KVM: PPC: Book3S HV: Fix decrementer migration
Subsystems (2)
PATCH 1/1 KVM: PPC: Book3S HV: Fix decrementer migration
We used to have a workaround[1] for a hang during migration that was made ineffective when we converted the decrementer expiry to be relative to guest timebase.
The point of the workaround was that in the absence of an explicit decrementer expiry value provided by userspace during migration, KVM needs to initialize dec_expires to a value that will result in an expired decrementer after subtracting the current guest timebase. That stops the vcpu from hanging after migration due to a decrementer that's too large.
If the dec_expires is now relative to guest timebase, its initialization needs to be guest timebase-relative as well, otherwise we end up with a decrementer expiry that is still larger than the guest timebase.
1- https://git.kernel.org/torvalds/c/5855564c8ab2
--- arch/powerpc/kvm/book3s_hv.c | 18 ++++++++++++++++-- arch/powerpc/kvm/powerpc.c | 1 - 2 files changed, 16 insertions(+), 3 deletions(-)
arch/powerpc/kvm/book3s_hv.c +16 -1
@@ -2517,10 +2517,24 @@ static int kvmppc_set_one_reg_hv(struct kvm_vcpu *vcpu, u64 id, r =set_vpa (vcpu, &vcpu->arch.dtl, addr, len);break ;case KVM_REG_PPC_TB_OFFSET : + {/* round up to multiple of 2^24 */ - vcpu->arch.vcore->tb_offset -ALIGN (set_reg_val (id, *val),1UL <<24 ); +u64 tb_offset =ALIGN (set_reg_val (id, *val),1UL <<24 ); + +/* +* Now that we know the timebase offset, update the +* decrementer expiry with a guest timebase value. If +* the userspace does not set DEC_EXPIRY, this ensures +* a migrated vcpu at least starts with an expired +* decrementer, which is better than a large one that +* causes a hang. +*/ +if (!vcpu->arch.dec_expires && tb_offset) + vcpu->arch.dec_expires =get_tb () + tb_offset; + + vcpu->arch.vcore->tb_offset = tb_offset;break ; + }case KVM_REG_PPC_LPCR :kvmppc_set_lpcr (vcpu,set_reg_val (id, *val), true);break ;
arch/powerpc/kvm/powerpc.c +0 -1
@@ -786,7 +786,6 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)hrtimer_init (&vcpu->arch.dec_timer,CLOCK_REALTIME ,HRTIMER_MODE_ABS ); vcpu->arch.dec_timer.function = kvmppc_decrementer_wakeup; - vcpu->arch.dec_expires =get_tb ();#ifdef CONFIG_KVM_EXIT_TIMING mutex_init (&vcpu->arch.exit_timing_lock);
Signature · 1 line
2.35.3
Michael Ellerman
AppliedQuoted 4 lines · Fabiano Rosas
We used to have a workaround[1] for a hang during migration that was made ineffective when we converted the decrementer expiry to be relative to guest timebase.
Quoted 8 lines
The point of the workaround was that in the absence of an explicit decrementer expiry value provided by userspace during migration, KVM needs to initialize
dec_expiresto a value that will result in an expired decrementer after subtracting the current guest timebase. That stops the vcpu from hanging after migration due to a decrementer that's too large.[...]
Applied to powerpc/topic/ppc-kvm.
[1/1] KVM: PPC: Book3S HV: Fix decrementer migration
https://git.kernel.org/powerpc/c/0a5bfb824a6ea35e54b7e5ac6f881beea5e309d2
cheers