From: Yan Zhao <yan.y.zhao@intel.com>
To: Rick Edgecombe <rick.p.edgecombe@intel.com>
Cc: <kas@kernel.org>, <bp@alien8.de>, <chao.gao@intel.com>,
<dave.hansen@linux.intel.com>, <isaku.yamahata@intel.com>,
<kai.huang@intel.com>, <kvm@vger.kernel.org>,
<linux-coco@lists.linux.dev>, <linux-kernel@vger.kernel.org>,
<mingo@redhat.com>, <pbonzini@redhat.com>, <seanjc@google.com>,
<tglx@linutronix.de>, <x86@kernel.org>, <vannapurve@google.com>
Subject: Re: [PATCH v3 11/16] KVM: TDX: Add x86 ops for external spt cache
Date: Tue, 23 Sep 2025 15:03:27 +0800 [thread overview]
Message-ID: <aNJGP6lwO9WOqjfh@yzhao56-desk.sh.intel.com> (raw)
In-Reply-To: <20250918232224.2202592-12-rick.p.edgecombe@intel.com>
On Thu, Sep 18, 2025 at 04:22:19PM -0700, Rick Edgecombe wrote:
> Move mmu_external_spt_cache behind x86 ops.
>
> In the mirror/external MMU concept, the KVM MMU manages a non-active EPT
> tree for private memory (the mirror). The actual active EPT tree the
> private memory is protected inside the TDX module. Whenever the mirror EPT
> is changed, it needs to call out into one of a set of x86 opts that
> implement various update operation with TDX specific SEAMCALLs and other
> tricks. These implementations operate on the TDX S-EPT (the external).
>
> In reality these external operations are designed narrowly with respect to
> TDX particulars. On the surface, what TDX specific things are happening to
> fulfill these update operations are mostly hidden from the MMU, but there
> is one particular area of interest where some details leak through.
>
> The S-EPT needs pages to use for the S-EPT page tables. These page tables
> need to be allocated before taking the mmu lock, like all the rest. So the
> KVM MMU pre-allocates pages for TDX to use for the S-EPT in the same place
> where it pre-allocates the other page tables. It’s not too bad and fits
> nicely with the others.
>
> However, Dynamic PAMT will need even more pages for the same operations.
> Further, these pages will need to be handed to the arch/86 side which used
> them for DPAMT updates, which is hard for the existing KVM based cache.
> The details living in core MMU code start to add up.
>
> So in preparation to make it more complicated, move the external page
> table cache into TDX code by putting it behind some x86 ops. Have one for
> topping up and one for allocation. Don’t go so far to try to hide the
> existence of external page tables completely from the generic MMU, as they
> are currently stores in their mirror struct kvm_mmu_page and it’s quite
> handy.
>
> To plumb the memory cache operations through tdx.c, export some of
> the functions temporarily. This will be removed in future changes.
>
> Signed-off-by: Rick Edgecombe <rick.p.edgecombe@intel.com>
> ---
> v3:
> - New patch
> ---
> arch/x86/include/asm/kvm-x86-ops.h | 2 ++
> arch/x86/include/asm/kvm_host.h | 11 ++++++-----
> arch/x86/kvm/mmu/mmu.c | 4 +---
> arch/x86/kvm/mmu/mmu_internal.h | 2 +-
> arch/x86/kvm/vmx/tdx.c | 17 +++++++++++++++++
> arch/x86/kvm/vmx/tdx.h | 2 ++
> virt/kvm/kvm_main.c | 2 ++
> 7 files changed, 31 insertions(+), 9 deletions(-)
>
> diff --git a/arch/x86/include/asm/kvm-x86-ops.h b/arch/x86/include/asm/kvm-x86-ops.h
> index 62c3e4de3303..a4e4c1333224 100644
> --- a/arch/x86/include/asm/kvm-x86-ops.h
> +++ b/arch/x86/include/asm/kvm-x86-ops.h
> @@ -98,6 +98,8 @@ 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_OPTIONAL(alloc_external_fault_cache)
> +KVM_X86_OP_OPTIONAL(topup_external_fault_cache)
> KVM_X86_OP(has_wbinvd_exit)
> KVM_X86_OP(get_l2_tsc_offset)
> KVM_X86_OP(get_l2_tsc_multiplier)
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index cb86f3cca3e9..e4cf0f40c757 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -855,11 +855,6 @@ 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 external 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.
