This patch set improves the shadow TLB handling of our e500
target.

The really important bit here is that with these patches applied,
we can map guest TLB1 entries into the host's TLB0. This gives a
significant performance improvement as you can see below.

Alex

v1 -> v2:

  - new patch: Move write_stlbe higher
  - new patch: Explicitly mark shadow maps invalid
  - new patch: Propagate errors when shadow mapping
  - remove force init logic, it's handled by the explicit invalidate now
  - leave stlbe uninitialized
  - allow mixing of TLB0 and TLB1_BITMAP backing of TLB1 pages

---

without patch, using 4k backed memory:

$ time for i in {1..1000}; do /bin/echo > /dev/null; done
real    0m12.947s
user    0m1.076s
sys     0m9.720s


with hugetlbfs:

$ time for i in {1..1000}; do /bin/echo > /dev/null; done

real    0m3.262s
user    0m0.464s
sys     0m0.236s


with patches applied, using 4k backed memory:

$ time for i in {1..1000}; do /bin/echo > /dev/null; done

real    0m4.446s
user    0m0.380s
sys     0m0.644s

Alexander Graf (6):
  KVM: PPC: E500: Move write_stlbe higher
  KVM: PPC: E500: Explicitly mark shadow maps invalid
  KVM: PPC: E500: Propagate errors when shadow mapping
  KVM: PPC: e500: Call kvmppc_mmu_map for initial mapping
  KVM: PPC: E500: Split host and guest MMU parts
  KVM: PPC: e500: Implement TLB1-in-TLB0 mapping

 arch/powerpc/kvm/Makefile        |    9 +-
 arch/powerpc/kvm/e500.h          |    1 +
 arch/powerpc/kvm/e500_mmu.c      |  811 +++++++++++++++++++++
 arch/powerpc/kvm/e500_mmu_host.c |  699 +++++++++++++++++++
 arch/powerpc/kvm/e500_mmu_host.h |   20 +
 arch/powerpc/kvm/e500_tlb.c      | 1430 --------------------------------------
 6 files changed, 1537 insertions(+), 1433 deletions(-)
 create mode 100644 arch/powerpc/kvm/e500_mmu.c
 create mode 100644 arch/powerpc/kvm/e500_mmu_host.c
 create mode 100644 arch/powerpc/kvm/e500_mmu_host.h
 delete mode 100644 arch/powerpc/kvm/e500_tlb.c

--
To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to