation of private/shared EPT into separate roots. A future breakout series will include the changes to actually interact with the TDX module to actually map private memory. The purpose of sending out a smaller series is to focus review, and hopefully rapidly iterate. We would like the series to go into kvm-coco-queue when it is ready.

I think patches are in pretty good shape at this point.

There is a larger team working on TDX KVM base enabling. The patches were originally authored by Sean Christopherson and Isaku Yamahata, but otherwise it especially represents the work of Isaku and Yan Y Zhao and myself.

The series has been tested as part of a development branch for the TDX base series [1]. The testing of this series consists TDX kvm-unit-tests [2], regular KVM selftests, and booting a Linux TD. Rebasing the TDX selftests is still in progress.

Updates from v3
===============
For v2, Paolo did an extensive review. Most of that feedback was non-functional, but internally we found two issues with the TDP MMU iterator changes implemented in v2:
- Shared bits not applied in try_step_side(), which affects the math in 4-level paging
- Shared bit passed into tdp_iter_start() for fast page fault path

Besides those fixes some other changes of note:
- "KVM: x86/tdp_mmu: Add a helper function to walk down the TDP MMU" is pushed out of this series for later.
- New patch "KVM: x86/tdp_mmu: Take a GFN in kvm_tdp_mmu_fast_pf_get_last_sptep()" to fix one of the previously mentioned bugs found earlier.
- New patch "KVM: x86/tdp_mmu: Rename REMOVED_SPTE to FROZEN_SPTE", which Paolo mentioned might be a candidate for 6.11

With respect to other series in progress:
- The "Introduce a quirk to control memslot zap behavior" [3] series will need a change in order to work with this. The attr_filter field (previously process) added in this series will need to be set in that series' kvm_mmu_zap_memslot_leafs() function.

I will post an additional patch, that goes on top of that series and can be added when they are combined. Any branch with either of those series should be fine functionally until we can create VM of the TDX type.

- The series to make max gfn configurable is still outstanding, but not required functionally.

This series is on top of kvm-coco-queue and two fixes to that branch. The only critical one is the patch at [4].

Private/shared memory in TDX
============================
Confidential computing solutions have concepts of private and shared memory. Often the guest accesses either private or shared memory via a bit in the PTE. Solutions like SEV treat this bit more like a permission bit, where solutions like TDX and ARM CCA treat it more like a GPA bit. In the latter case, the host maps private memory in one half of the address space and shared in another. For TDX these two halves are mapped by different EPT roots. The private half (also called Secure EPT in Intel documentation) gets managed by the privileged TDX Module. The shared half is managed by the untrusted part of the VMM (KVM).

In addition to the separate roots for private and shared, there are limitations on what operations can be done on the private side. TDX wants to protect against protected memory being reset or otherwise scrambled by the host. In order to prevent this, the guest has to take specific action to “accept” memory after changes are made by the VMM to the private EPT. This prevents the VMM from performing many of the usual memory management operations that involve zapping and refaulting memory. The private memory also is always RWX and cannot have VMM specified cache attribute attributes applied.

TDX KVM MMU Design For Private Memory
=====================================

Private/shared split
--------------------
The operations that actually change the private half of the EPT are limited and relatively slow compared to reading a PTE. For this reason the design for KVM is to keep a “mirrored” copy of the private EPT in KVM’s memory. This will allow KVM to quickly walk the EPT and only perform the slower private EPT operations when it needs to actually modify mid-level private PTEs.

To clarify the definitions of the three EPT trees at this point:
external EPT - Protected by the TDX module, modified via TDX module
calls.
mirror EPT - Bookkeeping tree used as an optimization by KVM, not
mapped.
shared EPT - Normal EPT that maps unencrypted shared memory.
Managed like the EPT of a normal VM.

It’s worth noting that we are making an effort to remove optimizations that have complexity for the base enabling. Although keeping a mirrored copy of the private page tables kind of fits into that category, it has been so fundamental to the design for so long, dropping it would be too disruptive.

Mirror EPT
------------
The mirror EPT needs to keep a mirrored version of the private EPT maintained in the TDX module in order to be able to find out if a GPA’s mid-level pagetable have already been installed. So this mirrored copy has the same structure as the private EPT, having a page table present for every GPA range and level in the mirrored EPT where a page table is present private. The private page tables also cannot be zapped while the range has anything mapped, so the mirrored/private page tables need to be protected from KVM operations that zap any non-leaf PTEs, for example kvm_mmu_reset_context() or kvm_mmu_zap_all_fast()

Modifications to the mirrored page tables need to also perform the same operations to the private page tables. The actual TDX module calls to do this are not covered in this prep series.

For convenience SPs for private page tables are tracked with a role bit out of convenience. (Note to reviewers, please consider if this is really needed).

Zapping Changes
---------------
For normal VMs, guest memory is zapped for several reasons, like user memory getting paged out by the guest, memslots getting deleted or virtualization operations like MTRRs, and attachment of non-coherent DMA. For TDX (and SNP) there is also zapping associated with the conversion of memory between shared and privates. These operations need to take care to do two things:
1. Not zap any private memory that is in use by the guest.
2. Not zap any memory alias unnecessarily (i.e. Don’t zap anything more than needed). The purpose of this is to not have any unnecessary behavior userspace could grow to rely on.

For 1, this is possible because the zapping that is out of the control of KVM/userspace (paging out of userspace memory) will only apply to shared memory. Guest mem fd operations are protected from mmu notifier operations. During TD runtime, zapping of private memory will only be from memslot deletion and from conversion between private and shared memory which is triggered by the guest.

For 2, KVM needs to be taught which operations will operate on which aliases. An enum based scheme is introduced such that operations can target specific aliases like:
Memslot deletion - Private and shared
MMU notifier based zapping - Shared only

Conversion to shared       - Private only
Conversion to private      - Shared only
MTRRs, etc                 - Zapping will be avoided all together

For zapping arising from other virtualization based operations, there are four scenarios:
1. MTRR update
2. CR0.CD update
3. APICv update
4. Non-coherent DMA status update

KVM TDX will not support 1-3. In future changes (after this series) the features will not be supported for TDX. For 4, there isn’t an easy way to not support the feature as the notification is just passed to KVM and it has to act accordingly. However, other proposed changes [5] will avoid the need for zapping on non-coherent DMA notification for selfsnoop CPUs. So KVM can follow this logic and just always honor guest PAT for shared memory.

Atomically updating private EPT
-------------------------------
Although this prep series does not interact with the TDX module at all to actually configure the private EPT, it does lay the ground work for doing this. In some ways updating the private EPT is as simple as plumbing PTE modifications through to also call into the TDX module, but there is one tricky property that is worth elaborating on. That is how to handle that the TDP MMU allows modification of PTEs with the mmu_lock held only for read and uses the PTEs themselves to perform synchronization.

Unfortunately while operating on a single PTE can be done atomically, operating on both the mirrored and private PTEs at the same time needs additional solution. To handle this situation, REMOVED_SPTE is used to prevent concurrent operations while a call to the TDX module updates the private EPT.

The series is based on kvm-coco-queue.

[0] https://lore.kernel.org/kvm/[email protected]/
[1] https://github.com/intel/tdx/tree/tdx_kvm_dev-2024-06-19
[2] https://lore.kernel.org/kvm/[email protected]/
[3] https://lore.kernel.org/kvm/[email protected]/
[4] https://lore.kernel.org/lkml/[email protected]/
[5] https://lore.kernel.org/kvm/[email protected]/

Isaku Yamahata (13):
KVM: Add member to struct kvm_gfn_range for target alias
KVM: x86/mmu: Add an external pointer to struct kvm_mmu_page
KVM: x86/mmu: Add an is_mirror member for union kvm_mmu_page_role
KVM: x86/tdp_mmu: Take struct kvm in iter loops
KVM: x86/mmu: Support GFN direct bits
KVM: x86/tdp_mmu: Extract root invalid check from tdx_mmu_next_root()
KVM: x86/tdp_mmu: Introduce KVM MMU root types to specify page table
type
KVM: x86/tdp_mmu: Take root in tdp_mmu_for_each_pte()
KVM: x86/tdp_mmu: Support mirror root for TDP MMU
KVM: x86/tdp_mmu: Propagate attr_filter to MMU notifier callbacks
KVM: x86/tdp_mmu: Propagate building mirror page tables
KVM: x86/tdp_mmu: Propagate tearing down mirror page tables
KVM: x86/tdp_mmu: Take root types for
kvm_tdp_mmu_invalidate_all_roots()

Rick Edgecombe (4):
KVM: x86/tdp_mmu: Rename REMOVED_SPTE to FROZEN_SPTE
KVM: x86: Add a VM type define for TDX
KVM: x86/mmu: Make kvm_tdp_mmu_alloc_root() return void
KVM: x86/tdp_mmu: Take a GFN in kvm_tdp_mmu_fast_pf_get_last_sptep()

 arch/x86/include/asm/kvm-x86-ops.h |   4 +
 arch/x86/include/asm/kvm_host.h    |  26 ++-
 arch/x86/include/uapi/asm/kvm.h    |   1 +
 arch/x86/kvm/mmu.h                 |  17 ++
 arch/x86/kvm/mmu/mmu.c             |  41 +++-
 arch/x86/kvm/mmu/mmu_internal.h    |  64 +++++-
 arch/x86/kvm/mmu/spte.c            |   2 +-
 arch/x86/kvm/mmu/spte.h            |  15 +-
 arch/x86/kvm/mmu/tdp_iter.c        |  10 +-
 arch/x86/kvm/mmu/tdp_iter.h        |  16 +-
 arch/x86/kvm/mmu/tdp_mmu.c         | 328 +++++++++++++++++++++--------
 arch/x86/kvm/mmu/tdp_mmu.h         |  51 ++++-
 include/linux/kvm_host.h           |   6 +
 virt/kvm/guest_memfd.c             |   2 +
 virt/kvm/kvm_main.c                |  14 ++
 15 files changed, 481 insertions(+), 116 deletions(-)

base-commit: 698ca1e403579ca00e16a5b28ae4d576d9f1b20e
prerequisite-patch-id: c10f666d6a21c1485a32f393a50f38aa69c25caa
prerequisite-patch-id: a525ee37f0c8f41e74909343517d57a553724152

Signature · 1 line
2.34.1
#1 lore raw

PATCH v3 1/17 KVM: x86/tdp_mmu: Rename REMOVED_SPTE to FROZEN_SPTE

Rename REMOVED_SPTE to FROZEN_SPTE so that it can be used for other multi-part operations.

REMOVED_SPTE is used as a non-present intermediate value for multi-part operations that can happen when a thread doesn't have an MMU write lock. Today these operations are when removing PTEs.

However, future changes will want to use the same concept for setting a PTE. In that case the REMOVED_SPTE name does not quite fit. So rename it to FROZEN_SPTE so it can be used for both types of operations.

Also rename the relevant helpers and comments that refer to "removed" within the context of the SPTE value. Take care to not update naming referring the "remove" operations, which are still distinct.

---
TDX MMU Prep v3:
- New patch

---
 arch/x86/kvm/mmu/mmu.c     |  2 +-
 arch/x86/kvm/mmu/spte.c    |  2 +-
 arch/x86/kvm/mmu/spte.h    | 10 ++++-----
 arch/x86/kvm/mmu/tdp_mmu.c | 42 +++++++++++++++++++-------------------
 4 files changed, 28 insertions(+), 28 deletions(-)
arch/x86/kvm/mmu/mmu.c +1 -1
index 61982da8c8b2..828c70ead96f 100644
@@ -3455,7 +3455,7 @@ static int fast_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 		 * available as the vCPU holds a reference to its root(s).
 		 */
 		if (WARN_ON_ONCE(!sptep))
-			spte = REMOVED_SPTE;
+			spte = FROZEN_SPTE;
 
 		if (!is_shadow_present_pte(spte))
 			break;
arch/x86/kvm/mmu/spte.c +1 -1
index a5e014d7bc62..59cac37615b6 100644
@@ -383,7 +383,7 @@ void kvm_mmu_set_mmio_spte_mask(u64 mmio_value, u64 mmio_mask, u64 access_mask)
 	 * not set any RWX bits.
 	 */
 	if (WARN_ON((mmio_value & mmio_mask) != mmio_value) ||
-	    WARN_ON(mmio_value && (REMOVED_SPTE & mmio_mask) == mmio_value))
+	    WARN_ON(mmio_value && (FROZEN_SPTE & mmio_mask) == mmio_value))
 		mmio_value = 0;
 
 	if (!mmio_value)
arch/x86/kvm/mmu/spte.h +5 -5
index 5dd5405fa07a..86e5259aa824 100644
@@ -200,7 +200,7 @@ extern u64 __read_mostly shadow_nonpresent_or_rsvd_mask;
 
 /*
  * If a thread running without exclusive control of the MMU lock must perform a
- * multi-part operation on an SPTE, it can set the SPTE to REMOVED_SPTE as a
+ * multi-part operation on an SPTE, it can set the SPTE to FROZEN_SPTE as a
  * non-present intermediate value. Other threads which encounter this value
  * should not modify the SPTE.
  *
@@ -210,14 +210,14 @@ extern u64 __read_mostly shadow_nonpresent_or_rsvd_mask;
  *
  * Only used by the TDP MMU.
  */
-#define REMOVED_SPTE	(SHADOW_NONPRESENT_VALUE | 0x5a0ULL)
+#define FROZEN_SPTE	(SHADOW_NONPRESENT_VALUE | 0x5a0ULL)
 
 /* Removed SPTEs must not be misconstrued as shadow present PTEs. */
-static_assert(!(REMOVED_SPTE & SPTE_MMU_PRESENT_MASK));
+static_assert(!(FROZEN_SPTE & SPTE_MMU_PRESENT_MASK));
 
-static inline bool is_removed_spte(u64 spte)
+static inline bool is_frozen_spte(u64 spte)
 {
-	return spte == REMOVED_SPTE;
+	return spte == FROZEN_SPTE;
 }
 
 /* Get an SPTE's index into its parent's page table (and the spt array). */
arch/x86/kvm/mmu/tdp_mmu.c +21 -21

121 lines not on this page · raw message

Signature · 1 line
2.34.1
#2 lore raw

PATCH v3 2/17 KVM: Add member to struct kvm_gfn_range for target alias

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Add new members to strut kvm_gfn_range to indicate which mapping (private-vs-shared) to operate on: enum kvm_gfn_range_filter attr_filter. Update the core zapping operations to set them appropriately.

TDX utilizes two GPA aliases for the same memslots, one for memory that is for private memory and one that is for shared. For private memory, KVM cannot always perform the same operations it does on memory for default VMs, such as zapping pages and having them be faulted back in, as this requires guest coordination. However, some operations such as guest driven conversion of memory between private and shared should zap private memory.

Internally to the MMU, private and shared mappings are tracked on separate roots. Mapping and zapping operations will operate on the respective GFN alias for each root (private or shared). So zapping operations will by default zap both aliases. Add fields in struct kvm_gfn_range to allow callers to specify which aliases so they can only target the aliases appropriate for their specific operation.

There was feedback that target aliases should be specified such that the default value (0) is to operate on both aliases. Several options were considered. Several variations of having separate bools defined such that the default behavior was to process both aliases. They either allowed nonsensical configurations, or were confusing for the caller. A simple enum was also explored and was close, but was hard to process in the caller. Instead, use an enum with the default value (0) reserved as a disallowed value. Catch ranges that didn't have the target aliases specified by looking for that specific value.

Set target alias with enum appropriately for these MMU operations:
- For KVM's mmu notifier callbacks, zap shared pages only because private pages won't have a userspace mapping
- For setting memory attributes, kvm_arch_pre_set_memory_attributes() chooses the aliases based on the attribute.
- For guest_memfd invalidations, zap private only.

---
TDX MMU Prep v3:
- Fix typo in comment (Paolo)
- Remove KVM_PROCESS_PRIVATE_AND_SHARED (Paolo)
- Remove outdated reference to exclude_{private,shared} (Paolo)
- Set process member in new kvm_mmu_zap_memslot_leafs() function
- Rename process -> filter (Paolo)

TDX MMU Prep:
- Replaced KVM_PROCESS_BASED_ON_ARG with BUGGY_KVM_INVALIDATION to follow the original suggestion and not populte kvm_handle_gfn_range(). And add WARN_ON_ONCE().
- Move attribute specific logic into kvm_vm_set_mem_attributes()
- Drop Sean's suggested-by tag as the solution has changed
- Re-write commit log

v18:
- rebased to kvm-next

v3:
- Drop the KVM_GFN_RANGE flags
- Updated struct kvm_gfn_range
- Change kvm_arch_set_memory_attributes() to return bool for flush
- Added set_memory_attributes x86 op for vendor backends
- Refined commit message to describe TDX care concretely

---
 arch/x86/kvm/mmu/mmu.c   |  6 ++++++
 include/linux/kvm_host.h |  6 ++++++
 virt/kvm/guest_memfd.c   |  2 ++
 virt/kvm/kvm_main.c      | 14 ++++++++++++++
 4 files changed, 28 insertions(+)
arch/x86/kvm/mmu/mmu.c +6 -0
index 828c70ead96f..f41c498fcdb5 100644
@@ -7451,6 +7451,12 @@ bool kvm_arch_pre_set_memory_attributes(struct kvm *kvm,
 	if (WARN_ON_ONCE(!kvm_arch_has_private_mem(kvm)))
 		return false;
 
+	/* Unmap the old attribute page. */
+	if (range->arg.attributes & KVM_MEMORY_ATTRIBUTE_PRIVATE)
+		range->attr_filter = KVM_FILTER_SHARED;
+	else
+		range->attr_filter = KVM_FILTER_PRIVATE;
+
 	return kvm_unmap_gfn_range(kvm, range);
 }
 
include/linux/kvm_host.h +6 -0
index c3c922bf077f..8dce85962583 100644
@@ -260,11 +260,17 @@ union kvm_mmu_notifier_arg {
 	unsigned long attributes;
 };
 
+enum kvm_gfn_range_filter {
+	KVM_FILTER_SHARED		= BIT(0),
+	KVM_FILTER_PRIVATE		= BIT(1),
+};
+
 struct kvm_gfn_range {
 	struct kvm_memory_slot *slot;
 	gfn_t start;
 	gfn_t end;
 	union kvm_mmu_notifier_arg arg;
+	enum kvm_gfn_range_filter attr_filter;
 	bool may_block;
 };
 bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range);
virt/kvm/guest_memfd.c +2 -0
index 9714add38852..86aaf26c1144 100644
@@ -109,6 +109,8 @@ static void kvm_gmem_invalidate_begin(struct kvm_gmem *gmem, pgoff_t start,
 			.end = slot->base_gfn + min(pgoff + slot->npages, end) - pgoff,
 			.slot = slot,
 			.may_block = true,
+			/* guest memfd is relevant to only private mappings. */
+			.attr_filter = KVM_FILTER_PRIVATE,
 		};
 
 		if (!found_memslot) {
virt/kvm/kvm_main.c +14 -0
index 81b90bf03f2f..93c7b227aae0 100644
@@ -635,6 +635,11 @@ static __always_inline kvm_mn_ret_t __kvm_handle_hva_range(struct kvm *kvm,
 			 */
 			gfn_range.arg = range->arg;
 			gfn_range.may_block = range->may_block;
+			/*
+			 * HVA-based notifications aren't relevant to private
+			 * mappings as they don't have a userspace mapping.
+			 */
+			gfn_range.attr_filter = KVM_FILTER_SHARED;
 
 			/*
 			 * {gfn(page) | page intersects with [hva_start, hva_end)} =
@@ -2450,6 +2455,14 @@ static __always_inline void kvm_handle_gfn_range(struct kvm *kvm,
 	gfn_range.arg = range->arg;
 	gfn_range.may_block = range->may_block;
 
+	/*
+	 * If/when KVM supports more attributes beyond private .vs shared, this
+	 * _could_ set KVM_FILTER_{SHARED,PRIVATE} appropriately if the entire target
+	 * range already has the desired private vs. shared state (it's unclear
+	 * if that is a net win).  For now, KVM reaches this point if and only
+	 * if the private flag is being toggled, i.e. all mappings are in play.
+	 */
+
 	for (i = 0; i < kvm_arch_nr_memslot_as_ids(kvm); i++) {
 		slots = __kvm_memslots(kvm, i);
 
@@ -2506,6 +2519,7 @@ static int kvm_vm_set_mem_attributes(struct kvm *kvm, gfn_t start, gfn_t end,
 	struct kvm_mmu_notifier_range pre_set_range = {
 		.start = start,
 		.end = end,
+		.arg.attributes = attributes,
 		.handler = kvm_pre_set_memory_attributes,
 		.on_lock = kvm_mmu_invalidate_begin,
 		.flush_on_ret = true,
Signature · 1 line
2.34.1
#3 lore raw

PATCH v3 3/17 KVM: x86: Add a VM type define for TDX

In reply to Rick Edgecombe

Add a VM type define for TDX.

Future changes will need to lay the ground work for TDX support by making some behavior conditional on the VM being a TDX guest.

---
TDX MMU Prep:
- New patch, split from main series

---
 arch/x86/include/uapi/asm/kvm.h | 1 +
 1 file changed, 1 insertion(+)
arch/x86/include/uapi/asm/kvm.h +1 -0
index 988b5204d636..4dea0cfeee51 100644
@@ -922,5 +922,6 @@ struct kvm_hyperv_eventfd {
 #define KVM_X86_SEV_VM		2
 #define KVM_X86_SEV_ES_VM	3
 #define KVM_X86_SNP_VM		4
+#define KVM_X86_TDX_VM		5
 
 #endif /* _ASM_X86_KVM_H */
Signature · 1 line
2.34.1
#4 lore raw

PATCH v3 4/17 KVM: x86/mmu: Add an external pointer to struct kvm_mmu_page

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Add a external pointer to struct kvm_mmu_page for TDX's private page table and add helper functions to allocate/initialize/free a private page table page. TDX will only be supported with the TDP MMU. Because KVM TDP MMU doesn't use unsync_children and write_flooding_count, pack them to have room for a pointer and use a union to avoid memory overhead.

For private GPA, CPU refers to a private page table whose contents are encrypted. The dedicated APIs to operate on it (e.g. updating/reading its PTE entry) are used, and their cost is expensive.

When KVM resolves the KVM page fault, it walks the page tables. To reuse the existing KVM MMU code and mitigate the heavy cost of directly walking the private page table allocate two sets of page tables for the private half of the GPA space.

For the page tables that KVM will walk, allocate them like normal and refer to them as mirror page tables. Additionally allocate one more page for the page tables the CPU will walk, and call them external page tables. Resolve the KVM page fault with the existing code, and do additional operations necessary for modifying the external page table in future patches.

The relationship of the types of page tables in this scheme is depicted below:

              KVM page fault                     |
                     |                           |
                     V                           |
        -------------+----------                 |
        |                      |                 |
        V                      V                 |
     shared GPA           private GPA            |
        |                      |                 |
        V                      V                 |
    shared PT root      mirror PT root           |    private PT root
        |                      |                 |           |
        V                      V                 |           V
     shared PT           mirror PT        --propagate--> external PT
        |                      |                 |           |
        |                      \-----------------+------\    |
        |                                        |      |    |
        V                                        |      V    V
  shared guest page                              |    private guest page
                                                 |
                           non-encrypted memory  |    encrypted memory
                                                 |

PT - Page table
Shared PT - Visible to KVM, and the CPU uses it for shared mappings.
External PT - The CPU uses it, but it is invisible to KVM. TDX module
updates this table to map private guest pages.
Mirror PT - It is visible to KVM, but the CPU doesn't use it. KVM uses
it to propagate PT change to the actual private PT.

Add a helper kvm_has_mirrored_tdp() to trigger this behavior and wire it to the TDX vm type.

