From: David Woodhouse <dwmw2@infradead.org>
To: "Jörg Rödel" <joro@8bytes.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
Sean Christopherson <seanjc@google.com>,
Tom Lendacky <thomas.lendacky@amd.com>,
ashish.kalra@amd.com, michael.roth@amd.com, nsaenz@amazon.com,
anelkz@amazon.de, James.Bottomley@HansenPartnership.com,
Melody Wang <huibo.wang@amd.com>,
kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
kvmarm@lists.linux.dev, loongarch@lists.linux.dev,
linux-mips@vger.kernel.org, linuxppc-dev@lists.ozlabs.org,
kvm-riscv@lists.infradead.org, x86@kernel.org,
coconut-svsm@lists.linux.dev, joerg.roedel@amd.com,
Mushahid Hussain <hmushi@amazon.co.uk>,
Sriram Nambakam <snambakam@linux.microsoft.com>,
David Woodhouse <dwmw@amazon.co.uk>
Subject: [RFC PATCH 08/13] KVM: planes: Enlarge every twin's fpstate for dynamic xfeatures
Date: Fri, 13 Nov 2026 11:26:02 -0700 [thread overview]
Message-ID: <20261113182607.82813-9-dwmw2@infradead.org> (raw)
In-Reply-To: <20261113182607.82813-1-dwmw2@infradead.org>
From: David Woodhouse <dwmw@amazon.co.uk>
Plane twins share CPUID, so a twin's guest can set the XCR0 bits for a
dynamic xfeature such as AMX's XTILEDATA with its own XSETBV. But
kvm_check_cpuid() calls fpu_enable_guest_xfd_features() only for the
vCPU that KVM_SET_CPUID was issued on, and a twin created afterwards
starts with the default fpstate. A twin that then runs AMX code loses
its tile data on every fpstate save, because the fpstate it saves into
has no room for it. On a host that exposes AMX to the guest, any twin
using AMX hits this.
Whether the XRSTORS on the next entry can also leave a previous
context's tile registers in place for that twin to read is an open
question; it has not been verified.
Enlarge the fpstate of every twin in kvm_check_cpuid(), and give a twin
created after KVM_SET_CPUID the dynamic xfeatures of its plane-0
sibling's fpstate.
Only a twin that has not run may have its fpstate enlarged:
fpstate_realloc() hands a guest fpstate that is not loaded a fresh
image and copies nothing, so a twin that has run would lose its XSAVE
state. kvm_set_cpuid() already accepts only identical CPUID once the
vCPU has run, but it checked only the vCPU the ioctl was issued on.
With plane 0 held while a higher plane runs, KVM_SET_CPUID2 on plane 0
would pass that check and resize the plane that ran. Since the twins
share CPUID, take the identical-CPUID path once any twin has run.
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
---
arch/x86/kvm/cpuid.c | 21 +++++++++++++++++++--
arch/x86/kvm/x86.c | 17 +++++++++++++++++
2 files changed, 36 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index c5fa0551ddd5..971ebd781c6e 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -141,7 +141,9 @@ EXPORT_SYMBOL_FOR_KVM_INTERNAL(kvm_find_cpuid_entry2);
static int kvm_check_cpuid(struct kvm_vcpu *vcpu)
{
struct kvm_cpuid_entry2 *best;
+ struct kvm_vcpu *v;
u64 xfeatures;
+ int i, r;
/*
* The existing code assumes virtual address is 48-bit or 57-bit in the
@@ -168,7 +170,18 @@ static int kvm_check_cpuid(struct kvm_vcpu *vcpu)
if (!xfeatures)
return 0;
- return fpu_enable_guest_xfd_features(&vcpu->arch.guest_fpu, xfeatures);
+ /*
+ * The planes share CPUID, so any twin's guest can enable these
+ * xfeatures in its XCR0: enlarge every twin's fpstate, not just
+ * this vCPU's.
