On Tue, Sep 1, 2015 at 3:30 PM, Wanpeng Li <wanpeng.li@xxxxxxxxxxx> wrote:
On 9/2/15 5:45 AM, David Matlack wrote:
On Thu, Aug 27, 2015 at 2:47 AM, Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
v3 -> v4:I posted a patchset which adds dynamic poll toggling (on/off switch). I
* bring back grow vcpu->halt_poll_ns when interrupt arrives and shrinks
when idle VCPU is detected
v2 -> v3:
* grow/shrink vcpu->halt_poll_ns by *halt_poll_ns_grow or
* drop the macros and hard coding the numbers in the param definitions
* update the comments "5-7 us"
* remove halt_poll_ns_max and use halt_poll_ns as the max halt_poll_ns
vcpu->halt_poll_ns start at zero
* drop the wrappers
* move the grow/shrink logic before "out:" w/ "if (waited)"
this gives you a good place to build your dynamic growth patch on top. The
toggling patch has close to zero overhead for idle VMs and equivalent
performance VMs doing message passing as always-poll. It's a patch that's
in my queue for a few weeks but just haven't had the time to send out. We
win even more with your patchset by only polling as much as we need (via
dynamic growth/shrink). It also gives us a better place to stand for
a default for halt_poll_ns. (We can run experiments and see how high
vcpu->halt_poll_ns tends to grow.)
The reason I posted a separate patch for toggling is because it adds
to kvm_vcpu_block and deals with a weird edge case (kvm_vcpu_block can get
called multiple times for one halt). To do dynamic poll adjustment
we have to time the length of each halt. Otherwise we hit some bad edge
v3: v3 had lots of idle overhead. It's because vcpu->halt_poll_ns grew
time we had a long halt. So idle VMs looked like: 0 us -> 500 us -> 1
2 ms -> 4 ms -> 0 us. Ideally vcpu->halt_poll_ns should just stay at 0
the halts are long.
v4: v4 fixed the idle overhead problem but broke dynamic growth for
passing VMs. Every time a VM did a short halt, vcpu->halt_poll_ns would
That means vcpu->halt_poll_ns will always be maxed out, even when the
time is much less than the max.
I think we can fix both edge cases if we make grow/shrink decisions based
the length of kvm_vcpu_block rather than the arrival of a guest interrupt
Some thoughts for dynamic growth:
* Given Windows 10 timer tick (1 ms), let's set the maximum poll time
less than 1ms. 200 us has been a good value for always-poll. We can
probably go a bit higher once we have your patch. Maybe 500 us?
Install the netperf package, or build from here:
* The base case of dynamic growth (the first grow() after being at 0)
be small. 500 us is too big. When I run TCP_RR in my guest I see poll
of < 10 us. TCP_RR is on the lower-end of message passing workload
so 10 us would be a good base case.
How to get your TCP_RR benchmark?
In the vm:
# ./netperf -t TCP_RR
Be sure to use an SMP guest (we want TCP_RR to be a cross-core message
passing workload in order to test halt-polling).