[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200318152837.GB26585@xz-x1>
Date: Wed, 18 Mar 2020 11:28:37 -0400
From: Peter Xu <peterx@...hat.com>
To: maobibo <maobibo@...ngson.cn>
Cc: linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
Sean Christopherson <sean.j.christopherson@...el.com>,
Paolo Bonzini <pbonzini@...hat.com>,
linux-mips@...r.kernel.org, Vitaly Kuznetsov <vkuznets@...hat.com>
Subject: Re: [PATCH RFC 4/4] KVM: MIPS: Define arch-specific
kvm_flush_remote_tlbs()
On Wed, Mar 18, 2020 at 11:03:13AM +0800, maobibo wrote:
>
>
> On 02/08/2020 06:35 AM, Peter Xu wrote:
> > Select HAVE_KVM_ARCH_TLB_FLUSH_ALL for MIPS to define its own
> > kvm_flush_remote_tlbs(). It's as simple as calling the
> > flush_shadow_all(kvm) hook. Then replace all the flush_shadow_all()
> > calls to use this KVM generic API to do TLB flush.
> >
> > Signed-off-by: Peter Xu <peterx@...hat.com>
> > ---
> > arch/mips/kvm/Kconfig | 1 +
> > arch/mips/kvm/mips.c | 22 ++++++++++------------
> > 2 files changed, 11 insertions(+), 12 deletions(-)
> >
> > diff --git a/arch/mips/kvm/Kconfig b/arch/mips/kvm/Kconfig
> > index eac25aef21e0..8a06f660d39e 100644
> > --- a/arch/mips/kvm/Kconfig
> > +++ b/arch/mips/kvm/Kconfig
> > @@ -26,6 +26,7 @@ config KVM
> > select KVM_MMIO
> > select MMU_NOTIFIER
> > select SRCU
> > + select HAVE_KVM_ARCH_TLB_FLUSH_ALL
> > ---help---
> > Support for hosting Guest kernels.
> >
> > diff --git a/arch/mips/kvm/mips.c b/arch/mips/kvm/mips.c
> > index 1d5e7ffda746..518020b466bf 100644
> > --- a/arch/mips/kvm/mips.c
> > +++ b/arch/mips/kvm/mips.c
> > @@ -194,13 +194,16 @@ int kvm_arch_create_memslot(struct kvm *kvm, struct kvm_memory_slot *slot,
> > return 0;
> > }
> >
> > +void kvm_flush_remote_tlbs(struct kvm *kvm)
> > +{
> > + kvm_mips_callbacks->flush_shadow_all(kvm);
> > +}
> > +
> Hi Peter,
Hi, Bibo,
>
> Although I do not understand mip VZ fully, however it changes behavior of
> MIPS VZ, since kvm_flush_remote_tlbs is also called in function
> kvm_mmu_notifier_change_pte/kvm_mmu_notifier_invalidate_range_start
I'm not familiar with MIPS either, however... I would start to suspect
MIPS could be problematic with MMU notifiers when cpu_has_guestid is
not set. If that's the case, then this series might instead fix it.
Thanks,
--
Peter Xu
Powered by blists - more mailing lists