> @@ -1856,6 +1851,12 @@ struct kvm_x86_ops {
> int (*remove_external_spte)(struct kvm *kvm, gfn_t gfn, enum pg_level level,
> kvm_pfn_t pfn_for_gfn);
>
> + /* Allocation a pages from the external page cache. */
> + void *(*alloc_external_fault_cache)(struct kvm_vcpu *vcpu);
> +
> + /* Top up extra pages needed for faulting in external page tables. */
> + int (*topup_external_fault_cache)(struct kvm_vcpu *vcpu);
> +
> bool (*has_wbinvd_exit)(void);
>
> u64 (*get_l2_tsc_offset)(struct kvm_vcpu *vcpu);
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index 55335dbd70ce..b3feaee893b2 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -601,8 +601,7 @@ static int mmu_topup_memory_caches(struct kvm_vcpu *vcpu, bool maybe_indirect)
> 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);
> + r = kvm_x86_call(topup_external_fault_cache)(vcpu);
> if (r)
> return r;
> }
> @@ -625,7 +624,6 @@ 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);
Though pre-allocated pages are eventually freed in tdx_vcpu_free() in patch 13,
looks they are leaked in this patch.
BTW, why not invoke kvm_x86_call(free_external_fault_cache)(vcpu) here?
It looks more natural to free the remaining pre-allocated pages in
mmu_free_memory_caches(), which is invoked after kvm_mmu_unload(vcpu) while
tdx_vcpu_free() is before it.
> kvm_mmu_free_memory_cache(&vcpu->arch.mmu_page_header_cache);
> }
>
> diff --git a/arch/x86/kvm/mmu/mmu_internal.h b/arch/x86/kvm/mmu/mmu_internal.h
> index ed5c01df21ba..1fa94ab100be 100644
> --- a/arch/x86/kvm/mmu/mmu_internal.h
> +++ b/arch/x86/kvm/mmu/mmu_internal.h
> @@ -175,7 +175,7 @@ static inline void kvm_mmu_alloc_external_spt(struct kvm_vcpu *vcpu, struct kvm_
> * Therefore, KVM does not need to initialize or access external_spt.
> * KVM only interacts with sp->spt for private EPT operations.
> */
> - sp->external_spt = kvm_mmu_memory_cache_alloc(&vcpu->arch.mmu_external_spt_cache);
> + sp->external_spt = kvm_x86_call(alloc_external_fault_cache)(vcpu);
> }
>
> static inline gfn_t kvm_gfn_root_bits(const struct kvm *kvm, const struct kvm_mmu_page *root)
> diff --git a/arch/x86/kvm/vmx/tdx.c b/arch/x86/kvm/vmx/tdx.c
> index dd2be7bedd48..6c9e11be9705 100644
> --- a/arch/x86/kvm/vmx/tdx.c
> +++ b/arch/x86/kvm/vmx/tdx.c
> @@ -1590,6 +1590,21 @@ static void tdx_unpin(struct kvm *kvm, struct page *page)
> put_page(page);
> }
>
> +static void *tdx_alloc_external_fault_cache(struct kvm_vcpu *vcpu)
> +{
> + struct vcpu_tdx *tdx = to_tdx(vcpu);
> +
> + return kvm_mmu_memory_cache_alloc(&tdx->mmu_external_spt_cache);
> +}
> +
> +static int tdx_topup_external_fault_cache(struct kvm_vcpu *vcpu)
> +{
> + struct vcpu_tdx *tdx = to_tdx(vcpu);
> +
> + return kvm_mmu_topup_memory_cache(&tdx->mmu_external_spt_cache,
> + PT64_ROOT_MAX_LEVEL);
> +}
> +
> static int tdx_mem_page_aug(struct kvm *kvm, gfn_t gfn,
> enum pg_level level, struct page *page)
> {
> @@ -3647,4 +3662,6 @@ void __init tdx_hardware_setup(void)
> vt_x86_ops.free_external_spt = tdx_sept_free_private_spt;
> vt_x86_ops.remove_external_spte = tdx_sept_remove_private_spte;
> vt_x86_ops.protected_apic_has_interrupt = tdx_protected_apic_has_interrupt;
