Commit 2852ebfa authored by Nicholas Piggin's avatar Nicholas Piggin Committed by Michael Ellerman

KVM: PPC: Book3S HV Nested: L2 LPCR should inherit L1 LPES setting

The L1 should not be able to adjust LPES mode for the L2. Setting LPES
if the L0 needs it clear would cause external interrupts to be sent to
L2 and missed by the L0.

Clearing LPES when it may be set, as typically happens with XIVE enabled
could cause a performance issue despite having no native XIVE support in
the guest, because it will cause mediated interrupts for the L2 to be
taken in HV mode, which then have to be injected.
Signed-off-by: default avatarNicholas Piggin <npiggin@gmail.com>
Reviewed-by: default avatarFabiano Rosas <farosas@linux.ibm.com>
Signed-off-by: default avatarMichael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220303053315.1056880-7-npiggin@gmail.com
parent 11681b79
...@@ -5317,6 +5317,10 @@ static int kvmppc_core_init_vm_hv(struct kvm *kvm) ...@@ -5317,6 +5317,10 @@ static int kvmppc_core_init_vm_hv(struct kvm *kvm)
kvm->arch.host_lpcr = lpcr = mfspr(SPRN_LPCR); kvm->arch.host_lpcr = lpcr = mfspr(SPRN_LPCR);
lpcr &= LPCR_PECE | LPCR_LPES; lpcr &= LPCR_PECE | LPCR_LPES;
} else { } else {
/*
* The L2 LPES mode will be set by the L0 according to whether
* or not it needs to take external interrupts in HV mode.
*/
lpcr = 0; lpcr = 0;
} }
lpcr |= (4UL << LPCR_DPFD_SH) | LPCR_HDICE | lpcr |= (4UL << LPCR_DPFD_SH) | LPCR_HDICE |
......
...@@ -261,8 +261,7 @@ static void load_l2_hv_regs(struct kvm_vcpu *vcpu, ...@@ -261,8 +261,7 @@ static void load_l2_hv_regs(struct kvm_vcpu *vcpu,
/* /*
* Don't let L1 change LPCR bits for the L2 except these: * Don't let L1 change LPCR bits for the L2 except these:
*/ */
mask = LPCR_DPFD | LPCR_ILE | LPCR_TC | LPCR_AIL | LPCR_LD | mask = LPCR_DPFD | LPCR_ILE | LPCR_TC | LPCR_AIL | LPCR_LD | LPCR_MER;
LPCR_LPES | LPCR_MER;
/* /*
* Additional filtering is required depending on hardware * Additional filtering is required depending on hardware
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment