On 12/01/2016 01:28 PM, David Hildenbrand wrote: > Am 28.11.2016 um 10:06 schrieb Christian Borntraeger: >> Right now we save the host access registers in kvm_arch_vcpu_load >> and load them in kvm_arch_vcpu_put. Vice versa for the guest access >> registers. On schedule this means, that we load/save access registers >> multiple times. > > Wasn't the main point about this to also protect from scheduling > another VCPU (maybe even from another VM)? (which in return would have > its own set of access registers and therefore would now use the wrong > set of access registers) The thing is that the scheduler already loads and save old/new registers (see arch/s390/include/asm/switch_to.h), so the new process will get its correct registers. The only difference is that we store in prev->thread.acrs[0] now the guest registers of prev instead of the host registers of prev. But the host registers have been taken care of by KVM. > > Must likely I am forgetting something very important here :) > -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
