Skip to contentSkip to conversation

KVM: arm64: timer fixes and optimisations

PATCH Applied Marc Zyngier 2 participants 5 messages Jan 12, 2023 – Jan 30, 2023

Subsystems (3)
lore
Share & tools
lore

Marc Zyngier

author maintainer
#0 lore raw

cover KVM: arm64: timer fixes and optimisations

Having been busy on the NV front the past few weeks, I collected a small set of fixes/improvements that make sense even outside of NV.

The first one is an interesting fix (actually a regression introduced by the initial set of NV-related patches) reported by Scott and Ganapatrao, where we fail to recognise that a timer that has fired doesn't need to fire again. And again.

The second patch is a definite performance improvement on nVHE systems when accessing the emulated physical timer. It also makes NV bearable in some conditions (with FEAT_ECV, for example).

The last patch is more of a sanity check. There is no reason to BUG() if we can avoid it at all and kill the guest instead!

These patches are 6.3 candidates, although the first one could be a 6.2 fix.

	M.

Marc Zyngier (3):
KVM: arm64: Don't arm a hrtimer for an already pending timer
KVM: arm64: Reduce overhead of trapped timer sysreg accesses
KVM: arm64: timers: Don't BUG() on unhandled timer trap

 arch/arm64/kvm/arch_timer.c | 77 +++++++++++++++++++++++--------------
 arch/arm64/kvm/sys_regs.c   |  4 +-
 2 files changed, 52 insertions(+), 29 deletions(-)
author maintainer
#1 lore raw

PATCH 1/3 KVM: arm64: Don't arm a hrtimer for an already pending timer

When fully emulating a timer, we back it with a hrtimer that is armver on vcpu_load(). However, we do this even if the timer is already pending.

This causes spurious interrupts to be taken, though the guest doesn't observe them (the interrupt is already pending).

Although this is a waste of precious cycles, this isn't the end of the world with the current state of KVM. However, this can lead to a situation where a guest doesn't make forward progress anymore with NV.

Fix it by checking that if the timer is already pending before arming a new hrtimer. Also drop the hrtimer cancelling, which is useless, by construction.

Fixes bee038a67487 ("KVM: arm/arm64: Rework the timer code to use a timer_map")
---
 arch/arm64/kvm/arch_timer.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
arch/arm64/kvm/arch_timer.c +1 -3
index bb24a76b4224..587d87aec33f 100644
@@ -428,10 +428,8 @@ static void timer_emulate(struct arch_timer_context *ctx)
 	 * scheduled for the future.  If the timer cannot fire at all,
 	 * then we also don't need a soft timer.
 	 */
-	if (!kvm_timer_irq_can_fire(ctx)) {
-		soft_timer_cancel(&ctx->hrtimer);
+	if (should_fire || !kvm_timer_irq_can_fire(ctx))
 		return;
-	}
 
 	soft_timer_start(&ctx->hrtimer, kvm_timer_compute_delta(ctx));
 }
Signature · 1 line
2.34.1
author maintainer
#2 lore raw

PATCH 2/3 KVM: arm64: Reduce overhead of trapped timer sysreg accesses

In reply to Marc Zyngier

Each read/write to a trapped timer system register results in a whole kvm_timer_vcpu_put/load() cycle which affects all of the timers, and a bit more.

There is no need for such a thing, and we can limit the impact to the timer being affected, and only this one.

This drastically simplifies the emulated case, and limits the damage for trapped accesses. This also brings some performance back for NV.

Whilst we're at it, fix a comment that didn't quite capture why we always set CNTVOFF_EL2 to 0 when disabling the virtual timer.