> + vt_x86_ops.topup_external_fault_cache = tdx_topup_external_fault_cache;
> + vt_x86_ops.alloc_external_fault_cache = tdx_alloc_external_fault_cache;
> }
> diff --git a/arch/x86/kvm/vmx/tdx.h b/arch/x86/kvm/vmx/tdx.h
> index f4e609a745ee..cd7993ef056e 100644
> --- a/arch/x86/kvm/vmx/tdx.h
> +++ b/arch/x86/kvm/vmx/tdx.h
> @@ -70,6 +70,8 @@ struct vcpu_tdx {
>
> u64 map_gpa_next;
> u64 map_gpa_end;
> +
> + struct kvm_mmu_memory_cache mmu_external_spt_cache;
> };
>
> void tdh_vp_rd_failed(struct vcpu_tdx *tdx, char *uclass, u32 field, u64 err);
> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index fee108988028..f05e6d43184b 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c
> @@ -404,6 +404,7 @@ int kvm_mmu_topup_memory_cache(struct kvm_mmu_memory_cache *mc, int min)
> {
> return __kvm_mmu_topup_memory_cache(mc, KVM_ARCH_NR_OBJS_PER_MEMORY_CACHE, min);
> }
> +EXPORT_SYMBOL_GPL(kvm_mmu_topup_memory_cache);
>
> int kvm_mmu_memory_cache_nr_free_objects(struct kvm_mmu_memory_cache *mc)
> {
> @@ -436,6 +437,7 @@ void *kvm_mmu_memory_cache_alloc(struct kvm_mmu_memory_cache *mc)
> BUG_ON(!p);
> return p;
> }
> +EXPORT_SYMBOL_GPL(kvm_mmu_memory_cache_alloc);
> #endif
>
> static void kvm_vcpu_init(struct kvm_vcpu *vcpu, struct kvm *kvm, unsigned id)
> --
> 2.51.0
>
next prev parent reply other threads:[~2025-09-23 7:04 UTC|newest]
Thread overview: 97+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-18 23:22 [PATCH v3 00/16] TDX: Enable Dynamic PAMT Rick Edgecombe
2025-09-18 23:22 ` [PATCH v3 01/16] x86/tdx: Move all TDX error defines into <asm/shared/tdx_errno.h> Rick Edgecombe
2025-09-19 1:29 ` Huang, Kai
2025-09-25 23:23 ` Edgecombe, Rick P
2025-09-25 23:32 ` Huang, Kai
2025-09-23 5:49 ` Binbin Wu
2025-09-25 23:09 ` Edgecombe, Rick P
2025-09-26 5:36 ` Binbin Wu
2025-09-26 4:52 ` Xiaoyao Li
2025-09-26 19:53 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 02/16] x86/tdx: Add helpers to check return status codes Rick Edgecombe
2025-09-19 1:26 ` Huang, Kai
2025-09-25 23:27 ` Edgecombe, Rick P
2025-09-23 6:19 ` Binbin Wu
2025-09-25 23:24 ` Edgecombe, Rick P
2025-09-26 6:32 ` Xiaoyao Li
2025-09-26 21:27 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 03/16] x86/virt/tdx: Simplify tdmr_get_pamt_sz() Rick Edgecombe
2025-09-19 0:50 ` Huang, Kai
2025-09-19 19:26 ` Edgecombe, Rick P
2025-09-29 11:44 ` Xiaoyao Li
2025-09-29 17:47 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 04/16] x86/virt/tdx: Allocate page bitmap for Dynamic PAMT Rick Edgecombe
2025-09-23 7:15 ` Binbin Wu
2025-09-25 23:28 ` Edgecombe, Rick P
2025-09-26 8:41 ` Xiaoyao Li
2025-09-26 21:57 ` Edgecombe, Rick P
2025-09-26 22:06 ` Dave Hansen
2025-10-06 19:34 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 05/16] x86/virt/tdx: Allocate reference counters for PAMT memory Rick Edgecombe
2025-09-23 7:45 ` Binbin Wu
2025-09-29 17:41 ` Edgecombe, Rick P
2025-09-29 18:08 ` Dave Hansen
2025-09-30 1:04 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 06/16] x86/virt/tdx: Improve PAMT refcounters allocation for sparse memory Rick Edgecombe
2025-09-19 7:25 ` Huang, Kai
2025-09-23 9:38 ` Binbin Wu
2025-09-24 6:50 ` Huang, Kai
2025-09-24 8:57 ` Binbin Wu
