linux-kernel - [PATCH v7 3/3] KVM: SVM: Flush cache only on CPUs running SEV guest

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-Id: <20250128015345.7929-4-szy0127@sjtu.edu.cn>
Date: Tue, 28 Jan 2025 09:53:45 +0800
From: Zheyun Shen <szy0127@...u.edu.cn>
To: thomas.lendacky@....com,
	seanjc@...gle.com,
	pbonzini@...hat.com,
	tglx@...utronix.de,
	kevinloughlin@...gle.com,
	mingo@...hat.com,
	bp@...en8.de
Cc: kvm@...r.kernel.org,
	linux-kernel@...r.kernel.org,
	Zheyun Shen <szy0127@...u.edu.cn>
Subject: [PATCH v7 3/3] KVM: SVM: Flush cache only on CPUs running SEV guest

On AMD CPUs without ensuring cache consistency, each memory page
reclamation in an SEV guest triggers a call to wbinvd_on_all_cpus(),
thereby affecting the performance of other programs on the host.

Typically, an AMD server may have 128 cores or more, while the SEV guest
might only utilize 8 of these cores. Meanwhile, host can use qemu-affinity
to bind these 8 vCPUs to specific physical CPUs.

Therefore, keeping a record of the physical core numbers each time a vCPU
runs can help avoid flushing the cache for all CPUs every time.

Signed-off-by: Zheyun Shen <szy0127@...u.edu.cn>
---
 arch/x86/kvm/svm/sev.c | 30 +++++++++++++++++++++++++++---
 arch/x86/kvm/svm/svm.c |  2 ++
 arch/x86/kvm/svm/svm.h |  5 ++++-
 3 files changed, 33 insertions(+), 4 deletions(-)

diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index 1ce67de9d..4b80ecbe7 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -252,6 +252,27 @@ static void sev_asid_free(struct kvm_sev_info *sev)
 	sev->misc_cg = NULL;
 }
 
+void sev_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
+{
+	/*
+	 * To optimize cache flushes when memory is reclaimed from an SEV VM,
+	 * track physical CPUs that enter the guest for SEV VMs and thus can
+	 * have encrypted, dirty data in the cache, and flush caches only for
+	 * CPUs that have entered the guest.
+	 */
+	cpumask_set_cpu(cpu, to_kvm_sev_info(vcpu->kvm)->wbinvd_dirty_mask);
+}
+
+static void sev_do_wbinvd(struct kvm *kvm)
+{
+	/*
+	 * TODO: Clear CPUs from the bitmap prior to flushing.  Doing so
+	 * requires serializing multiple calls and having CPUs mark themselves
+	 * "dirty" if they are currently running a vCPU for the VM.
+	 */
+	wbinvd_on_many_cpus(to_kvm_sev_info(kvm)->wbinvd_dirty_mask);
+}
+
 static void sev_decommission(unsigned int handle)
 {
 	struct sev_data_decommission decommission;
@@ -448,6 +469,8 @@ static int __sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp,
 	ret = sev_platform_init(&init_args);
 	if (ret)
 		goto e_free;
+	if (!zalloc_cpumask_var(&sev->wbinvd_dirty_mask, GFP_KERNEL_ACCOUNT))
+		goto e_free;
 
 	/* This needs to happen after SEV/SNP firmware initialization. */
 	if (vm_type == KVM_X86_SNP_VM) {
@@ -2778,7 +2801,7 @@ int sev_mem_enc_unregister_region(struct kvm *kvm,
 	 * releasing the pages back to the system for use. CLFLUSH will
 	 * not do this, so issue a WBINVD.
 	 */
-	wbinvd_on_all_cpus();
+	sev_do_wbinvd(kvm);
 
 	__unregister_enc_region_locked(kvm, region);
 
@@ -2926,6 +2949,7 @@ void sev_vm_destroy(struct kvm *kvm)
 	}
 
 	sev_asid_free(sev);
+	free_cpumask_var(sev->wbinvd_dirty_mask);
 }
 
 void __init sev_set_cpu_caps(void)
@@ -3129,7 +3153,7 @@ static void sev_flush_encrypted_page(struct kvm_vcpu *vcpu, void *va)
 	return;
 
 do_wbinvd:
-	wbinvd_on_all_cpus();
+	sev_do_wbinvd(vcpu->kvm);
 }
 
 void sev_guest_memory_reclaimed(struct kvm *kvm)
@@ -3143,7 +3167,7 @@ void sev_guest_memory_reclaimed(struct kvm *kvm)
 	if (!sev_guest(kvm) || sev_snp_guest(kvm))
 		return;
 
-	wbinvd_on_all_cpus();
+	sev_do_wbinvd(kvm);
 }
 
 void sev_free_vcpu(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
index dd15cc635..f3b03b0d8 100644
--- a/arch/x86/kvm/svm/svm.c
+++ b/arch/x86/kvm/svm/svm.c
@@ -1565,6 +1565,8 @@ static void svm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	}
 	if (kvm_vcpu_apicv_active(vcpu))
 		avic_vcpu_load(vcpu, cpu);
+	if (sev_guest(vcpu->kvm))
+		sev_vcpu_load(vcpu, cpu);
 }
 
 static void svm_vcpu_put(struct kvm_vcpu *vcpu)
diff --git a/arch/x86/kvm/svm/svm.h b/arch/x86/kvm/svm/svm.h
index 43fa6a16e..82ec80cf4 100644
--- a/arch/x86/kvm/svm/svm.h
+++ b/arch/x86/kvm/svm/svm.h
@@ -112,6 +112,8 @@ struct kvm_sev_info {
 	void *guest_req_buf;    /* Bounce buffer for SNP Guest Request input */
 	void *guest_resp_buf;   /* Bounce buffer for SNP Guest Request output */
 	struct mutex guest_req_mutex; /* Must acquire before using bounce buffers */
+	/* CPUs invoked VMRUN call wbinvd after guest memory is reclaimed */
+	struct cpumask *wbinvd_dirty_mask;
 };
 
 struct kvm_svm {
@@ -763,6 +765,7 @@ void sev_snp_init_protected_guest_state(struct kvm_vcpu *vcpu);
 int sev_gmem_prepare(struct kvm *kvm, kvm_pfn_t pfn, gfn_t gfn, int max_order);
 void sev_gmem_invalidate(kvm_pfn_t start, kvm_pfn_t end);
 int sev_private_max_mapping_level(struct kvm *kvm, kvm_pfn_t pfn);
+void sev_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
 #else
 static inline struct page *snp_safe_alloc_page_node(int node, gfp_t gfp)
 {
@@ -793,7 +796,7 @@ static inline int sev_private_max_mapping_level(struct kvm *kvm, kvm_pfn_t pfn)
 {
 	return 0;
 }
-
+static inline void sev_vcpu_load(struct kvm_vcpu *vcpu, int cpu) {}
 #endif
 
 /* vmenter.S */
-- 
2.34.1