---
 arch/arm64/kvm/arch_timer.c | 73 ++++++++++++++++++++++++-------------
 1 file changed, 48 insertions(+), 25 deletions(-)
arch/arm64/kvm/arch_timer.c +48 -25
index 587d87aec33f..1a1d7e258aba 100644
@@ -434,6 +434,11 @@ static void timer_emulate(struct arch_timer_context *ctx)
 	soft_timer_start(&ctx->hrtimer, kvm_timer_compute_delta(ctx));
 }
 
+static void set_cntvoff(u64 cntvoff)
+{
+	kvm_call_hyp(__kvm_timer_set_cntvoff, cntvoff);
+}
+
 static void timer_save_state(struct arch_timer_context *ctx)
 {
 	struct arch_timer_cpu *timer = vcpu_timer(ctx->vcpu);
@@ -457,6 +462,22 @@ static void timer_save_state(struct arch_timer_context *ctx)
 		write_sysreg_el0(0, SYS_CNTV_CTL);
 		isb();
 
+		/*
+		 * The kernel may decide to run userspace after
+		 * calling vcpu_put, so we reset cntvoff to 0 to
+		 * ensure a consistent read between user accesses to
+		 * the virtual counter and kernel access to the
+		 * physical counter of non-VHE case.
+		 *
+		 * For VHE, the virtual counter uses a fixed virtual
+		 * offset of zero, so no need to zero CNTVOFF_EL2
+		 * register, but this is actually useful when switching
+		 * between EL1/vEL2 with NV.
+		 *
+		 * Do it unconditionally, as this is either unavoidable
+		 * or dirt cheap.
+		 */
+		set_cntvoff(0);
 		break;
 	case TIMER_PTIMER:
 		timer_set_ctl(ctx, read_sysreg_el0(SYS_CNTP_CTL));
@@ -530,6 +551,7 @@ static void timer_restore_state(struct arch_timer_context *ctx)
 
 	switch (index) {
 	case TIMER_VTIMER:
+		set_cntvoff(timer_get_offset(ctx));
 		write_sysreg_el0(timer_get_cval(ctx), SYS_CNTV_CVAL);
 		isb();
 		write_sysreg_el0(timer_get_ctl(ctx), SYS_CNTV_CTL);
@@ -550,11 +572,6 @@ static void timer_restore_state(struct arch_timer_context *ctx)
 	local_irq_restore(flags);
 }
 
-static void set_cntvoff(u64 cntvoff)
-{
-	kvm_call_hyp(__kvm_timer_set_cntvoff, cntvoff);
-}
-
 static inline void set_timer_irq_phys_active(struct arch_timer_context *ctx, bool active)
 {
 	int r;
@@ -629,8 +646,6 @@ void kvm_timer_vcpu_load(struct kvm_vcpu *vcpu)
 		kvm_timer_vcpu_load_nogic(vcpu);
 	}
 
-	set_cntvoff(timer_get_offset(map.direct_vtimer));
-
 	kvm_timer_unblocking(vcpu);
 
 	timer_restore_state(map.direct_vtimer);
@@ -686,15 +701,6 @@ void kvm_timer_vcpu_put(struct kvm_vcpu *vcpu)
 
 	if (kvm_vcpu_is_blocking(vcpu))
 		kvm_timer_blocking(vcpu);
-
-	/*
-	 * The kernel may decide to run userspace after calling vcpu_put, so
-	 * we reset cntvoff to 0 to ensure a consistent read between user
-	 * accesses to the virtual counter and kernel access to the physical
-	 * counter of non-VHE case. For VHE, the virtual counter uses a fixed
-	 * virtual offset of zero, so no need to zero CNTVOFF_EL2 register.
-	 */
-	set_cntvoff(0);
 }
 
 /*
@@ -924,14 +930,22 @@ u64 kvm_arm_timer_read_sysreg(struct kvm_vcpu *vcpu,
 			      enum kvm_arch_timers tmr,
 			      enum kvm_arch_timer_regs treg)
 {
+	struct arch_timer_context *timer;
+	struct timer_map map;
 	u64 val;
 
