linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Zachary Amsden <zamsden@redhat.com>
To: KVM <kvm@vger.kernel.org>, Avi Kivity <avi@redhat.com>,
	Marcelo Tosatti <mtosatti@redhat.com>,
	Glauber Costa <glommer@redhat.com>
Cc: Linux-kernel <linux-kernel@vger.kernel.org>,
	Zachary Amsden <zamsden@redhat.com>
Subject: [PATCH 07/18] Unify TSC logic
Date: Mon, 12 Jul 2010 16:08:01 -1000	[thread overview]
Message-ID: <1278986892-11733-8-git-send-email-zamsden@redhat.com> (raw)
In-Reply-To: <1278986892-11733-1-git-send-email-zamsden@redhat.com>

Move the TSC control logic from the vendor backends into x86.c
by adding adjust_tsc_offset to x86 ops.  Now all TSC decisions
can be done in one place.

Also, rename some variable in the VCPU structure to more accurately
reflect their actual content.

Signed-off-by: Zachary Amsden <zamsden@redhat.com>
---
 arch/x86/include/asm/kvm_host.h |    5 +++--
 arch/x86/kvm/svm.c              |   26 ++++++++++----------------
 arch/x86/kvm/vmx.c              |   22 ++++++++--------------
 arch/x86/kvm/x86.c              |   18 +++++++++++++++---
 4 files changed, 36 insertions(+), 35 deletions(-)

diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 4b42893..324e892 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -255,7 +255,6 @@ struct kvm_mmu {
 };
 
 struct kvm_vcpu_arch {
-	u64 host_tsc;
 	/*
 	 * rip and regs accesses must go through
 	 * kvm_{register,rip}_{read,write} functions.
@@ -336,9 +335,10 @@ struct kvm_vcpu_arch {
 
 	gpa_t time;
 	struct pvclock_vcpu_time_info hv_clock;
-	unsigned int hv_clock_tsc_khz;
+	unsigned int hw_tsc_khz;
 	unsigned int time_offset;
 	struct page *time_page;
+	u64 last_host_tsc;
 
 	bool nmi_pending;
 	bool nmi_injected;
@@ -520,6 +520,7 @@ struct kvm_x86_ops {
 	u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
 	int (*get_lpage_level)(void);
 	bool (*rdtscp_supported)(void);
+	void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment);
 
 	void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
 
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index 7e7889d..b52222c 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -948,20 +948,6 @@ static void svm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	int i;
 
 	if (unlikely(cpu != vcpu->cpu)) {
-		u64 delta;
-
-		if (check_tsc_unstable()) {
-			/*
-			 * Make sure that the guest sees a monotonically
-			 * increasing TSC.
-			 */
-			delta = vcpu->arch.host_tsc - native_read_tsc();
-			svm->vmcb->control.tsc_offset += delta;
-			if (is_nested(svm))
-				svm->nested.hsave->control.tsc_offset += delta;
-		}
-		vcpu->cpu = cpu;
-		kvm_migrate_timers(vcpu);
 		svm->asid_generation = 0;
 	}
 
@@ -977,8 +963,6 @@ static void svm_vcpu_put(struct kvm_vcpu *vcpu)
 	++vcpu->stat.host_state_reload;
 	for (i = 0; i < NR_HOST_SAVE_USER_MSRS; i++)
 		wrmsrl(host_save_user_msrs[i], svm->host_user_msrs[i]);
-
-	vcpu->arch.host_tsc = native_read_tsc();
 }
 
 static unsigned long svm_get_rflags(struct kvm_vcpu *vcpu)
@@ -3432,6 +3416,15 @@ static void svm_write_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
 	svm->vmcb->control.tsc_offset = offset + g_tsc_offset;
 }
 
+static void svm_adjust_tsc_offset(struct kvm_vcpu *vcpu, s64 adjustment)
+{
+	struct vcpu_svm *svm = to_svm(vcpu);
+
+	svm->vmcb->control.tsc_offset += adjustment;
+	if (is_nested(svm))
+		svm->nested.hsave->control.tsc_offset += adjustment;
+}
+
 static void svm_fpu_deactivate(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_svm *svm = to_svm(vcpu);
@@ -3520,6 +3513,7 @@ static struct kvm_x86_ops svm_x86_ops = {
 	.has_wbinvd_exit = svm_has_wbinvd_exit,
 
