This is an RFC series for fast migration of confidential guests using an in-guest migration helper that lives in OVMF. QEMU VM live migration needs to read source VM's RAM and write it in the target VM; this mechanism doesn't work when the guest memory is encrypted or QEMU is prevented from reading it in another way. In order to support live migration in such scenarios, we introduce an in-guest migration helper which can securely extract RAM content from the guest in order to send it to the target. The migration helper is implemented as part of the VM's firmware in OVMF.
We've implemented and tested this on AMD SEV, but expect most of the processes can be used with other technologies that prevent direct access of hypervisor to the guest's memory. Specifically, we don't use SEV's PSP migration commands (SEV_SEND_START, SEV_RECEIVE_START, etc) at all. Corresponding RFC patches for OVMF have been posted by my colleague Tobin Feldman-Fitzthum on edk2-devel. Those include the crux of the migration helper: a mailbox protocol over a shared memory page which allows communication between QEMU and the migration helper. In the source VM this is used to read a page and encrypt it for transport; in the target it is used to decrypt the incoming page and storing the content in the correct address in the guest memory. All encryption and decryption operations occur inside the trusted context in the VM, and therefore the VM's memory plaintext content is never accessible to the hosts participating in the migration. In order to allow OVMF to run the migration helper in parallel to the guest OS, we introduce the notion of auxiliary vcpus, which are usable for OVMF but are hidden from the guest OS. These might have other future uses for in-guest operations/agents. In the target VM we need the migration handler running to receive incoming RAM pages; to achieve that, we boot the VM into OVMF with a special fw_cfg value that causes OVMF to not boot the guest OS; we then allow QEMU to receive an incoming migration by issuing a new start-migrate-incoming QMP command. The confidential RAM migration requires checking whether a given guest RAM page is encrypted or not. This is currently achieved using AMD's patches which track the encryption status of guest pages in KVM, using hypercalls from OVMF and guest Linux to report changes of such status. The QEMU side of these patches is included as the first two patches in this series. The concrete implementation of this page encryption tracking is currently in flux in the KVM mailing list, but the underlying implementation doesn't affect our confidential RAM migration as long as it can check whether a given guest address is encrypted. List of patches in this series: 1-2: reposting AMD encrypted page bitmap support. 3-11: introduce the notion of auxiliary vcpus. 12-21: introduce the migration specifics. 22-23: fix devices issues when loading state into a live VM 24: introduce the start-migrate-incoming QMP command to switch the target into accepting the incoming migration. 25: remove SEV migration blocker 26: add documentation Brijesh Singh (1): kvm: add support to sync the page encryption state bitmap Dov Murik (21): linux-headers: Add definitions of KVM page encryption bitmap ioctls machine: Add auxcpus=N suboption to -smp hw/boards: Add aux flag to CPUArchId hw/i386: Mark auxiliary vcpus in possible_cpus cpu: Add boolean aux field to CPUState hw/i386: Set CPUState.aux=true for auxiliary vcpus softmmu: Don't sync aux vcpus in pre_loadvm softmmu: Add cpu_synchronize_without_aux_post_init migration: Add helpers to save confidential RAM migration: Add helpers to load confidential RAM migration: Introduce gpa_inside_migration_helper_shared_area migration: Save confidential guest RAM using migration helper migration: Load confidential guest RAM using migration helper migration: Stop VM after loading confidential RAM migration: Don't sync vcpus when migrating confidential guests migration: When starting target, don't sync auxiliary vcpus hw/isa/lpc_ich9: Allow updating an already-running VM target/i386: Re-sync kvm-clock after confidential guest migration migration: Add start-migrate-incoming QMP command target/i386: SEV: Allow migration unless there are no aux vcpus docs: Add confidential guest live migration documentation Tobin Feldman-Fitzthum (4): hw/acpi: Don't include auxiliary vcpus in ACPI tables softmmu: Add pause_all_vcpus_except_aux migration: Stop non-aux vcpus before copying the last pages migration: Call migration handler cleanup routines docs/confidential-guest-live-migration.rst | 142 ++++++++++++ docs/confidential-guest-support.txt | 5 + docs/index.rst | 1 + qapi/migration.json | 26 +++ include/exec/ram_addr.h | 197 ++++++++++++++++ include/exec/ramblock.h | 3 + include/exec/ramlist.h | 3 +- include/hw/boards.h | 3 + include/hw/core/cpu.h | 2 + include/hw/i386/x86.h | 2 +- include/sysemu/cpus.h | 2 + linux-headers/linux/kvm.h | 13 ++ migration/confidential-ram.h | 23 ++ accel/kvm/kvm-all.c | 43 ++++ hw/acpi/cpu.c | 10 + hw/core/cpu.c | 1 + hw/core/machine.c | 7 + hw/i386/acpi-build.c | 5 + hw/i386/acpi-common.c | 5 + hw/i386/pc.c | 7 + hw/i386/x86.c | 10 +- hw/isa/lpc_ich9.c | 3 +- migration/confidential-ram.c | 258 +++++++++++++++++++++ migration/migration.c | 18 +- migration/ram.c | 135 ++++++++++- migration/savevm.c | 13 +- softmmu/cpus.c | 68 +++++- softmmu/runstate.c | 1 + softmmu/vl.c | 3 + target/i386/machine.c | 9 + target/i386/sev.c | 25 +- migration/meson.build | 6 +- migration/trace-events | 4 + 33 files changed, 1027 insertions(+), 26 deletions(-) create mode 100644 docs/confidential-guest-live-migration.rst create mode 100644 migration/confidential-ram.h create mode 100644 migration/confidential-ram.c base-commit: 00d8ba9e0d62ea1c7459c25aeabf9c8bb7659462 -- 2.20.1