+	get_timer_map(vcpu, &map);
+	timer = vcpu_get_timer(vcpu, tmr);
+
+	if (timer == map.emul_ptimer)
+		return kvm_arm_timer_read(vcpu, timer, treg);
+
 	preempt_disable();
-	kvm_timer_vcpu_put(vcpu);
+	timer_save_state(timer);
 
-	val = kvm_arm_timer_read(vcpu, vcpu_get_timer(vcpu, tmr), treg);
+	val = kvm_arm_timer_read(vcpu, timer, treg);
 
-	kvm_timer_vcpu_load(vcpu);
+	timer_restore_state(timer);
 	preempt_enable();
 
 	return val;
@@ -965,13 +979,22 @@ void kvm_arm_timer_write_sysreg(struct kvm_vcpu *vcpu,
 				enum kvm_arch_timer_regs treg,
 				u64 val)
 {
-	preempt_disable();
-	kvm_timer_vcpu_put(vcpu);
-
-	kvm_arm_timer_write(vcpu, vcpu_get_timer(vcpu, tmr), treg, val);
+	struct arch_timer_context *timer;
+	struct timer_map map;
 
-	kvm_timer_vcpu_load(vcpu);
-	preempt_enable();
+	get_timer_map(vcpu, &map);
+	timer = vcpu_get_timer(vcpu, tmr);
+	if (timer == map.emul_ptimer) {
+		soft_timer_cancel(&timer->hrtimer);
+		kvm_arm_timer_write(vcpu, timer, treg, val);
+		timer_emulate(timer);
+	} else {
+		preempt_disable();
+		timer_save_state(timer);
+		kvm_arm_timer_write(vcpu, timer, treg, val);
+		timer_restore_state(timer);
+		preempt_enable();
+	}
 }
 
 static int kvm_timer_starting_cpu(unsigned int cpu)
Signature · 1 line
2.34.1
author maintainer
#3 lore raw

PATCH 3/3 KVM: arm64: timers: Don't BUG() on unhandled timer trap

In reply to Marc Zyngier

Although not handling a trap is a pretty bad situation to be in, panicing the kernel isn't useful and provides no valuable information to help debugging the situation.

Instead, dump the encoding of the unhandled sysreg, and inject an UNDEF in the guest. At least, this gives a user an opportunity to report the issue with some information to help debugging it.

---
 arch/arm64/kvm/sys_regs.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kvm/sys_regs.c +3 -1
index d5ee52d6bf73..32f4e424b9a5 100644
@@ -1049,7 +1049,9 @@ static bool access_arch_timer(struct kvm_vcpu *vcpu,
 		treg = TIMER_REG_CVAL;
 		break;
 	default:
-		BUG();
+		print_sys_reg_msg(p, "%s", "Unhandled trapped timer register");
+		kvm_inject_undefined(vcpu);
+		return false;
 	}
 
 	if (p->is_write)
Signature · 1 line
2.34.1
Applied
#4 lore raw
In reply to Marc Zyngier
Quoted 9 lines · Marc Zyngier

Having been busy on the NV front the past few weeks, I collected a small set of fixes/improvements that make sense even outside of NV.

The first one is an interesting fix (actually a regression introduced by the initial set of NV-related patches) reported by Scott and Ganapatrao, where we fail to recognise that a timer that has fired doesn't need to fire again. And again.

[...]

Applied to kvmarm/next, thanks!

[1/3] KVM: arm64: Don't arm a hrtimer for an already pending timer

      https://git.kernel.org/kvmarm/kvmarm/c/4d74ecfa6458
[2/3] KVM: arm64: Reduce overhead of trapped timer sysreg accesses
      https://git.kernel.org/kvmarm/kvmarm/c/fc6ee952cf00
[3/3] KVM: arm64: timers: Don't BUG() on unhandled timer trap
      https://git.kernel.org/kvmarm/kvmarm/c/ba82e06cf7f4

--
Best,
Oliver