The PSCI v1.0 specification describes a call, SYSTEM_SUSPEND, which allows software to request that the system be placed into the lowest possible power state and await a wakeup event. This call is optional in v1.0 and v1.1. KVM does not currently support this optional call. This series adds support for the PSCI SYSTEM_SUSPEND call to KVM/arm64. For reasons best described in patch 8, it is infeasible to correctly implement PSCI SYSTEM_SUSPEND (or any system-wide event for that matter) in a split design between kernel/userspace. As such, this series cheaply exits to userspace so it can decide what to do with the call. This series also gives userspace some help to emulate suspension with a new MP state that awaits an unmasked pending interrupt. Patches 1-5 are small reworks to more easily shoehorn the new features into the kernel. Patch 6 stands up the new suspend MP state, allowing userspace to emulate the PSCI call. Patch 7 actually allows userspace to enable the PSCI call, which requires explicit opt-in for the new KVM_EXIT_SYSTEM_EVENT type. Patches 8-11 clean up the way PSCI is tested in selftests to more easily add new test cases. Finally, the last patch actually tests that PSCI SYSTEM_SUSPEND calls within the guest result in userspace exits. Applies cleanly to 5.18-rc5. I'm sure you're already aware of it Marc, but for the sake of everyone else there's some light conflict with Raghu's patches that you've got queued up [1]. Tested with the included selftest and a hacked up kvmtool [2] with support for the new UAPI. [1]: https://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms.git/log/?h=kvm-arm64/hcall-selection [2]: https://lore.kernel.org/all/20220311175717.616958-1-oupton@xxxxxxxxxx/ v5: http://lore.kernel.org/r/20220311174001.605719-1-oupton@xxxxxxxxxx v5 -> v6: - Rebase to 5.18-rc5 - Collect Reiji's R-b's - Drop the system_event helper. Since we now have variadic data returning to userspace it doesn't make much sense to roll it up into a helper. Meh. - Put back the pointless kvm_vcpu_request() in kvm_arm_vcpu_suspend(). We'll rip out the reliance on vCPU requests for power state later on. It is entirely benign, even when a vCPU targets itself. Oliver Upton (12): KVM: arm64: Don't depend on fallthrough to hide SYSTEM_RESET2 KVM: arm64: Dedupe vCPU power off helpers KVM: arm64: Track vCPU power state using MP state values KVM: arm64: Rename the KVM_REQ_SLEEP handler KVM: arm64: Return a value from check_vcpu_requests() KVM: arm64: Add support for userspace to suspend a vCPU KVM: arm64: Implement PSCI SYSTEM_SUSPEND selftests: KVM: Rename psci_cpu_on_test to psci_test selftests: KVM: Create helper for making SMCCC calls selftests: KVM: Use KVM_SET_MP_STATE to power off vCPU in psci_test selftests: KVM: Refactor psci_test to make it amenable to new tests selftests: KVM: Test SYSTEM_SUSPEND PSCI call Documentation/virt/kvm/api.rst | 76 ++++++- arch/arm64/include/asm/kvm_host.h | 10 +- arch/arm64/kvm/arm.c | 104 +++++++-- arch/arm64/kvm/psci.c | 65 ++++-- include/uapi/linux/kvm.h | 4 + tools/testing/selftests/kvm/.gitignore | 2 +- tools/testing/selftests/kvm/Makefile | 2 +- .../selftests/kvm/aarch64/psci_cpu_on_test.c | 121 ---------- .../testing/selftests/kvm/aarch64/psci_test.c | 213 ++++++++++++++++++ .../selftests/kvm/include/aarch64/processor.h | 22 ++ .../selftests/kvm/lib/aarch64/processor.c | 25 ++ tools/testing/selftests/kvm/steal_time.c | 13 +- 12 files changed, 480 insertions(+), 177 deletions(-) delete mode 100644 tools/testing/selftests/kvm/aarch64/psci_cpu_on_test.c create mode 100644 tools/testing/selftests/kvm/aarch64/psci_test.c -- 2.36.0.464.gb9c8b46e94-goog