---
TDX MMU Prep v3:
- mirrored->external rename (Paolo)
- Remove accidentally included kvm_mmu_alloc_private_spt() (Paolo)
- Those -> These (Paolo)
- Change log updates to make external/mirrored naming more clear

TDX MMU Prep v2:
- Rename private->mirror
- Don't trigger off of shared mask

TDX MMU Prep:
- Rename terminology, dummy PT => mirror PT. and updated the commit message By Rick and Kai.
- Added a comment on union of private_spt by Rick.
- Don't handle the root case in kvm_mmu_alloc_private_spt(), it will not be needed in future patches. (Rick)
- Update comments (Yan)
- Remove kvm_mmu_init_private_spt(), open code it in later patches (Yan)

v19:
- typo in the comment in kvm_mmu_alloc_private_spt()
- drop CONFIG_KVM_MMU_PRIVATE

---
 arch/x86/include/asm/kvm_host.h |  5 +++++
 arch/x86/kvm/mmu.h              |  5 +++++
 arch/x86/kvm/mmu/mmu.c          |  7 +++++++
 arch/x86/kvm/mmu/mmu_internal.h | 31 +++++++++++++++++++++++++++----
 arch/x86/kvm/mmu/tdp_mmu.c      |  1 +
 5 files changed, 45 insertions(+), 4 deletions(-)
arch/x86/include/asm/kvm_host.h +5 -0
index aabf1648a56a..9e35fe32f500 100644
@@ -817,6 +817,11 @@ struct kvm_vcpu_arch {
 	struct kvm_mmu_memory_cache mmu_shadow_page_cache;
 	struct kvm_mmu_memory_cache mmu_shadowed_info_cache;
 	struct kvm_mmu_memory_cache mmu_page_header_cache;
+	/*
+	 * This cache is to allocate private page table. E.g. private EPT used
+	 * by the TDX module.
+	 */
+	struct kvm_mmu_memory_cache mmu_external_spt_cache;
 
 	/*
 	 * QEMU userspace and the guest each have their own FPU state.
arch/x86/kvm/mmu.h +5 -0
index dc80e72e4848..0c3bf89cf7db 100644
@@ -318,4 +318,9 @@ static inline gpa_t kvm_translate_gpa(struct kvm_vcpu *vcpu,
 		return gpa;
 	return translate_nested_gpa(vcpu, gpa, access, exception);
 }
+
+static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
+{
+	return kvm->arch.vm_type == KVM_X86_TDX_VM;
+}
 #endif
arch/x86/kvm/mmu/mmu.c +7 -0
index f41c498fcdb5..8023cebeefaa 100644
@@ -685,6 +685,12 @@ static int mmu_topup_memory_caches(struct kvm_vcpu *vcpu, bool maybe_indirect)
 				       1 + PT64_ROOT_MAX_LEVEL + PTE_PREFETCH_NUM);
 	if (r)
 		return r;
+	if (kvm_has_mirrored_tdp(vcpu->kvm)) {
+		r = kvm_mmu_topup_memory_cache(&vcpu->arch.mmu_external_spt_cache,
+					       PT64_ROOT_MAX_LEVEL);
+		if (r)
+			return r;
+	}
 	r = kvm_mmu_topup_memory_cache(&vcpu->arch.mmu_shadow_page_cache,
 				       PT64_ROOT_MAX_LEVEL);
 	if (r)
@@ -704,6 +710,7 @@ static void mmu_free_memory_caches(struct kvm_vcpu *vcpu)
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_pte_list_desc_cache);
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_shadow_page_cache);
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_shadowed_info_cache);
+	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_external_spt_cache);
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_page_header_cache);
 }
 
arch/x86/kvm/mmu/mmu_internal.h +27 -4
index 706f0ce8784c..d2837f796f34 100644
@@ -101,7 +101,22 @@ struct kvm_mmu_page {
 		int root_count;
 		refcount_t tdp_mmu_root_count;
 	};
-	unsigned int unsync_children;
+	union {
+		/* These two members aren't used for TDP MMU */
+		struct {
+			unsigned int unsync_children;
+			/*
+			 * Number of writes since the last time traversal
+			 * visited this page.
+			 */
+			atomic_t write_flooding_count;
+		};
+		/*
+		 * Page table page of private PT.
+		 * Passed to TDX module, not accessed by KVM.
+		 */
+		void *external_spt;
+	};
 	union {
 		struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
 		tdp_ptep_t ptep;
@@ -124,9 +139,6 @@ struct kvm_mmu_page {
 	int clear_spte_count;
 #endif
 
-	/* Number of writes since the last time traversal visited this page.  */
-	atomic_t write_flooding_count;
-
 #ifdef CONFIG_X86_64
 	/* Used for freeing the page asynchronously if it is a TDP MMU page. */
 	struct rcu_head rcu_head;
@@ -145,6 +157,17 @@ static inline int kvm_mmu_page_as_id(struct kvm_mmu_page *sp)
 	return kvm_mmu_role_as_id(sp->role);
 }
 
+static inline void kvm_mmu_alloc_external_spt(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
+{
+	/*
+	 * external_spt is allocated for TDX module to hold private EPT mappings,
+	 * TDX module will initialize the page by itself.
+	 * Therefore, KVM does not need to initialize or access external_spt.
+	 * KVM only interacts with sp->spt for external EPT operations.
+	 */
+	sp->external_spt = kvm_mmu_memory_cache_alloc(&vcpu->arch.mmu_external_spt_cache);
+}
+
 static inline bool kvm_mmu_page_ad_need_write_protect(struct kvm_mmu_page *sp)
 {
 	/*
arch/x86/kvm/mmu/tdp_mmu.c +1 -0
index 16b54208e8d7..35249555b585 100644
@@ -53,6 +53,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
 
 static void tdp_mmu_free_sp(struct kvm_mmu_page *sp)
 {
+	free_page((unsigned long)sp->external_spt);
 	free_page((unsigned long)sp->spt);
 	kmem_cache_free(mmu_page_header_cache, sp);
 }
Signature · 1 line
2.34.1
#5 lore raw

Re: [PATCH v3 04/17] KVM: x86/mmu: Add an external pointer to struct kvm_mmu_page

Quoted 17 lines · Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Add a external pointer to struct kvm_mmu_page for TDX's private page table and add helper functions to allocate/initialize/free a private page table page. TDX will only be supported with the TDP MMU. Because KVM TDP MMU doesn't use unsync_children and write_flooding_count, pack them to have room for a pointer and use a union to avoid memory overhead.

For private GPA, CPU refers to a private page table whose contents are encrypted. The dedicated APIs to operate on it (e.g. updating/reading its PTE entry) are used, and their cost is expensive.

When KVM resolves the KVM page fault, it walks the page tables. To reuse the existing KVM MMU code and mitigate the heavy cost of directly walking the private page table allocate two sets of page tables for the private half of the GPA space.

Quoted 5 lines

For the page tables that KVM will walk, allocate them like normal and refer to them as mirror page tables. Additionally allocate one more page for the page tables the CPU will walk, and call them external page tables. Resolve the KVM page fault with the existing code, and do additional operations necessary for modifying the external page table in future patches.

There should be only one page table for mirror page table and one page table for external page table.

What about the below change?

For the page table that KVM will walk, allocate it like normal and refer to it as mirror page table. Additionally allocate one more page table that the CPU will walk, and call it external page table. Resolve the KVM page fault with the existing code, and do additional operations necessary for modifying the external page table in future patches.

Quoted 13 lines

The relationship of the types of page tables in this scheme is depicted below:

              KVM page fault                     |
                     |                           |
                     V                           |
        -------------+----------                 |
        |                      |                 |
        V                      V                 |
     shared GPA           private GPA            |
        |                      |                 |
        V                      V                 |
    shared PT root      mirror PT root           |    private PT root
                                                            ^

Should we use "private PT" or "external PT" here?
"private PT", which is the term TDX uses, looks more appropriate, but a legend is lacked for it below, e.g.

Private PT - TDX's terminology of the external PT referred in KVM.

Or just use "private EPT"?
But there're already lots of co-existence of "private EPT", "private page table" in the code comments below. Not sure if it matters.

Quoted 62 lines
        |                      |                 |           |
        V                      V                 |           V
     shared PT           mirror PT        --propagate--> external PT
        |                      |                 |           |
        |                      \-----------------+------\    |
        |                                        |      |    |
        V                                        |      V    V
  shared guest page                              |    private guest page
                                                 |
                           non-encrypted memory  |    encrypted memory
                                                 |
PT          - Page table
Shared PT   - Visible to KVM, and the CPU uses it for shared mappings.

External PT - The CPU uses it, but it is invisible to KVM. TDX module
updates this table to map private guest pages.
Mirror PT - It is visible to KVM, but the CPU doesn't use it. KVM uses
it to propagate PT change to the actual private PT.

Add a helper kvm_has_mirrored_tdp() to trigger this behavior and wire it to the TDX vm type.

---
TDX MMU Prep v3:
- mirrored->external rename (Paolo)
- Remove accidentally included kvm_mmu_alloc_private_spt() (Paolo)
- Those -> These (Paolo)
- Change log updates to make external/mirrored naming more clear

TDX MMU Prep v2:
- Rename private->mirror
- Don't trigger off of shared mask

TDX MMU Prep:
- Rename terminology, dummy PT => mirror PT. and updated the commit message By Rick and Kai.
- Added a comment on union of private_spt by Rick.
- Don't handle the root case in kvm_mmu_alloc_private_spt(), it will not be needed in future patches. (Rick)
- Update comments (Yan)
- Remove kvm_mmu_init_private_spt(), open code it in later patches (Yan)

v19:
- typo in the comment in kvm_mmu_alloc_private_spt()
- drop CONFIG_KVM_MMU_PRIVATE

---
 arch/x86/include/asm/kvm_host.h |  5 +++++
 arch/x86/kvm/mmu.h              |  5 +++++
 arch/x86/kvm/mmu/mmu.c          |  7 +++++++
 arch/x86/kvm/mmu/mmu_internal.h | 31 +++++++++++++++++++++++++++----
 arch/x86/kvm/mmu/tdp_mmu.c      |  1 +
 5 files changed, 45 insertions(+), 4 deletions(-)
arch/x86/include/asm/kvm_host.h +0 -0
index aabf1648a56a..9e35fe32f500 100644

arch/x86/include/asm/kvm_host.h struct kvm_vcpu_arch { · line 821

Quoted 5 lines
arch/x86/include/asm/kvm_host.h +2 -0
@@ -817,6 +817,11 @@ struct kvm_vcpu_arch {
 	struct kvm_mmu_memory_cache mmu_shadow_page_cache;
 	struct kvm_mmu_memory_cache mmu_shadowed_info_cache;
 	struct kvm_mmu_memory_cache mmu_page_header_cache;
+	/*
+	 * This cache is to allocate private page table. E.g. private EPT used

s/private page table/external page table

Quoted 56 lines
arch/x86/include/asm/kvm_host.h +3 -0
+	 * by the TDX module.
+	 */
+	struct kvm_mmu_memory_cache mmu_external_spt_cache;
 
 	/*
 	 * QEMU userspace and the guest each have their own FPU state.
arch/x86/kvm/mmu.h +5 -0
index dc80e72e4848..0c3bf89cf7db 100644
@@ -318,4 +318,9 @@ static inline gpa_t kvm_translate_gpa(struct kvm_vcpu *vcpu,
 		return gpa;
 	return translate_nested_gpa(vcpu, gpa, access, exception);
 }
+
+static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
+{
+	return kvm->arch.vm_type == KVM_X86_TDX_VM;
+}
 #endif
arch/x86/kvm/mmu/mmu.c +7 -0
index f41c498fcdb5..8023cebeefaa 100644
@@ -685,6 +685,12 @@ static int mmu_topup_memory_caches(struct kvm_vcpu *vcpu, bool maybe_indirect)
 				       1 + PT64_ROOT_MAX_LEVEL + PTE_PREFETCH_NUM);
 	if (r)
 		return r;
+	if (kvm_has_mirrored_tdp(vcpu->kvm)) {
+		r = kvm_mmu_topup_memory_cache(&vcpu->arch.mmu_external_spt_cache,
+					       PT64_ROOT_MAX_LEVEL);
+		if (r)
+			return r;
+	}
 	r = kvm_mmu_topup_memory_cache(&vcpu->arch.mmu_shadow_page_cache,
 				       PT64_ROOT_MAX_LEVEL);
 	if (r)
@@ -704,6 +710,7 @@ static void mmu_free_memory_caches(struct kvm_vcpu *vcpu)
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_pte_list_desc_cache);
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_shadow_page_cache);
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_shadowed_info_cache);
+	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_external_spt_cache);
 	kvm_mmu_free_memory_cache(&vcpu->arch.mmu_page_header_cache);
 }
 
arch/x86/kvm/mmu/mmu_internal.h +2 -1
index 706f0ce8784c..d2837f796f34 100644
@@ -101,7 +101,22 @@ struct kvm_mmu_page {
 		int root_count;
 		refcount_t tdp_mmu_root_count;
 	};
-	unsigned int unsync_children;
+	union {
+		/* These two members aren't used for TDP MMU */

arch/x86/kvm/mmu/mmu_internal.h struct kvm_mmu_page { · line 115

Quoted 10 lines
arch/x86/kvm/mmu/mmu_internal.h +10 -0
+		struct {
+			unsigned int unsync_children;
+			/*
+			 * Number of writes since the last time traversal
+			 * visited this page.
+			 */
+			atomic_t write_flooding_count;
+		};
+		/*
+		 * Page table page of private PT.

s/private/external

Quoted 18 lines
arch/x86/kvm/mmu/mmu_internal.h +4 -3
+		 * Passed to TDX module, not accessed by KVM.
+		 */
+		void *external_spt;
+	};
 	union {
 		struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */
 		tdp_ptep_t ptep;
@@ -124,9 +139,6 @@ struct kvm_mmu_page {
 	int clear_spte_count;
 #endif
 
-	/* Number of writes since the last time traversal visited this page.  */
-	atomic_t write_flooding_count;
-
 #ifdef CONFIG_X86_64
 	/* Used for freeing the page asynchronously if it is a TDP MMU page. */
 	struct rcu_head rcu_head;

arch/x86/kvm/mmu/mmu_internal.h static inline int kvm_mmu_page_as_id(struct kvm_mmu_page *sp) · line 166

Quoted 10 lines
arch/x86/kvm/mmu/mmu_internal.h +7 -0
@@ -145,6 +157,17 @@ static inline int kvm_mmu_page_as_id(struct kvm_mmu_page *sp)
 	return kvm_mmu_role_as_id(sp->role);
 }
 
+static inline void kvm_mmu_alloc_external_spt(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
+{
+	/*
+	 * external_spt is allocated for TDX module to hold private EPT mappings,
+	 * TDX module will initialize the page by itself.
+	 * Therefore, KVM does not need to initialize or access external_spt.
+	 * KVM only interacts with sp->spt for external EPT operations.

s/external EPT/private EPT
or
s/external EPT/external page table

Quoted 22 lines
arch/x86/kvm/mmu/mmu_internal.h +4 -0
+	 */
+	sp->external_spt = kvm_mmu_memory_cache_alloc(&vcpu->arch.mmu_external_spt_cache);
+}
+
 static inline bool kvm_mmu_page_ad_need_write_protect(struct kvm_mmu_page *sp)
 {
 	/*
arch/x86/kvm/mmu/tdp_mmu.c +1 -0
index 16b54208e8d7..35249555b585 100644
@@ -53,6 +53,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
 
 static void tdp_mmu_free_sp(struct kvm_mmu_page *sp)
 {
+	free_page((unsigned long)sp->external_spt);
 	free_page((unsigned long)sp->spt);
 	kmem_cache_free(mmu_page_header_cache, sp);
 }
Signature · 1 line
2.34.1
#6 lore raw

PATCH v3 5/17 KVM: x86/mmu: Add an is_mirror member for union kvm_mmu_page_role

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Introduce a "is_mirror" member to the kvm_mmu_page_role union to identify SPTEs associated with the mirrored EPT.

The TDX module maintains the private half of the EPT mapped in the TD in its protected memory. KVM keeps a copy of the private GPAs in a mirrored EPT tree within host memory. This "is_mirror" attribute enables vCPUs to find and get the root page of mirrored EPT from the MMU root list for a guest TD. This also allows KVM MMU code to detect changes in mirrored EPT according to the "is_mirror" mmu page role and propagate the changes to the private EPT managed by TDX module.

---
TDX MMU Prep v3:
- Rename role mirror_pt -> is_mirror (Paolo)
- Remove unnessary helpers that just access a member (Paolo)

TDX MMU Prep v2:
- Rename private -> mirrored

TDX MMU Prep:
- Remove warning and NULL check in is_private_sptep() (Rick)
- Update commit log (Yan)

v19:
- Fix is_private_sptep() when NULL case.
- drop CONFIG_KVM_MMU_PRIVATE

---
 arch/x86/include/asm/kvm_host.h | 3 ++-
 arch/x86/kvm/mmu/mmu_internal.h | 5 +++++
 arch/x86/kvm/mmu/spte.h         | 5 +++++
 3 files changed, 12 insertions(+), 1 deletion(-)
arch/x86/include/asm/kvm_host.h +2 -1
index 9e35fe32f500..6c59b129f382 100644
@@ -351,7 +351,8 @@ union kvm_mmu_page_role {
 		unsigned ad_disabled:1;
 		unsigned guest_mode:1;
 		unsigned passthrough:1;
-		unsigned :5;
+		unsigned is_mirror:1;
+		unsigned :4;
 
 		/*
 		 * This is left at the top of the word so that
arch/x86/kvm/mmu/mmu_internal.h +5 -0
index d2837f796f34..5a2c9be23627 100644
@@ -157,6 +157,11 @@ static inline int kvm_mmu_page_as_id(struct kvm_mmu_page *sp)
 	return kvm_mmu_role_as_id(sp->role);
 }
 
+static inline bool is_mirror_sp(const struct kvm_mmu_page *sp)
+{
+	return sp->role.is_mirror;
+}
+
 static inline void kvm_mmu_alloc_external_spt(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
 {
 	/*
arch/x86/kvm/mmu/spte.h +5 -0
index 86e5259aa824..4883d139761b 100644
@@ -265,6 +265,11 @@ static inline struct kvm_mmu_page *root_to_sp(hpa_t root)
 	return spte_to_child_sp(root);
 }
 
+static inline bool is_mirror_sptep(u64 *sptep)
+{
+	return is_mirror_sp(sptep_to_sp(sptep));
+}
+
 static inline bool is_mmio_spte(struct kvm *kvm, u64 spte)
 {
 	return (spte & shadow_mmio_mask) == kvm->arch.shadow_mmio_value &&
Signature · 1 line
2.34.1
#7 lore raw

PATCH v3 6/17 KVM: x86/mmu: Make kvm_tdp_mmu_alloc_root() return void

In reply to Rick Edgecombe

The kvm_tdp_mmu_alloc_root() function currently always returns 0. This allows for the caller, mmu_alloc_direct_roots(), to call kvm_tdp_mmu_alloc_root() and also return 0 in one line:
return kvm_tdp_mmu_alloc_root(vcpu);

So it is useful even though the return value of kvm_tdp_mmu_alloc_root() is always the same. However, in future changes, kvm_tdp_mmu_alloc_root() will be called twice in mmu_alloc_direct_roots(). This will force the first call to either awkwardly handle the return value that will always be zero or ignore it. So change kvm_tdp_mmu_alloc_root() to return void. Do it in a separate change so the future change will be cleaner.

---
TDX MMU Prep v3:
- Add Paolo's reviewed-by

TDX MMU Prep:
- New patch

---
 arch/x86/kvm/mmu/mmu.c     | 6 ++++--
 arch/x86/kvm/mmu/tdp_mmu.c | 3 +--
 arch/x86/kvm/mmu/tdp_mmu.h | 2 +-
 3 files changed, 6 insertions(+), 5 deletions(-)
arch/x86/kvm/mmu/mmu.c +4 -2
index 8023cebeefaa..138e7bbcda1e 100644
@@ -3700,8 +3700,10 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
 	unsigned i;
 	int r;
 
-	if (tdp_mmu_enabled)
-		return kvm_tdp_mmu_alloc_root(vcpu);
+	if (tdp_mmu_enabled) {
+		kvm_tdp_mmu_alloc_root(vcpu);
+		return 0;
+	}
 
 	write_lock(&vcpu->kvm->mmu_lock);
 	r = make_mmu_pages_available(vcpu);
arch/x86/kvm/mmu/tdp_mmu.c +1 -2
index 35249555b585..07b0b884e246 100644
@@ -224,7 +224,7 @@ static void tdp_mmu_init_child_sp(struct kvm_mmu_page *child_sp,
 	tdp_mmu_init_sp(child_sp, iter->sptep, iter->gfn, role);
 }
 
-int kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
 {
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
 	union kvm_mmu_page_role role = mmu->root_role;
@@ -285,7 +285,6 @@ int kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
 	 */
 	mmu->root.hpa = __pa(root->spt);
 	mmu->root.pgd = 0;
-	return 0;
 }
 
 static void handle_changed_spte(struct kvm *kvm, int as_id, gfn_t gfn,
arch/x86/kvm/mmu/tdp_mmu.h +1 -1
index 58b55e61bd33..437ddd4937a9 100644
@@ -10,7 +10,7 @@
 void kvm_mmu_init_tdp_mmu(struct kvm *kvm);
 void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm);
 
-int kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu);
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu);
 
 __must_check static inline bool kvm_tdp_mmu_get_root(struct kvm_mmu_page *root)
 {
Signature · 1 line
2.34.1
#8 lore raw

PATCH v3 7/17 KVM: x86/tdp_mmu: Take struct kvm in iter loops

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Add a struct kvm argument to the TDP MMU iterators.

Future changes will want to change how the iterator behaves based on a member of struct kvm. Change the signature and callers of the iterator loop helpers in a separate patch to make the future one easier to review.

---
TDX MMU Prep v3:
- Split from "KVM: x86/mmu: Support GFN direct mask" (Paolo)

---
 arch/x86/kvm/mmu/tdp_iter.h |  6 +++---
 arch/x86/kvm/mmu/tdp_mmu.c  | 36 ++++++++++++++++++------------------
 2 files changed, 21 insertions(+), 21 deletions(-)
arch/x86/kvm/mmu/tdp_iter.h +3 -3
index fae559559a80..62c9ca32d922 100644
@@ -120,13 +120,13 @@ struct tdp_iter {
  * Iterates over every SPTE mapping the GFN range [start, end) in a
  * preorder traversal.
  */
-#define for_each_tdp_pte_min_level(iter, root, min_level, start, end) \
+#define for_each_tdp_pte_min_level(iter, kvm, root, min_level, start, end)		  \
 	for (tdp_iter_start(&iter, root, min_level, start); \
 	     iter.valid && iter.gfn < end;		     \
 	     tdp_iter_next(&iter))
 
-#define for_each_tdp_pte(iter, root, start, end) \
-	for_each_tdp_pte_min_level(iter, root, PG_LEVEL_4K, start, end)
+#define for_each_tdp_pte(iter, kvm, root, start, end)				\
+	for_each_tdp_pte_min_level(iter, kvm, root, PG_LEVEL_4K, start, end)
 
 tdp_ptep_t spte_to_child_pt(u64 pte, int level);
 
arch/x86/kvm/mmu/tdp_mmu.c +18 -18
index 07b0b884e246..4a7518c9ba7e 100644
@@ -674,18 +674,18 @@ static inline void tdp_mmu_iter_set_spte(struct kvm *kvm, struct tdp_iter *iter,
 					  iter->gfn, iter->level);
 }
 
-#define tdp_root_for_each_pte(_iter, _root, _start, _end) \
-	for_each_tdp_pte(_iter, _root, _start, _end)
+#define tdp_root_for_each_pte(_iter, _kvm, _root, _start, _end)	\
+	for_each_tdp_pte(_iter, _kvm, _root, _start, _end)
 
-#define tdp_root_for_each_leaf_pte(_iter, _root, _start, _end)	\
-	tdp_root_for_each_pte(_iter, _root, _start, _end)		\
+#define tdp_root_for_each_leaf_pte(_iter, _kvm, _root, _start, _end)	\
+	tdp_root_for_each_pte(_iter, _kvm, _root, _start, _end)		\
 		if (!is_shadow_present_pte(_iter.old_spte) ||		\
 		    !is_last_spte(_iter.old_spte, _iter.level))		\
 			continue;					\
 		else
 
-#define tdp_mmu_for_each_pte(_iter, _mmu, _start, _end)		\
-	for_each_tdp_pte(_iter, root_to_sp(_mmu->root.hpa), _start, _end)
+#define tdp_mmu_for_each_pte(_iter, _kvm, _mmu, _start, _end)		\
+	for_each_tdp_pte(_iter, _kvm, root_to_sp(_mmu->root.hpa), _start, _end)
 
 /*
  * Yield if the MMU lock is contended or this thread needs to return control
@@ -751,7 +751,7 @@ static void __tdp_mmu_zap_root(struct kvm *kvm, struct kvm_mmu_page *root,
 	gfn_t end = tdp_mmu_max_gfn_exclusive();
 	gfn_t start = 0;
 
-	for_each_tdp_pte_min_level(iter, root, zap_level, start, end) {
+	for_each_tdp_pte_min_level(iter, kvm, root, zap_level, start, end) {
 retry:
 		if (tdp_mmu_iter_cond_resched(kvm, &iter, false, shared))
 			continue;
@@ -855,7 +855,7 @@ static bool tdp_mmu_zap_leafs(struct kvm *kvm, struct kvm_mmu_page *root,
 
 	rcu_read_lock();
 
-	for_each_tdp_pte_min_level(iter, root, PG_LEVEL_4K, start, end) {
+	for_each_tdp_pte_min_level(iter, kvm, root, PG_LEVEL_4K, start, end) {
 		if (can_yield &&
 		    tdp_mmu_iter_cond_resched(kvm, &iter, flush, false)) {
 			flush = false;
@@ -1116,7 +1116,7 @@ int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 
 	rcu_read_lock();
 
-	tdp_mmu_for_each_pte(iter, mmu, fault->gfn, fault->gfn + 1) {
+	tdp_mmu_for_each_pte(iter, kvm, mmu, fault->gfn, fault->gfn + 1) {
 		int r;
 
 		if (fault->nx_huge_page_workaround_enabled)
@@ -1214,7 +1214,7 @@ static __always_inline bool kvm_tdp_mmu_handle_gfn(struct kvm *kvm,
 	for_each_tdp_mmu_root(kvm, root, range->slot->as_id) {
 		rcu_read_lock();
 
