On 10.02.2012, at 00:18, Christoffer Dall wrote:
> On Thu, Feb 9, 2012 at 3:02 PM, Alexander Graf <agraf@xxxxxxx> wrote:
>>
>> On 09.02.2012, at 23:33, Christoffer Dall wrote:
>>
>>> From: Christoffer Dall <c.dall@xxxxxxxxxxxxxxxxxxxxxx>
>>>
>>> The kvm_vcpu_kick function performs roughly the same funcitonality on
>>> most all architectures, so we shouldn't have separate copies.
>>>
>>> PowerPC keeps a pointer to interchanging waitqueues on the vcpu_arch
>>> structure and to accomodate this special need a
>>> __KVM_HAVE_ARCH_VCPU_GET_WQ define and accompanying function
>>> kvm_arch_vcpu_wq have been defined. For all other architectures this
>>> is a generic inline that just returns &vcpu->wq;
>>>
>>> This patch applies to Linus' tree on the Linux 3.3-rc1 tag.
>>
>> Please work against kvm.git for kvm patches :). Makes life easier for everyone involved. It takes a while for patches to land in Linus' tree. I don't think it matters for this patch, but it might in the future.
>>
>
> ok
>
>>>
>>> Changes since v2:
>>> - Restore arch-specific vcpu->cpu assignment to arch-specific code
>>> - Set vcpu mode in PowerPC arch-specific kvm code
>>>
>>> Changes since v1:
>>> - Abstact CPU mode check into arch-specific function
>>> - Remove redundant vcpu->cpu assignment
>>>
>>> Signed-off-by: Christoffer Dall <c.dall@xxxxxxxxxxxxxxxxxxxxxx>
>>> ---
>>> arch/ia64/include/asm/kvm_host.h | 1 +
>>> arch/ia64/kvm/kvm-ia64.c | 15 ---------------
>>> arch/powerpc/include/asm/kvm_host.h | 6 ++++++
>>> arch/powerpc/kvm/book3s_pr.c | 4 ++++
>>> arch/powerpc/kvm/booke.c | 2 ++
>>> arch/powerpc/kvm/powerpc.c | 17 +++++++----------
>>> arch/s390/kvm/kvm-s390.c | 8 ++++++++
>>> arch/x86/kvm/x86.c | 16 ++--------------
>>> include/linux/kvm_host.h | 9 +++++++++
>>> virt/kvm/kvm_main.c | 22 ++++++++++++++++++++++
>>> 10 files changed, 61 insertions(+), 39 deletions(-)
>>>
>>> diff --git a/arch/ia64/include/asm/kvm_host.h b/arch/ia64/include/asm/kvm_host.h
>>> index 2689ee5..06a5e91 100644
>>> --- a/arch/ia64/include/asm/kvm_host.h
>>> +++ b/arch/ia64/include/asm/kvm_host.h
>>> @@ -365,6 +365,7 @@ struct thash_cb {
>>> };
>>>
>>> struct kvm_vcpu_stat {
>>> + u32 halt_wakeup;
>>> };
>>>
>>> struct kvm_vcpu_arch {
>>> diff --git a/arch/ia64/kvm/kvm-ia64.c b/arch/ia64/kvm/kvm-ia64.c
>>> index 4050520..3f6a3cc 100644
>>> --- a/arch/ia64/kvm/kvm-ia64.c
>>> +++ b/arch/ia64/kvm/kvm-ia64.c
>>> @@ -1849,21 +1849,6 @@ void kvm_arch_hardware_unsetup(void)
>>> {
>>> }
>>>
>>> -void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
>>> -{
>>> - int me;
>>> - int cpu = vcpu->cpu;
>>> -
>>> - if (waitqueue_active(&vcpu->wq))
>>> - wake_up_interruptible(&vcpu->wq);
>>> -
>>> - me = get_cpu();
>>> - if (cpu != me && (unsigned) cpu < nr_cpu_ids && cpu_online(cpu))
>>> - if (!test_and_set_bit(KVM_REQ_KICK, &vcpu->requests))
>>> - smp_send_reschedule(cpu);
>>> - put_cpu();
>>> -}
>>> -
>>> int kvm_apic_set_irq(struct kvm_vcpu *vcpu, struct kvm_lapic_irq *irq)
>>> {
>>> return __apic_accept_irq(vcpu, irq->vector);
>>> diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
>>> index bf8af5d..b687444 100644
>>> --- a/arch/powerpc/include/asm/kvm_host.h
>>> +++ b/arch/powerpc/include/asm/kvm_host.h
>>> @@ -438,4 +438,10 @@ struct kvm_vcpu_arch {
>>> #define KVMPPC_VCPU_BUSY_IN_HOST 1
>>> #define KVMPPC_VCPU_RUNNABLE 2
>>>
>>> +#define __KVM_HAVE_ARCH_VCPU_GET_WQ 1
>>> +static inline wait_queue_head *kvm_arch_vcpu_wq(struct kvm_vcpu *vcpu)
>>> +{
>>> + return vcpu->arch.wqp;
>>> +}
>>> +
>>> #endif /* __POWERPC_KVM_HOST_H__ */
>>> diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
>>> index e2cfb9e..43466aa 100644
>>> --- a/arch/powerpc/kvm/book3s_pr.c
>>> +++ b/arch/powerpc/kvm/book3s_pr.c
>>> @@ -968,6 +968,8 @@ int kvmppc_vcpu_run(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
>>> if (vcpu->arch.shared->msr & MSR_FP)
>>> kvmppc_handle_ext(vcpu, BOOK3S_INTERRUPT_FP_UNAVAIL, MSR_FP);
>>>
>>> + vcpu->mode = IN_GUEST_MODE;
>>
>> This is racy. I'll just write a patch myself then :)
>>
>
> I couldn't quite figure out the atomicity here, so please fix it for me...