 	.write_tsc_offset = svm_write_tsc_offset,
+	.adjust_tsc_offset = svm_adjust_tsc_offset,
 };
 
 static int __init svm_init(void)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 9055ca6..d8e23ab 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -504,7 +504,6 @@ static void __vcpu_clear(void *arg)
 		vmcs_clear(vmx->vmcs);
 	if (per_cpu(current_vmcs, cpu) == vmx->vmcs)
 		per_cpu(current_vmcs, cpu) = NULL;
-	rdtscll(vmx->vcpu.arch.host_tsc);
 	list_del(&vmx->local_vcpus_link);
 	vmx->vcpu.cpu = -1;
 	vmx->launched = 0;
@@ -887,7 +886,6 @@ static void vmx_load_host_state(struct vcpu_vmx *vmx)
 static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
-	u64 tsc_this, delta, new_offset;
 	u64 phys_addr = __pa(per_cpu(vmxarea, cpu));
 
 	if (!vmm_exclusive)
@@ -904,14 +902,12 @@ static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 		struct desc_ptr dt;
 		unsigned long sysenter_esp;
 
-		kvm_migrate_timers(vcpu);
 		kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
 		local_irq_disable();
 		list_add(&vmx->local_vcpus_link,
 			 &per_cpu(vcpus_on_cpu, cpu));
 		local_irq_enable();
 
-		vcpu->cpu = cpu;
 		/*
 		 * Linux uses per-cpu TSS and GDT, so set these when switching
 		 * processors.
@@ -922,16 +918,6 @@ static void vmx_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 
 		rdmsrl(MSR_IA32_SYSENTER_ESP, sysenter_esp);
 		vmcs_writel(HOST_IA32_SYSENTER_ESP, sysenter_esp); /* 22.2.3 */
-
-		/*
-		 * Make sure the time stamp counter is monotonous.
-		 */
-		rdtscll(tsc_this);
-		if (tsc_this < vcpu->arch.host_tsc) {
-			delta = vcpu->arch.host_tsc - tsc_this;
-			new_offset = vmcs_read64(TSC_OFFSET) + delta;
-			vmcs_write64(TSC_OFFSET, new_offset);
-		}
 	}
 }
 
@@ -1160,6 +1146,12 @@ static void vmx_write_tsc_offset(struct kvm_vcpu *vcpu, u64 offset)
 	vmcs_write64(TSC_OFFSET, offset);
 }
 
+static void vmx_adjust_tsc_offset(struct kvm_vcpu *vcpu, s64 adjustment)
+{
+	u64 offset = vmcs_read64(TSC_OFFSET);
+	vmcs_write64(TSC_OFFSET, offset + adjustment);
+}
+
 /*
  * Reads an msr value (of 'msr_index') into 'pdata'.
  * Returns 0 on success, non-0 otherwise.
@@ -4113,6 +4105,7 @@ static struct kvm_vcpu *vmx_create_vcpu(struct kvm *kvm, unsigned int id)
 
 	cpu = get_cpu();
 	vmx_vcpu_load(&vmx->vcpu, cpu);
+	vmx->vcpu.cpu = cpu;
 	err = vmx_vcpu_setup(vmx);
 	vmx_vcpu_put(&vmx->vcpu);
 	put_cpu();
@@ -4352,6 +4345,7 @@ static struct kvm_x86_ops vmx_x86_ops = {
 	.has_wbinvd_exit = cpu_has_vmx_wbinvd_exit,
 
 	.write_tsc_offset = vmx_write_tsc_offset,
+	.adjust_tsc_offset = vmx_adjust_tsc_offset,
 };
 
 static int __init vmx_init(void)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 045b2d6..9e71222 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -979,9 +979,9 @@ static int kvm_write_guest_time(struct kvm_vcpu *v)
 		return 1;
 	}
 