-		tdp_root_for_each_leaf_pte(iter, root, range->start, range->end)
+		tdp_root_for_each_leaf_pte(iter, kvm, root, range->start, range->end)
 			ret |= handler(kvm, &iter, range);
 
 		rcu_read_unlock();
@@ -1297,7 +1297,7 @@ static bool wrprot_gfn_range(struct kvm *kvm, struct kvm_mmu_page *root,
 
 	BUG_ON(min_level > KVM_MAX_HUGEPAGE_LEVEL);
 
-	for_each_tdp_pte_min_level(iter, root, min_level, start, end) {
+	for_each_tdp_pte_min_level(iter, kvm, root, min_level, start, end) {
 retry:
 		if (tdp_mmu_iter_cond_resched(kvm, &iter, false, true))
 			continue;
@@ -1460,7 +1460,7 @@ static int tdp_mmu_split_huge_pages_root(struct kvm *kvm,
 	 * level above the target level (e.g. splitting a 1GB to 512 2MB pages,
 	 * and then splitting each of those to 512 4KB pages).
 	 */
-	for_each_tdp_pte_min_level(iter, root, target_level + 1, start, end) {
+	for_each_tdp_pte_min_level(iter, kvm, root, target_level + 1, start, end) {
 retry:
 		if (tdp_mmu_iter_cond_resched(kvm, &iter, false, shared))
 			continue;
@@ -1545,7 +1545,7 @@ static bool clear_dirty_gfn_range(struct kvm *kvm, struct kvm_mmu_page *root,
 
 	rcu_read_lock();
 
-	tdp_root_for_each_pte(iter, root, start, end) {
+	tdp_root_for_each_pte(iter, kvm, root, start, end) {
 retry:
 		if (!is_shadow_present_pte(iter.old_spte) ||
 		    !is_last_spte(iter.old_spte, iter.level))
@@ -1600,7 +1600,7 @@ static void clear_dirty_pt_masked(struct kvm *kvm, struct kvm_mmu_page *root,
 
 	rcu_read_lock();
 
-	tdp_root_for_each_leaf_pte(iter, root, gfn + __ffs(mask),
+	tdp_root_for_each_leaf_pte(iter, kvm, root, gfn + __ffs(mask),
 				    gfn + BITS_PER_LONG) {
 		if (!mask)
 			break;
@@ -1657,7 +1657,7 @@ static void zap_collapsible_spte_range(struct kvm *kvm,
 
 	rcu_read_lock();
 
-	for_each_tdp_pte_min_level(iter, root, PG_LEVEL_2M, start, end) {
+	for_each_tdp_pte_min_level(iter, kvm, root, PG_LEVEL_2M, start, end) {
 retry:
 		if (tdp_mmu_iter_cond_resched(kvm, &iter, false, true))
 			continue;
@@ -1727,7 +1727,7 @@ static bool write_protect_gfn(struct kvm *kvm, struct kvm_mmu_page *root,
 
 	rcu_read_lock();
 
-	for_each_tdp_pte_min_level(iter, root, min_level, gfn, gfn + 1) {
+	for_each_tdp_pte_min_level(iter, kvm, root, min_level, gfn, gfn + 1) {
 		if (!is_shadow_present_pte(iter.old_spte) ||
 		    !is_last_spte(iter.old_spte, iter.level))
 			continue;
@@ -1782,7 +1782,7 @@ int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
 
 	*root_level = vcpu->arch.mmu->root_role.level;
 
-	tdp_mmu_for_each_pte(iter, mmu, gfn, gfn + 1) {
+	tdp_mmu_for_each_pte(iter, vcpu->kvm, mmu, gfn, gfn + 1) {
 		leaf = iter.level;
 		sptes[leaf] = iter.old_spte;
 	}
@@ -1809,7 +1809,7 @@ u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, u64 addr,
 	gfn_t gfn = addr >> PAGE_SHIFT;
 	tdp_ptep_t sptep = NULL;
 
-	tdp_mmu_for_each_pte(iter, mmu, gfn, gfn + 1) {
+	tdp_mmu_for_each_pte(iter, vcpu->kvm, mmu, gfn, gfn + 1) {
 		*spte = iter.old_spte;
 		sptep = iter.sptep;
 	}
Signature · 1 line
2.34.1
#9 lore raw

PATCH v3 8/17 KVM: x86/tdp_mmu: Take a GFN in kvm_tdp_mmu_fast_pf_get_last_sptep()

In reply to Rick Edgecombe

Pass fault->gfn into kvm_tdp_mmu_fast_pf_get_last_sptep(), instead of passing fault->addr and then converting it to a GFN.

Future changes will make fault->addr and fault->gfn differ when running TDX guests. The GFN will be conceptually the same as it is for normal VMs, but fault->addr may contain a TDX specific bit that differentiates between "shared" and "private" memory. This bit will be used to direct faults to be handled on different roots, either the normal "direct" root or a new type of root that handles private memory. The TDP iterators will process the traditional GFN concept and apply the required TDX specifics depending on the root type. For this reason, it needs to operate on regular GFN and not the addr, which may contain these special TDX specific bits.

Today kvm_tdp_mmu_fast_pf_get_last_sptep() takes fault->addr and then immediately converts it to a GFN with a bit shift. However, this would unfortunately retain the TDX specific bits in what is supposed to be a traditional GFN. Excluding TDX's needs, it is also is unnecessary to pass fault->addr and convert it to a GFN when the GFN is already on hand.

So instead just pass the GFN into kvm_tdp_mmu_fast_pf_get_last_sptep() and use it directly.

---
TDX MMU Prep v3:
- New patch

---
 arch/x86/kvm/mmu/mmu.c     | 2 +-
 arch/x86/kvm/mmu/tdp_mmu.c | 3 +--
 arch/x86/kvm/mmu/tdp_mmu.h | 2 +-
 3 files changed, 3 insertions(+), 4 deletions(-)
arch/x86/kvm/mmu/mmu.c +1 -1
index 138e7bbcda1e..e9c1783a8743 100644
@@ -3452,7 +3452,7 @@ static int fast_page_fault(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 		u64 new_spte;
 
 		if (tdp_mmu_enabled)
-			sptep = kvm_tdp_mmu_fast_pf_get_last_sptep(vcpu, fault->addr, &spte);
+			sptep = kvm_tdp_mmu_fast_pf_get_last_sptep(vcpu, fault->gfn, &spte);
 		else
 			sptep = fast_pf_get_last_sptep(vcpu, fault->addr, &spte);
 
arch/x86/kvm/mmu/tdp_mmu.c +1 -2
index 4a7518c9ba7e..067249dbbb5e 100644
@@ -1801,12 +1801,11 @@ int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
  *
  * WARNING: This function is only intended to be called during fast_page_fault.
  */
-u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, u64 addr,
+u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, gfn_t gfn,
 					u64 *spte)
 {
 	struct tdp_iter iter;
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
-	gfn_t gfn = addr >> PAGE_SHIFT;
 	tdp_ptep_t sptep = NULL;
 
 	tdp_mmu_for_each_pte(iter, vcpu->kvm, mmu, gfn, gfn + 1) {
arch/x86/kvm/mmu/tdp_mmu.h +1 -1
index 437ddd4937a9..1ba84487f3b7 100644
@@ -64,7 +64,7 @@ static inline void kvm_tdp_mmu_walk_lockless_end(void)
 
 int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
 			 int *root_level);
-u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, u64 addr,
+u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, gfn_t gfn,
 					u64 *spte);
 
 #ifdef CONFIG_X86_64
Signature · 1 line
2.34.1
#10 lore raw

PATCH v3 9/17 KVM: x86/mmu: Support GFN direct bits

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Teach the MMU to map guest GFNs at a massaged position on the TDP, to aid in implementing TDX shared memory.

Like other Coco technologies, TDX has the concept of private and shared memory. For TDX the private and shared mappings are managed on separate EPT roots. The private half is managed indirectly though calls into a protected runtime environment called the TDX module, where the shared half is managed within KVM in normal page tables.

For TDX, the shared half will be mapped in the higher alias, with a "shared bit" set in the GPA. However, KVM will still manage it with the same memslots as the private half. This means memslot looks ups and zapping operations will be provided with a GFN without the shared bit set.

So KVM will either need to apply or strip the shared bit before mapping or zapping the shared EPT. Having GFNs sometimes have the shared bit and sometimes not would make the code confusing.

So instead arrange the code such that GFNs never have shared bit set. Create a concept of "direct bits", that is stripped from the fault address when setting fault->gfn, and applied within the TDP MMU iterator. Calling code will behave as if is operating on the PTE mapping the GFN (without shared bits) but within the iterator, the actual mappings will be shifted using bits specific for the root. SPs will have the GFN set without the shared bit. In the end the TDP MMU will behave like it is mapping things at the GFN without the shared bit but with a strange page table format where everything is offset by the shared bit.

Since TDX only needs to shift the mapping like this for the shared bit, which is mapped as the normal TDP root, add a "gfn_direct_bits" field to the kvm_arch structure for each VM with a default value of 0. It will have the bit set at the position of the GPA shared bit in GFN through TD specific initialization code. Keep TDX specific concepts out of the MMU code by not naming it "shared".

Ranged TLB flushes (i.e. flush_remote_tlbs_range()) target specific GFN ranges. In convention established above, these would need to target the shifted GFN range. It won't matter functionally, since the actual implementation will always result in a full flush for the only planned user (TDX). For correctness reasons, future changes can provide a TDX x86_ops.flush_remote_tlbs_range implementation to return -EOPNOTSUPP and force the full flush for TDs.

This leaves one drawback. Some operations use a concept of max gfn (i.e. kvm_mmu_max_gfn()), to iterate over the whole TDP range. These would then exceed the range actually covered by each root. It should only result in a bit of extra iterating, and not cause functional problems. This will be addressed in a future change.

---
TDX MMU Prep v3:
- Add comment for kvm_gfn_root_mask() (Paolo)
- Change names mask -> bits (Paolo)
- Add comment in struct definition for fault->gfn not containing shared bit. (Paolo)
- Drop special handling in kvm_arch_flush_remote_tlbs_range(), implement kvm_x86_ops.flush_remote_tlbs_range in a future patch. (Paolo)
- Do addition of kvm arg to iterator in previous patch (Paolo)
- OR gfn_bits in try_step_side() too, because of issue seen with 4 level EPT
- Add warning for GFN bits in wrong arg in tdp_iter_start()

TDX MMU Prep v2:
- Rename from "KVM: x86/mmu: Add address conversion functions for TDX shared bit of GPA"
- Dropped Binbin's reviewed-by tag because of the extend of the changes
- Rename gfn_shared_mask to gfn_direct_mask.
- Don't include shared bits in GFNs, hide the existence in the TDP MMU iterator.
- Don't do range flushes if a gfn_direct_mask is present.

---
 arch/x86/include/asm/kvm_host.h |  2 ++
 arch/x86/kvm/mmu.h              |  5 +++++
 arch/x86/kvm/mmu/mmu_internal.h | 28 ++++++++++++++++++++++++++--
 arch/x86/kvm/mmu/tdp_iter.c     | 10 ++++++----
 arch/x86/kvm/mmu/tdp_iter.h     | 10 ++++++----
 5 files changed, 45 insertions(+), 10 deletions(-)
arch/x86/include/asm/kvm_host.h +2 -0
index 6c59b129f382..6e07eff06a58 100644
@@ -1525,6 +1525,8 @@ struct kvm_arch {
 	 */
 #define SPLIT_DESC_CACHE_MIN_NR_OBJECTS (SPTE_ENT_PER_PAGE + 1)
 	struct kvm_mmu_memory_cache split_desc_cache;
+
+	gfn_t gfn_direct_bits;
 };
 
 struct kvm_vm_stat {
arch/x86/kvm/mmu.h +5 -0
index 0c3bf89cf7db..63179a4fba7b 100644
@@ -323,4 +323,9 @@ static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
 {
 	return kvm->arch.vm_type == KVM_X86_TDX_VM;
 }
+
+static inline gfn_t kvm_gfn_direct_bits(const struct kvm *kvm)
+{
+	return kvm->arch.gfn_direct_bits;
+}
 #endif
arch/x86/kvm/mmu/mmu_internal.h +26 -2
index 5a2c9be23627..a19a4a05566a 100644
@@ -6,6 +6,8 @@
 #include <linux/kvm_host.h>
 #include <asm/kvm_host.h>
 
+#include "mmu.h"
+
 #ifdef CONFIG_KVM_PROVE_MMU
 #define KVM_MMU_WARN_ON(x) WARN_ON_ONCE(x)
 #else
@@ -173,6 +175,18 @@ static inline void kvm_mmu_alloc_external_spt(struct kvm_vcpu *vcpu, struct kvm_
 	sp->external_spt = kvm_mmu_memory_cache_alloc(&vcpu->arch.mmu_external_spt_cache);
 }
 
+static inline gfn_t kvm_gfn_root_bits(const struct kvm *kvm, const struct kvm_mmu_page *root)
+{
+	/*
+	 * Since mirror SPs are used only for TDX, which maps private memory
+	 * at its "natural" GFN, no mask needs to be applied to them - and, dually,
+	 * we expect that the bits is only used for the shared PT.
+	 */
+	if (is_mirror_sp(root))
+		return 0;
+	return kvm_gfn_direct_bits(kvm);
+}
+
 static inline bool kvm_mmu_page_ad_need_write_protect(struct kvm_mmu_page *sp)
 {
 	/*
@@ -257,7 +271,12 @@ struct kvm_page_fault {
 	 */
 	u8 goal_level;
 
-	/* Shifted addr, or result of guest page table walk if addr is a gva.  */
+	/*
+	 * Shifted addr, or result of guest page table walk if addr is a gva. In
+	 * the case of VM where memslot's can be mapped at multiple GPA aliases
+	 * (i.e. TDX), the gfn field does not contain the bit that selects between
+	 * the aliases (i.e. the shared bit for TDX).
+	 */
 	gfn_t gfn;
 
 	/* The memslot containing gfn. May be NULL. */
@@ -343,7 +362,12 @@ static inline int __kvm_mmu_do_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gp
 	int r;
 
 	if (vcpu->arch.mmu->root_role.direct) {
-		fault.gfn = fault.addr >> PAGE_SHIFT;
+		/*
+		 * Things like memslots don't understand the concept of a shared
+		 * bit. Strip it so that the GFN can be used like normal, and the
+		 * fault.addr can be used when the shared bit is needed.
+		 */
+		fault.gfn = gpa_to_gfn(fault.addr) & ~kvm_gfn_direct_bits(vcpu->kvm);
 		fault.slot = kvm_vcpu_gfn_to_memslot(vcpu, fault.gfn);
 	}
 
arch/x86/kvm/mmu/tdp_iter.c +6 -4
index 04c247bfe318..9e17bfa80901 100644
@@ -12,7 +12,7 @@
 static void tdp_iter_refresh_sptep(struct tdp_iter *iter)
 {
 	iter->sptep = iter->pt_path[iter->level - 1] +
-		SPTE_INDEX(iter->gfn << PAGE_SHIFT, iter->level);
+		SPTE_INDEX((iter->gfn | iter->gfn_bits) << PAGE_SHIFT, iter->level);
 	iter->old_spte = kvm_tdp_mmu_read_spte(iter->sptep);
 }
 
@@ -37,15 +37,17 @@ void tdp_iter_restart(struct tdp_iter *iter)
  * rooted at root_pt, starting with the walk to translate next_last_level_gfn.
  */
 void tdp_iter_start(struct tdp_iter *iter, struct kvm_mmu_page *root,
-		    int min_level, gfn_t next_last_level_gfn)
+		    int min_level, gfn_t next_last_level_gfn, gfn_t gfn_bits)
 {
 	if (WARN_ON_ONCE(!root || (root->role.level < 1) ||
-			 (root->role.level > PT64_ROOT_MAX_LEVEL))) {
+			 (root->role.level > PT64_ROOT_MAX_LEVEL) ||
+			 (gfn_bits && next_last_level_gfn >= gfn_bits))) {
 		iter->valid = false;
 		return;
 	}
 
 	iter->next_last_level_gfn = next_last_level_gfn;
+	iter->gfn_bits = gfn_bits;
 	iter->root_level = root->role.level;
 	iter->min_level = min_level;
 	iter->pt_path[iter->root_level - 1] = (tdp_ptep_t)root->spt;
@@ -113,7 +115,7 @@ static bool try_step_side(struct tdp_iter *iter)
 	 * Check if the iterator is already at the end of the current page
 	 * table.
 	 */
-	if (SPTE_INDEX(iter->gfn << PAGE_SHIFT, iter->level) ==
+	if (SPTE_INDEX((iter->gfn | iter->gfn_bits) << PAGE_SHIFT, iter->level) ==
 	    (SPTE_ENT_PER_PAGE - 1))
 		return false;
 
arch/x86/kvm/mmu/tdp_iter.h +6 -4
index 62c9ca32d922..e9b4dff7c129 100644
@@ -91,8 +91,10 @@ struct tdp_iter {
 	tdp_ptep_t pt_path[PT64_ROOT_MAX_LEVEL];
 	/* A pointer to the current SPTE */
 	tdp_ptep_t sptep;
-	/* The lowest GFN mapped by the current SPTE */
+	/* The lowest GFN (mask bits excluded) mapped by the current SPTE */
 	gfn_t gfn;
+	/* Mask applied to convert the GFN to the mapping GPA */
+	gfn_t gfn_bits;
 	/* The level of the root page given to the iterator */
 	int root_level;
 	/* The lowest level the iterator should traverse to */
@@ -121,8 +123,8 @@ struct tdp_iter {
  * preorder traversal.
  */
 #define for_each_tdp_pte_min_level(iter, kvm, root, min_level, start, end)		  \
-	for (tdp_iter_start(&iter, root, min_level, start); \
-	     iter.valid && iter.gfn < end;		     \
+	for (tdp_iter_start(&iter, root, min_level, start, kvm_gfn_root_bits(kvm, root)); \
+	     iter.valid && iter.gfn < end;						  \
 	     tdp_iter_next(&iter))
 
 #define for_each_tdp_pte(iter, kvm, root, start, end)				\
@@ -131,7 +133,7 @@ struct tdp_iter {
 tdp_ptep_t spte_to_child_pt(u64 pte, int level);
 
 void tdp_iter_start(struct tdp_iter *iter, struct kvm_mmu_page *root,
-		    int min_level, gfn_t next_last_level_gfn);
+		    int min_level, gfn_t next_last_level_gfn, gfn_t gfn_bits);
 void tdp_iter_next(struct tdp_iter *iter);
 void tdp_iter_restart(struct tdp_iter *iter);
 
Signature · 1 line
2.34.1
#11 lore raw

PATCH v3 10/17 KVM: x86/tdp_mmu: Extract root invalid check from tdx_mmu_next_root()

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Extract tdp_mmu_root_match() to check if the root has given types and use it for the root page table iterator. It checks only_invalid now.

TDX KVM operates on a shared page table only (Shared-EPT), a mirrored page table only (Secure-EPT), or both based on the operation. KVM MMU notifier operations only on shared page table. KVM guest_memfd invalidation operations only on mirrored page table, and so on. Introduce a centralized matching function instead of open coding matching logic in the iterator. The next step is to extend the function to check whether the page is shared or private

---
TDX MMU Prep:
- New patch

---
 arch/x86/kvm/mmu/tdp_mmu.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
arch/x86/kvm/mmu/tdp_mmu.c +10 -2
index 067249dbbb5e..cecc25947001 100644
@@ -92,6 +92,14 @@ void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root)
 	call_rcu(&root->rcu_head, tdp_mmu_free_sp_rcu_callback);
 }
 
+static bool tdp_mmu_root_match(struct kvm_mmu_page *root, bool only_valid)
+{
+	if (only_valid && root->role.invalid)
+		return false;
+
+	return true;
+}
+
 /*
  * Returns the next root after @prev_root (or the first root if @prev_root is
  * NULL).  A reference to the returned root is acquired, and the reference to
@@ -125,7 +133,7 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
 						   typeof(*next_root), link);
 
 	while (next_root) {
-		if ((!only_valid || !next_root->role.invalid) &&
+		if (tdp_mmu_root_match(next_root, only_valid) &&
 		    kvm_tdp_mmu_get_root(next_root))
 			break;
 
@@ -176,7 +184,7 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
 	list_for_each_entry(_root, &_kvm->arch.tdp_mmu_roots, link)		\
 		if (kvm_lockdep_assert_mmu_lock_held(_kvm, false) &&		\
 		    ((_as_id >= 0 && kvm_mmu_page_as_id(_root) != _as_id) ||	\
-		     ((_only_valid) && (_root)->role.invalid))) {		\
+		     !tdp_mmu_root_match((_root), (_only_valid)))) {		\
 		} else
 
 #define for_each_tdp_mmu_root(_kvm, _root, _as_id)			\
Signature · 1 line
2.34.1
#12 lore raw

PATCH v3 11/17 KVM: x86/tdp_mmu: Introduce KVM MMU root types to specify page table type

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Define an enum kvm_tdp_mmu_root_types to specify the KVM MMU root type [1] so that the iterator on the root page table can consistently filter the root page table type instead of only_valid.

TDX KVM will operate on KVM page tables with specified types. Shared page table, private page table, or both. Introduce an enum instead of bool only_valid so that we can easily enhance page table types applicable to shared, private, or both in addition to valid or not. Replace only_valid=false with KVM_ANY_ROOTS and only_valid=true with KVM_ANY_VALID_ROOTS. Use KVM_ANY_ROOTS and KVM_ANY_VALID_ROOTS to wrap KVM_VALID_ROOTS to avoid further code churn when direct vs mirror root concepts are introduced in future patches.

---
TDX MMU Prep v3:
- Drop KVM_ANY_ROOTS, KVM_ANY_VALID_ROOTS and switch to KVM_VALID_ROOTS and KVM_ALL_ROOTS. (Paolo)

TDX MMU Prep:
- Newly introduced.

---
 arch/x86/kvm/mmu/tdp_mmu.c | 41 +++++++++++++++++++-------------------
 arch/x86/kvm/mmu/tdp_mmu.h |  7 +++++++
 2 files changed, 28 insertions(+), 20 deletions(-)
arch/x86/kvm/mmu/tdp_mmu.c +21 -20
index cecc25947001..c8e5e779967e 100644
@@ -92,27 +92,28 @@ void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root)
 	call_rcu(&root->rcu_head, tdp_mmu_free_sp_rcu_callback);
 }
 
-static bool tdp_mmu_root_match(struct kvm_mmu_page *root, bool only_valid)
+static bool tdp_mmu_root_match(struct kvm_mmu_page *root,
+			       enum kvm_tdp_mmu_root_types types)
 {
-	if (only_valid && root->role.invalid)
-		return false;
+	if (root->role.invalid)
+		return types & KVM_INVALID_ROOTS;
 
 	return true;
 }
 
 /*
  * Returns the next root after @prev_root (or the first root if @prev_root is
- * NULL).  A reference to the returned root is acquired, and the reference to
- * @prev_root is released (the caller obviously must hold a reference to
- * @prev_root if it's non-NULL).
+ * NULL) that matches with @types.  A reference to the returned root is
+ * acquired, and the reference to @prev_root is released (the caller obviously
+ * must hold a reference to @prev_root if it's non-NULL).
  *
- * If @only_valid is true, invalid roots are skipped.
+ * Roots that doesn't match with @types are skipped.
  *
  * Returns NULL if the end of tdp_mmu_roots was reached.
  */
 static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
 					      struct kvm_mmu_page *prev_root,
-					      bool only_valid)
+					      enum kvm_tdp_mmu_root_types types)
 {
 	struct kvm_mmu_page *next_root;
 
@@ -133,7 +134,7 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
 						   typeof(*next_root), link);
 
 	while (next_root) {
-		if (tdp_mmu_root_match(next_root, only_valid) &&
+		if (tdp_mmu_root_match(next_root, types) &&
 		    kvm_tdp_mmu_get_root(next_root))
 			break;
 
@@ -158,20 +159,20 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
  * If shared is set, this function is operating under the MMU lock in read
  * mode.
  */
-#define __for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _only_valid)	\
-	for (_root = tdp_mmu_next_root(_kvm, NULL, _only_valid);		\
+#define __for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _types)	\
+	for (_root = tdp_mmu_next_root(_kvm, NULL, _types);		\
 	     ({ lockdep_assert_held(&(_kvm)->mmu_lock); }), _root;		\
-	     _root = tdp_mmu_next_root(_kvm, _root, _only_valid))		\
+	     _root = tdp_mmu_next_root(_kvm, _root, _types))		\
 		if (_as_id >= 0 && kvm_mmu_page_as_id(_root) != _as_id) {	\
 		} else
 
 #define for_each_valid_tdp_mmu_root_yield_safe(_kvm, _root, _as_id)	\
-	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, true)
+	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, KVM_VALID_ROOTS)
 
 #define for_each_tdp_mmu_root_yield_safe(_kvm, _root)			\
-	for (_root = tdp_mmu_next_root(_kvm, NULL, false);		\
+	for (_root = tdp_mmu_next_root(_kvm, NULL, KVM_ALL_ROOTS);		\
 	     ({ lockdep_assert_held(&(_kvm)->mmu_lock); }), _root;	\
-	     _root = tdp_mmu_next_root(_kvm, _root, false))
+	     _root = tdp_mmu_next_root(_kvm, _root, KVM_ALL_ROOTS))
 
 /*
  * Iterate over all TDP MMU roots.  Requires that mmu_lock be held for write,
@@ -180,18 +181,18 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
  * Holding mmu_lock for write obviates the need for RCU protection as the list
  * is guaranteed to be stable.
  */
-#define __for_each_tdp_mmu_root(_kvm, _root, _as_id, _only_valid)		\
+#define __for_each_tdp_mmu_root(_kvm, _root, _as_id, _types)			\
 	list_for_each_entry(_root, &_kvm->arch.tdp_mmu_roots, link)		\
 		if (kvm_lockdep_assert_mmu_lock_held(_kvm, false) &&		\
 		    ((_as_id >= 0 && kvm_mmu_page_as_id(_root) != _as_id) ||	\
-		     !tdp_mmu_root_match((_root), (_only_valid)))) {		\
+		     !tdp_mmu_root_match((_root), (_types)))) {			\
 		} else
 
 #define for_each_tdp_mmu_root(_kvm, _root, _as_id)			\
-	__for_each_tdp_mmu_root(_kvm, _root, _as_id, false)
+	__for_each_tdp_mmu_root(_kvm, _root, _as_id, KVM_ALL_ROOTS)
 
 #define for_each_valid_tdp_mmu_root(_kvm, _root, _as_id)		\
-	__for_each_tdp_mmu_root(_kvm, _root, _as_id, true)
+	__for_each_tdp_mmu_root(_kvm, _root, _as_id, KVM_VALID_ROOTS)
 
 static struct kvm_mmu_page *tdp_mmu_alloc_sp(struct kvm_vcpu *vcpu)
 {
@@ -1197,7 +1198,7 @@ bool kvm_tdp_mmu_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range,
 {
 	struct kvm_mmu_page *root;
 
-	__for_each_tdp_mmu_root_yield_safe(kvm, root, range->slot->as_id, false)
+	__for_each_tdp_mmu_root_yield_safe(kvm, root, range->slot->as_id, KVM_ALL_ROOTS)
 		flush = tdp_mmu_zap_leafs(kvm, root, range->start, range->end,
 					  range->may_block, flush);
 
arch/x86/kvm/mmu/tdp_mmu.h +7 -0
index 1ba84487f3b7..b887c225ff24 100644
@@ -19,6 +19,13 @@ __must_check static inline bool kvm_tdp_mmu_get_root(struct kvm_mmu_page *root)
 
 void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root);
 
+enum kvm_tdp_mmu_root_types {
+	KVM_INVALID_ROOTS = BIT(0),
+
+	KVM_VALID_ROOTS = BIT(1),
+	KVM_ALL_ROOTS = KVM_VALID_ROOTS | KVM_INVALID_ROOTS,
+};
+
 bool kvm_tdp_mmu_zap_leafs(struct kvm *kvm, gfn_t start, gfn_t end, bool flush);
 bool kvm_tdp_mmu_zap_sp(struct kvm *kvm, struct kvm_mmu_page *sp);
 void kvm_tdp_mmu_zap_all(struct kvm *kvm);
Signature · 1 line
2.34.1
#13 lore raw

PATCH v3 12/17 KVM: x86/tdp_mmu: Take root in tdp_mmu_for_each_pte()

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Take the root as an argument of tdp_mmu_for_each_pte() instead of looking it up in the mmu. With no other purpose of passing the mmu, drop it.