2025-10-01 0:32 ` Edgecombe, Rick P
2025-10-01 10:40 ` Huang, Kai
2025-10-01 19:00 ` Edgecombe, Rick P
2025-10-01 20:49 ` Huang, Kai
2025-10-15 1:35 ` Huang, Kai
2025-09-18 23:22 ` [PATCH v3 07/16] x86/virt/tdx: Add tdx_alloc/free_page() helpers Rick Edgecombe
2025-09-22 11:27 ` Huang, Kai
2025-09-26 22:41 ` Edgecombe, Rick P
2025-09-29 7:56 ` Yan Zhao
2025-09-29 17:19 ` Edgecombe, Rick P
2025-09-30 14:03 ` Xiaoyao Li
2025-09-30 17:38 ` Dave Hansen
2025-09-30 17:47 ` Edgecombe, Rick P
2025-09-30 15:25 ` Dave Hansen
2025-09-30 17:00 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 08/16] x86/virt/tdx: Optimize " Rick Edgecombe
2025-09-19 9:39 ` Kiryl Shutsemau
2025-09-24 6:15 ` Binbin Wu
2025-09-18 23:22 ` [PATCH v3 09/16] KVM: TDX: Allocate PAMT memory for TD control structures Rick Edgecombe
2025-09-18 23:22 ` [PATCH v3 10/16] KVM: TDX: Allocate PAMT memory for vCPU " Rick Edgecombe
2025-09-18 23:22 ` [PATCH v3 11/16] KVM: TDX: Add x86 ops for external spt cache Rick Edgecombe
2025-09-19 9:44 ` Kiryl Shutsemau
2025-09-23 7:03 ` Yan Zhao [this message]
2025-09-26 22:10 ` Edgecombe, Rick P
2025-09-28 8:35 ` Yan Zhao
2025-09-24 7:58 ` Binbin Wu
2025-09-30 1:02 ` Yan Zhao
2025-09-30 17:54 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 12/16] x86/virt/tdx: Add helpers to allow for pre-allocating pages Rick Edgecombe
2025-09-19 9:55 ` Kiryl Shutsemau
2025-10-01 19:48 ` Edgecombe, Rick P
2025-09-22 11:20 ` Huang, Kai
2025-09-26 23:47 ` Edgecombe, Rick P
2025-09-28 22:56 ` Huang, Kai
2025-09-29 12:10 ` Huang, Kai
2025-09-26 1:44 ` Yan Zhao
2025-09-26 22:05 ` Edgecombe, Rick P
2025-09-28 1:40 ` Yan Zhao
2025-09-26 15:19 ` Dave Hansen
2025-09-26 15:49 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 13/16] KVM: TDX: Handle PAMT allocation in fault path Rick Edgecombe
2025-09-30 1:09 ` Yan Zhao
2025-09-30 18:11 ` Edgecombe, Rick P
2025-09-18 23:22 ` [PATCH v3 14/16] KVM: TDX: Reclaim PAMT memory Rick Edgecombe
2025-09-18 23:22 ` [PATCH v3 15/16] x86/virt/tdx: Enable Dynamic PAMT Rick Edgecombe
2025-09-18 23:22 ` [PATCH v3 16/16] Documentation/x86: Add documentation for TDX's " Rick Edgecombe
2025-09-26 2:28 ` [PATCH v3 00/16] TDX: Enable " Yan Zhao
2025-09-26 14:09 ` Dave Hansen
2025-09-26 16:02 ` Edgecombe, Rick P
2025-09-26 16:11 ` Dave Hansen
2025-09-26 19:00 ` Edgecombe, Rick P
2025-09-26 19:03 ` Dave Hansen
2025-09-26 19:52 ` Edgecombe, Rick P
2025-09-28 1:34 ` Yan Zhao
2025-09-29 11:17 ` Kiryl Shutsemau
2025-09-29 16:22 ` Dave Hansen
2025-09-29 16:58 ` Edgecombe, Rick P
2025-09-30 18:29 ` Edgecombe, Rick P
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=aNJGP6lwO9WOqjfh@yzhao56-desk.sh.intel.com \
--to=yan.y.zhao@intel.com \
--cc=bp@alien8.de \
--cc=chao.gao@intel.com \
--cc=dave.hansen@linux.intel.com \
--cc=isaku.yamahata@intel.com \
--cc=kai.huang@intel.com \
--cc=kas@kernel.org \
--cc=kvm@vger.kernel.org \
--cc=linux-coco@lists.linux.dev \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=pbonzini@redhat.com \
--cc=rick.p.edgecombe@intel.com \
--cc=seanjc@google.com \
--cc=tglx@linutronix.de \
--cc=vannapurve@google.com \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®