+ */
+ vcpu_for_each_plane(vcpu->common, i, v) {
+ r = fpu_enable_guest_xfd_features(&v->arch.guest_fpu, xfeatures);
+ if (r)
+ return r;
+ }
+
+ return 0;
}
static u32 kvm_apply_cpuid_pv_features_quirk(struct kvm_vcpu *vcpu);
@@ -545,8 +558,12 @@ static int kvm_set_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *e2,
* VMMs (e.g. QEMU) reuse vCPU fds for CPU hotplug/unplug and do
* KVM_SET_CPUID{,2} again. To support this legacy behavior, check
* whether the supplied CPUID data is equal to what's already set.
+ * Plane twins share CPUID, so this applies once any twin has run.
*/
- if (!kvm_can_set_cpuid_and_feature_msrs(vcpu)) {
+ vcpu_for_each_plane(common, i, v) {
+ if (kvm_can_set_cpuid_and_feature_msrs(v))
+ continue;
+
r = kvm_cpuid_check_equal(vcpu, e2, nent);
if (r)
goto err;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 6d3d93e85a65..af28e73e4b9b 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -9771,6 +9771,23 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
vcpu->arch.hv_root_tdp = INVALID_PAGE;
#endif
+ /*
+ * A plane twin created after KVM_SET_CPUID shares CPUID that may
+ * already allow dynamic xfeatures, which KVM_SET_CPUID enabled only
+ * in the twins that existed at the time. Give it the dynamic
+ * xfeatures of its plane-0 sibling's fpstate.
+ */
+ if (vcpu->plane_level && vcpu->common->vcpus[0]) {
+ u64 dyn = vcpu->common->vcpus[0]->arch.guest_fpu.fpstate->xfeatures &
+ XFEATURE_MASK_USER_DYNAMIC;
+
+ if (dyn) {
+ r = fpu_enable_guest_xfd_features(&vcpu->arch.guest_fpu, dyn);
+ if (r)
+ goto free_guest_fpu;
+ }
+ }
+
/*
* A plane twin created after KVM_SET_CPUID shares CPUID that may
* already advertise Hyper-V. Allocate the Hyper-V context now so
--
2.50.1
next prev parent reply other threads:[~2026-11-13 18:26 UTC|newest]
Thread overview: 113+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-08 14:41 [PATCH 00/60] KVM Planes + SEV-SNP Support Jörg Rödel
2026-06-08 14:41 ` [PATCH 01/60] x86/sev: Define the #HV doorbell page structure Jörg Rödel
2026-06-08 14:41 ` [PATCH 02/60] KVM: SVM: Add support for the SEV-SNP #HV doorbell page NAE event Jörg Rödel
2026-06-10 13:28 ` Tom Lendacky
2026-06-10 15:05 ` Tom Lendacky
2026-06-08 14:41 ` [PATCH 03/60] KVM: SVM: Inject #HV when Restricted Injection is active Jörg Rödel
2026-06-08 14:41 ` [PATCH 04/60] KVM: SVM: Inject NMIs " Jörg Rödel
2026-06-08 14:41 ` [PATCH 05/60] KVM: SVM: Inject MCEs " Jörg Rödel
2026-06-08 14:41 ` [PATCH 06/60] KVM: SVM: Enable Restricted Injection for an SEV-SNP guest Jörg Rödel
2026-06-08 14:41 ` [PATCH 07/60] KVM: SVM: Add support for the SEV-SNP #HV IPI NAE event Jörg Rödel
2026-06-10 13:45 ` Tom Lendacky
2026-06-08 14:42 ` [PATCH 08/60] Documentation: kvm: introduce "VM plane" concept Jörg Rödel
2026-06-08 14:42 ` [PATCH 09/60] kvm: Introduce struct kvm_plane Jörg Rödel
2026-06-08 14:42 ` [PATCH 10/60] kvm: Move vcpu_array to " Jörg Rödel
2026-06-08 14:42 ` [PATCH 11/60] kvm: Introduce struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 12/60] kvm: Move vcpu accounting to " Jörg Rödel
