The Arm SMCCC is rather prescriptive in regards to the allocation of SMCCC function ID ranges. Many of the hypercall ranges have an associated specification from Arm (FF-A, PSCI, SDEI, etc.) with some room for vendor-specific implementations. The ever-expanding hypercall surface leaves a lot of work within KVM for providing new features. Furthermore, KVM implements its own vendor-specific ABI, leaving little room for other implementations (like Hyper-V, for example). This series takes a stab at both by allowing SMCCC ranges to shunt to userspace. The meat of the UAPI is a bitmap (conveyed through a VM capability) that creates trap bits for known SMCCC ranges. Exits are done by reusing the KVM_EXIT_HYPERCALL interface, with its hilariously x86-specific kvm_run structure (no longmode here!) => We have these new hypercall bitmap registers, why not use that? The hypercall bitmap registers aren't necessarily aimed at the same problem. The bitmap registers allow a VMM to preserve the ABI the guest gets from KVM by default when migrating between hosts. By default KVM exposes the entire feature set to the guest, whereas user hypercalls need explicit opt-in from userspace. TODO: - SMCCC_ARCH_FEATURES is pitifully overloaded to allow discovery of paravirtualized time features. Queries to SMCCC_ARCH_FEATURES against PV time should go to userspace if userspace has PV time trapped. - May want to use a completion on reentry for propagating the return value to avoid extra ioctls (SMCCC calls can return more than a single register value). Tossing up on the list as an RFC to gauge interest in such an interface. Gently tested with the included selftest and that's about it. Applies to v6.1-rc3 + the header relocation [1]. No firm reason for it besides the fact that I initially did the refactoring as part of this series and am too lazy to rebase. [1] https://lore.kernel.org/kvmarm/20221110010603.3382385-1-oliver.upton@xxxxxxxxx/ Cc: Marc Zyngier <maz@xxxxxxxxxx> Cc: James Morse <james.morse@xxxxxxx> Cc: Alexandru Elisei <alexandru.elisei@xxxxxxx> Cc: Suzuki K Poulose <suzuki.poulose@xxxxxxx> Cc: kvmarm@xxxxxxxxxxxxxxx Cc: kvmarm@xxxxxxxxxxxxxxxxxxxxx Oliver Upton (3): KVM: arm64: Use a generalized accessor for SMCCC args KVM: arm64: Allow userspace to trap SMCCC sub-ranges KVM: selftests: Test user hypercalls arch/arm64/include/asm/kvm_host.h | 5 + arch/arm64/include/asm/kvm_hypercalls.h | 22 ++- arch/arm64/include/uapi/asm/kvm.h | 15 ++ arch/arm64/kvm/arm.c | 10 ++ arch/arm64/kvm/hypercalls.c | 52 ++++++- arch/arm64/kvm/psci.c | 14 +- arch/arm64/kvm/pvtime.c | 2 +- arch/arm64/kvm/trng.c | 4 +- include/uapi/linux/kvm.h | 1 + tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 1 + .../selftests/kvm/aarch64/user_hypercalls.c | 130 ++++++++++++++++++ 12 files changed, 231 insertions(+), 26 deletions(-) create mode 100644 tools/testing/selftests/kvm/aarch64/user_hypercalls.c -- 2.38.1.431.g37b22c650d-goog _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