Yeah, my head is already spinning :). This is getting more tricky than I'd like it to be.
>
>>> +
>>> kvm_guest_enter();
>>>
>>> ret = __kvmppc_vcpu_run(kvm_run, vcpu);
>>> @@ -976,6 +978,8 @@ int kvmppc_vcpu_run(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
>>>
>>> local_irq_disable();
>>>
>>> + vcpu->mode = OUTSIDE_GUEST_MODE;
>>> +
>>> current->thread.regs->msr = ext_msr;
>>>
>>> /* Make sure we save the guest FPU/Altivec/VSX state */
>>> diff --git a/arch/powerpc/kvm/booke.c b/arch/powerpc/kvm/booke.c
>>> index bb6c988..f909289 100644
>>> --- a/arch/powerpc/kvm/booke.c
>>> +++ b/arch/powerpc/kvm/booke.c
>>> @@ -321,11 +321,13 @@ int kvmppc_vcpu_run(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
>>> return -EINVAL;
>>> }
>>>
>>> + vcpu->mode = IN_GUEST_MODE;
>>> local_irq_disable();
>>> kvm_guest_enter();
>>> ret = __kvmppc_vcpu_run(kvm_run, vcpu);
>>> kvm_guest_exit();
>>> local_irq_enable();
>>> + vcpu->mode = OUTSIDE_GUEST_MODE;
>>>
>>> return ret;
>>> }
>>> diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
>>> index 607fbdf..3b9b5cd 100644
>>> --- a/arch/powerpc/kvm/powerpc.c
>>> +++ b/arch/powerpc/kvm/powerpc.c
>>> @@ -42,6 +42,11 @@ int kvm_arch_vcpu_runnable(struct kvm_vcpu *v)
>>> !!(v->arch.pending_exceptions);
>>> }
>>>
>>> +int kvm_arch_vcpu_in_guest_mode(struct kvm_vcpu *vcpu)
>>> +{
>>> + return kvm_vcpu_exiting_guest_mode(vcpu) == IN_GUEST_MODE;
>>> +}
>>> +
>>> int kvmppc_kvm_pv(struct kvm_vcpu *vcpu)
>>> {
>>> int nr = kvmppc_get_gpr(vcpu, 11);
>>> @@ -311,10 +316,7 @@ static void kvmppc_decrementer_func(unsigned long data)
>>>
>>> kvmppc_core_queue_dec(vcpu);
>>>
>>> - if (waitqueue_active(vcpu->arch.wqp)) {
>>> - wake_up_interruptible(vcpu->arch.wqp);
>>> - vcpu->stat.halt_wakeup++;
>>> - }
>>> + kvm_vcpu_kick(vcpu);
>>> }
>>>
>>> /*
>>> @@ -572,12 +574,7 @@ int kvm_vcpu_ioctl_interrupt(struct kvm_vcpu *vcpu, struct kvm_interrupt *irq)
>>>
>>> kvmppc_core_queue_external(vcpu, irq);
>>>
>>> - if (waitqueue_active(vcpu->arch.wqp)) {
>>> - wake_up_interruptible(vcpu->arch.wqp);
>>> - vcpu->stat.halt_wakeup++;
>>> - } else if (vcpu->cpu != -1) {
>>> - smp_send_reschedule(vcpu->cpu);
>>> - }
>>> + kvm_vcpu_kick(vcpu);
>>>
>>> return 0;
>>> }
>>> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
>>> index d1c44573..5e2217f 100644
>>> --- a/arch/s390/kvm/kvm-s390.c
>>> +++ b/arch/s390/kvm/kvm-s390.c
>>> @@ -380,6 +380,14 @@ int kvm_arch_vcpu_runnable(struct kvm_vcpu *vcpu)
>>> return 0;
>>> }
>>>
>>> +int kvm_arch_vcpu_in_guest_mode(struct kvm_vcpu *vcpu)
>>> +{
>>> + /* kvm common code refers to this, but never calls it */
>>> + BUG();
>>> + return 0;
>>> +}
>>> +
>>> +
>>> static int kvm_arch_vcpu_ioctl_initial_reset(struct kvm_vcpu *vcpu)
>>> {
>>> kvm_s390_vcpu_initial_reset(vcpu);
>>> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>>> index 43c4dab..57d13a6 100644
>>> --- a/arch/x86/kvm/x86.c
>>> +++ b/arch/x86/kvm/x86.c
>>> @@ -6135,21 +6135,9 @@ int kvm_arch_vcpu_runnable(struct kvm_vcpu *vcpu)
>>> kvm_cpu_has_interrupt(vcpu));
>>> }
>>>
>>> -void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
>>> +int kvm_arch_vcpu_in_guest_mode(struct kvm_vcpu *vcpu)
>>> {
>>> - int me;
>>> - int cpu = vcpu->cpu;
>>> -
>>> - if (waitqueue_active(&vcpu->wq)) {
>>> - wake_up_interruptible(&vcpu->wq);