2026-06-08 14:42 ` [PATCH 13/60] kvm: Add read accessors for kvm_vcpu scheduling state Jörg Rödel
2026-06-08 14:42 ` [PATCH 14/60] kvm: Make kvm_running_vcpus point to struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 15/60] kvm: Move VCPU scheduling state " Jörg Rödel
2026-06-08 14:42 ` [PATCH 16/60] kvm: Add accessors for kvm_vcpu->mutex Jörg Rödel
2026-06-08 14:42 ` [PATCH 17/60] kvm: Move VCPU locking to struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 18/60] kvm: Move kvm_vcpu->rcuwait " Jörg Rödel
2026-06-08 14:42 ` [PATCH 19/60] kvm: Introduce accessors for kvm_vcpu->mode Jörg Rödel
2026-06-08 14:42 ` [PATCH 20/60] kvm: Move kvm_vcpu mode and requests field to struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 21/60] kvm: Introduce per-plane VCPU requests Jörg Rödel
2026-06-08 14:42 ` [PATCH 22/60] kvm: Move kvm_vcpu pid members to struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 23/60] kvm: Move kvm_vcpu sigset " Jörg Rödel
2026-06-08 14:42 ` [PATCH 24/60] kvm: Move kvm_vcpu spinloop " Jörg Rödel
2026-06-08 14:42 ` [PATCH 25/60] kvm: Move kvm_vcpu->dirty_ring " Jörg Rödel
2026-06-08 14:42 ` [PATCH 26/60] kvm: Introduce arch-specific plane state Jörg Rödel
2026-06-08 14:42 ` [PATCH 27/60] kvm: Introduce arch-specific part of struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 28/60] kvm: Implement KVM_CAP_PLANES Jörg Rödel
2026-06-08 14:42 ` [PATCH 29/60] kvm: Implement KVM_CREATE_PLANE ioctl Jörg Rödel
2026-09-11 17:54 ` Serge Hallyn
2026-06-08 14:42 ` [PATCH 30/60] kvm: Add KVM_EXIT_PLANE_EVENT Jörg Rödel
2026-06-08 14:42 ` [PATCH 31/60] kvm: Allocate struct kvm_plane in architecture code Jörg Rödel
2026-06-08 14:42 ` [PATCH 32/60] kvm: Allocate struct kvm_run only for struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 33/60] KVM: Implement KVM_CREATE_VCPU ioctl for planes Jörg Rödel
2026-06-08 14:42 ` [PATCH 34/60] kvm: Keep track of plane VCPUs in struct kvm_vcpu_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 35/60] kvm: Add VCPU plane-scheduling state and helpers Jörg Rödel
2026-06-08 16:47 ` Paolo Bonzini
2026-06-08 17:52 ` Jörg Rödel
2026-06-08 17:58 ` Paolo Bonzini
2026-06-09 12:37 ` Jörg Rödel
2026-06-09 12:59 ` James Bottomley
2026-06-09 14:27 ` Jörg Rödel
2026-06-09 15:06 ` James Bottomley
2026-07-17 13:48 ` Saenz Julienne, Nicolas
2026-07-17 15:56 ` James Bottomley
2026-07-17 17:33 ` Saenz Julienne, Nicolas
2026-07-17 18:26 ` James Bottomley
2026-07-17 14:35 ` James Bottomley
2026-09-24 14:10 ` Jörg Rödel
2026-06-08 14:42 ` [PATCH 36/60] kvm: Add plane_level to kvm_kernel_irq_routing_entry Jörg Rödel
2026-06-08 14:42 ` [PATCH 37/60] kvm: Pass plane_level to kvm_set_routing_entry() Jörg Rödel
2026-06-08 14:42 ` [PATCH 38/60] kvm: Make KVM_SIGNAL_MSI per plane Jörg Rödel
2026-06-08 14:42 ` [PATCH 39/60] kvm: Make KVM_SET_GSI_ROUTING " Jörg Rödel
2026-06-08 14:42 ` [PATCH 40/60] kvm: x86: Handle IOAPIC EOIs " Jörg Rödel
2026-06-08 14:42 ` [PATCH 41/60] kvm: x86: Make apic_map " Jörg Rödel
2026-06-08 14:42 ` [PATCH 42/60] kvm: x86: Make local APIC code aware of planes Jörg Rödel