Future changes will want to change which root is used based on the context of the MMU operation. So change the callers to pass in the root currently used, mmu->root.hpa in a preparatory patch to make the later one smaller and easier to review.

---
TDX MMU Prep v3:
- Split from "KVM: x86/mmu: Support GFN direct mask" (Paolo)

---
 arch/x86/kvm/mmu/tdp_mmu.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
arch/x86/kvm/mmu/tdp_mmu.c +8 -8
index c8e5e779967e..2200bdc7681f 100644
@@ -693,8 +693,8 @@ static inline void tdp_mmu_iter_set_spte(struct kvm *kvm, struct tdp_iter *iter,
 			continue;					\
 		else
 
-#define tdp_mmu_for_each_pte(_iter, _kvm, _mmu, _start, _end)		\
-	for_each_tdp_pte(_iter, _kvm, root_to_sp(_mmu->root.hpa), _start, _end)
+#define tdp_mmu_for_each_pte(_iter, _kvm, _root, _start, _end)	\
+	for_each_tdp_pte(_iter, _kvm, _root, _start, _end)
 
 /*
  * Yield if the MMU lock is contended or this thread needs to return control
@@ -1113,8 +1113,8 @@ static int tdp_mmu_split_huge_page(struct kvm *kvm, struct tdp_iter *iter,
  */
 int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 {
-	struct kvm_mmu *mmu = vcpu->arch.mmu;
 	struct kvm *kvm = vcpu->kvm;
+	struct kvm_mmu_page *root = root_to_sp(vcpu->arch.mmu->root.hpa);
 	struct tdp_iter iter;
 	struct kvm_mmu_page *sp;
 	int ret = RET_PF_RETRY;
@@ -1125,7 +1125,7 @@ int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 
 	rcu_read_lock();
 
-	tdp_mmu_for_each_pte(iter, kvm, mmu, fault->gfn, fault->gfn + 1) {
+	tdp_mmu_for_each_pte(iter, kvm, root, fault->gfn, fault->gfn + 1) {
 		int r;
 
 		if (fault->nx_huge_page_workaround_enabled)
@@ -1784,14 +1784,14 @@ bool kvm_tdp_mmu_write_protect_gfn(struct kvm *kvm,
 int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
 			 int *root_level)
 {
+	struct kvm_mmu_page *root = root_to_sp(vcpu->arch.mmu->root.hpa);
 	struct tdp_iter iter;
-	struct kvm_mmu *mmu = vcpu->arch.mmu;
 	gfn_t gfn = addr >> PAGE_SHIFT;
 	int leaf = -1;
 
 	*root_level = vcpu->arch.mmu->root_role.level;
 
-	tdp_mmu_for_each_pte(iter, vcpu->kvm, mmu, gfn, gfn + 1) {
+	tdp_mmu_for_each_pte(iter, vcpu->kvm, root, gfn, gfn + 1) {
 		leaf = iter.level;
 		sptes[leaf] = iter.old_spte;
 	}
@@ -1813,11 +1813,11 @@ int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
 u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, gfn_t gfn,
 					u64 *spte)
 {
+	struct kvm_mmu_page *root = root_to_sp(vcpu->arch.mmu->root.hpa);
 	struct tdp_iter iter;
-	struct kvm_mmu *mmu = vcpu->arch.mmu;
 	tdp_ptep_t sptep = NULL;
 
-	tdp_mmu_for_each_pte(iter, vcpu->kvm, mmu, gfn, gfn + 1) {
+	tdp_mmu_for_each_pte(iter, vcpu->kvm, root, gfn, gfn + 1) {
 		*spte = iter.old_spte;
 		sptep = iter.sptep;
 	}
Signature · 1 line
2.34.1
#14 lore raw

PATCH v3 13/17 KVM: x86/tdp_mmu: Support mirror root for TDP MMU

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Add the ability for the TDP MMU to maintain a mirror of a separate mapping.

Like other Coco technologies, TDX has the concept of private and shared memory. For TDX the private and shared mappings are managed on separate EPT roots. The private half is managed indirectly though calls into a protected runtime environment called the TDX module, where the shared half is managed within KVM in normal page tables.

In order to handle both shared and private memory, KVM needs to learn to handle faults and other operations on the correct root for the operation. KVM could learn the concept of private roots, and operate on them by calling out to operations that call into the TDX module. But there are two problems with that:
1. Calls into the TDX module are relatively slow compared to the simple accesses required to read a PTE managed directly by KVM.
2. Other Coco technologies deal with private memory completely differently and it will make the code confusing when being read from their perspective. Special operations added for TDX that set private or zap private memory will have nothing to do with these other private memory technologies. (SEV, etc).

To handle these, instead teach the TDP MMU about a new concept "mirror roots". Such roots maintain page tables that are not actually mapped, and are just used to traverse quickly to determine if the mid level page tables need to be installed. When the memory be mirrored needs to actually be changed, calls can be made to via x86_ops.

  private KVM page fault   |
      |                    |
      V                    |
 private GPA               |     CPU protected EPTP
      |                    |           |
      V                    |           V
 mirror PT root            |     private PT root
      |                    |           |
      V                    |           V
   mirror PT   --hook to propagate-->private PT
      |                    |           |
      \--------------------+------\    |
                           |      |    |
                           |      V    V
                           |    private guest page
                           |
                           |
     non-encrypted memory  |    encrypted memory
                           |

Leave calling out to actually update the private page tables that are being mirrored for later changes. Just implement the handling of MMU operations on to mirrored roots.

In order to direct operations to correct root, add root types KVM_DIRECT_ROOTS and KVM_MIRROR_ROOTS. Tie the usage of mirrored/direct roots to private/shared with conditionals. It could also be implemented by making the kvm_tdp_mmu_root_types and kvm_gfn_range_filter enum bits line up such that conversion could be a direct assignment with a case. Don't do this because the mapping of private to mirrored is confusing enough. So it is worth not hiding the logic in type casting.

Cleanup the mirror root in kvm_mmu_destroy() instead of the normal place in kvm_mmu_free_roots(), because the private root that is being cannot be rebuilt like a normal root. It needs to persist for the lifetime of the VM.

The TDX module will also need to be provided with page tables to use for the actual mapping being mirrored by the mirrored page tables. Allocate these in the mapping path using the recently added kvm_mmu_alloc_private_spt().

Don't support 2M page for now. This is avoided by forcing 4k pages in the fault. Add a KVM_BUG_ON() to verify.

---
TDX MMU Prep v3:
- Remove handle_changed_spte() changes
- Drop kvm_on_mirror()/kvm_on_direct(), open code (Paolo)
- Rename tdp_mmu_get_fault_root_type() -> tdp_mmu_get_root_for_fault() (Paolo)
- Use fault->addr (in helper) to determine direct vs mirror root (Paolo)
- Rename process -> filter (Paolo)

TDX MMU Prep v2:
- Rename private->mirror
- Split apart from "KVM: x86/tdp_mmu: Support TDX private mapping for TDP MMU"
- Update log
- Sprinkle a few comments
- Use kvm_on_*() helpers to direct iterator to proper root
- Drop BUGGY_KVM_ROOTS because the translation between the process enum is no longer automatic, and the warn already happens elsewhere.

TDX MMU Prep:
- Remove unnecessary gfn, access twist in tdp_mmu_map_handle_target_level(). (Chao Gao)
- Open code call to kvm_mmu_alloc_private_spt() instead oCf doing it in tdp_mmu_alloc_sp()
- Update comment in set_private_spte_present() (Yan)
- Open code call to kvm_mmu_init_private_spt() (Yan)
- Add comments on TDX MMU hooks (Yan)
- Fix various whitespace alignment (Yan)
- Remove pointless warnings and conditionals in handle_removed_private_spte() (Yan)
- Remove redundant lockdep assert in tdp_mmu_set_spte() (Yan)
- Remove incorrect comment in handle_changed_spte() (Yan)
- Remove unneeded kvm_pfn_to_refcounted_page() and is_error_noslot_pfn() check in kvm_tdp_mmu_map() (Yan)
- Do kvm_gfn_for_root() branchless (Rick)
- Update kvm_tdp_mmu_alloc_root() callers to not check error code (Rick)
- Add comment for stripping shared bit for fault.gfn (Chao)

v19:
- drop CONFIG_KVM_MMU_PRIVATE

---
 arch/x86/include/asm/kvm_host.h |  1 +
 arch/x86/kvm/mmu.h              |  7 ++++++
 arch/x86/kvm/mmu/mmu.c          | 11 ++++++++-
 arch/x86/kvm/mmu/tdp_mmu.c      | 34 ++++++++++++++++++++------
 arch/x86/kvm/mmu/tdp_mmu.h      | 43 ++++++++++++++++++++++++++++++---
 5 files changed, 84 insertions(+), 12 deletions(-)
arch/x86/include/asm/kvm_host.h +1 -0
index 6e07eff06a58..d67e88a69fc4 100644
@@ -460,6 +460,7 @@ struct kvm_mmu {
 	int (*sync_spte)(struct kvm_vcpu *vcpu,
 			 struct kvm_mmu_page *sp, int i);
 	struct kvm_mmu_root_info root;
+	hpa_t mirror_root_hpa;
 	union kvm_cpu_role cpu_role;
 	union kvm_mmu_page_role root_role;
 
arch/x86/kvm/mmu.h +7 -0
index 63179a4fba7b..7b12ba761c51 100644
@@ -328,4 +328,11 @@ static inline gfn_t kvm_gfn_direct_bits(const struct kvm *kvm)
 {
 	return kvm->arch.gfn_direct_bits;
 }
+
+static inline bool kvm_is_addr_direct(struct kvm *kvm, gpa_t gpa)
+{
+	gpa_t gpa_direct_bits = gfn_to_gpa(kvm_gfn_direct_bits(kvm));
+
+	return !gpa_direct_bits || (gpa & gpa_direct_bits);
+}
 #endif
arch/x86/kvm/mmu/mmu.c +10 -1
index e9c1783a8743..287dcc2685e4 100644
@@ -3701,7 +3701,9 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
 	int r;
 
 	if (tdp_mmu_enabled) {
-		kvm_tdp_mmu_alloc_root(vcpu);
+		if (kvm_has_mirrored_tdp(vcpu->kvm))
+			kvm_tdp_mmu_alloc_root(vcpu, true);
+		kvm_tdp_mmu_alloc_root(vcpu, false);
 		return 0;
 	}
 
@@ -6245,6 +6247,7 @@ static int __kvm_mmu_create(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu)
 
 	mmu->root.hpa = INVALID_PAGE;
 	mmu->root.pgd = 0;
+	mmu->mirror_root_hpa = INVALID_PAGE;
 	for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
 		mmu->prev_roots[i] = KVM_MMU_ROOT_INFO_INVALID;
 
@@ -7220,6 +7223,12 @@ int kvm_mmu_vendor_module_init(void)
 void kvm_mmu_destroy(struct kvm_vcpu *vcpu)
 {
 	kvm_mmu_unload(vcpu);
+	if (tdp_mmu_enabled) {
+		read_lock(&vcpu->kvm->mmu_lock);
+		mmu_free_root_page(vcpu->kvm, &vcpu->arch.mmu->mirror_root_hpa,
+				   NULL);
+		read_unlock(&vcpu->kvm->mmu_lock);
+	}
 	free_mmu_pages(&vcpu->arch.root_mmu);
 	free_mmu_pages(&vcpu->arch.guest_mmu);
 	mmu_free_memory_caches(vcpu);
arch/x86/kvm/mmu/tdp_mmu.c +26 -8
index 2200bdc7681f..a0010c62425f 100644
@@ -95,10 +95,15 @@ void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root)
 static bool tdp_mmu_root_match(struct kvm_mmu_page *root,
 			       enum kvm_tdp_mmu_root_types types)
 {
+	if (WARN_ON_ONCE(!(types & KVM_VALID_ROOTS)))
+		return false;
+
 	if (root->role.invalid)
 		return types & KVM_INVALID_ROOTS;
+	if (likely(!is_mirror_sp(root)))
+		return types & KVM_DIRECT_ROOTS;
 
-	return true;
+	return types & KVM_MIRROR_ROOTS;
 }
 
 /*
@@ -233,7 +238,7 @@ static void tdp_mmu_init_child_sp(struct kvm_mmu_page *child_sp,
 	tdp_mmu_init_sp(child_sp, iter->sptep, iter->gfn, role);
 }
 
-void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool mirror)
 {
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
 	union kvm_mmu_page_role role = mmu->root_role;
@@ -241,6 +246,9 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_mmu_page *root;
 
+	if (mirror)
+		role.is_mirror = 1;
+
 	/*
 	 * Check for an existing root before acquiring the pages lock to avoid
 	 * unnecessary serialization if multiple vCPUs are loading a new root.
@@ -292,8 +300,12 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
 	 * and actually consuming the root if it's invalidated after dropping
 	 * mmu_lock, and the root can't be freed as this vCPU holds a reference.
 	 */
-	mmu->root.hpa = __pa(root->spt);
-	mmu->root.pgd = 0;
+	if (mirror) {
+		mmu->mirror_root_hpa = __pa(root->spt);
+	} else {
+		mmu->root.hpa = __pa(root->spt);
+		mmu->root.pgd = 0;
+	}
 }
 
 static void handle_changed_spte(struct kvm *kvm, int as_id, gfn_t gfn,
@@ -1113,8 +1125,8 @@ static int tdp_mmu_split_huge_page(struct kvm *kvm, struct tdp_iter *iter,
  */
 int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 {
+	struct kvm_mmu_page *root = tdp_mmu_get_root_for_fault(vcpu, fault);
 	struct kvm *kvm = vcpu->kvm;
-	struct kvm_mmu_page *root = root_to_sp(vcpu->arch.mmu->root.hpa);
 	struct tdp_iter iter;
 	struct kvm_mmu_page *sp;
 	int ret = RET_PF_RETRY;
@@ -1152,13 +1164,18 @@ int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 		 */
 		sp = tdp_mmu_alloc_sp(vcpu);
 		tdp_mmu_init_child_sp(sp, &iter);
+		if (is_mirror_sp(sp))
+			kvm_mmu_alloc_external_spt(vcpu, sp);
 
 		sp->nx_huge_page_disallowed = fault->huge_page_disallowed;
 
-		if (is_shadow_present_pte(iter.old_spte))
+		if (is_shadow_present_pte(iter.old_spte)) {
+			/* Don't support large page for mirrored roots (TDX) */
+			KVM_BUG_ON(is_mirror_sptep(iter.sptep), vcpu->kvm);
 			r = tdp_mmu_split_huge_page(kvm, &iter, sp, true);
-		else
+		} else {
 			r = tdp_mmu_link_sp(kvm, &iter, sp, true);
+		}
 
 		/*
 		 * Force the guest to retry if installing an upper level SPTE
@@ -1813,7 +1830,8 @@ int kvm_tdp_mmu_get_walk(struct kvm_vcpu *vcpu, u64 addr, u64 *sptes,
 u64 *kvm_tdp_mmu_fast_pf_get_last_sptep(struct kvm_vcpu *vcpu, gfn_t gfn,
 					u64 *spte)
 {
-	struct kvm_mmu_page *root = root_to_sp(vcpu->arch.mmu->root.hpa);
+	/* Fast pf is not supported for mirrored roots  */
+	struct kvm_mmu_page *root = tdp_mmu_get_root(vcpu, KVM_DIRECT_ROOTS);
 	struct tdp_iter iter;
 	tdp_ptep_t sptep = NULL;
 
arch/x86/kvm/mmu/tdp_mmu.h +40 -3

58 lines not on this page · raw message

Signature · 1 line
2.34.1
#15 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 5 lines · Rick Edgecombe
arch/x86/kvm/mmu/mmu.c +0 -0
index e9c1783a8743..287dcc2685e4 100644

arch/x86/kvm/mmu/mmu.c static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu) · line 3708

Quoted 9 lines
arch/x86/kvm/mmu/mmu.c +3 -1
@@ -3701,7 +3701,9 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
 	int r;
 
 	if (tdp_mmu_enabled) {
-		kvm_tdp_mmu_alloc_root(vcpu);
+		if (kvm_has_mirrored_tdp(vcpu->kvm))
+			kvm_tdp_mmu_alloc_root(vcpu, true);
+		kvm_tdp_mmu_alloc_root(vcpu, false);
 		return 0;
 	}

mmu_alloc_direct_roots() is called when vcpu->arch.mmu->root.hpa is INVALID_PAGE in kvm_mmu_reload(), which could happen after direct root is invalidated.

arch/x86/kvm/mmu/mmu.c void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu) · line 251

Quoted 12 lines
arch/x86/kvm/mmu/mmu.c +1 -1
-void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool mirror)

{

arch/x86/kvm/mmu/mmu.c +3 -0
 	struct kvm_mmu *mmu = vcpu->arch.mmu;
 	union kvm_mmu_page_role role = mmu->root_role;
@@ -241,6 +246,9 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_mmu_page *root;
 
+	if (mirror)
+		role.is_mirror = 1;
+

Could we add a validity check of mirror_root_hpa to prevent an incorrect ref count increment of the mirror root?

+ if (mirror) {
+ if (mmu->mirror_root_hpa != INVALID_PAGE)
+ return;
+

                role.is_mirror = true;

+ }

Quoted 16 lines
arch/x86/kvm/mmu/mmu.c +6 -2
 	/*
 	 * Check for an existing root before acquiring the pages lock to avoid
 	 * unnecessary serialization if multiple vCPUs are loading a new root.
@@ -292,8 +300,12 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
 	 * and actually consuming the root if it's invalidated after dropping
 	 * mmu_lock, and the root can't be freed as this vCPU holds a reference.
 	 */
-	mmu->root.hpa = __pa(root->spt);
-	mmu->root.pgd = 0;
+	if (mirror) {
+		mmu->mirror_root_hpa = __pa(root->spt);
+	} else {
+		mmu->root.hpa = __pa(root->spt);
+		mmu->root.pgd = 0;
+	}
 }
#16 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 20 lines · Yan Zhao
Quoted 7 lines · Rick Edgecombe
arch/x86/kvm/mmu/mmu.c +0 -0
index e9c1783a8743..287dcc2685e4 100644
@@ -3701,7 +3701,9 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu
 *vcpu)
         int r;
Quoted 8 lines


        if (tdp_mmu_enabled) {
-               kvm_tdp_mmu_alloc_root(vcpu);
+               if (kvm_has_mirrored_tdp(vcpu->kvm)) +                       kvm_tdp_mmu_alloc_root(vcpu, true); +               kvm_tdp_mmu_alloc_root(vcpu, false); return 0;
        }

mmu_alloc_direct_roots() is called when vcpu->arch.mmu->root.hpa is INVALID_PAGE
in kvm_mmu_reload(), which could happen after direct root is invalidated.

Quoted 14 lines
Quoted 8 lines
arch/x86/kvm/mmu/mmu.c +1 -1
-void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool mirror)

  {
        struct kvm_mmu *mmu = vcpu->arch.mmu; union kvm_mmu_page_role role = mmu->root_role;

arch/x86/kvm/mmu/mmu.c +0 -0
@@ -241,6 +246,9 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
         struct kvm *kvm = vcpu->kvm;
         struct kvm_mmu_page *root;
Quoted 4 lines


+       if (mirror)
+               role.is_mirror = 1;
+

Could we add a validity check of mirror_root_hpa to prevent an incorrect ref count increment of the mirror root?

I was originally suspicious of the asymmetry of the tear down of mirror and direct roots vs the allocation. Do you see a concrete problem, or just advocating for safety?

Quoted 7 lines

+       if (mirror) {
+               if (mmu->mirror_root_hpa != INVALID_PAGE) +                       return;
+
                role.is_mirror = true;
+       }

#17 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 36 lines · Rick P
Quoted 20 lines · Yan Zhao
Quoted 7 lines · Rick Edgecombe
arch/x86/kvm/mmu/mmu.c +0 -0
index e9c1783a8743..287dcc2685e4 100644
@@ -3701,7 +3701,9 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu
 *vcpu)
         int r;
Quoted 8 lines


        if (tdp_mmu_enabled) {
-               kvm_tdp_mmu_alloc_root(vcpu);
+               if (kvm_has_mirrored_tdp(vcpu->kvm)) +                       kvm_tdp_mmu_alloc_root(vcpu, true); +               kvm_tdp_mmu_alloc_root(vcpu, false); return 0;
        }

mmu_alloc_direct_roots() is called when vcpu->arch.mmu->root.hpa is INVALID_PAGE
in kvm_mmu_reload(), which could happen after direct root is invalidated.

Quoted 14 lines
Quoted 8 lines
arch/x86/kvm/mmu/mmu.c +1 -1
-void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool mirror)

  {
        struct kvm_mmu *mmu = vcpu->arch.mmu; union kvm_mmu_page_role role = mmu->root_role;

arch/x86/kvm/mmu/mmu.c +0 -0
@@ -241,6 +246,9 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu)
         struct kvm *kvm = vcpu->kvm;
         struct kvm_mmu_page *root;
Quoted 4 lines


+       if (mirror)
+               role.is_mirror = 1;
+

Could we add a validity check of mirror_root_hpa to prevent an incorrect ref count increment of the mirror root?

Quoted 3 lines

I was originally suspicious of the asymmetry of the tear down of mirror and direct roots vs the allocation. Do you see a concrete problem, or just advocating for safety?

IMO it's a concrete problem, though rare up to now. e.g.

After repeatedly hot-plugping and hot-unplugping memory, which increases memslots generation, kvm_mmu_zap_all_fast() will be called to invalidate direct roots when the memslots generation wraps around.

Quoted 9 lines
Quoted 8 lines

+       if (mirror) {
+               if (mmu->mirror_root_hpa != INVALID_PAGE) +                       return;
+
                role.is_mirror = true;
+       }

#18 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 10 lines · Yan Zhao
Quoted 10 lines
Quoted 4 lines
Quoted 4 lines

I was originally suspicious of the asymmetry of the tear down of mirror and
direct roots vs the allocation. Do you see a concrete problem, or just advocating for safety?