>>> - ++vcpu->stat.halt_wakeup;
>>> - }
>>> -
>>> - me = get_cpu();
>>> - if (cpu != me && (unsigned)cpu < nr_cpu_ids && cpu_online(cpu))
>>> - if (kvm_vcpu_exiting_guest_mode(vcpu) == IN_GUEST_MODE)
>>> - smp_send_reschedule(cpu);
>>> - put_cpu();
>>> + return kvm_vcpu_exiting_guest_mode(vcpu) == IN_GUEST_MODE;
>>> }
>>>
>>> int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu)
>>> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
>>> index 41973e2..638fa64 100644
>>> --- a/include/linux/kvm_host.h
>>> +++ b/include/linux/kvm_host.h
>>> @@ -440,6 +440,7 @@ void mark_page_dirty_in_slot(struct kvm *kvm, struct kvm_memory_slot *memslot,
>>> gfn_t gfn);
>>>
>>> void kvm_vcpu_block(struct kvm_vcpu *vcpu);
>>> +void kvm_vcpu_kick(struct kvm_vcpu *vcpu);
>>> void kvm_vcpu_on_spin(struct kvm_vcpu *vcpu);
>>> void kvm_resched(struct kvm_vcpu *vcpu);
>>> void kvm_load_guest_fpu(struct kvm_vcpu *vcpu);
>>> @@ -507,6 +508,7 @@ int kvm_arch_hardware_setup(void);
>>> void kvm_arch_hardware_unsetup(void);
>>> void kvm_arch_check_processor_compat(void *rtn);
>>> int kvm_arch_vcpu_runnable(struct kvm_vcpu *vcpu);
>>> +int kvm_arch_vcpu_in_guest_mode(struct kvm_vcpu *vcpu);
>>>
>>> void kvm_free_physmem(struct kvm *kvm);
>>>
>>> @@ -522,6 +524,13 @@ static inline void kvm_arch_free_vm(struct kvm *kvm)
>>> }
>>> #endif
>>>
>>> +#ifndef __KVM_HAVE_ARCH_VCPU_GET_WQ
>>> +static inline wait_queue_head_t *kvm_arch_vcpu_wq(struct kvm_vcpu *vcpu)
>>> +{
>>> + return &vcpu->wq;
>>> +}
>>> +#endif
>>> +
>>> int kvm_arch_init_vm(struct kvm *kvm);
>>> void kvm_arch_destroy_vm(struct kvm *kvm);
>>> void kvm_free_all_assigned_devices(struct kvm *kvm);
>>> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
>>> index 7287bf5..3d72008 100644
>>> --- a/virt/kvm/kvm_main.c
>>> +++ b/virt/kvm/kvm_main.c
>>> @@ -1581,6 +1581,28 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
>>> finish_wait(&vcpu->wq, &wait);
>>> }
>>>
>>> +/*
>>> + * Kick a sleeping VCPU, or a guest VCPU in guest mode, into host kernel mode.
>>> + */
>>> +void kvm_vcpu_kick(struct kvm_vcpu *vcpu)
>>> +{
>>> + int me;
>>> + int cpu = vcpu->cpu;
>>> + wait_queue_head_t *wqp;
>>> +
>>> + wqp = kvm_arch_vcpu_wq(vcpu);
>>> + if (waitqueue_active(wqp)) {
>>> + wake_up_interruptible(wqp);
>>> + ++vcpu->stat.halt_wakeup;
>>> + }
>>> +
>>> + me = get_cpu();
>>> + if (cpu != me && (unsigned)cpu < nr_cpu_ids && cpu_online(cpu))
>>> + if (kvm_arch_vcpu_in_guest_mode(vcpu))
>>
>> Didn't the last version did an atomic cmpxchg here? Also, that check certainly wasn't there for ia64 before. Are you sure it's happy with it? :)
>>
>>
>
> no, there was no cmpxchg here. I'm not sure why you think it's necessary.
>
> and no, ia64 didn't check for this before, and I'm not sure it's happy
> with it, which is why we have these reviews I guess.
>
> how about calling that functions kvm_arch_vcpu_should_kick and
> performing the (!test_and_set_bit(KVM_REQ_KICK, &vcpu->requests))
> check from ia64 in there and do the vcpu->mode check for other archs?
How about we return 1 for kvm_arch_vcpu_in_guest_mode() on !x86 always, just like it's done today basically? Then we can worry about needless IPIs later and don't regress from the respective current kick implementations.
Alex
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
