[MODERATED] [RFC PATCH 0/6] kvm: handle host mode irqs 0

[Nicolai Stange <nstange@suse.de>]

Hi all,

there had been some discussion on whether or not IRQs that happened between
vmexit and the next vmenter should trigger L1d flushes, c.f. [1] and [2].

This series implements what Paolo suggested at [2]:

> handle at least the common case of softirqs by adding some kind of
> "# of times softirqs were run" percpu variable.

I chose to track irqs instead of softirqs only.

[1/6] - [3/6] are preparatory
[4/6] implements the "# of time irqs were run" percpu variable
[5/6] makes vmx issue L1d flushes if that count has changed
[6/6] is a followup cleanup to that

Each patch has been compile-tested on i686 and x86_64. However, only the
latter received some runtime testing.

Thanks,

Nicolai

[1] dc14f2c1-f69c-e690-dad2-8b4f1b772ee3@suse.de
[2] 3270ad99-2008-ccea-8e0a-d3fbf7fc4e50@redhat.com


Nicolai Stange (6):
  x86/KVM/VMX: don't set ->l1tf_flush_l1d to true from vmx_l1d_flush()
  x86/KVM/VMX: replace 'vmx_l1d_flush_always' with 'vmx_l1d_flush_cond'
  x86/KVM/VMX: move the ->l1tf_flush_l1d test to vmx_l1d_flush()
  x86: implement kernel mode irq tracking
  x86/KVM/VMX: flush L1d upon interrupt after VMEXIT
  x86/KVM/VMX: don't set ->l1tf_flush_l1d from
    vmx_handle_external_intr()

 arch/x86/Kconfig                 |  3 +++
 arch/x86/entry/entry_32.S        | 12 ++++++++++
 arch/x86/entry/entry_64.S        | 12 ++++++++++
 arch/x86/include/asm/kvm_host.h  |  1 +
 arch/x86/include/asm/processor.h |  4 ++++
 arch/x86/kernel/cpu/common.c     |  5 ++++
 arch/x86/kvm/Kconfig             |  1 +
 arch/x86/kvm/vmx.c               | 51 +++++++++++++++++++++++++++-------------
 8 files changed, 73 insertions(+), 16 deletions(-)

-- 
2.13.7