IMO it's a concrete problem, though rare up to now. e.g.

After repeatedly hot-plugping and hot-unplugping memory, which increases memslots generation, kvm_mmu_zap_all_fast() will be called to invalidate > direct
roots when the memslots generation wraps around.

Hmm, yes. I'm not sure about putting the check there though. It adds even more confusion to the lifecycle.
- mirror_root_hpa != INVALID_PAGE check in a different placed than root.hpa != INVALID_PAGE check.
- they get allocated in the same place
- they are torn down in the different places.

Can you think of clearer fix for it. Maybe we can just move the mirror root allocation such that it's not subjected to the reload path? Like something that matches the tear down in kvm_mmu_destroy()?

#19 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 19 lines · Rick P
Quoted 10 lines · Yan Zhao
Quoted 10 lines
Quoted 4 lines
Quoted 4 lines

I was originally suspicious of the asymmetry of the tear down of mirror and
direct roots vs the allocation. Do you see a concrete problem, or just advocating for safety?

IMO it's a concrete problem, though rare up to now. e.g.

After repeatedly hot-plugping and hot-unplugping memory, which increases memslots generation, kvm_mmu_zap_all_fast() will be called to invalidate > direct
roots when the memslots generation wraps around.

Hmm, yes. I'm not sure about putting the check there though. It adds even more confusion to the lifecycle.
- mirror_root_hpa != INVALID_PAGE check in a different placed than root.hpa != INVALID_PAGE check.
- they get allocated in the same place
- they are torn down in the different places.

Quoted 3 lines

Can you think of clearer fix for it. Maybe we can just move the mirror root allocation such that it's not subjected to the reload path? Like something that matches the tear down in kvm_mmu_destroy()?

But we still need the reload path to have each vcpu to hold a ref count of the mirror root.
What about below fix?

arch/x86/kvm/mmu.h +20 -6
index 026e8edfb0bd..4decd13457ec 100644
@@ -127,9 +127,28 @@ void kvm_mmu_sync_prev_roots(struct kvm_vcpu *vcpu);
 void kvm_mmu_track_write(struct kvm_vcpu *vcpu, gpa_t gpa, const u8 *new,
                         int bytes);
 
+static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
+{
+       return kvm->arch.vm_type == KVM_X86_TDX_VM;
+}
+
+static inline bool kvm_mmu_root_hpa_is_invalid(struct kvm_vcpu *vcpu)
+{
+       return vcpu->arch.mmu->root.hpa == INVALID_PAGE;
+}
+
+static inline bool kvm_mmu_mirror_root_hpa_is_invalid(struct kvm_vcpu *vcpu)
+{
+       if (!kvm_has_mirrored_tdp(vcpu->kvm))
+               return false;
+
+       return vcpu->arch.mmu->mirror_root_hpa == INVALID_PAGE;
+}
+
 static inline int kvm_mmu_reload(struct kvm_vcpu *vcpu)
 {
-       if (likely(vcpu->arch.mmu->root.hpa != INVALID_PAGE))
+       if (!kvm_mmu_root_hpa_is_invalid(vcpu) &&
+           !kvm_mmu_mirror_root_hpa_is_invalid(vcpu))
                return 0;
 
        return kvm_mmu_load(vcpu);
@@ -322,11 +341,6 @@ static inline gpa_t kvm_translate_gpa(struct kvm_vcpu *vcpu,
        return translate_nested_gpa(vcpu, gpa, access, exception);
 }
 
-static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
-{
-       return kvm->arch.vm_type == KVM_X86_TDX_VM;
-}
-
 static inline gfn_t kvm_gfn_direct_bits(const struct kvm *kvm)
 {
        return kvm->arch.gfn_direct_bits;
arch/x86/kvm/mmu/mmu.c +4 -2
index e1299eb03e63..5e7d92074f70 100644
@@ -3702,9 +3702,11 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
        int r;
 
        if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm))
+               if (kvm_mmu_mirror_root_hpa_is_invalid(vcpu))
                        kvm_tdp_mmu_alloc_root(vcpu, true);
-               kvm_tdp_mmu_alloc_root(vcpu, false);
+
+               if (kvm_mmu_root_hpa_is_invalid(vcpu))
+                       kvm_tdp_mmu_alloc_root(vcpu, false);
                return 0;
        }
 
#20 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 25 lines · Yan Zhao

But we still need the reload path to have each vcpu to hold a ref count of the mirror root.
What about below fix?

arch/x86/kvm/mmu.h +13 -0
index 026e8edfb0bd..4decd13457ec 100644
@@ -127,9 +127,28 @@ void kvm_mmu_sync_prev_roots(struct kvm_vcpu *vcpu);
  void kvm_mmu_track_write(struct kvm_vcpu *vcpu, gpa_t gpa, const u8 *new,
                          int bytes);
 
+static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
+{
+       return kvm->arch.vm_type == KVM_X86_TDX_VM;
+}
+
+static inline bool kvm_mmu_root_hpa_is_invalid(struct kvm_vcpu *vcpu)
+{
+       return vcpu->arch.mmu->root.hpa == INVALID_PAGE;
+}
+
+static inline bool kvm_mmu_mirror_root_hpa_is_invalid(struct kvm_vcpu *vcpu)
+{
+       if (!kvm_has_mirrored_tdp(vcpu->kvm))

arch/x86/kvm/mmu.h void kvm_mmu_sync_prev_roots(struct kvm_vcpu *vcpu); · line 151

Quoted 10 lines
arch/x86/kvm/mmu.h +7 -1
+               return false;
+
+       return vcpu->arch.mmu->mirror_root_hpa == INVALID_PAGE;
+}
+
  static inline int kvm_mmu_reload(struct kvm_vcpu *vcpu)
  {
-       if (likely(vcpu->arch.mmu->root.hpa != INVALID_PAGE))
+       if (!kvm_mmu_root_hpa_is_invalid(vcpu) &&
+           !kvm_mmu_mirror_root_hpa_is_invalid(vcpu))

If we go this way, we should keep the likely. But, I'm not convinced.

Quoted 22 lines
arch/x86/kvm/mmu.h +0 -5
                 return 0;
 
         return kvm_mmu_load(vcpu);
@@ -322,11 +341,6 @@ static inline gpa_t kvm_translate_gpa(struct kvm_vcpu
 *vcpu,
         return translate_nested_gpa(vcpu, gpa, access, exception);
  }
 
-static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
-{
-       return kvm->arch.vm_type == KVM_X86_TDX_VM;
-}
-
  static inline gfn_t kvm_gfn_direct_bits(const struct kvm *kvm)
  {

        return kvm->arch.gfn_direct_bits;

arch/x86/kvm/mmu/mmu.c +0 -0
index e1299eb03e63..5e7d92074f70 100644
@@ -3702,9 +3702,11 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu
 *vcpu)

arch/x86/kvm/mmu/mmu.c static int mmu_alloc_direct_roots(struct kvm_vcpu · line 3710

Quoted 10 lines
arch/x86/kvm/mmu/mmu.c +4 -2
         int r;
 
         if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm))
+               if (kvm_mmu_mirror_root_hpa_is_invalid(vcpu))
                         kvm_tdp_mmu_alloc_root(vcpu, true);
-               kvm_tdp_mmu_alloc_root(vcpu, false);
+
+               if (kvm_mmu_root_hpa_is_invalid(vcpu))
+                       kvm_tdp_mmu_alloc_root(vcpu, false);

So we can have either:
mirror_root_hpa = INVALID_PAGE
root.hpa = INVALID_PAGE

or:
mirror_root_hpa = root
root.hpa = INVALID_PAGE

We don't ever have:
mirror_root_hpa = INVALID_PAGE
root.hpa = root

Because mirror_root_hpa is special.

arch/x86/kvm/mmu/mmu.c static int mmu_alloc_direct_roots(struct kvm_vcpu · line 3712

Quoted 3 lines
arch/x86/kvm/mmu/mmu.c +0 -0
                 return 0;
         }

Having the check in kvm_mmu_reload() and here both is really ugly.  Right now we have kvm_mmu_reload() which only allocates new roots if needed, then kvm_mmu_load() goes and allocates/references them. Now mmu_alloc_direct_roots() has the same logic to only reload as needed.

So could we just leave the kvm_mmu_reload() logic as is, and just have special logic in mmu_alloc_direct_roots() to not avoid re-allocating mirror roots? This is actually what v19 had, but it was thought to be a historical relic:
"Historically we needed it. We don't need it now. We can drop it."
https://lore.kernel.org/kvm/[email protected]/

So just have this small fixup instead?

arch/x86/kvm/mmu/mmu.c +2 -1
index ae5d5dee86af..2e062178222e 100644
@@ -3705,7 +3705,8 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
        int r;
 
        if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm))
+               if (kvm_has_mirrored_tdp(vcpu->kvm) &&
+                   !VALID_PAGE(mmu->mirror_root_hpa))
                        kvm_tdp_mmu_alloc_root(vcpu, true);
                kvm_tdp_mmu_alloc_root(vcpu, false);
                return 0;
#21 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 80 lines · Rick P
Quoted 25 lines · Yan Zhao

But we still need the reload path to have each vcpu to hold a ref count of the mirror root.
What about below fix?

arch/x86/kvm/mmu.h +13 -0
index 026e8edfb0bd..4decd13457ec 100644
@@ -127,9 +127,28 @@ void kvm_mmu_sync_prev_roots(struct kvm_vcpu *vcpu);
  void kvm_mmu_track_write(struct kvm_vcpu *vcpu, gpa_t gpa, const u8 *new,
                          int bytes);
 
+static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
+{
+       return kvm->arch.vm_type == KVM_X86_TDX_VM;
+}
+
+static inline bool kvm_mmu_root_hpa_is_invalid(struct kvm_vcpu *vcpu)
+{
+       return vcpu->arch.mmu->root.hpa == INVALID_PAGE;
+}
+
+static inline bool kvm_mmu_mirror_root_hpa_is_invalid(struct kvm_vcpu *vcpu)
+{
+       if (!kvm_has_mirrored_tdp(vcpu->kvm))

arch/x86/kvm/mmu.h void kvm_mmu_sync_prev_roots(struct kvm_vcpu *vcpu); · line 151

Quoted 10 lines
arch/x86/kvm/mmu.h +7 -1
+               return false;
+
+       return vcpu->arch.mmu->mirror_root_hpa == INVALID_PAGE;
+}
+
  static inline int kvm_mmu_reload(struct kvm_vcpu *vcpu)
  {
-       if (likely(vcpu->arch.mmu->root.hpa != INVALID_PAGE))
+       if (!kvm_mmu_root_hpa_is_invalid(vcpu) &&
+           !kvm_mmu_mirror_root_hpa_is_invalid(vcpu))

If we go this way, we should keep the likely. But, I'm not convinced.

Quoted 22 lines
arch/x86/kvm/mmu.h +0 -5
                 return 0;
 
         return kvm_mmu_load(vcpu);
@@ -322,11 +341,6 @@ static inline gpa_t kvm_translate_gpa(struct kvm_vcpu
 *vcpu,
         return translate_nested_gpa(vcpu, gpa, access, exception);
  }
 
-static inline bool kvm_has_mirrored_tdp(const struct kvm *kvm)
-{
-       return kvm->arch.vm_type == KVM_X86_TDX_VM;
-}
-
  static inline gfn_t kvm_gfn_direct_bits(const struct kvm *kvm)
  {

        return kvm->arch.gfn_direct_bits;

arch/x86/kvm/mmu/mmu.c +0 -0
index e1299eb03e63..5e7d92074f70 100644
@@ -3702,9 +3702,11 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu
 *vcpu)

arch/x86/kvm/mmu/mmu.c static int mmu_alloc_direct_roots(struct kvm_vcpu · line 3710

Quoted 10 lines
arch/x86/kvm/mmu/mmu.c +4 -2
         int r;
 
         if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm))
+               if (kvm_mmu_mirror_root_hpa_is_invalid(vcpu))
                         kvm_tdp_mmu_alloc_root(vcpu, true);
-               kvm_tdp_mmu_alloc_root(vcpu, false);
+
+               if (kvm_mmu_root_hpa_is_invalid(vcpu))
+                       kvm_tdp_mmu_alloc_root(vcpu, false);

So we can have either:
mirror_root_hpa = INVALID_PAGE
root.hpa = INVALID_PAGE

or:
mirror_root_hpa = root
root.hpa = INVALID_PAGE

Quoted 6 lines

We don't ever have:
mirror_root_hpa = INVALID_PAGE
root.hpa = root

Because mirror_root_hpa is special.

Good point.

Quoted 9 lines

arch/x86/kvm/mmu/mmu.c static int mmu_alloc_direct_roots(struct kvm_vcpu · line 3712

Quoted 3 lines
arch/x86/kvm/mmu/mmu.c +0 -0
                 return 0;
         }

Having the check in kvm_mmu_reload() and here both is really ugly.  Right now we have kvm_mmu_reload() which only allocates new roots if needed, then kvm_mmu_load() goes and allocates/references them. Now mmu_alloc_direct_roots() has the same logic to only reload as needed.

Quoted 7 lines

So could we just leave the kvm_mmu_reload() logic as is, and just have special logic in mmu_alloc_direct_roots() to not avoid re-allocating mirror roots? This is actually what v19 had, but it was thought to be a historical relic:
"Historically we needed it. We don't need it now. We can drop it."
https://lore.kernel.org/kvm/[email protected]/

So just have this small fixup instead?

Yes.

Quoted 6 lines
arch/x86/kvm/mmu/mmu.c +0 -0
index ae5d5dee86af..2e062178222e 100644

arch/x86/kvm/mmu/mmu.c static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu) · line 3712

Quoted 9 lines
arch/x86/kvm/mmu/mmu.c +2 -1
@@ -3705,7 +3705,8 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
        int r;
 
        if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm))
+               if (kvm_has_mirrored_tdp(vcpu->kvm) &&
+                   !VALID_PAGE(mmu->mirror_root_hpa))
                        kvm_tdp_mmu_alloc_root(vcpu, true);
                kvm_tdp_mmu_alloc_root(vcpu, false);
                return 0;

Perhaps also a comment in kvm_mmu_reload() to address concerns like why checking only root.hpa in kvm_mmu_reload() is enough.

arch/x86/kvm/mmu.h +9 -0
index 03737f3aaeeb..aba98c8cc67d 100644
@@ -129,6 +129,15 @@ void kvm_mmu_track_write(struct kvm_vcpu *vcpu, gpa_t gpa, const u8 *new,
 
 static inline int kvm_mmu_reload(struct kvm_vcpu *vcpu)
 {
+       /*
+        * Checking root.hpa is sufficient even when KVM has mirror root.
+        * We can have either:
+        * (1) mirror_root_hpa = INVALID_PAGE, root.hpa = INVALID_PAGE
+        * (2) mirror_root_hpa = root ,        root.hpa = INVALID_PAGE
+        * (3) mirror_root_hpa = root1,        root.hpa = root2
+        * We don't ever have:
+        *     mirror_root_hpa = INVALID_PAGE, root.hpa = root
+        */
        if (likely(vcpu->arch.mmu->root.hpa != INVALID_PAGE))
                return 0;
 
arch/x86/kvm/mmu/mmu.c +2 -1
index a5f803f1d17e..eee35e958971 100644
@@ -3705,7 +3705,8 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
        int r;
 
        if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm))
+               if (kvm_has_mirrored_tdp(vcpu->kvm) &&
+                   !VALID_PAGE(mmu->mirror_root_hpa))
                        kvm_tdp_mmu_alloc_root(vcpu, true);
                kvm_tdp_mmu_alloc_root(vcpu, false);
                return 0;
#22 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 3 lines · Yan Zhao

Perhaps also a comment in kvm_mmu_reload() to address concerns like why checking
only root.hpa in kvm_mmu_reload() is enough.

Sounds good, and thanks again for catching this.

Quoted 7 lines
arch/x86/kvm/mmu.h +0 -0
index 03737f3aaeeb..aba98c8cc67d 100644
@@ -129,6 +129,15 @@ void kvm_mmu_track_write(struct kvm_vcpu *vcpu, gpa_t
 gpa, const u8 *new,
Quoted 8 lines


 static inline int kvm_mmu_reload(struct kvm_vcpu *vcpu)
 {
+       /*
+        * Checking root.hpa is sufficient even when KVM has mirror root. +        * We can have either:
+        * (1) mirror_root_hpa = INVALID_PAGE, root.hpa = INVALID_PAGE +        * (2) mirror_root_hpa = root ,        root.hpa = INVALID_PAGE

Looks good to me except for the space ^

Quoted 21 lines

+        * (3) mirror_root_hpa = root1,        root.hpa = root2 +        * We don't ever have:
+        *     mirror_root_hpa = INVALID_PAGE, root.hpa = root +        */
        if (likely(vcpu->arch.mmu->root.hpa != INVALID_PAGE)) return 0;

arch/x86/kvm/mmu/mmu.c +0 -0
index a5f803f1d17e..eee35e958971 100644
@@ -3705,7 +3705,8 @@ static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
         int r;


        if (tdp_mmu_enabled) {
-               if (kvm_has_mirrored_tdp(vcpu->kvm)) +               if (kvm_has_mirrored_tdp(vcpu->kvm) && +                   !VALID_PAGE(mmu->mirror_root_hpa)) kvm_tdp_mmu_alloc_root(vcpu, true); kvm_tdp_mmu_alloc_root(vcpu, false); return 0;

#23 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

In reply to Rick Edgecombe

arch/x86/kvm/mmu/tdp_mmu.h line 15

Quoted 12 lines · Rick Edgecombe
arch/x86/kvm/mmu/tdp_mmu.h +1 -1
index b887c225ff24..2903f03a34be 100644
@@ -10,7 +10,7 @@
 void kvm_mmu_init_tdp_mmu(struct kvm *kvm);
 void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm);
 
-void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu);
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool private);
 
 __must_check static inline bool kvm_tdp_mmu_get_root(struct kvm_mmu_page *root)

This function name is very similar to below tdp_mmu_get_root().

Quoted 7 lines
arch/x86/kvm/mmu/tdp_mmu.h +7 -0
+static inline struct kvm_mmu_page *tdp_mmu_get_root_for_fault(struct kvm_vcpu *vcpu,
+							      struct kvm_page_fault *fault)
+{
+	if (unlikely(!kvm_is_addr_direct(vcpu->kvm, fault->addr)))
+		return root_to_sp(vcpu->arch.mmu->mirror_root_hpa);
+
+	return root_to_sp(vcpu->arch.mmu->root.hpa);

arch/x86/kvm/mmu/tdp_mmu.h void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root); · line 64

Quoted 10 lines
arch/x86/kvm/mmu/tdp_mmu.h +10 -0
+}
+
+static inline struct kvm_mmu_page *tdp_mmu_get_root(struct kvm_vcpu *vcpu,
+						    enum kvm_tdp_mmu_root_types type)
+{
+	if (unlikely(type == KVM_MIRROR_ROOTS))
+		return root_to_sp(vcpu->arch.mmu->mirror_root_hpa);
+
+	return root_to_sp(vcpu->arch.mmu->root.hpa);
+}

No need to put the two functions in tdp_mmu.h as they are not called in mmu.c.

Could we move them to tdp_mmu.c and rename them to something like tdp_mmu_type_to_root() and tdp_mmu_fault_to_root() ?

#24 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 15 lines · Yan Zhao
Quoted 10 lines · Rick Edgecombe
arch/x86/kvm/mmu/tdp_mmu.h +0 -0
index b887c225ff24..2903f03a34be 100644
@@ -10,7 +10,7 @@
   void kvm_mmu_init_tdp_mmu(struct kvm *kvm);
   void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm);

arch/x86/kvm/mmu/tdp_mmu.h +1 -1
-void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu);
+void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool private);
Quoted 3 lines


  __must_check static inline bool kvm_tdp_mmu_get_root(struct kvm_mmu_page *root)

This function name is very similar to below tdp_mmu_get_root().

That is true.

Quoted 14 lines
Quoted 14 lines

+static inline struct kvm_mmu_page *tdp_mmu_get_root_for_fault(struct kvm_vcpu *vcpu,
+                                                             struct kvm_page_fault *fault)

arch/x86/kvm/mmu/tdp_mmu.h +9 -0
+{
+       if (unlikely(!kvm_is_addr_direct(vcpu->kvm, fault->addr)))
+               return root_to_sp(vcpu->arch.mmu->mirror_root_hpa);
+
+       return root_to_sp(vcpu->arch.mmu->root.hpa);
+}
+
+static inline struct kvm_mmu_page *tdp_mmu_get_root(struct kvm_vcpu *vcpu,
+                                                   enum
Quoted 8 lines

arch/x86/kvm/mmu/tdp_mmu.h void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root); · line 64

Quoted 7 lines

kvm_tdp_mmu_root_types type)

arch/x86/kvm/mmu/tdp_mmu.h +6 -0
+{
+       if (unlikely(type == KVM_MIRROR_ROOTS))
+               return root_to_sp(vcpu->arch.mmu->mirror_root_hpa);
+
+       return root_to_sp(vcpu->arch.mmu->root.hpa);
+}

No need to put the two functions in tdp_mmu.h as they are not called in mmu.c.

I think it is nice to have the root/enum logic above close together.

Quoted 3 lines

Could we move them to tdp_mmu.c and rename them to something like tdp_mmu_type_to_root() and tdp_mmu_fault_to_root() ?

tdp_mmu_get_root_for_fault() was proposed by Paolo, and tdp_mmu_get_root() was discussed without comment. Not to say there is anything wrong with the names proposed, but I think this is wading into bikeshedding territory at this stage.

#25 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 7 lines · Rick Edgecombe
Quoted 2 lines

Could we move them to tdp_mmu.c and rename them to something like tdp_mmu_type_to_root() and tdp_mmu_fault_to_root() ?

tdp_mmu_get_root_for_fault() was proposed by Paolo, and tdp_mmu_get_root() was discussed without comment. Not to say there is anything wrong with the names proposed, but I think this is wading into bikeshedding territory at this stage.

I kept thinking about this comment. On more consideration tdp_mmu_get_root() is a problematically terrible name since "get" usually means take a reference to something which is something you can do to roots and (and what kvm_tdp_mmu_get_root() is doing). So people might think tdp_mmu_get_root() is taking a reference.

#26 lore raw

Re: [PATCH v3 13/17] KVM: x86/tdp_mmu: Support mirror root for TDP MMU

Quoted 9 lines · Rick P
Quoted 7 lines · Rick Edgecombe
Quoted 2 lines

Could we move them to tdp_mmu.c and rename them to something like tdp_mmu_type_to_root() and tdp_mmu_fault_to_root() ?

tdp_mmu_get_root_for_fault() was proposed by Paolo, and tdp_mmu_get_root() was discussed without comment. Not to say there is anything wrong with the names proposed, but I think this is wading into bikeshedding territory at this stage.

Quoted 5 lines

I kept thinking about this comment. On more consideration tdp_mmu_get_root() is a problematically terrible name since "get" usually means take a reference to something which is something you can do to roots and (and what kvm_tdp_mmu_get_root() is doing). So people might think tdp_mmu_get_root() is taking a reference.

Right, I have exactly the same feeling to "get".

#27 lore raw

PATCH v3 14/17 KVM: x86/tdp_mmu: Propagate attr_filter to MMU notifier callbacks

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Teach the MMU notifier callbacks how to check kvm_gfn_range.process to filter which KVM MMU root types to operate on.

The private GPAs are backed by guest memfd. Such memory is not subjected to MMU notifier callbacks because it can't be mapped into the host user address space. Now kvm_gfn_range conveys info about which root to operate on. Enhance the callback to filter the root page table type.

The KVM MMU notifier comes down to two functions. kvm_tdp_mmu_unmap_gfn_range() and kvm_tdp_mmu_handle_gfn().

For VM's without a private/shared split in the EPT, all operations should target the normal(direct) root.

invalidate_range_start() comes into kvm_tdp_mmu_unmap_gfn_range(). invalidate_range_end() doesn't come into arch code.

With the switch from for_each_tdp_mmu_root() to __for_each_tdp_mmu_root() in kvm_tdp_mmu_handle_gfn(), there are no longer any users of for_each_tdp_mmu_root(). Remove it.

