[PATCH v2 0/8] x86/irq: KVM: Optimize KVM's PIR harvesting

Sean Christopherson posted 8 patches 10 months, 1 week ago
arch/x86/include/asm/posted_intr.h | 78 +++++++++++++++++++++++++++---
arch/x86/kernel/irq.c              | 63 ++++--------------------
arch/x86/kvm/lapic.c               | 20 ++++----
arch/x86/kvm/lapic.h               |  4 +-
arch/x86/kvm/vmx/posted_intr.h     |  2 +-
5 files changed, 95 insertions(+), 72 deletions(-)
[PATCH v2 0/8] x86/irq: KVM: Optimize KVM's PIR harvesting
Posted by Sean Christopherson 10 months, 1 week ago
Optimizing KVM's PIR harvesting using the same techniques as posted MSIs,
most notably to use 8-byte accesses on 64-bit kernels (/facepalm).

Fix a few warts along the way, and finish up by adding a helper to dedup
the PIR harvesting code between KVM and posted MSIs.

v2:
 - Collect a review. [tglx]
 - Use an "unsigned long" with a bitwise-OR to gather PIR. [tglx]

v1: https://lore.kernel.org/all/20250315030630.2371712-1-seanjc@google.com

Sean Christopherson (8):
  x86/irq: Ensure initial PIR loads are performed exactly once
  x86/irq: Track if IRQ was found in PIR during initial loop (to load
    PIR vals)
  KVM: VMX: Ensure vIRR isn't reloaded at odd times when sync'ing PIR
  x86/irq: KVM: Track PIR bitmap as an "unsigned long" array
  KVM: VMX: Process PIR using 64-bit accesses on 64-bit kernels
  KVM: VMX: Isolate pure loads from atomic XCHG when processing PIR
  KVM: VMX: Use arch_xchg() when processing PIR to avoid instrumentation
  x86/irq: KVM: Add helper for harvesting PIR to deduplicate KVM and
    posted MSIs

 arch/x86/include/asm/posted_intr.h | 78 +++++++++++++++++++++++++++---
 arch/x86/kernel/irq.c              | 63 ++++--------------------
 arch/x86/kvm/lapic.c               | 20 ++++----
 arch/x86/kvm/lapic.h               |  4 +-
 arch/x86/kvm/vmx/posted_intr.h     |  2 +-
 5 files changed, 95 insertions(+), 72 deletions(-)


base-commit: 782f9feaa9517caf33186dcdd6b50a8f770ed29b
-- 
2.49.0.472.ge94155a9ec-goog
Re: [PATCH v2 0/8] x86/irq: KVM: Optimize KVM's PIR harvesting
Posted by Sean Christopherson 9 months, 2 weeks ago
On Tue, 01 Apr 2025 09:34:39 -0700, Sean Christopherson wrote:
> Optimizing KVM's PIR harvesting using the same techniques as posted MSIs,
> most notably to use 8-byte accesses on 64-bit kernels (/facepalm).
> 
> Fix a few warts along the way, and finish up by adding a helper to dedup
> the PIR harvesting code between KVM and posted MSIs.
> 
> v2:
>  - Collect a review. [tglx]
>  - Use an "unsigned long" with a bitwise-OR to gather PIR. [tglx]
> 
> [...]

Applied to kvm-x86 pir.

Thomas and other x86 maintainers, please holler if you object to taking this
through KVM (x86), or to any of the patches.  I want to start getting coverage
in -next, and deliberately put this in its own topic branch so I can rewrite or
drop things as needed.

[1/8] x86/irq: Ensure initial PIR loads are performed exactly once
      https://github.com/kvm-x86/linux/commit/600e9606046a
[2/8] x86/irq: Track if IRQ was found in PIR during initial loop (to load PIR vals)
      https://github.com/kvm-x86/linux/commit/3cdb8261504c
[3/8] KVM: VMX: Ensure vIRR isn't reloaded at odd times when sync'ing PIR
      https://github.com/kvm-x86/linux/commit/6433fc01f9f1
[4/8] x86/irq: KVM: Track PIR bitmap as an "unsigned long" array
      https://github.com/kvm-x86/linux/commit/f1459315f4d2
[5/8] KVM: VMX: Process PIR using 64-bit accesses on 64-bit kernels
      https://github.com/kvm-x86/linux/commit/06b4d0ea226c
[6/8] KVM: VMX: Isolate pure loads from atomic XCHG when processing PIR
      https://github.com/kvm-x86/linux/commit/b41f8638b9d3
[7/8] KVM: VMX: Use arch_xchg() when processing PIR to avoid instrumentation
      https://github.com/kvm-x86/linux/commit/baf68a0e3bd6
[8/8] x86/irq: KVM: Add helper for harvesting PIR to deduplicate KVM and posted MSIs
      https://github.com/kvm-x86/linux/commit/edaf3eded386

--
https://github.com/kvm-x86/linux/tree/next