On 2017/7/3 18:06, Thomas Gleixner wrote:
On Mon, 3 Jul 2017, Yang Zhang wrote:The background is that we(Alibaba Cloud) do get more and more complaints from our customers in both KVM and Xen compare to bare-mental.After investigations, the root cause is known to us: big cost in message passing workload(David show it in KVM forum 2015) A typical message workload like below: vcpu 0 vcpu 1 1. send ipi 2. doing hlt 3. go into idle 4. receive ipi and wake up from hlt 5. write APIC time twice 6. write APIC time twice to to stop sched timer reprogram sched timer 7. doing hlt 8. handle task and send ipi to vcpu 0 9. same to 4. 10. same to 3 One transaction will introduce about 12 vmexits(2 hlt and 10 msr write). The cost of such vmexits will degrades performance severely. Linux kernel already provide idle=poll to mitigate the trend. But it only eliminates the IPI and hlt vmexit. It has nothing to do with start/stop sched timer. A compromise would be to turn off NOHZ kernel, but it is not the default config for new distributions.You still can turn if off on the kernel command line via nohz=off
You are right. Senior users will turn off it manually. But it only solve the sched timer. They still have the IPI/hlt problem. Another point is we release the distribution image to customer without any extra configuration to avoid mismatch between VM and bare-metal. To change such configuration needs reboot, but some customer's business cannot be interrupted after they start the service(like online gaming). It would be better if we can provide the sysctl interface to allow run-time modification. By the way, idle=poll seems too heavy to use.
Thanks, tglx
-- Yang Alibaba Cloud Computing -- To unsubscribe from this list: send the line "unsubscribe linux-doc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