---
TDX MMU Prep v3:
- Change subject from "Make mmu notifier callbacks to check kvm_process" to "Propagate attr_filter to MMU notifier callbacks" (Paolo)
- Remove no longer used for_each_tdp_mmu_root() (Binbin)

TDX MMU Prep v2:
- Use newly added kvm_process_to_root_types()

TDX MMU Prep:
- Remove warning (Rick)
- Remove confusing mention of mapping flags (Chao)
- Re-write coverletter

v19:
- type: test_gfn() => test_young()

---
 arch/x86/kvm/mmu/tdp_mmu.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
arch/x86/kvm/mmu/tdp_mmu.c +9 -5
index a0010c62425f..582e5a045bb7 100644
@@ -193,9 +193,6 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
 		     !tdp_mmu_root_match((_root), (_types)))) {			\
 		} else
 
-#define for_each_tdp_mmu_root(_kvm, _root, _as_id)			\
-	__for_each_tdp_mmu_root(_kvm, _root, _as_id, KVM_ALL_ROOTS)
-
 #define for_each_valid_tdp_mmu_root(_kvm, _root, _as_id)		\
 	__for_each_tdp_mmu_root(_kvm, _root, _as_id, KVM_VALID_ROOTS)
 
@@ -1210,12 +1207,16 @@ int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault)
 	return ret;
 }
 
+/* Used by mmu notifier via kvm_unmap_gfn_range() */
 bool kvm_tdp_mmu_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range,
 				 bool flush)
 {
+	enum kvm_tdp_mmu_root_types types;
 	struct kvm_mmu_page *root;
 
-	__for_each_tdp_mmu_root_yield_safe(kvm, root, range->slot->as_id, KVM_ALL_ROOTS)
+	types = kvm_gfn_range_filter_to_root_types(kvm, range->attr_filter);
+
+	__for_each_tdp_mmu_root_yield_safe(kvm, root, range->slot->as_id, types)
 		flush = tdp_mmu_zap_leafs(kvm, root, range->start, range->end,
 					  range->may_block, flush);
 
@@ -1229,15 +1230,18 @@ static __always_inline bool kvm_tdp_mmu_handle_gfn(struct kvm *kvm,
 						   struct kvm_gfn_range *range,
 						   tdp_handler_t handler)
 {
+	enum kvm_tdp_mmu_root_types types;
 	struct kvm_mmu_page *root;
 	struct tdp_iter iter;
 	bool ret = false;
 
+	types = kvm_gfn_range_filter_to_root_types(kvm, range->attr_filter);
+
 	/*
 	 * Don't support rescheduling, none of the MMU notifiers that funnel
 	 * into this helper allow blocking; it'd be dead, wasteful code.
 	 */
-	for_each_tdp_mmu_root(kvm, root, range->slot->as_id) {
+	__for_each_tdp_mmu_root(kvm, root, range->slot->as_id, types) {
 		rcu_read_lock();
 
 		tdp_root_for_each_leaf_pte(iter, kvm, root, range->start, range->end)
Signature · 1 line
2.34.1
#28 lore raw

PATCH v3 15/17 KVM: x86/tdp_mmu: Propagate building mirror page tables

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Integrate hooks for mirroring page table operations for cases where TDX will set PTEs or link page tables.

Like other Coco technologies, TDX has the concept of private and shared memory. For TDX the private and shared mappings are managed on separate EPT roots. The private half is managed indirectly though calls into a protected runtime environment called the TDX module, where the shared half is managed within KVM in normal page tables.

Since calls into the TDX module are relatively slow, walking private page tables by making calls into the TDX module would not be efficient. Because of this, previous changes have taught the TDP MMU to keep a mirror root, which is separate, unmapped TDP root that private operations can be directed to. Currently this root is disconnected from any actual guest mapping. Now add plumbing to propagate changes to the "external" page tables being mirrored. Just create the x86_ops for now, leave plumbing the operations into the TDX module for future patches.

Add two operations for setting up external page tables, one for linking new page tables and one for setting leaf PTEs. Don't add any op for configuring the root PFN, as TDX handles this itself. Don't provide a way to set permissions on the PTEs also, as TDX doesn't support it.

This results is MMU "mirroring" support that is very targeted towards TDX. Since it is likely there will be no other user, the main benefit of making the support generic is to keep TDX specific *looking* code outside of the MMU. As a generic feature it will make enough sense from TDX's perspective. For developers unfamiliar with TDX arch it can express the general concepts such that they can continue to work in the code.

TDX MMU support will exclude certain MMU operations, so only plug in the mirroring x86 ops where they will be needed. For setting/linking, only hook tdp_mmu_set_spte_atomic() which is use used for mapping and linking PTs. Don't bother hooking tdp_mmu_iter_set_spte() as it is only used for setting PTEs in operations unsupported by TDX: splitting huge pages and write protecting. Sprinkle a KVM_BUG_ON()s to document as code that these paths are not supported for mirrored page tables. For zapping operations, leave those for near future changes.

Many operations in the TDP MMU depend on atomicity of the PTE update. While the mirror PTE on KVM's side can be updated atomically, the update that happens inside the external operations (S-EPT updates via TDX module call) can't happen atomically with the mirror update. The following race could result during two vCPU's populating private memory:

* vcpu 1: atomically update 2M level mirror EPT entry to be present
* vcpu 2: read 2M level EPT entry that is present
* vcpu 2: walk down into 4K level EPT
* vcpu 2: atomically update 4K level mirror EPT entry to be present
* vcpu 2: set_exterma;_spte() to update 4K secure EPT entry => error because 2M secure EPT entry is not populated yet
* vcpu 1: link_external_spt() to update 2M secure EPT entry

Prevent this by setting the mirror PTE to FROZEN_SPTE while the reflect operations are performed. Only write the actual mirror PTE value once the reflect operations has completed. When trying to set a PTE to present and encountering a removed SPTE, retry the fault.

By doing this the race is prevented as follows:
* vcpu 1: atomically update 2M level EPT entry to be FROZEN_SPTE
* vcpu 2: read 2M level EPT entry that is FROZEN_SPTE
* vcpu 2: find that the EPT entry is frozen
abandon page table walk to resume guest execution
* vcpu 1: link_external_spt() to update 2M secure EPT entry
* vcpu 1: atomically update 2M level EPT entry to be present (unfreeze)
* vcpu 2: resume guest execution
Depending on vcpu 1 state, vcpu 2 may result in EPT violation again or make progress on guest execution

---
TDX MMU Prep v3:
- Rename mirrored->external (Paolo)
- Better comment on logic that bugs if doing tdp_mmu_set_spte() on present PTE. (Paolo)
- Move zapping KVM_BUG_ON() to proper patch
- Use spte_to_child_sp() (Paolo)
- Drop unnessary comment in __tdp_mmu_set_spte_atomic() (Paolo)
- Rename pfn->pfn_for_gfn to match remove_external_pte in next patch.
- Rename REMOVED_SPTE to FROZEN_SPTE (Paolo)

TDX MMU Prep v2:
- Split from "KVM: x86/tdp_mmu: Support TDX private mapping for TDP MMU"
- Rename x86_ops from "private" to "reflect"
- In response to "sp->mirrored_spt" rename helpers to "mirrored"
- Drop unused old_pfn and new_pfn in handle_changed_spte()
- Drop redundant is_shadow_present_pte() check in __tdp_mmu_set_spte_atomic
- Adjust some warnings and KVM_BUG_ONs

---
 arch/x86/include/asm/kvm-x86-ops.h |  2 +
 arch/x86/include/asm/kvm_host.h    |  7 +++
 arch/x86/kvm/mmu/tdp_mmu.c         | 98 ++++++++++++++++++++++++++----
 3 files changed, 94 insertions(+), 13 deletions(-)
arch/x86/include/asm/kvm-x86-ops.h +2 -0
index 566d19b02483..3ef19fcb5e42 100644
@@ -95,6 +95,8 @@ KVM_X86_OP_OPTIONAL_RET0(set_tss_addr)
 KVM_X86_OP_OPTIONAL_RET0(set_identity_map_addr)
 KVM_X86_OP_OPTIONAL_RET0(get_mt_mask)
 KVM_X86_OP(load_mmu_pgd)
+KVM_X86_OP_OPTIONAL(link_external_spt)
+KVM_X86_OP_OPTIONAL(set_external_spte)
 KVM_X86_OP(has_wbinvd_exit)
 KVM_X86_OP(get_l2_tsc_offset)
 KVM_X86_OP(get_l2_tsc_multiplier)
arch/x86/include/asm/kvm_host.h +7 -0
index d67e88a69fc4..12ff04135a0e 100644
@@ -1738,6 +1738,13 @@ struct kvm_x86_ops {
 	void (*load_mmu_pgd)(struct kvm_vcpu *vcpu, hpa_t root_hpa,
 			     int root_level);
 
+	/* Update external mapping with page table link */
+	int (*link_external_spt)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				void *mirrored_spt);
+	/* Update the external page table from spte getting set */
+	int (*set_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				 kvm_pfn_t pfn_for_gfn);
+
 	bool (*has_wbinvd_exit)(void);
 
 	u64 (*get_l2_tsc_offset)(struct kvm_vcpu *vcpu);
arch/x86/kvm/mmu/tdp_mmu.c +85 -13

140 lines not on this page · raw message

Signature · 1 line
2.34.1
#29 lore raw

Re: [PATCH v3 15/17] KVM: x86/tdp_mmu: Propagate building mirror page tables

Quoted 20 lines · Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Integrate hooks for mirroring page table operations for cases where TDX will set PTEs or link page tables.

Like other Coco technologies, TDX has the concept of private and shared memory. For TDX the private and shared mappings are managed on separate EPT roots. The private half is managed indirectly though calls into a protected runtime environment called the TDX module, where the shared half is managed within KVM in normal page tables.

Since calls into the TDX module are relatively slow, walking private page tables by making calls into the TDX module would not be efficient. Because of this, previous changes have taught the TDP MMU to keep a mirror root, which is separate, unmapped TDP root that private operations can be directed to. Currently this root is disconnected from any actual guest mapping. Now add plumbing to propagate changes to the "external" page tables being mirrored. Just create the x86_ops for now, leave plumbing the operations into the TDX module for future patches.

Quoted 6 lines

Add two operations for setting up external page tables, one for linking new page tables and one for setting leaf PTEs. Don't add any op for configuring the root PFN, as TDX handles this itself. Don't provide a way to set permissions on the PTEs also, as TDX doesn't support it.

This results is MMU "mirroring" support that is very targeted towards TDX.

               ^
               extra "is"?

Quoted 6 lines

Since it is likely there will be no other user, the main benefit of making the support generic is to keep TDX specific *looking* code outside of the MMU. As a generic feature it will make enough sense from TDX's perspective. For developers unfamiliar with TDX arch it can express the general concepts such that they can continue to work in the code.

Quoted 6 lines

TDX MMU support will exclude certain MMU operations, so only plug in the mirroring x86 ops where they will be needed. For setting/linking, only hook tdp_mmu_set_spte_atomic() which is use used for mapping and linking PTs. Don't bother hooking tdp_mmu_iter_set_spte() as it is only used for setting PTEs in operations unsupported by TDX: splitting huge pages and write protecting. Sprinkle a KVM_BUG_ON()s to document as code that these

                            ^
                            extra "a"

Quoted 17 lines

paths are not supported for mirrored page tables. For zapping operations, leave those for near future changes.

Many operations in the TDP MMU depend on atomicity of the PTE update. While the mirror PTE on KVM's side can be updated atomically, the update that happens inside the external operations (S-EPT updates via TDX module call) can't happen atomically with the mirror update. The following race could result during two vCPU's populating private memory:

* vcpu 1: atomically update 2M level mirror EPT entry to be present
* vcpu 2: read 2M level EPT entry that is present
* vcpu 2: walk down into 4K level EPT
* vcpu 2: atomically update 4K level mirror EPT entry to be present
* vcpu 2: set_exterma;_spte() to update 4K secure EPT entry => error because 2M secure EPT entry is not populated yet
* vcpu 1: link_external_spt() to update 2M secure EPT entry

Quoted 2 lines

Prevent this by setting the mirror PTE to FROZEN_SPTE while the reflect operations are performed. Only write the actual mirror PTE value once the

            ^
            maybe "are being"

Quoted 2 lines

reflect operations has completed. When trying to set a PTE to present and encountering a removed SPTE, retry the fault.

                ^
                frozen

Quoted 61 lines

By doing this the race is prevented as follows:
* vcpu 1: atomically update 2M level EPT entry to be FROZEN_SPTE
* vcpu 2: read 2M level EPT entry that is FROZEN_SPTE
* vcpu 2: find that the EPT entry is frozen
abandon page table walk to resume guest execution
* vcpu 1: link_external_spt() to update 2M secure EPT entry
* vcpu 1: atomically update 2M level EPT entry to be present (unfreeze)
* vcpu 2: resume guest execution
Depending on vcpu 1 state, vcpu 2 may result in EPT violation again or make progress on guest execution

---
TDX MMU Prep v3:
- Rename mirrored->external (Paolo)
- Better comment on logic that bugs if doing tdp_mmu_set_spte() on present PTE. (Paolo)
- Move zapping KVM_BUG_ON() to proper patch
- Use spte_to_child_sp() (Paolo)
- Drop unnessary comment in __tdp_mmu_set_spte_atomic() (Paolo)
- Rename pfn->pfn_for_gfn to match remove_external_pte in next patch.
- Rename REMOVED_SPTE to FROZEN_SPTE (Paolo)

TDX MMU Prep v2:
- Split from "KVM: x86/tdp_mmu: Support TDX private mapping for TDP MMU"
- Rename x86_ops from "private" to "reflect"
- In response to "sp->mirrored_spt" rename helpers to "mirrored"
- Drop unused old_pfn and new_pfn in handle_changed_spte()
- Drop redundant is_shadow_present_pte() check in __tdp_mmu_set_spte_atomic
- Adjust some warnings and KVM_BUG_ONs

---
  arch/x86/include/asm/kvm-x86-ops.h |  2 +
  arch/x86/include/asm/kvm_host.h    |  7 +++
  arch/x86/kvm/mmu/tdp_mmu.c         | 98 ++++++++++++++++++++++++++----
  3 files changed, 94 insertions(+), 13 deletions(-)
arch/x86/include/asm/kvm-x86-ops.h +2 -0
index 566d19b02483..3ef19fcb5e42 100644
@@ -95,6 +95,8 @@ KVM_X86_OP_OPTIONAL_RET0(set_tss_addr)
  KVM_X86_OP_OPTIONAL_RET0(set_identity_map_addr)
  KVM_X86_OP_OPTIONAL_RET0(get_mt_mask)
  KVM_X86_OP(load_mmu_pgd)
+KVM_X86_OP_OPTIONAL(link_external_spt)
+KVM_X86_OP_OPTIONAL(set_external_spte)
  KVM_X86_OP(has_wbinvd_exit)
  KVM_X86_OP(get_l2_tsc_offset)
  KVM_X86_OP(get_l2_tsc_multiplier)
arch/x86/include/asm/kvm_host.h +0 -0
index d67e88a69fc4..12ff04135a0e 100644

arch/x86/include/asm/kvm_host.h struct kvm_x86_ops { · line 1741

Quoted 4 lines
arch/x86/include/asm/kvm_host.h +1 -0
@@ -1738,6 +1738,13 @@ struct kvm_x86_ops {
  	void (*load_mmu_pgd)(struct kvm_vcpu *vcpu, hpa_t root_hpa,
  			     int root_level);
 
+	/* Update external mapping with page table link */

Nit: Add "." at the end of the sentence.

arch/x86/include/asm/kvm_host.h struct kvm_x86_ops { · line 1744

Quoted 3 lines
arch/x86/include/asm/kvm_host.h +3 -0
+	int (*link_external_spt)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				void *mirrored_spt);
+	/* Update the external page table from spte getting set */

Ditto.

Quoted 11 lines
arch/x86/include/asm/kvm_host.h +3 -0
+	int (*set_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				 kvm_pfn_t pfn_for_gfn);
+
  	bool (*has_wbinvd_exit)(void);
 
  	u64 (*get_l2_tsc_offset)(struct kvm_vcpu *vcpu);
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
index 582e5a045bb7..bc1ad127046d 100644

arch/x86/kvm/mmu/tdp_mmu.c void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool mirror) · line 247

Quoted 5 lines
arch/x86/kvm/mmu/tdp_mmu.c +1 -1
@@ -244,7 +244,7 @@ void kvm_tdp_mmu_alloc_root(struct kvm_vcpu *vcpu, bool mirror)
  	struct kvm_mmu_page *root;
 
  	if (mirror)
-		role.is_mirror = 1;
+		role.is_mirror = true;

If this is needed, it's better to be done in "KVM: x86/tdp_mmu: Support mirror root for TDP MMU".

Quoted 7 lines
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
  	/*
  	 * Check for an existing root before acquiring the pages lock to avoid
@@ -440,6 +440,59 @@ static void handle_removed_pt(struct kvm *kvm, tdp_ptep_t pt, bool shared)
  	call_rcu(&sp->rcu_head, tdp_mmu_free_sp_rcu_callback);
  }
 

[...]

#30 lore raw

Re: [PATCH v3 15/17] KVM: x86/tdp_mmu: Propagate building mirror page tables

Quoted 4 lines · Binbin Wu
Quoted 2 lines

-               role.is_mirror = 1;
+               role.is_mirror = true;

If this is needed, it's better to be done in "KVM: x86/tdp_mmu: Support mirror root for TDP MMU".

Whoops, and agreed on the textual issues. Thanks.

#31 lore raw

PATCH v3 16/17 KVM: x86/tdp_mmu: Propagate tearing down mirror page tables

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Integrate hooks for mirroring page table operations for cases where TDX will zap PTEs or free page tables.

Like other Coco technologies, TDX has the concept of private and shared memory. For TDX the private and shared mappings are managed on separate EPT roots. The private half is managed indirectly though calls into a protected runtime environment called the TDX module, where the shared half is managed within KVM in normal page tables.

Since calls into the TDX module are relatively slow, walking private page tables by making calls into the TDX module would not be efficient. Because of this, previous changes have taught the TDP MMU to keep a mirror root, which is separate, unmapped TDP root that private operations can be directed to. Currently this root is disconnected from the guest. Now add plumbing to propagate changes to the "external" page tables being mirrored. Just create the x86_ops for now, leave plumbing the operations into the TDX module for future patches.

Add two operations for tearing down page tables, one for freeing page tables (free_external_spt) and one for zapping PTEs (remove_external_spte). Define them such that remove_external_spte will perform a TLB flush as well. (in TDX terms "ensure there are no active translations").

TDX MMU support will exclude certain MMU operations, so only plug in the mirroring x86 ops where they will be needed. For zapping/freeing, only hook tdp_mmu_iter_set_spte() which is use used for mapping and linking PTs. Don't bother hooking tdp_mmu_set_spte_atomic() as it is only used for zapping PTEs in operations unsupported by TDX: zapping collapsible PTEs and kvm_mmu_zap_all_fast().

In previous changes to address races around concurrent populating using tdp_mmu_set_spte_atomic(), a solution was introduced to temporarily set REMOVED_SPTE in the mirrored page tables while performing the external operations. Such a solution is not needed for the tear down paths in TDX as these will always be performed with the mmu_lock held for write. Sprinkle some KVM_BUG_ON()s to reflect this.

---
TDX MMU Prep v3:
- Rename mirrored->external (Paolo)
- Drop new_spte arg from reflect_removed_spte() (Paolo)
- ...and drop was_present and is_present bools (Paolo)
- Use base_gfn instead of sp->gfn (Paolo)
- Better comment on logic that bugs if doing tdp_mmu_set_spte() on present PTE. (Paolo)
- Move comment around KVM_BUG_ON() in __tdp_mmu_set_spte_atomic() to this patch, and add better comment. (Paolo)
- In remove_external_spte(), remove was_leaf bool, skip duplicates present check and add comment.
- Rename REMOVED_SPTE to FROZEN_SPTE (Paolo)

TDX MMU Prep v2:
- Split from "KVM: x86/tdp_mmu: Support TDX private mapping for TDP MMU"
- Rename x86_ops from "private" to "reflect"
- In response to "sp->mirrored_spt" rename helpers to "mirrored"
- Remove unused present mirroring support in tdp_mmu_set_spte()
- Merge reflect_zap_spte() into reflect_remove_spte()
- Move mirror zapping logic out of handle_changed_spte()
- Add some KVM_BUG_ONs

---
 arch/x86/include/asm/kvm-x86-ops.h |  2 ++
 arch/x86/include/asm/kvm_host.h    |  8 +++++
 arch/x86/kvm/mmu/tdp_mmu.c         | 51 +++++++++++++++++++++++++++++-
 3 files changed, 60 insertions(+), 1 deletion(-)
arch/x86/include/asm/kvm-x86-ops.h +2 -0
index 3ef19fcb5e42..18a83b211c90 100644
@@ -97,6 +97,8 @@ KVM_X86_OP_OPTIONAL_RET0(get_mt_mask)
 KVM_X86_OP(load_mmu_pgd)
 KVM_X86_OP_OPTIONAL(link_external_spt)
 KVM_X86_OP_OPTIONAL(set_external_spte)
+KVM_X86_OP_OPTIONAL(free_external_spt)
+KVM_X86_OP_OPTIONAL(remove_external_spte)
 KVM_X86_OP(has_wbinvd_exit)
 KVM_X86_OP(get_l2_tsc_offset)
 KVM_X86_OP(get_l2_tsc_multiplier)
arch/x86/include/asm/kvm_host.h +8 -0
index 12ff04135a0e..dca623ffa903 100644
@@ -1745,6 +1745,14 @@ struct kvm_x86_ops {
 	int (*set_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
 				 kvm_pfn_t pfn_for_gfn);
 
+	/* Update external page tables for page table about to be freed */
+	int (*free_external_spt)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				 void *external_spt);
+
+	/* Update external page table from spte getting removed, and flush TLB */
+	int (*remove_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				    kvm_pfn_t pfn_for_gfn);
+
 	bool (*has_wbinvd_exit)(void);
 
 	u64 (*get_l2_tsc_offset)(struct kvm_vcpu *vcpu);
arch/x86/kvm/mmu/tdp_mmu.c +50 -1

76 lines not on this page · raw message

Signature · 1 line
2.34.1
#32 lore raw

Re: [PATCH v3 16/17] KVM: x86/tdp_mmu: Propagate tearing down mirror page tables

Quoted 20 lines · Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Integrate hooks for mirroring page table operations for cases where TDX will zap PTEs or free page tables.

Like other Coco technologies, TDX has the concept of private and shared memory. For TDX the private and shared mappings are managed on separate EPT roots. The private half is managed indirectly though calls into a protected runtime environment called the TDX module, where the shared half is managed within KVM in normal page tables.

Since calls into the TDX module are relatively slow, walking private page tables by making calls into the TDX module would not be efficient. Because of this, previous changes have taught the TDP MMU to keep a mirror root, which is separate, unmapped TDP root that private operations can be directed to. Currently this root is disconnected from the guest. Now add plumbing to propagate changes to the "external" page tables being mirrored. Just create the x86_ops for now, leave plumbing the operations into the TDX module for future patches.

Quoted 8 lines

Add two operations for tearing down page tables, one for freeing page tables (free_external_spt) and one for zapping PTEs (remove_external_spte). Define them such that remove_external_spte will perform a TLB flush as well. (in TDX terms "ensure there are no active translations").

TDX MMU support will exclude certain MMU operations, so only plug in the mirroring x86 ops where they will be needed. For zapping/freeing, only hook tdp_mmu_iter_set_spte() which is use used for mapping and linking

                                        ^ extra "use"

Also, this sentence is a bit confusing about "used for mapping and linking".

Quoted 7 lines

PTs. Don't bother hooking tdp_mmu_set_spte_atomic() as it is only used for zapping PTEs in operations unsupported by TDX: zapping collapsible PTEs and kvm_mmu_zap_all_fast().

In previous changes to address races around concurrent populating using tdp_mmu_set_spte_atomic(), a solution was introduced to temporarily set REMOVED_SPTE in the mirrored page tables while performing the external

  ^
 FROZEN_SPTE

Quoted 57 lines

operations. Such a solution is not needed for the tear down paths in TDX as these will always be performed with the mmu_lock held for write. Sprinkle some KVM_BUG_ON()s to reflect this.

---
TDX MMU Prep v3:
- Rename mirrored->external (Paolo)
- Drop new_spte arg from reflect_removed_spte() (Paolo)
- ...and drop was_present and is_present bools (Paolo)
- Use base_gfn instead of sp->gfn (Paolo)
- Better comment on logic that bugs if doing tdp_mmu_set_spte() on present PTE. (Paolo)
- Move comment around KVM_BUG_ON() in __tdp_mmu_set_spte_atomic() to this patch, and add better comment. (Paolo)
- In remove_external_spte(), remove was_leaf bool, skip duplicates present check and add comment.
- Rename REMOVED_SPTE to FROZEN_SPTE (Paolo)

TDX MMU Prep v2:
- Split from "KVM: x86/tdp_mmu: Support TDX private mapping for TDP MMU"
- Rename x86_ops from "private" to "reflect"
- In response to "sp->mirrored_spt" rename helpers to "mirrored"
- Remove unused present mirroring support in tdp_mmu_set_spte()
- Merge reflect_zap_spte() into reflect_remove_spte()
- Move mirror zapping logic out of handle_changed_spte()
- Add some KVM_BUG_ONs

---
  arch/x86/include/asm/kvm-x86-ops.h |  2 ++
  arch/x86/include/asm/kvm_host.h    |  8 +++++
  arch/x86/kvm/mmu/tdp_mmu.c         | 51 +++++++++++++++++++++++++++++-
  3 files changed, 60 insertions(+), 1 deletion(-)
arch/x86/include/asm/kvm-x86-ops.h +2 -0
index 3ef19fcb5e42..18a83b211c90 100644
@@ -97,6 +97,8 @@ KVM_X86_OP_OPTIONAL_RET0(get_mt_mask)
  KVM_X86_OP(load_mmu_pgd)
  KVM_X86_OP_OPTIONAL(link_external_spt)
  KVM_X86_OP_OPTIONAL(set_external_spte)
+KVM_X86_OP_OPTIONAL(free_external_spt)
+KVM_X86_OP_OPTIONAL(remove_external_spte)
  KVM_X86_OP(has_wbinvd_exit)
  KVM_X86_OP(get_l2_tsc_offset)
  KVM_X86_OP(get_l2_tsc_multiplier)
arch/x86/include/asm/kvm_host.h +0 -0
index 12ff04135a0e..dca623ffa903 100644

arch/x86/include/asm/kvm_host.h struct kvm_x86_ops { · line 1748

Quoted 4 lines
arch/x86/include/asm/kvm_host.h +1 -0
@@ -1745,6 +1745,14 @@ struct kvm_x86_ops {
  	int (*set_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
  				 kvm_pfn_t pfn_for_gfn);
 
+	/* Update external page tables for page table about to be freed */

Nit: Add "." at the end of the sentence.

arch/x86/include/asm/kvm_host.h struct kvm_x86_ops { · line 1752

Quoted 4 lines
arch/x86/include/asm/kvm_host.h +4 -0
+	int (*free_external_spt)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				 void *external_spt);
+
+	/* Update external page table from spte getting removed, and flush TLB */

Ditto

Quoted 6 lines
arch/x86/include/asm/kvm_host.h +3 -0
+	int (*remove_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
+				    kvm_pfn_t pfn_for_gfn);
+
  	bool (*has_wbinvd_exit)(void);
 
  	u64 (*get_l2_tsc_offset)(struct kvm_vcpu *vcpu);

[...]