2026-06-08 14:42 ` [PATCH 43/60] kvm: x86: Move CPUID state to struct kvm_vcpu_arch_common Jörg Rödel
2026-06-08 14:42 ` [PATCH 44/60] kvm: x86: Move cpu_caps " Jörg Rödel
2026-06-08 14:42 ` [PATCH 45/60] kvm: x86: Update state for all plane VCPUs after CPUID update Jörg Rödel
2026-06-08 14:42 ` [PATCH 46/60] kvm: x86: Share MTRR state across planes Jörg Rödel
2026-06-08 14:42 ` [PATCH 47/60] kvm: x86: Select a plane to run Jörg Rödel
2026-06-08 14:42 ` [PATCH 48/60] kvm: x86: Make event injection VCPU requests per-plane Jörg Rödel
2026-06-08 14:42 ` [PATCH 49/60] kvm: x86: Allow hardware backend to overwrite struct kvm_plane allocation Jörg Rödel
2026-06-08 14:42 ` [PATCH 50/60] kvm: x86: Make KVM_REQ_UPDATE_PROTECTED_GUEST_STATE per plane Jörg Rödel
2026-06-08 14:42 ` [PATCH 51/60] kvm: x86: Share pio_data across planes Jörg Rödel
2026-06-08 14:42 ` [PATCH 52/60] kvm: x86: Switch to plane0 if it has events Jörg Rödel
2026-06-08 14:42 ` [PATCH 53/60] kvm: x86: Introduce max_planes x86-op Jörg Rödel
2026-06-08 14:42 ` [PATCH 54/60] kvm: x86: Restrict KVM planes support to KVM_IRQCHIP_SPLIT Jörg Rödel
2026-06-08 14:42 ` [PATCH 55/60] kvm: svm: Track vmsa_features per plane Jörg Rödel
2026-06-08 14:42 ` [PATCH 56/60] kvm: svm: Implement GET_AP_APIC_IDS NAE event Jörg Rödel
2026-06-08 14:42 ` [PATCH 57/60] kvm: sev: Allow for VMPL level specification in AP create Jörg Rödel
2026-06-08 14:42 ` [PATCH 58/60] kvm: svm: Invoke a specified VMPL level VMSA for the vCPU Jörg Rödel
2026-06-08 14:42 ` [PATCH 59/60] kvm: svm: Implement max_planes x86 operation Jörg Rödel
2026-06-08 14:42 ` [PATCH 60/60] kvm: svm: Advertise full multi-VMPL support to the SNP guest Jörg Rödel
2026-06-09 9:29 ` [syzbot ci] Re: KVM Planes + SEV-SNP Support syzbot ci
2026-06-10 18:13 ` [PATCH 00/60] " Melody Wang
2026-11-11 22:28 ` [PATCH 00/16] KVM: Fixes for the Planes + SEV-SNP series David Woodhouse
2026-11-11 22:28 ` [PATCH 01/16] KVM: s390: arm64: Pass plane_level to the kvm_set_routing_entry() stub David Woodhouse
2026-11-11 22:28 ` [PATCH 02/16] KVM: SVM: Derive the plane limit from the immutable VM type David Woodhouse
2026-11-11 22:28 ` [PATCH 03/16] KVM: planes: Report KVM_MAX_PLANES from the system fd David Woodhouse
2026-11-11 22:28 ` [PATCH 04/16] KVM: Guard the per-plane KVM_SIGNAL_MSI cases with __KVM_HAVE_MSI David Woodhouse
2026-11-11 22:29 ` [PATCH 05/16] KVM: x86: Initialise Hyper-V state on every plane vCPU David Woodhouse
2026-11-11 22:29 ` [PATCH 06/16] KVM: x86: Widen the plane vCPU fd ioctl allowlist David Woodhouse
2026-11-11 22:29 ` [PATCH 07/16] KVM: x86/mmu: Free obsolete roots of all planes David Woodhouse
2026-11-11 22:29 ` [PATCH 08/16] KVM: planes: Publish a new plane only once KVM_CREATE_PLANE cannot fail David Woodhouse
2026-11-11 22:29 ` [PATCH 09/16] KVM: planes: Kick the running twin's pCPU, not the request target's David Woodhouse
2026-11-11 22:29 ` [PATCH 10/16] KVM: planes: Keep current_vcpu in sync in vcpu_load() David Woodhouse
2026-11-11 22:29 ` [PATCH 11/16] KVM: x86: Load each plane twin while updating it after KVM_SET_CPUID David Woodhouse
