linux.git/arch/x86/kernel/kvm.c, branch v4.18

kvm: rename KVM_HINTS_DEDICATED to KVM_HINTS_REALTIME

2018-05-17T17:12:13+00:00

KVM_HINTS_DEDICATED seems to be somewhat confusing:

Guest doesn't really care whether it's the only task running on a host
CPU as long as it's not preempted.

And there are more reasons for Guest to be preempted than host CPU
sharing, for example, with memory overcommit it can get preempted on a
memory access, post copy migration can cause preemption, etc.

Let's call it KVM_HINTS_REALTIME which seems to better
match what guests expect.

Also, the flag most be set on all vCPUs - current guests assume this.
Note so in the documentation.

Signed-off-by: Michael S. Tsirkin 
Signed-off-by: Paolo Bonzini

Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

2018-04-09T18:42:31+00:00

Pull kvm updates from Paolo Bonzini:
 "ARM:
   - VHE optimizations

   - EL2 address space randomization

   - speculative execution mitigations ("variant 3a", aka execution past
     invalid privilege register access)

   - bugfixes and cleanups

  PPC:
   - improvements for the radix page fault handler for HV KVM on POWER9

  s390:
   - more kvm stat counters

   - virtio gpu plumbing

   - documentation

   - facilities improvements

  x86:
   - support for VMware magic I/O port and pseudo-PMCs

   - AMD pause loop exiting

   - support for AMD core performance extensions

   - support for synchronous register access

   - expose nVMX capabilities to userspace

   - support for Hyper-V signaling via eventfd

   - use Enlightened VMCS when running on Hyper-V

   - allow userspace to disable MWAIT/HLT/PAUSE vmexits

   - usual roundup of optimizations and nested virtualization bugfixes

  Generic:
   - API selftest infrastructure (though the only tests are for x86 as
     of now)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (174 commits)
  kvm: x86: fix a prototype warning
  kvm: selftests: add sync_regs_test
  kvm: selftests: add API testing infrastructure
  kvm: x86: fix a compile warning
  KVM: X86: Add Force Emulation Prefix for "emulate the next instruction"
  KVM: X86: Introduce handle_ud()
  KVM: vmx: unify adjacent #ifdefs
  x86: kvm: hide the unused 'cpu' variable
  KVM: VMX: remove bogus WARN_ON in handle_ept_misconfig
  Revert "KVM: X86: Fix SMRAM accessing even if VM is shutdown"
  kvm: Add emulation for movups/movupd
  KVM: VMX: raise internal error for exception during invalid protected mode state
  KVM: nVMX: Optimization: Dont set KVM_REQ_EVENT when VMExit with nested_run_pending
  KVM: nVMX: Require immediate-exit when event reinjected to L2 and L1 event pending
  KVM: x86: Fix misleading comments on handling pending exceptions
  KVM: x86: Rename interrupt.pending to interrupt.injected
  KVM: VMX: No need to clear pending NMI/interrupt on inject realmode interrupt
  x86/kvm: use Enlightened VMCS when running on Hyper-V
  x86/hyper-v: detect nested features
  x86/hyper-v: define struct hv_enlightened_vmcs and clean field bits
  ...

KVM: X86: Fix setup the virt_spin_lock_key before static key get initialized

2018-03-28T20:47:06+00:00

 static_key_disable_cpuslocked(): static key 'virt_spin_lock_key+0x0/0x20' used before call to jump_label_init()
 WARNING: CPU: 0 PID: 0 at kernel/jump_label.c:161 static_key_disable_cpuslocked+0x61/0x80
 RIP: 0010:static_key_disable_cpuslocked+0x61/0x80
 Call Trace:
  static_key_disable+0x16/0x20
  start_kernel+0x192/0x4b3
  secondary_startup_64+0xa5/0xb0

Qspinlock will be choosed when dedicated pCPUs are available, however, the
static virt_spin_lock_key is set in kvm_spinlock_init() before jump_label_init()
has been called, which will result in a WARN(). This patch fixes it by delaying
the virt_spin_lock_key setup to .smp_prepare_cpus().

Reported-by: Davidlohr Bueso 
Cc: Paolo Bonzini 
Cc: Radim Krčmář 
Cc: Davidlohr Bueso 
Signed-off-by: Wanpeng Li 
Fixes: b2798ba0b876 ("KVM: X86: Choose qspinlock when dedicated physical CPUs are available")
Signed-off-by: Radim Krčmář

KVM: x86: Fix pv tlb flush dependencies

2018-03-28T13:53:34+00:00

PV TLB FLUSH can only be turned on when steal time is enabled.
The condition got reversed during conflict resolution.