#33 lore raw

Re: [PATCH v3 16/17] KVM: x86/tdp_mmu: Propagate tearing down mirror page tables

Quoted 7 lines · Binbin Wu
Quoted 3 lines

TDX MMU support will exclude certain MMU operations, so only plug in the mirroring x86 ops where they will be needed. For zapping/freeing, only hook tdp_mmu_iter_set_spte() which is use used for mapping and linking

                                         ^ extra "use"

Also, this sentence is a bit confusing about "used for mapping and linking".

Yes. Is this more clear? "...tdp_mmu_iter_set_spte(), which is use used for setting leaf PTEs and linking non-leaf PTEs."

Quoted 10 lines
Quoted 8 lines

PTs. Don't bother hooking tdp_mmu_set_spte_atomic() as it is only used for zapping PTEs in operations unsupported by TDX: zapping collapsible PTEs and kvm_mmu_zap_all_fast().

In previous changes to address races around concurrent populating using tdp_mmu_set_spte_atomic(), a solution was introduced to temporarily set REMOVED_SPTE in the mirrored page tables while performing the external

   ^
  FROZEN_SPTE

Oops. And agreed on the other nits. Thanks.

#34 lore raw

PATCH v3 17/17 KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

In reply to Rick Edgecombe

From: Isaku Yamahata <[email protected]>

Rename kvm_tdp_mmu_invalidate_all_roots() to kvm_tdp_mmu_invalidate_roots(), and make it enum kvm_tdp_mmu_root_types as an argument.

kvm_tdp_mmu_invalidate_roots() is called with different root types. For kvm_mmu_zap_all_fast() it only operates on shared roots. But when tearing down a VM it needs to invalidate all roots. Have the callers only invalidate the required roots instead of all roots.

Within kvm_tdp_mmu_invalidate_roots(), respect the root type passed by checking the root type in root iterator.

---
TDX MMU Prep v3:
- Use root enum instead of process enum (Paolo)
- Squash with "KVM: x86/tdp_mmu: Invalidate correct roots" (Paolo)
- Update comment in kvm_mmu_zap_all_fast() (Paolo)
- Add warning for attempting to invalidate invalid roots (Paolo)

TDX MMU Prep v2:
- Use process enum instead of root

TDX MMU Prep:
- New patch

---
 arch/x86/kvm/mmu/mmu.c     |  9 +++++++--
 arch/x86/kvm/mmu/tdp_mmu.c | 15 +++++++++++++--
 arch/x86/kvm/mmu/tdp_mmu.h |  3 ++-
 3 files changed, 22 insertions(+), 5 deletions(-)
arch/x86/kvm/mmu/mmu.c +7 -2
index 287dcc2685e4..c5255aa62146 100644
@@ -6414,8 +6414,13 @@ static void kvm_mmu_zap_all_fast(struct kvm *kvm)
 	 * write and in the same critical section as making the reload request,
 	 * e.g. before kvm_zap_obsolete_pages() could drop mmu_lock and yield.
 	 */
-	if (tdp_mmu_enabled)
-		kvm_tdp_mmu_invalidate_all_roots(kvm);
+	if (tdp_mmu_enabled) {
+		/*
+		 * External page tables don't support fast zapping, therefore
+		 * their mirrors must be invalidated separately by the caller.
+		 */
+		kvm_tdp_mmu_invalidate_roots(kvm, KVM_DIRECT_ROOTS);
+	}
 
 	/*
 	 * Notify all vcpus to reload its shadow page table and flush TLB.
arch/x86/kvm/mmu/tdp_mmu.c +13 -2
index 630e6b6d4bf2..a1ab67a4f41f 100644
@@ -37,7 +37,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
 	 * for zapping and thus puts the TDP MMU's reference to each root, i.e.
 	 * ultimately frees all roots.
 	 */
-	kvm_tdp_mmu_invalidate_all_roots(kvm);
+	kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);
 	kvm_tdp_mmu_zap_invalidated_roots(kvm);
 
 	WARN_ON(atomic64_read(&kvm->arch.tdp_mmu_pages));
@@ -1110,10 +1110,18 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
  * Note, kvm_tdp_mmu_zap_invalidated_roots() is gifted the TDP MMU's reference.
  * See kvm_tdp_mmu_alloc_root().
  */
-void kvm_tdp_mmu_invalidate_all_roots(struct kvm *kvm)
+void kvm_tdp_mmu_invalidate_roots(struct kvm *kvm,
+				  enum kvm_tdp_mmu_root_types root_types)
 {
 	struct kvm_mmu_page *root;
 
+	/*
+	 * Invalidating invalid roots doesn't make sense, prevent developers from
+	 * having to think about it.
+	 */
+	if (WARN_ON_ONCE(root_types & KVM_INVALID_ROOTS))
+		root_types &= ~KVM_INVALID_ROOTS;
+
 	/*
 	 * mmu_lock must be held for write to ensure that a root doesn't become
 	 * invalid while there are active readers (invalidating a root while
@@ -1135,6 +1143,9 @@ void kvm_tdp_mmu_invalidate_all_roots(struct kvm *kvm)
 	 * or get/put references to roots.
 	 */
 	list_for_each_entry(root, &kvm->arch.tdp_mmu_roots, link) {
+		if (!tdp_mmu_root_match(root, root_types))
+			continue;
+
 		/*
 		 * Note, invalid roots can outlive a memslot update!  Invalid
 		 * roots must be *zapped* before the memslot update completes,
arch/x86/kvm/mmu/tdp_mmu.h +2 -1
index 2903f03a34be..56741d31048a 100644
@@ -66,7 +66,8 @@ static inline struct kvm_mmu_page *tdp_mmu_get_root(struct kvm_vcpu *vcpu,
 bool kvm_tdp_mmu_zap_leafs(struct kvm *kvm, gfn_t start, gfn_t end, bool flush);
 bool kvm_tdp_mmu_zap_sp(struct kvm *kvm, struct kvm_mmu_page *sp);
 void kvm_tdp_mmu_zap_all(struct kvm *kvm);
-void kvm_tdp_mmu_invalidate_all_roots(struct kvm *kvm);
+void kvm_tdp_mmu_invalidate_roots(struct kvm *kvm,
+				  enum kvm_tdp_mmu_root_types root_types);
 void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm);
 
 int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
Signature · 1 line
2.34.1
#35 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

arch/x86/kvm/mmu/tdp_mmu.c void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm) · line 40

Quoted 10 lines · Rick Edgecombe
arch/x86/kvm/mmu/tdp_mmu.c +1 -1
index 630e6b6d4bf2..a1ab67a4f41f 100644
@@ -37,7 +37,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
 	 * for zapping and thus puts the TDP MMU's reference to each root, i.e.
 	 * ultimately frees all roots.
 	 */
-	kvm_tdp_mmu_invalidate_all_roots(kvm);
+	kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);

all roots (mirror + direct) are invalidated here.

arch/x86/kvm/mmu/tdp_mmu.c void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm) · line 41

Quoted 1 line
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
 	kvm_tdp_mmu_zap_invalidated_roots(kvm);

kvm_tdp_mmu_zap_invalidated_roots() will zap invalidated mirror root with mmu_lock held for read, which should trigger KVM_BUG_ON() in __tdp_mmu_set_spte_atomic(), which assumes "atomic zapping don't operate on mirror roots".

But up to now, the KVM_BUG_ON() is not triggered because kvm_mmu_notifier_release() is called earlier than kvm_destroy_vm() (as in below call trace), and kvm_arch_flush_shadow_all() in kvm_mmu_notifier_release() has zapped all mirror SPTEs before kvm_mmu_uninit_vm() called in kvm_destroy_vm().

kvm_mmu_notifier_release
kvm_flush_shadow_all

    kvm_arch_flush_shadow_all
      static_call_cond(kvm_x86_flush_shadow_all_private)(kvm);
      kvm_mmu_zap_all  ==>hold mmu_lock for write
        kvm_tdp_mmu_zap_all ==>zap KVM_ALL_ROOTS with mmu_lock held for write

kvm_destroy_vm
kvm_arch_destroy_vm

    kvm_mmu_uninit_vm
      kvm_mmu_uninit_tdp_mmu
        kvm_tdp_mmu_invalidate_roots ==>invalid all KVM_VALID_ROOTS
        kvm_tdp_mmu_zap_invalidated_roots ==> zap all roots with mmu_lock held for read

A question is that kvm_mmu_notifier_release(), as a callback of primary MMU notifier, why does it zap mirrored tdp when all other callbacks are with KVM_FILTER_SHARED?

Could we just zap all KVM_DIRECT_ROOTS (valid | invalid) in kvm_mmu_notifier_release() and move mirrord tdp related stuffs from kvm_arch_flush_shadow_all() to kvm_mmu_uninit_tdp_mmu(), ensuring mmu_lock is held for write?

Quoted 2 lines
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
 	WARN_ON(atomic64_read(&kvm->arch.tdp_mmu_pages));
#36 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 13 lines · Yan Zhao
Quoted 11 lines · Rick Edgecombe
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
index 630e6b6d4bf2..a1ab67a4f41f 100644
@@ -37,7 +37,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
          * for zapping and thus puts the TDP MMU's reference to each root,

i.e.
         * ultimately frees all roots.
         */
-       kvm_tdp_mmu_invalidate_all_roots(kvm);
+       kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);

all roots (mirror + direct) are invalidated here.

Right.

Quoted 30 lines
Quoted 2 lines

        kvm_tdp_mmu_zap_invalidated_roots(kvm);

kvm_tdp_mmu_zap_invalidated_roots() will zap invalidated mirror root with mmu_lock held for read, which should trigger KVM_BUG_ON() in __tdp_mmu_set_spte_atomic(), which assumes "atomic zapping don't operate on mirror roots".

But up to now, the KVM_BUG_ON() is not triggered because kvm_mmu_notifier_release() is called earlier than kvm_destroy_vm() (as in below
call trace), and kvm_arch_flush_shadow_all() in kvm_mmu_notifier_release() has zapped all mirror SPTEs before kvm_mmu_uninit_vm() called in kvm_destroy_vm().

kvm_mmu_notifier_release
  kvm_flush_shadow_all
    kvm_arch_flush_shadow_all static_call_cond(kvm_x86_flush_shadow_all_private)(kvm); kvm_mmu_zap_all  ==>hold mmu_lock for write kvm_tdp_mmu_zap_all ==>zap KVM_ALL_ROOTS with mmu_lock held for write

kvm_destroy_vm
  kvm_arch_destroy_vm
    kvm_mmu_uninit_vm
      kvm_mmu_uninit_tdp_mmu
        kvm_tdp_mmu_invalidate_roots ==>invalid all KVM_VALID_ROOTS kvm_tdp_mmu_zap_invalidated_roots ==> zap all roots with mmu_lock held for read

Quoted 8 lines

A question is that kvm_mmu_notifier_release(), as a callback of primary MMU notifier, why does it zap mirrored tdp when all other callbacks are with KVM_FILTER_SHARED?

Could we just zap all KVM_DIRECT_ROOTS (valid | invalid) in kvm_mmu_notifier_release() and move mirrord tdp related stuffs from kvm_arch_flush_shadow_all() to kvm_mmu_uninit_tdp_mmu(), ensuring mmu_lock is held for write?

Sigh, thanks for flagging this. I agree it seems weird to free private memory from an MMU notifier callback. I also found this old thread where Sean NAKed the current approach (free hkid in mmu release):
https://lore.kernel.org/kvm/[email protected]/#t

One challenge is that flush_shadow_all_private() needs to be done before kvm_destroy_vcpus(), where it gets into tdx_vcpu_free(). So kvm_mmu_uninit_vm() is too late. Perhaps this is why it was shoved into mmu notifier release (which happens long before as you noted). Isaku, do you recall any other reasons?

But static_call_cond(kvm_x86_vm_destroy) happens before kvm_destroy_vcpus, so we could maybe actually just do the tdx_mmu_release_hkid() part there. Then drop the flush_shadow_all_private x86 op. See the (not thoroughly checked) diff at the bottom of this mail.

But most of what is being discussed is in future patches where it starts to get into the TDX module interaction. So I wonder if we should drop this patch 17 from "part 1" and include it with the next series so it can all be considered together.

arch/x86/include/asm/kvm-x86- renamed from arch/x86/include/asm/kvm-x86-ops.h +0 -0

ops.h
index 2adf36b74910..3927731aa947 100644

arch/x86/include/asm/kvm-x86-ops.h +0 -1
@@ -23,7 +23,6 @@ KVM_X86_OP(has_emulated_msr)
 KVM_X86_OP(vcpu_after_set_cpuid)
 KVM_X86_OP_OPTIONAL(vm_enable_cap)
 KVM_X86_OP(vm_init)
-KVM_X86_OP_OPTIONAL(flush_shadow_all_private)
 KVM_X86_OP_OPTIONAL(vm_destroy)
 KVM_X86_OP_OPTIONAL(vm_free)
 KVM_X86_OP_OPTIONAL_RET0(vcpu_precreate)
arch/x86/include/asm/kvm_host.h +0 -1
index 8a72e5873808..8b2b79b39d0f 100644
@@ -1647,7 +1647,6 @@ struct kvm_x86_ops {
        unsigned int vm_size;
        int (*vm_enable_cap)(struct kvm *kvm, struct kvm_enable_cap *cap);
        int (*vm_init)(struct kvm *kvm);
-       void (*flush_shadow_all_private)(struct kvm *kvm);
        void (*vm_destroy)(struct kvm *kvm);
        void (*vm_free)(struct kvm *kvm);
 
arch/x86/kvm/mmu/mmu.c +1 -8
index e1299eb03e63..4deeeac14324 100644
@@ -6446,7 +6446,7 @@ static void kvm_mmu_zap_all_fast(struct kvm *kvm)
         * lead to use-after-free.
         */
        if (tdp_mmu_enabled)
-               kvm_tdp_mmu_zap_invalidated_roots(kvm);
+               kvm_tdp_mmu_zap_invalidated_roots(kvm, true);
 }
 
 static bool kvm_has_zapped_obsolete_pages(struct kvm *kvm)
@@ -6977,13 +6977,6 @@ static void kvm_mmu_zap_all(struct kvm *kvm)
 
 void kvm_arch_flush_shadow_all(struct kvm *kvm)
 {
-       /*
-        * kvm_mmu_zap_all() zaps both private and shared page tables.  Before
-        * tearing down private page tables, TDX requires some TD resources to
-        * be destroyed (i.e. keyID must have been reclaimed, etc).  Invoke
-        * kvm_x86_flush_shadow_all_private() for this.
-        */
-       static_call_cond(kvm_x86_flush_shadow_all_private)(kvm);
        kvm_mmu_zap_all(kvm);
 }
 
arch/x86/kvm/mmu/tdp_mmu.c +12 -6
index 68dfcdb46ab7..9e8b012aa8cc 100644
@@ -38,7 +38,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
         * ultimately frees all roots.
         */
        kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);
-       kvm_tdp_mmu_zap_invalidated_roots(kvm);
+       kvm_tdp_mmu_zap_invalidated_roots(kvm, false);
 
        WARN_ON(atomic64_read(&kvm->arch.tdp_mmu_pages));
        WARN_ON(!list_empty(&kvm->arch.tdp_mmu_roots));
@@ -1057,7 +1057,7 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
         * KVM_RUN is unreachable, i.e. no vCPUs will ever service the request.
         */
        lockdep_assert_held_write(&kvm->mmu_lock);
-       for_each_tdp_mmu_root_yield_safe(kvm, root)
+       __for_each_tdp_mmu_root_yield_safe(kvm, root, -1, KVM_DIRECT_ROOTS)
                tdp_mmu_zap_root(kvm, root, false);
 }
 
@@ -1065,11 +1065,14 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
  * Zap all invalidated roots to ensure all SPTEs are dropped before the "fast
  * zap" completes.
  */
-void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
+void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm, bool shared)
 {
        struct kvm_mmu_page *root;
 
-       read_lock(&kvm->mmu_lock);
+       if (shared)
+               read_lock(&kvm->mmu_lock);
+       else
+               write_lock(&kvm->mmu_lock);
 
        for_each_tdp_mmu_root_yield_safe(kvm, root) {
                if (!root->tdp_mmu_scheduled_root_to_zap)
@@ -1087,7 +1090,7 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
                 * that may be zapped, as such entries are associated with the
                 * ASID on both VMX and SVM.
                 */
-               tdp_mmu_zap_root(kvm, root, true);
+               tdp_mmu_zap_root(kvm, root, shared);
 
                /*
                 * The referenced needs to be put *after* zapping the root, as
@@ -1097,7 +1100,10 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
                kvm_tdp_mmu_put_root(kvm, root);
        }
 
-       read_unlock(&kvm->mmu_lock);
+       if (shared)
+               read_unlock(&kvm->mmu_lock);
+       else
+               write_unlock(&kvm->mmu_lock);
 }
 
 /*
arch/x86/kvm/mmu/tdp_mmu.h +0 -0
index 56741d31048a..7927fa4a96e0 100644
@@ -68,7 +68,7 @@ bool kvm_tdp_mmu_zap_sp(struct kvm *kvm, struct kvm_mmu_page

*sp);
void kvm_tdp_mmu_zap_all(struct kvm *kvm); void kvm_tdp_mmu_invalidate_roots(struct kvm *kvm,

                                  enum kvm_tdp_mmu_root_types root_types);

-void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm); +void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm, bool shared);

int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);

arch/x86/kvm/vmx/main.c +3 -8
index b6828e35eb17..3f9bfcd3e152 100644
@@ -98,16 +98,12 @@ static int vt_vm_init(struct kvm *kvm)
        return vmx_vm_init(kvm);
 }
 
-static void vt_flush_shadow_all_private(struct kvm *kvm)
-{
-       if (is_td(kvm))
-               tdx_mmu_release_hkid(kvm);
-}
-
 static void vt_vm_destroy(struct kvm *kvm)
 {
-       if (is_td(kvm))
+       if (is_td(kvm)) {
+               tdx_mmu_release_hkid(kvm);
                return;
+       }
 
        vmx_vm_destroy(kvm);
 }
@@ -980,7 +976,6 @@ struct kvm_x86_ops vt_x86_ops __initdata = {
        .vm_size = sizeof(struct kvm_vmx),
        .vm_enable_cap = vt_vm_enable_cap,
        .vm_init = vt_vm_init,
-       .flush_shadow_all_private = vt_flush_shadow_all_private,
        .vm_destroy = vt_vm_destroy,
        .vm_free = vt_vm_free,
 
#37 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 65 lines · Rick P
Quoted 13 lines · Yan Zhao
Quoted 11 lines · Rick Edgecombe
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
index 630e6b6d4bf2..a1ab67a4f41f 100644
@@ -37,7 +37,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
          * for zapping and thus puts the TDP MMU's reference to each root,

i.e.
         * ultimately frees all roots.
         */
-       kvm_tdp_mmu_invalidate_all_roots(kvm);
+       kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);

all roots (mirror + direct) are invalidated here.

Right.

Quoted 30 lines
Quoted 2 lines

        kvm_tdp_mmu_zap_invalidated_roots(kvm);

kvm_tdp_mmu_zap_invalidated_roots() will zap invalidated mirror root with mmu_lock held for read, which should trigger KVM_BUG_ON() in __tdp_mmu_set_spte_atomic(), which assumes "atomic zapping don't operate on mirror roots".

But up to now, the KVM_BUG_ON() is not triggered because kvm_mmu_notifier_release() is called earlier than kvm_destroy_vm() (as in below
call trace), and kvm_arch_flush_shadow_all() in kvm_mmu_notifier_release() has zapped all mirror SPTEs before kvm_mmu_uninit_vm() called in kvm_destroy_vm().

kvm_mmu_notifier_release
  kvm_flush_shadow_all
    kvm_arch_flush_shadow_all static_call_cond(kvm_x86_flush_shadow_all_private)(kvm); kvm_mmu_zap_all  ==>hold mmu_lock for write kvm_tdp_mmu_zap_all ==>zap KVM_ALL_ROOTS with mmu_lock held for write

kvm_destroy_vm
  kvm_arch_destroy_vm
    kvm_mmu_uninit_vm
      kvm_mmu_uninit_tdp_mmu
        kvm_tdp_mmu_invalidate_roots ==>invalid all KVM_VALID_ROOTS kvm_tdp_mmu_zap_invalidated_roots ==> zap all roots with mmu_lock held for read

Quoted 8 lines

A question is that kvm_mmu_notifier_release(), as a callback of primary MMU notifier, why does it zap mirrored tdp when all other callbacks are with KVM_FILTER_SHARED?

Could we just zap all KVM_DIRECT_ROOTS (valid | invalid) in kvm_mmu_notifier_release() and move mirrord tdp related stuffs from kvm_arch_flush_shadow_all() to kvm_mmu_uninit_tdp_mmu(), ensuring mmu_lock is held for write?

Sigh, thanks for flagging this. I agree it seems weird to free private memory from an MMU notifier callback. I also found this old thread where Sean NAKed the current approach (free hkid in mmu release):
https://lore.kernel.org/kvm/[email protected]/#t

One challenge is that flush_shadow_all_private() needs to be done before kvm_destroy_vcpus(), where it gets into tdx_vcpu_free(). So kvm_mmu_uninit_vm() is too late. Perhaps this is why it was shoved into mmu notifier release (which happens long before as you noted). Isaku, do you recall any other reasons?

Quoted 4 lines

But static_call_cond(kvm_x86_vm_destroy) happens before kvm_destroy_vcpus, so we could maybe actually just do the tdx_mmu_release_hkid() part there. Then drop the flush_shadow_all_private x86 op. See the (not thoroughly checked) diff at the bottom of this mail.

It looks good to me.

Quoted 72 lines

But most of what is being discussed is in future patches where it starts to get into the TDX module interaction. So I wonder if we should drop this patch 17 from "part 1" and include it with the next series so it can all be considered together.

arch/x86/include/asm/kvm-x86- renamed from arch/x86/include/asm/kvm-x86-ops.h +0 -0

ops.h
index 2adf36b74910..3927731aa947 100644

arch/x86/include/asm/kvm-x86-ops.h +0 -1
@@ -23,7 +23,6 @@ KVM_X86_OP(has_emulated_msr)
 KVM_X86_OP(vcpu_after_set_cpuid)
 KVM_X86_OP_OPTIONAL(vm_enable_cap)
 KVM_X86_OP(vm_init)
-KVM_X86_OP_OPTIONAL(flush_shadow_all_private)
 KVM_X86_OP_OPTIONAL(vm_destroy)
 KVM_X86_OP_OPTIONAL(vm_free)
 KVM_X86_OP_OPTIONAL_RET0(vcpu_precreate)
arch/x86/include/asm/kvm_host.h +0 -1
index 8a72e5873808..8b2b79b39d0f 100644
@@ -1647,7 +1647,6 @@ struct kvm_x86_ops {
        unsigned int vm_size;
        int (*vm_enable_cap)(struct kvm *kvm, struct kvm_enable_cap *cap);
        int (*vm_init)(struct kvm *kvm);
-       void (*flush_shadow_all_private)(struct kvm *kvm);
        void (*vm_destroy)(struct kvm *kvm);
        void (*vm_free)(struct kvm *kvm);
 
arch/x86/kvm/mmu/mmu.c +1 -8
index e1299eb03e63..4deeeac14324 100644
@@ -6446,7 +6446,7 @@ static void kvm_mmu_zap_all_fast(struct kvm *kvm)
         * lead to use-after-free.
         */
        if (tdp_mmu_enabled)
-               kvm_tdp_mmu_zap_invalidated_roots(kvm);
+               kvm_tdp_mmu_zap_invalidated_roots(kvm, true);
 }
 
 static bool kvm_has_zapped_obsolete_pages(struct kvm *kvm)
@@ -6977,13 +6977,6 @@ static void kvm_mmu_zap_all(struct kvm *kvm)
 
 void kvm_arch_flush_shadow_all(struct kvm *kvm)
 {
-       /*
-        * kvm_mmu_zap_all() zaps both private and shared page tables.  Before
-        * tearing down private page tables, TDX requires some TD resources to
-        * be destroyed (i.e. keyID must have been reclaimed, etc).  Invoke
-        * kvm_x86_flush_shadow_all_private() for this.
-        */
-       static_call_cond(kvm_x86_flush_shadow_all_private)(kvm);
        kvm_mmu_zap_all(kvm);
 }
 
arch/x86/kvm/mmu/tdp_mmu.c +1 -1
index 68dfcdb46ab7..9e8b012aa8cc 100644
@@ -38,7 +38,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
         * ultimately frees all roots.
         */
        kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);
-       kvm_tdp_mmu_zap_invalidated_roots(kvm);
+       kvm_tdp_mmu_zap_invalidated_roots(kvm, false);
 
        WARN_ON(atomic64_read(&kvm->arch.tdp_mmu_pages));
        WARN_ON(!list_empty(&kvm->arch.tdp_mmu_roots));

arch/x86/kvm/mmu/tdp_mmu.c void kvm_tdp_mmu_zap_all(struct kvm *kvm) · line 1060

Quoted 5 lines
arch/x86/kvm/mmu/tdp_mmu.c +1 -1
@@ -1057,7 +1057,7 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
         * KVM_RUN is unreachable, i.e. no vCPUs will ever service the request.
         */
        lockdep_assert_held_write(&kvm->mmu_lock);
-       for_each_tdp_mmu_root_yield_safe(kvm, root)
+       __for_each_tdp_mmu_root_yield_safe(kvm, root, -1, KVM_DIRECT_ROOTS)

nit: update the comment of kvm_tdp_mmu_zap_all() and explain why it's KVM_DIRECT_ROOTS, not KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS.

