[PATCH 0/5] KVM: arm64: vgic-v3: Fix yet another lock ordering turd

[Oliver Upton <oliver.upton@linux.dev>]

syzkaller found yet another locking bug in the VGIC [*], this time due
to nesting a 'plain' spinlock (xa_lock) inside of a raw spinlock
(ap_list_lock). Given the way we do refcounts on LPIs it is possible
for this exact sort of issue to crop up where the last reference may
be dropped in unexpected places.

Small series to fix the issue by deferring xarray modifications outside
of the ap_list_lock critical section along with some slight lockdep
hinting to make these rare bugs a bit more obvious.

Applies to 6.17-rc4.

Oliver Upton (5):
  KVM: arm64: vgic-v3: Use bare refcount for VGIC LPIs
  KVM: arm64: Spin off release helper from vgic_put_irq()
  KVM: arm64: vgic-v3: Erase LPIs from xarray outside of raw spinlocks
  KVM: arm64: vgic-v3: Don't require IRQs be disabled for LPI xarray
    lock
  KVM: arm64: vgic-v3: Indicate vgic_put_irq() may take LPI xarray lock

 arch/arm64/kvm/vgic/vgic-debug.c |  2 +-
 arch/arm64/kvm/vgic/vgic-init.c  |  6 +--
 arch/arm64/kvm/vgic/vgic-its.c   | 15 +++---
 arch/arm64/kvm/vgic/vgic-v4.c    |  2 +-
 arch/arm64/kvm/vgic/vgic.c       | 78 +++++++++++++++++++++++---------
 arch/arm64/kvm/vgic/vgic.h       |  8 ++--
 include/kvm/arm_vgic.h           | 10 ++--
 7 files changed, 80 insertions(+), 41 deletions(-)


base-commit: b320789d6883cc00ac78ce83bccbfe7ed58afcf0
-- 
2.39.5