The time to switch a VM to Secure-VM, increases by the size of the VM.
A 100GB VM takes about 7minutes. This is unacceptable. This linear
increase is caused by a suboptimal behavior by the Ultravisor and the
Hypervisor. The Ultravisor unnecessarily migrates all the GFN of the
VM from normal-memory to secure-memory. It has to just migrate the
necessary and sufficient GFNs.
However when the optimization is incorporated in the Ultravisor, the
Hypervisor starts misbehaving. The Hypervisor has a inbuilt assumption
that the Ultravisor will explicitly request to migrate, each and every
GFN of the VM. If only necessary and sufficient GFNs are requested for
migration, the Hypervisor continues to manage the remaining GFNs as
normal GFNs. This leads to memory corruption; manifested
consistently when the SVM reboots.
The same is true, when a memory slot is hotplugged into a SVM. The
Hypervisor expects the ultravisor to request migration of all GFNs to
secure-GFN. But the hypervisor cannot handle any H_SVM_PAGE_IN
requests from the Ultravisor, done in the context of
UV_REGISTER_MEM_SLOT ucall. This problem manifests as random errors
in the SVM, when a memory-slot is hotplugged.
This patch series automatically migrates the non-migrated pages of a
SVM, and thus solves the problem.
Testing: Passed rigorous testing using various sized SVMs.
Changelog:
v5: . This patch series includes Laurent's fix for memory hotplug/unplug
. drop pages first and then delete the memslot. Otherwise
the memslot does not get cleanly deleted, causing
problems during reboot.
. recreatable through the following set of commands
. device_add pc-dimm,id=dimm1,memdev=mem1
. device_del dimm1
. device_add pc-dimm,id=dimm1,memdev=mem1
Further incorporates comments from Bharata:
. fix for off-by-one while disabling migration.
. code-reorganized to maximize sharing in init_start path
and in memory-hotplug path
. locking adjustments in mass-page migration during H_SVM_INIT_DONE.
. improved recovery on error paths.
. additional comments in the code for better understanding.
. removed the retry-on-migration-failure code.
. re-added the initial patch that adjust some prototype to overcome
a git problem, where it messes up the code context. Had
accidently dropped the patch in the last version.
v4: . Incorported Bharata's comments:
- Optimization -- replace write mmap semaphore with read mmap semphore.
- disable page-merge during memory hotplug.
- rearranged the patches. consolidated the page-migration-retry logic
in a single patch.
v3: . Optimized the page-migration retry-logic.
. Relax and relinquish the cpu regularly while bulk migrating
the non-migrated pages. This issue was causing soft-lockups.
Fixed it.
. Added a new patch, to retry page-migration a couple of times
before returning H_BUSY in H_SVM_PAGE_IN. This issue was
seen a few times in a 24hour continuous reboot test of the SVMs.
v2: . fixed a bug observed by Laurent. The state of the GFN's associated
with Secure-VMs were not reset during memslot flush.
. Re-organized the code, for easier review.
. Better description of the patch series.
v1: . fixed a bug observed by Bharata. Pages that where paged-in and later
paged-out must also be skipped from migration during H_SVM_INIT_DONE.
Laurent Dufour (3):
KVM: PPC: Book3S HV: migrate hot plugged memory
KVM: PPC: Book3S HV: move kvmppc_svm_page_out up
KVM: PPC: Book3S HV: rework secure mem slot dropping
Ram Pai (4):
KVM: PPC: Book3S HV: Fix function definition in book3s_hv_uvmem.c
KVM: PPC: Book3S HV: Disable page merging in H_SVM_INIT_START
KVM: PPC: Book3S HV: track the state GFNs associated with secure VMs
KVM: PPC: Book3S HV: in H_SVM_INIT_DONE, migrate remaining normal-GFNs
to secure-GFNs.
Documentation/powerpc/ultravisor.rst | 3 +
arch/powerpc/include/asm/kvm_book3s_uvmem.h | 16 +
arch/powerpc/kvm/book3s_hv.c | 10 +-
arch/powerpc/kvm/book3s_hv_uvmem.c | 690 +++++++++++++++++++++-------
4 files changed, 548 insertions(+), 171 deletions(-)
--
1.8.3.1