-	if (unlikely(vcpu->hv_clock_tsc_khz != this_tsc_khz)) {
+	if (unlikely(vcpu->hw_tsc_khz != this_tsc_khz)) {
 		kvm_set_time_scale(this_tsc_khz, &vcpu->hv_clock);
-		vcpu->hv_clock_tsc_khz = this_tsc_khz;
+		vcpu->hw_tsc_khz = this_tsc_khz;
 	}
 
 	/* With all the info we got, fill in the values */
@@ -1872,13 +1872,25 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	}
 
 	kvm_x86_ops->vcpu_load(vcpu, cpu);
-	kvm_request_guest_time_update(vcpu);
+	if (unlikely(vcpu->cpu != cpu)) {
+		/* Make sure TSC doesn't go backwards */
+		s64 tsc_delta = !vcpu->arch.last_host_tsc ? 0 : 
+				native_read_tsc() - vcpu->arch.last_host_tsc;
+		if (tsc_delta < 0)
+			mark_tsc_unstable("KVM discovered backwards TSC");
+		if (check_tsc_unstable())
+			kvm_x86_ops->adjust_tsc_offset(vcpu, -tsc_delta);
+		kvm_migrate_timers(vcpu);
+		kvm_request_guest_time_update(vcpu);
+		vcpu->cpu = cpu;
+	}
 }
 
 void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
 {
 	kvm_x86_ops->vcpu_put(vcpu);
 	kvm_put_guest_fpu(vcpu);
+	vcpu->arch.last_host_tsc = native_read_tsc();
 }
 
 static int is_efer_nx(void)
-- 
1.7.1


  parent reply	other threads:[~2010-07-13  2:12 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-07-13  2:07 KVM timekeeping fixes Zachary Amsden
2010-07-13  2:07 ` [PATCH 01/18] Make TSC offset writes non-preemptible Zachary Amsden
2010-07-13  2:07 ` [PATCH 02/18] Fix SVM VMCB reset Zachary Amsden
2010-07-13  2:07 ` [PATCH 03/18] TSC reset compensation Zachary Amsden
2010-07-13  2:07 ` [PATCH 04/18] Robust TSC compensation Zachary Amsden
2010-07-13  2:07 ` [PATCH 05/18] Make cpu_tsc_khz updates use local CPU Zachary Amsden
2010-07-13  2:08 ` [PATCH 06/18] Warn about unstable TSC Zachary Amsden
2010-07-13  2:08 ` Zachary Amsden [this message]
2010-07-13  2:08 ` [PATCH 08/18] Fix deep C-state TSC desynchronization Zachary Amsden
2010-07-13  2:08 ` [PATCH 09/18] Add helper functions for time computation Zachary Amsden
2010-07-13  2:08 ` [PATCH 10/18] Keep SMP VMs more in sync on unstable TSC Zachary Amsden
2010-07-13  2:08 ` [PATCH 11/18] Perform hardware_enable in CPU_STARTING callback Zachary Amsden
2010-07-13  2:08 ` [PATCH 12/18] Add clock sync request to hardware enable Zachary Amsden
2010-07-13  2:08 ` [PATCH 13/18] Move scale_delta into common header Zachary Amsden
2010-07-13  2:08 ` [PATCH 14/18] Fix a possible backwards warp of kvmclock Zachary Amsden
2010-07-13  2:08 ` [PATCH 15/18] Implement getnsboottime kernel API Zachary Amsden
2010-07-13  2:08 ` [PATCH 16/18] Use getnsboottime in KVM Zachary Amsden
2010-07-13  2:08 ` [PATCH 17/18] Indicate reliable TSC in kvmclock Zachary Amsden
2010-07-13  2:08 ` [PATCH 18/18] Add timekeeping documentation Zachary Amsden

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=1278986892-11733-8-git-send-email-zamsden@redhat.com \
    --to=zamsden@redhat.com \
    --cc=avi@redhat.com \
    --cc=glommer@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mtosatti@redhat.com \
    /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;
as well as URLs for NNTP newsgroup(s).