Re: [PATCH RFC] kvm: x86: add halt_poll module parameter

[Marcelo Tosatti <mtosatti@redhat.com>]

On Thu, Feb 05, 2015 at 05:05:25PM +0100, Paolo Bonzini wrote:
> This patch introduces a new module parameter for the KVM module; when it
> is present, KVM attempts a bit of polling on every HLT before scheduling
> itself out via kvm_vcpu_block.
> 
> This parameter helps a lot for latency-bound workloads---in particular
> I tested it with O_DSYNC writes with a battery-backed disk in the host.
> In this case, writes are fast (because the data doesn't have to go all
> the way to the platters) but they cannot be merged by either the host or
> the guest.  KVM's performance here is usually around 30% of bare metal,
> or 50% if you use cache=directsync or cache=writethrough (these
> parameters avoid that the guest sends pointless flush requests, and
> at the same time they are not slow because of the battery-backed cache).
> The bad performance happens because on every halt the host CPU decides
> to halt itself too.  When the interrupt comes, the vCPU thread is then
> migrated to a new physical CPU, and in general the latency is horrible
> because the vCPU thread has to be scheduled back in.
> 
> With this patch performance reaches 60-65% of bare metal and, more
> important, 99% of what you get if you use idle=poll in the guest.  This
> means that the tunable gets rid of this particular bottleneck, and more
> work can be done to improve performance in the kernel or QEMU.
> 
> Of course there is some price to pay; every time an otherwise idle vCPUs
> is interrupted by an interrupt, it will poll unnecessarily and thus
> impose a little load on the host.  The above results were obtained with
> a mostly random value of the parameter (2000000), and the load was around
> 1.5-2.5% CPU usage on one of the host's core for each idle guest vCPU.
> 
> The patch also adds a new stat, /sys/kernel/debug/kvm/halt_successful_poll,
> that can be used to tune the parameter.  It counts how many HLT
> instructions received an interrupt during the polling period; each
> successful poll avoids that Linux schedules the VCPU thread out and back
> in, and may also avoid a likely trip to C1 and back for the physical CPU.
> 
> While the VM is idle, a Linux 4 VCPU VM halts around 10 times per second.
> Of these halts, almost all are failed polls.  During the benchmark,
> instead, basically all halts end within the polling period, except a more
> or less constant stream of 50 per second coming from vCPUs that are not
> running the benchmark.  The wasted time is thus very low.  Things may
> be slightly different for Windows VMs, which have a ~10 ms timer tick.
> 
> The effect is also visible on Marcelo's recently-introduced latency
> test for the TSC deadline timer.  Though of course a non-RT kernel has
> awful latency bounds, the latency of the timer is around 8000-10000 clock
> cycles compared to 20000-120000 without setting halt_poll.  For the TSC
> deadline timer, thus, the effect is both a smaller average latency and
> a smaller variance.
> 
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  arch/x86/include/asm/kvm_host.h |  1 +
>  arch/x86/kvm/x86.c              | 28 ++++++++++++++++++++++++----
>  include/linux/kvm_host.h        |  1 +
>  virt/kvm/kvm_main.c             | 22 +++++++++++++++-------
>  4 files changed, 41 insertions(+), 11 deletions(-)
> 
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 848947ac6ade..a236e39cc385 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -655,6 +655,7 @@ struct kvm_vcpu_stat {
>  	u32 irq_window_exits;
>  	u32 nmi_window_exits;
>  	u32 halt_exits;
> +	u32 halt_successful_poll;
>  	u32 halt_wakeup;
>  	u32 request_irq_exits;
>  	u32 irq_exits;
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 1373e04e1f19..b7b20828f01c 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -96,6 +96,9 @@ EXPORT_SYMBOL_GPL(kvm_x86_ops);
>  static bool ignore_msrs = 0;
>  module_param(ignore_msrs, bool, S_IRUGO | S_IWUSR);
>  
> +unsigned int halt_poll = 0;
> +module_param(halt_poll, uint, S_IRUGO | S_IWUSR);
> +
>  unsigned int min_timer_period_us = 500;
>  module_param(min_timer_period_us, uint, S_IRUGO | S_IWUSR);
>  
> @@ -145,6 +148,7 @@ struct kvm_stats_debugfs_item debugfs_entries[] = {
>  	{ "irq_window", VCPU_STAT(irq_window_exits) },
>  	{ "nmi_window", VCPU_STAT(nmi_window_exits) },
>  	{ "halt_exits", VCPU_STAT(halt_exits) },
> +	{ "halt_successful_poll", VCPU_STAT(halt_successful_poll) },
>  	{ "halt_wakeup", VCPU_STAT(halt_wakeup) },
>  	{ "hypercalls", VCPU_STAT(hypercalls) },
>  	{ "request_irq", VCPU_STAT(request_irq_exits) },
> @@ -5819,13 +5823,29 @@ void kvm_arch_exit(void)
>  int kvm_emulate_halt(struct kvm_vcpu *vcpu)
>  {
>  	++vcpu->stat.halt_exits;
> -	if (irqchip_in_kernel(vcpu->kvm)) {
> -		vcpu->arch.mp_state = KVM_MP_STATE_HALTED;
> -		return 1;
> -	} else {
> +	if (!irqchip_in_kernel(vcpu->kvm)) {
>  		vcpu->run->exit_reason = KVM_EXIT_HLT;
>  		return 0;
>  	}
> +
> +	vcpu->arch.mp_state = KVM_MP_STATE_HALTED;
> +	if (halt_poll) {
> +		u64 start, curr;
> +		rdtscll(start);
> +		do {
> +			/*
> +			 * This sets KVM_REQ_UNHALT if an interrupt
> +			 * arrives.
> +			 */
> +			if (kvm_vcpu_check_block(vcpu) < 0) {
> +				++vcpu->stat.halt_successful_poll;
> +				break;
> +			}
> +			rdtscll(curr);
> +		} while(!need_resched() && curr - start < halt_poll);
> +	}
> +
> +	return 1;
>  }

You want at least a basic procedure to estimate a value
(its a function of the device after all).

Rather than halt_successful_poll's, i suppose the optimum 
can be estimated from a dataset containing entries
in the form:

interrupt time - hlt time

Then choose a given value from that table.

You can get the same out of halt_successful_poll, 
but requires multiple runs of the test:

Set halt_poll, run test, record halt_successful_poll.
Set halt_poll, run test, record halt_successful_poll.
Set halt_poll, run test, record halt_successful_poll.
...

A crude histogram also works, to avoid recording all "interrupt time -
hlt" entries and processing them in userspace.