Hi Reiji, On Thu, Dec 29, 2022 at 07:59:21PM -0800, Reiji Watanabe wrote: > The goal of this series is to allow userspace to limit the number > of PMU event counters on the vCPU. > > The number of PMU event counters is indicated in PMCR_EL0.N. > For a vCPU with PMUv3 configured, its value will be the same as > the host value by default. Userspace can set PMCR_EL0.N for the > vCPU to a lower value than the host value, using KVM_SET_ONE_REG. > However, it is practically unsupported, as KVM resets PMCR_EL0.N > to the host value on vCPU reset and some KVM code uses the host > value to identify (un)implemented event counters on the vCPU. > > This series will ensure that the PMCR_EL0.N value is preserved > on vCPU reset and that KVM doesn't use the host value > to identify (un)implemented event counters on the vCPU. > This allows userspace to limit the number of the PMU event > counters on the vCPU. I just wanted to bring up the conversation we had today on the list as it is a pretty relevant issue. KVM currently allows any value to be written to PMCR_EL0.N, meaning that userspace could advertize more PMCs than are supported by the system. IDK if Marc feels otherwise, but it doesn't seem like we should worry about ABI change here (i.e. userspace can no longer write junk to the register) as KVM has advertized the correct value to userspace. The only case that breaks would be a userspace that intentionally sets PMCR_EL0.N to something larger than the host. As accesses to unadvertized PMC indices is CONSTRAINED UNPRED behavior, I'm having a hard time coming up with a use case. -- Thanks, Oliver
