Date:   Fri, 10 Dec 2021 14:48:53 +0000
From:   Quentin Perret <qperret@google.com>
To:     Will Deacon <will@kernel.org>
Cc:     Marc Zyngier <maz@kernel.org>, James Morse <james.morse@arm.com>,
        Alexandru Elisei <alexandru.elisei@arm.com>,
        Suzuki K Poulose <suzuki.poulose@arm.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu,
        linux-kernel@vger.kernel.org, kernel-team@android.com
Subject: Re: [PATCH v3 15/15] KVM: arm64: pkvm: Unshare guest structs during
 teardown
Message-ID: <YbNo1SBn7ZNf89qL@google.com>
References: <20211201170411.1561936-1-qperret@google.com>
 <20211201170411.1561936-16-qperret@google.com>
 <20211209112233.GD1912@willie-the-truck>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211209112233.GD1912@willie-the-truck>
Precedence: bulk

On Thursday 09 Dec 2021 at 11:22:33 (+0000), Will Deacon wrote:
> On Wed, Dec 01, 2021 at 05:04:09PM +0000, Quentin Perret wrote:
> > Make use of the newly introduced unshare hypercall during guest teardown
> > to unmap guest-related data structures from the hyp stage-1.
> > 
> > Signed-off-by: Quentin Perret <qperret@google.com>
> > ---
> >  arch/arm64/include/asm/kvm_host.h |  2 ++
> >  arch/arm64/include/asm/kvm_mmu.h  |  1 +
> >  arch/arm64/kvm/arm.c              |  2 ++
> >  arch/arm64/kvm/fpsimd.c           | 34 ++++++++++++++++++++++---
> >  arch/arm64/kvm/mmu.c              | 42 +++++++++++++++++++++++++++++++
> >  arch/arm64/kvm/reset.c            |  8 +++++-
> >  6 files changed, 85 insertions(+), 4 deletions(-)
> 
> [...]
> 
> > diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
> > index d72566896755..8e506ba8988e 100644
> > --- a/arch/arm64/kvm/mmu.c
> > +++ b/arch/arm64/kvm/mmu.c
> > @@ -344,6 +344,32 @@ static int share_pfn_hyp(u64 pfn)
> >  	return ret;
> >  }
> >  
> > +static int unshare_pfn_hyp(u64 pfn)
> > +{
> > +	struct rb_node **node, *parent;
> > +	struct hyp_shared_pfn *this;
> > +	int ret = 0;
> > +
> > +	mutex_lock(&hyp_shared_pfns_lock);
> > +	this = find_shared_pfn(pfn, &node, &parent);
> > +	if (WARN_ON(!this)) {
> > +		ret = -EINVAL;
> 
> -ENOENT?

Sure.

> > +		goto unlock;
> > +	}
> > +
> > +	this->count--;
> > +	if (this->count)
> > +		goto unlock;
> 
> Again, if we did an RCU lookup then this could be converted to a refcount_t
> to take the mutex only when it hits zero. But for now I think it's fine.

No objection to do this in the future, but yeah I think we might as well
start simple :)

> > +
> > +	rb_erase(&this->node, &hyp_shared_pfns);
> > +	kfree(this);
> > +	ret = kvm_call_hyp_nvhe(__pkvm_host_unshare_hyp, pfn, 1);
> > +unlock:
> > +	mutex_unlock(&hyp_shared_pfns_lock);
> > +
> > +	return ret;
> > +}
> > +
> >  int kvm_share_hyp(void *from, void *to)
> >  {
> >  	phys_addr_t start, end, cur;
> > @@ -376,6 +402,22 @@ int kvm_share_hyp(void *from, void *to)
> >  	return 0;
> >  }
> >  
> > +void kvm_unshare_hyp(void *from, void *to)
> > +{
> > +	phys_addr_t start, end, cur;
> > +	u64 pfn;
> > +
> > +	if (is_kernel_in_hyp_mode() || kvm_host_owns_hyp_mappings() || !from)
> 
> I don't think you need the is_kernel_in_hyp_mode() check any more not that
> you've moved that into kvm_host_owns_hyp_mappings().

The logic is a little odd, but I think I still do as
kvm_host_owns_hyp_mappings() will return false if is_kernel_in_hyp_mode()
is true.