Cc: Paolo Bonzini 
Cc: Radim Krčmář 
Signed-off-by: Wanpeng Li 
Fixes: 4f2f61fc5071 ("KVM: X86: Avoid traversing all the cpus for pv tlb flush when steal time is disabled")
[Rebased on top of kvm/master and reworded the commit message. - Radim]
Signed-off-by: Radim Krčmář

KVM: X86: Don't use PV TLB flush with dedicated physical CPUs

2018-03-06T17:40:45+00:00

vCPUs are very unlikely to get preempted when they are the only task
running on a CPU.  PV TLB flush is slower that the native flush in that
case, so disable it.

Cc: Paolo Bonzini 
Cc: Radim Krčmář 
Cc: Eduardo Habkost 
Signed-off-by: Wanpeng Li 
Signed-off-by: Paolo Bonzini 
Signed-off-by: Radim Krčmář

KVM: X86: Choose qspinlock when dedicated physical CPUs are available

2018-03-06T17:40:45+00:00

Waiman Long mentioned that:
> Generally speaking, unfair lock performs well for VMs with a small
> number of vCPUs. Native qspinlock may perform better than pvqspinlock
> if there is vCPU pinning and there is no vCPU over-commitment.

This patch uses the KVM_HINTS_DEDICATED performance hint, which is
provided by the hypervisor admin, to choose the qspinlock algorithm
when a dedicated physical CPU is available.

PV_DEDICATED = 1, PV_UNHALT = anything: default is qspinlock
PV_DEDICATED = 0, PV_UNHALT = 1: default is Hybrid PV queued/unfair lock
PV_DEDICATED = 0, PV_UNHALT = 0: default is tas

Cc: Paolo Bonzini 
Cc: Radim Krčmář 
Cc: Eduardo Habkost 
Signed-off-by: Wanpeng Li 
Signed-off-by: Paolo Bonzini 
Signed-off-by: Radim Krčmář

KVM: Introduce paravirtualization hints and KVM_HINTS_DEDICATED

2018-03-06T17:40:44+00:00

This patch introduces kvm_para_has_hint() to query for hints about
the configuration of the guests.  The first hint KVM_HINTS_DEDICATED,
is set if the guest has dedicated physical CPUs for each vCPU (i.e.
pinning and no over-commitment).  This allows optimizing spinlocks
and tells the guest to avoid PV TLB flush.

Cc: Paolo Bonzini 
Cc: Radim Krčmář 
Cc: Eduardo Habkost 
Signed-off-by: Wanpeng Li 
Signed-off-by: Paolo Bonzini 
Signed-off-by: Radim Krčmář

KVM: X86: Avoid traversing all the cpus for pv tlb flush when steal time is disabled

2018-02-24T00:43:49+00:00

Avoid traversing all the cpus for pv tlb flush when steal time
is disabled since pv tlb flush depends on the field in steal time
for shared data.

Cc: Paolo Bonzini 
Cc: Radim KrÄmÃ¡Å™ 
Signed-off-by: Wanpeng Li 
Signed-off-by: Paolo Bonzini

x86/kvm: Make parse_no_xxx __init for kvm

2018-02-24T00:43:48+00:00

The early_param() is only called during kernel initialization, So Linux
marks the functions of it with __init macro to save memory.

But it forgot to mark the parse_no_kvmapf/stealacc/kvmclock_vsyscall,
So, Make them __init as well.

Cc: Paolo Bonzini 
Cc: rkrcmar@redhat.com
Cc: kvm@vger.kernel.org
Cc: Thomas Gleixner 
Cc: Ingo Molnar 
Cc: "H. Peter Anvin" 
Cc: Juergen Gross 
Cc: x86@kernel.org
Signed-off-by: Dou Liyang 
Signed-off-by: Paolo Bonzini

KVM: x86: fix backward migration with async_PF

2018-02-24T00:43:48+00:00

Guests on new hypersiors might set KVM_ASYNC_PF_DELIVERY_AS_PF_VMEXIT
bit when enabling async_PF, but this bit is reserved on old hypervisors,
which results in a failure upon migration.

To avoid breaking different cases, we are checking for CPUID feature bit
before enabling the feature and nothing else.

Fixes: 52a5c155cf79 ("KVM: async_pf: Let guest support delivery of async_pf from guest mode")
Cc: 
Reviewed-by: Wanpeng Li 
Reviewed-by: David Hildenbrand 
Signed-off-by: Radim Krčmář 
Signed-off-by: Paolo Bonzini