ljrcore
commented
1 month ago Perf kvm depends on this change, and perf related code will be contributed after optimization and cleanup.
Analyze events for all VMs, all VCPUs:
VM-EXIT Samples Samples% Time% Min Time Max Time Avg time
HC_WRMSR 183 33.46% 0.01% 0.28us 4.41us 0.66us ( +- 6.03% )
PF excp 117 21.39% 2.38% 0.78us 34187.94us 294.36us ( +- 99.26% )
ERETS 95 17.37% 0.01% 0.60us 12.55us 0.96us ( +- 12.96% )
HC_IRQ_HALT 76 13.89% 95.11% 1022.33us 351940.05us 18130.74us ( +- 29.91% )
HC_RDMSR 28 5.12% 0.00% 0.33us 0.78us 0.45us ( +- 4.89% )
INTERRUPT 27 4.94% 2.50% 0.73us 36012.79us 1339.30us ( +- 99.57% )
ERETU 14 2.56% 0.00% 0.46us 0.98us 0.61us ( +- 6.90% )
HC_TLB_INVLPG 3 0.55% 0.00% 0.88us 1.86us 1.53us ( +- 21.41% )
HC_LOAD_PGTBL 2 0.37% 0.00% 7.39us 18.36us 12.88us ( +- 42.60% )
HC_IRQ_WIN 1 0.18% 0.00% 1.11us 1.11us 1.11us ( +- 0.00% )
HC_LOAD_GS 1 0.18% 0.00% 0.55us 0.55us 0.55us ( +- 0.00% )
Total Samples:547, Total events handled time:1448802.46us.
Improve the handling of exit reasons for PVM hypercalls by adding more specific exit reasons based on the type of PVM hypercall. This enables a more fine-grained analysis of the performance and behavior of PVM-based virtual machines.
This change benefits performance analysis tools, such as perf, by allowing them to provide more detailed information about the execution of PVM hypercalls.
https://github.com/virt-pvm/linux/pull/9