2026-11-11 22:29 ` [PATCH 12/16] KVM: VMX: Leave the IPIv PID-table entry to the plane 0 vCPU David Woodhouse
2026-11-11 22:29 ` [PATCH 13/16] KVM: x86: Leave TSC matching to the plane 0 vCPUs David Woodhouse
2026-11-11 22:29 ` [PATCH 14/16] KVM: x86: Derive runtime CPUID bits from the executing plane David Woodhouse
2026-11-11 22:29 ` [PATCH 15/16] KVM: planes: Annotate the lockless twin-publication protocol David Woodhouse
2026-11-11 22:29 ` [PATCH 16/16] KVM: SVM: Make KVM_REQ_VMSA_PAGE_RELOAD per plane David Woodhouse
2026-11-13 0:29 ` [PATCH 17/16] Documentation: kvm: Renumber KVM_CREATE_PLANE and restore the kvm_run label David Woodhouse
2026-11-13 0:29 ` [PATCH 18/16] KVM: SVM: Bound the SNP plane count by the architectural VMPL limit David Woodhouse
2026-11-13 18:25 ` [RFC PATCH 00/13] KVM: planes: Userspace API David Woodhouse
2026-11-13 18:25 ` [RFC PATCH 01/13] KVM: x86: Let userspace opt in to multiple planes with KVM_CAP_PLANES David Woodhouse
2026-11-13 18:25 ` [RFC PATCH 02/13] KVM: selftests: Add VM plane infrastructure and tests David Woodhouse
2026-11-13 18:25 ` [RFC PATCH 03/13] KVM: selftests: Add x86 planes lifecycle test David Woodhouse
2026-11-13 18:25 ` [RFC PATCH 04/13] KVM: Add KVM_GET/SET_PLANE_RUN_STATE vCPU ioctls David Woodhouse
2026-11-13 18:25 ` [RFC PATCH 05/13] KVM: selftests: Cover plane run-state ioctls in planes_test David Woodhouse
2026-11-13 18:26 ` [RFC PATCH 06/13] KVM: planes: Fan out per-plane requests to twin vCPUs David Woodhouse
2026-11-13 18:26 ` [RFC PATCH 07/13] KVM: x86: Inhibit APICv when VM planes are enabled David Woodhouse
2026-11-13 18:26 ` David Woodhouse [this message]
2026-11-13 18:26 ` [RFC PATCH 09/13] KVM: x86: Let userspace choose state for plane twins to share David Woodhouse
2026-11-13 18:26 ` [RFC PATCH 10/13] KVM: planes: Let plane twins share XCR0 David Woodhouse
2026-11-13 18:26 ` [RFC PATCH 11/13] KVM: planes: Let plane twins share DR6 David Woodhouse
2026-11-13 18:26 ` [RFC PATCH 12/13] KVM: selftests: Check CPUID handling of plane twins David Woodhouse
2026-11-13 18:26 ` [RFC PATCH 13/13] KVM: selftests: Cover plane state sharing David Woodhouse
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=20261113182607.82813-9-dwmw2@infradead.org \
--to=dwmw2@infradead.org \
--cc=James.Bottomley@HansenPartnership.com \
--cc=anelkz@amazon.de \
--cc=ashish.kalra@amd.com \
--cc=coconut-svsm@lists.linux.dev \
--cc=dwmw@amazon.co.uk \
--cc=hmushi@amazon.co.uk \
--cc=huibo.wang@amd.com \
--cc=joerg.roedel@amd.com \
--cc=joro@8bytes.org \
--cc=kvm-riscv@lists.infradead.org \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.linux.dev \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mips@vger.kernel.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=loongarch@lists.linux.dev \
--cc=michael.roth@amd.com \
--cc=nsaenz@amazon.com \
--cc=pbonzini@redhat.com \
--cc=seanjc@google.com \
--cc=snambakam@linux.microsoft.com \
--cc=thomas.lendacky@amd.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®