Quoted 88 lines
arch/x86/kvm/mmu/tdp_mmu.c +10 -4
                tdp_mmu_zap_root(kvm, root, false);
 }
 
@@ -1065,11 +1065,14 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
  * Zap all invalidated roots to ensure all SPTEs are dropped before the "fast
  * zap" completes.
  */
-void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
+void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm, bool shared)
 {
        struct kvm_mmu_page *root;
 
-       read_lock(&kvm->mmu_lock);
+       if (shared)
+               read_lock(&kvm->mmu_lock);
+       else
+               write_lock(&kvm->mmu_lock);
 
        for_each_tdp_mmu_root_yield_safe(kvm, root) {
                if (!root->tdp_mmu_scheduled_root_to_zap)
@@ -1087,7 +1090,7 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
                 * that may be zapped, as such entries are associated with the
                 * ASID on both VMX and SVM.
                 */
-               tdp_mmu_zap_root(kvm, root, true);
+               tdp_mmu_zap_root(kvm, root, shared);
 
                /*
                 * The referenced needs to be put *after* zapping the root, as
@@ -1097,7 +1100,10 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
                kvm_tdp_mmu_put_root(kvm, root);
        }
 
-       read_unlock(&kvm->mmu_lock);
+       if (shared)
+               read_unlock(&kvm->mmu_lock);
+       else
+               write_unlock(&kvm->mmu_lock);
 }
 
 /*
arch/x86/kvm/mmu/tdp_mmu.h +1 -1
index 56741d31048a..7927fa4a96e0 100644
@@ -68,7 +68,7 @@ bool kvm_tdp_mmu_zap_sp(struct kvm *kvm, struct kvm_mmu_page
 *sp);
 void kvm_tdp_mmu_zap_all(struct kvm *kvm);
 void kvm_tdp_mmu_invalidate_roots(struct kvm *kvm,
                                  enum kvm_tdp_mmu_root_types root_types);
-void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm);
+void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm, bool shared);
 
 int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
arch/x86/kvm/vmx/main.c +3 -8
index b6828e35eb17..3f9bfcd3e152 100644
@@ -98,16 +98,12 @@ static int vt_vm_init(struct kvm *kvm)
        return vmx_vm_init(kvm);
 }
 
-static void vt_flush_shadow_all_private(struct kvm *kvm)
-{
-       if (is_td(kvm))
-               tdx_mmu_release_hkid(kvm);
-}
-
 static void vt_vm_destroy(struct kvm *kvm)
 {
-       if (is_td(kvm))
+       if (is_td(kvm)) {
+               tdx_mmu_release_hkid(kvm);
                return;
+       }
 
        vmx_vm_destroy(kvm);
 }
@@ -980,7 +976,6 @@ struct kvm_x86_ops vt_x86_ops __initdata = {
        .vm_size = sizeof(struct kvm_vmx),
        .vm_enable_cap = vt_vm_enable_cap,
        .vm_init = vt_vm_init,
-       .flush_shadow_all_private = vt_flush_shadow_all_private,
        .vm_destroy = vt_vm_destroy,
        .vm_free = vt_vm_free,
 
#38 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 8 lines · Yan Zhao
Quoted 7 lines
(quoted diff) +0 -0
@@ -1057,7 +1057,7 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
           * KVM_RUN is unreachable, i.e. no vCPUs will ever service the

request.
          */
         lockdep_assert_held_write(&kvm->mmu_lock);
-       for_each_tdp_mmu_root_yield_safe(kvm, root)
+       __for_each_tdp_mmu_root_yield_safe(kvm, root, -1, KVM_DIRECT_ROOTS)

nit: update the comment of kvm_tdp_mmu_zap_all()

Yea, good idea. It's definitely needs some explanation, considering the function is called "zap_all". A bit unfortunate actually.

Quoted 2 lines

and explain why it's
KVM_DIRECT_ROOTS, not KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS.

Explain why not to zap invalid mirrored roots?

#39 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 10 lines · Rick P
Quoted 8 lines · Yan Zhao
Quoted 7 lines
(quoted diff) +0 -0
@@ -1057,7 +1057,7 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
           * KVM_RUN is unreachable, i.e. no vCPUs will ever service the

request.
          */
         lockdep_assert_held_write(&kvm->mmu_lock);
-       for_each_tdp_mmu_root_yield_safe(kvm, root)
+       __for_each_tdp_mmu_root_yield_safe(kvm, root, -1, KVM_DIRECT_ROOTS)

nit: update the comment of kvm_tdp_mmu_zap_all()

Quoted 7 lines

Yea, good idea. It's definitely needs some explanation, considering the function is called "zap_all". A bit unfortunate actually.

Quoted 2 lines

and explain why it's
KVM_DIRECT_ROOTS, not KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS.

Explain why not to zap invalid mirrored roots?

No. Explain why not to zap invalid direct roots.
Just passing KVM_DIRECT_ROOTS will zap only valid direct roots, right? The original kvm_tdp_mmu_zap_all() "Zap all roots, including invalid roots". It might be better to explain why not to zap invalid direct roots here.

#40 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 5 lines · Yan Zhao
Quoted 1 line

Explain why not to zap invalid mirrored roots?

No. Explain why not to zap invalid direct roots.
Just passing KVM_DIRECT_ROOTS will zap only valid direct roots, right? The original kvm_tdp_mmu_zap_all() "Zap all roots, including invalid roots". It might be better to explain why not to zap invalid direct roots here.

Hmm, right. We don't actually have enum value to iterate all direct roots (valid and invalid). We could change tdp_mmu_root_match() to:

arch/x86/kvm/mmu/tdp_mmu.c +2 -2
index 8320b093fef9..bcd771a8835f 100644
@@ -98,8 +98,8 @@ static bool tdp_mmu_root_match(struct kvm_mmu_page *root,
        if (WARN_ON_ONCE(!(types & KVM_VALID_ROOTS)))
                return false;
 
-       if (root->role.invalid)
-               return types & KVM_INVALID_ROOTS;
+       if (root->role.invalid && !(types & KVM_INVALID_ROOTS))
+               return false;
        if (likely(!is_mirror_sp(root)))
                return types & KVM_DIRECT_ROOTS;
 

Maybe better than a comment...? Need to put the whole thing together and test it still.

#41 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 24 lines · Rick P
Quoted 5 lines · Yan Zhao
Quoted 1 line

Explain why not to zap invalid mirrored roots?

No. Explain why not to zap invalid direct roots.
Just passing KVM_DIRECT_ROOTS will zap only valid direct roots, right? The original kvm_tdp_mmu_zap_all() "Zap all roots, including invalid roots". It might be better to explain why not to zap invalid direct roots here.

Hmm, right. We don't actually have enum value to iterate all direct roots (valid and invalid). We could change tdp_mmu_root_match() to:

arch/x86/kvm/mmu/tdp_mmu.c +2 -2
index 8320b093fef9..bcd771a8835f 100644
@@ -98,8 +98,8 @@ static bool tdp_mmu_root_match(struct kvm_mmu_page *root,
        if (WARN_ON_ONCE(!(types & KVM_VALID_ROOTS)))
                return false;
 
-       if (root->role.invalid)
-               return types & KVM_INVALID_ROOTS;
+       if (root->role.invalid && !(types & KVM_INVALID_ROOTS))
+               return false;
        if (likely(!is_mirror_sp(root)))
                return types & KVM_DIRECT_ROOTS;
 
Quoted 2 lines

Maybe better than a comment...? Need to put the whole thing together and test it still.

Hmm, I think passing KVM_DIRECT_ROOTS only in kvm_tdp_mmu_zap_all() is ok, because kvm_mmu_uninit_tdp_mmu() will zap all invalidated roots eventually, though KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS matches more to the function name zap_all.

I'm not convinced that the change in tdp_mmu_root_match() above is needed. Once a root is marked invalid, it won't be restored later. Distinguishing between invalid direct roots and invalid mirror roots will only result in more unused roots lingering unnecessarily.

#42 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 8 lines · Yan Zhao
Quoted 3 lines

Maybe better than a comment...? Need to put the whole thing together and test it
still.

Hmm, I think passing KVM_DIRECT_ROOTS only in kvm_tdp_mmu_zap_all() is ok, because kvm_mmu_uninit_tdp_mmu() will zap all invalidated roots eventually, though KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS matches more to the function name zap_all.

Quoted 4 lines

I'm not convinced that the change in tdp_mmu_root_match() above is needed. Once a root is marked invalid, it won't be restored later. Distinguishing between invalid direct roots and invalid mirror roots will only result in more unused roots lingering unnecessarily.

The logic for direct roots is for normal VMs as well. In any case, we should not mix generic KVM MMU changes in with mirrored memory additions. So let's keep the existing direct root behavior the same.

#43 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 9 lines · Rick P
Quoted 8 lines · Yan Zhao
Quoted 3 lines

Maybe better than a comment...? Need to put the whole thing together and test it
still.

Hmm, I think passing KVM_DIRECT_ROOTS only in kvm_tdp_mmu_zap_all() is ok, because kvm_mmu_uninit_tdp_mmu() will zap all invalidated roots eventually, though KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS matches more to the function name zap_all.

Quoted 8 lines
Quoted 4 lines

I'm not convinced that the change in tdp_mmu_root_match() above is needed. Once a root is marked invalid, it won't be restored later. Distinguishing between invalid direct roots and invalid mirror roots will only result in more unused roots lingering unnecessarily.

The logic for direct roots is for normal VMs as well. In any case, we should not mix generic KVM MMU changes in with mirrored memory additions. So let's keep the existing direct root behavior the same.

To keep the existing direct root behavior the same, I think specifying KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS in kvm_tdp_mmu_zap_all() is enough.

No need to modify tdp_mmu_root_match() do distinguish between invalid direct roots and invalid mirror roots. As long as a root is invalid, guest is no longer affected by it and KVM will not use it any more. The last remaining operation to the invalid root is only zapping.

Distinguishing between invalid direct roots and invalid mirror roots would make the code to zap invalid roots unnecessarily complex, e.g.

kvm_tdp_mmu_zap_invalidated_roots() is called both in kvm_mmu_uninit_tdp_mmu() and kvm_mmu_zap_all_fast().

- When called in the former, both invalid direct and invalid mirror roots are required to zap;
- when called in the latter, only invalid direct roots are required to zap.

#44 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 2 lines · Yan Zhao

To keep the existing direct root behavior the same, I think specifying KVM_DIRECT_ROOTS | KVM_INVALID_ROOTS in kvm_tdp_mmu_zap_all() is enough.

Right.

Quoted 9 lines

No need to modify tdp_mmu_root_match() do distinguish between invalid direct roots and invalid mirror roots. As long as a root is invalid, guest is no longer
affected by it and KVM will not use it any more. The last remaining operation to the invalid root is only zapping.

Distinguishing between invalid direct roots and invalid mirror roots would make the code to zap invalid roots unnecessarily complex, e.g.

I'm not sure that it is more complicated. One requires a big comment to explain, and the other is self explanatory...

Quoted 7 lines

kvm_tdp_mmu_zap_invalidated_roots() is called both in kvm_mmu_uninit_tdp_mmu() and kvm_mmu_zap_all_fast().

- When called in the former, both invalid direct and invalid mirror roots are required to zap;
- when called in the latter, only invalid direct roots are required to zap.

It might help to put together a full fixup at this point. We have a couple diffs in this thread, and I'm not 100% which base patch we are talking about at this point.

#45 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

In reply to Edgecombe, Rick P
Quoted 60 lines · Rick P
Quoted 13 lines · Yan Zhao
Quoted 11 lines · Rick Edgecombe
arch/x86/kvm/mmu/tdp_mmu.c +0 -0
index 630e6b6d4bf2..a1ab67a4f41f 100644
@@ -37,7 +37,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
          * for zapping and thus puts the TDP MMU's reference to each root,

i.e.
         * ultimately frees all roots.
         */
-       kvm_tdp_mmu_invalidate_all_roots(kvm);
+       kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);

all roots (mirror + direct) are invalidated here.

Right.

Quoted 30 lines
Quoted 2 lines

        kvm_tdp_mmu_zap_invalidated_roots(kvm);

kvm_tdp_mmu_zap_invalidated_roots() will zap invalidated mirror root with mmu_lock held for read, which should trigger KVM_BUG_ON() in __tdp_mmu_set_spte_atomic(), which assumes "atomic zapping don't operate on mirror roots".

But up to now, the KVM_BUG_ON() is not triggered because kvm_mmu_notifier_release() is called earlier than kvm_destroy_vm() (as in below
call trace), and kvm_arch_flush_shadow_all() in kvm_mmu_notifier_release() has zapped all mirror SPTEs before kvm_mmu_uninit_vm() called in kvm_destroy_vm().

kvm_mmu_notifier_release
  kvm_flush_shadow_all
    kvm_arch_flush_shadow_all static_call_cond(kvm_x86_flush_shadow_all_private)(kvm); kvm_mmu_zap_all  ==>hold mmu_lock for write kvm_tdp_mmu_zap_all ==>zap KVM_ALL_ROOTS with mmu_lock held for write

kvm_destroy_vm
  kvm_arch_destroy_vm
    kvm_mmu_uninit_vm
      kvm_mmu_uninit_tdp_mmu
        kvm_tdp_mmu_invalidate_roots ==>invalid all KVM_VALID_ROOTS kvm_tdp_mmu_zap_invalidated_roots ==> zap all roots with mmu_lock held for read

Quoted 8 lines

A question is that kvm_mmu_notifier_release(), as a callback of primary MMU notifier, why does it zap mirrored tdp when all other callbacks are with KVM_FILTER_SHARED?

Could we just zap all KVM_DIRECT_ROOTS (valid | invalid) in kvm_mmu_notifier_release() and move mirrord tdp related stuffs from kvm_arch_flush_shadow_all() to kvm_mmu_uninit_tdp_mmu(), ensuring mmu_lock is held for write?

Sigh, thanks for flagging this. I agree it seems weird to free private memory from an MMU notifier callback. I also found this old thread where Sean NAKed the current approach (free hkid in mmu release):
https://lore.kernel.org/kvm/[email protected]/#t

Quoted 4 lines

One challenge is that flush_shadow_all_private() needs to be done before kvm_destroy_vcpus(), where it gets into tdx_vcpu_free(). So kvm_mmu_uninit_vm() is too late. Perhaps this is why it was shoved into mmu notifier release (which happens long before as you noted). Isaku, do you recall any other reasons?

Although it's a bit late, it's for record.

It's to optimize the destruction Secure-EPT. Free HKID early and destruct Secure-EPT by TDH.PHYMEM.PAGE.RECLAIM(). QEMU doesn't close any KVM file descriptors on exit. (gmem fd references KVM VM fd. so vm destruction happens after all gmem fds are closed. Closing gmem fd causes secure-EPT zapping befure releasing HKID.)

Because we're ignoring such optimization for now, we can simply defer releasing HKID following Seans's call.

Quoted 4 lines

But static_call_cond(kvm_x86_vm_destroy) happens before kvm_destroy_vcpus, so we could maybe actually just do the tdx_mmu_release_hkid() part there. Then drop the flush_shadow_all_private x86 op. See the (not thoroughly checked) diff at the bottom of this mail.

Yep, we can release HKID at vm destruction with potential too slow zapping of Secure-EPT. The following change basically looks good to me. (The callback for Secure-EPT can be simplified.)

Thanks,

Quoted 164 lines

But most of what is being discussed is in future patches where it starts to get into the TDX module interaction. So I wonder if we should drop this patch 17 from "part 1" and include it with the next series so it can all be considered together.

arch/x86/include/asm/kvm-x86- renamed from arch/x86/include/asm/kvm-x86-ops.h +0 -0

ops.h
index 2adf36b74910..3927731aa947 100644

arch/x86/include/asm/kvm-x86-ops.h +0 -1
@@ -23,7 +23,6 @@ KVM_X86_OP(has_emulated_msr)
 KVM_X86_OP(vcpu_after_set_cpuid)
 KVM_X86_OP_OPTIONAL(vm_enable_cap)
 KVM_X86_OP(vm_init)
-KVM_X86_OP_OPTIONAL(flush_shadow_all_private)
 KVM_X86_OP_OPTIONAL(vm_destroy)
 KVM_X86_OP_OPTIONAL(vm_free)
 KVM_X86_OP_OPTIONAL_RET0(vcpu_precreate)
arch/x86/include/asm/kvm_host.h +0 -1
index 8a72e5873808..8b2b79b39d0f 100644
@@ -1647,7 +1647,6 @@ struct kvm_x86_ops {
        unsigned int vm_size;
        int (*vm_enable_cap)(struct kvm *kvm, struct kvm_enable_cap *cap);
        int (*vm_init)(struct kvm *kvm);
-       void (*flush_shadow_all_private)(struct kvm *kvm);
        void (*vm_destroy)(struct kvm *kvm);
        void (*vm_free)(struct kvm *kvm);
 
arch/x86/kvm/mmu/mmu.c +1 -8
index e1299eb03e63..4deeeac14324 100644
@@ -6446,7 +6446,7 @@ static void kvm_mmu_zap_all_fast(struct kvm *kvm)
         * lead to use-after-free.
         */
        if (tdp_mmu_enabled)
-               kvm_tdp_mmu_zap_invalidated_roots(kvm);
+               kvm_tdp_mmu_zap_invalidated_roots(kvm, true);
 }
 
 static bool kvm_has_zapped_obsolete_pages(struct kvm *kvm)
@@ -6977,13 +6977,6 @@ static void kvm_mmu_zap_all(struct kvm *kvm)
 
 void kvm_arch_flush_shadow_all(struct kvm *kvm)
 {
-       /*
-        * kvm_mmu_zap_all() zaps both private and shared page tables.  Before
-        * tearing down private page tables, TDX requires some TD resources to
-        * be destroyed (i.e. keyID must have been reclaimed, etc).  Invoke
-        * kvm_x86_flush_shadow_all_private() for this.
-        */
-       static_call_cond(kvm_x86_flush_shadow_all_private)(kvm);
        kvm_mmu_zap_all(kvm);
 }
 
arch/x86/kvm/mmu/tdp_mmu.c +12 -6
index 68dfcdb46ab7..9e8b012aa8cc 100644
@@ -38,7 +38,7 @@ void kvm_mmu_uninit_tdp_mmu(struct kvm *kvm)
         * ultimately frees all roots.
         */
        kvm_tdp_mmu_invalidate_roots(kvm, KVM_VALID_ROOTS);
-       kvm_tdp_mmu_zap_invalidated_roots(kvm);
+       kvm_tdp_mmu_zap_invalidated_roots(kvm, false);
 
        WARN_ON(atomic64_read(&kvm->arch.tdp_mmu_pages));
        WARN_ON(!list_empty(&kvm->arch.tdp_mmu_roots));
@@ -1057,7 +1057,7 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
         * KVM_RUN is unreachable, i.e. no vCPUs will ever service the request.
         */
        lockdep_assert_held_write(&kvm->mmu_lock);
-       for_each_tdp_mmu_root_yield_safe(kvm, root)
+       __for_each_tdp_mmu_root_yield_safe(kvm, root, -1, KVM_DIRECT_ROOTS)
                tdp_mmu_zap_root(kvm, root, false);
 }
 
@@ -1065,11 +1065,14 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
  * Zap all invalidated roots to ensure all SPTEs are dropped before the "fast
  * zap" completes.
  */
-void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
+void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm, bool shared)
 {
        struct kvm_mmu_page *root;
 
-       read_lock(&kvm->mmu_lock);
+       if (shared)
+               read_lock(&kvm->mmu_lock);
+       else
+               write_lock(&kvm->mmu_lock);
 
        for_each_tdp_mmu_root_yield_safe(kvm, root) {
                if (!root->tdp_mmu_scheduled_root_to_zap)
@@ -1087,7 +1090,7 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
                 * that may be zapped, as such entries are associated with the
                 * ASID on both VMX and SVM.
                 */
-               tdp_mmu_zap_root(kvm, root, true);
+               tdp_mmu_zap_root(kvm, root, shared);
 
                /*
                 * The referenced needs to be put *after* zapping the root, as
@@ -1097,7 +1100,10 @@ void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
                kvm_tdp_mmu_put_root(kvm, root);
        }
 
-       read_unlock(&kvm->mmu_lock);
+       if (shared)
+               read_unlock(&kvm->mmu_lock);
+       else
+               write_unlock(&kvm->mmu_lock);
 }
 
 /*
arch/x86/kvm/mmu/tdp_mmu.h +1 -1
index 56741d31048a..7927fa4a96e0 100644
@@ -68,7 +68,7 @@ bool kvm_tdp_mmu_zap_sp(struct kvm *kvm, struct kvm_mmu_page
 *sp);
 void kvm_tdp_mmu_zap_all(struct kvm *kvm);
 void kvm_tdp_mmu_invalidate_roots(struct kvm *kvm,
                                  enum kvm_tdp_mmu_root_types root_types);
-void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm);
+void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm, bool shared);
 
 int kvm_tdp_mmu_map(struct kvm_vcpu *vcpu, struct kvm_page_fault *fault);
arch/x86/kvm/vmx/main.c +3 -8
index b6828e35eb17..3f9bfcd3e152 100644
@@ -98,16 +98,12 @@ static int vt_vm_init(struct kvm *kvm)
        return vmx_vm_init(kvm);
 }
 
-static void vt_flush_shadow_all_private(struct kvm *kvm)
-{
-       if (is_td(kvm))
-               tdx_mmu_release_hkid(kvm);
-}
-
 static void vt_vm_destroy(struct kvm *kvm)
 {
-       if (is_td(kvm))
+       if (is_td(kvm)) {
+               tdx_mmu_release_hkid(kvm);
                return;
+       }
 
        vmx_vm_destroy(kvm);
 }
@@ -980,7 +976,6 @@ struct kvm_x86_ops vt_x86_ops __initdata = {
        .vm_size = sizeof(struct kvm_vmx),
        .vm_enable_cap = vt_vm_enable_cap,
        .vm_init = vt_vm_init,
-       .flush_shadow_all_private = vt_flush_shadow_all_private,
        .vm_destroy = vt_vm_destroy,
        .vm_free = vt_vm_free,
 
Signature · 1 line
Isaku Yamahata <[email protected]>
#46 lore raw

Re: [PATCH v3 17/17] KVM: x86/tdp_mmu: Take root types for kvm_tdp_mmu_invalidate_all_roots()

Quoted 9 lines · Isaku Yamahata

Although it's a bit late, it's for record.

It's to optimize the destruction Secure-EPT.  Free HKID early and destruct Secure-EPT by TDH.PHYMEM.PAGE.RECLAIM().  QEMU doesn't close any KVM file descriptors on exit. (gmem fd references KVM VM fd. so vm destruction happens after all gmem fds are closed.  Closing gmem fd causes secure-EPT zapping befure
releasing HKID.)

Quoted 3 lines

Because we're ignoring such optimization for now, we can simply defer releasing
HKID following Seans's call.

Thanks for the background.

Quoted 13 lines
Quoted 9 lines

But static_call_cond(kvm_x86_vm_destroy) happens before kvm_destroy_vcpus, so we
could maybe actually just do the tdx_mmu_release_hkid() part there. Then drop
the flush_shadow_all_private x86 op. See the (not thoroughly checked) diff at
the bottom of this mail.

Yep, we can release HKID at vm destruction with potential too slow zapping of Secure-EPT.  The following change basically looks good to me. (The callback for Secure-EPT can be simplified.)