Re: [PATCH v4 28/34] KVM: arm64: Use page-table to track page ownership

[Quentin Perret <qperret@google.com>]

On Friday 12 Mar 2021 at 11:18:05 (+0000), Will Deacon wrote:
> On Fri, Mar 12, 2021 at 10:13:26AM +0000, Quentin Perret wrote:
> > On Friday 12 Mar 2021 at 09:32:06 (+0000), Will Deacon wrote:
> > > I'm not saying to use the VMID directly, just that allocating half of the
> > > pte feels a bit OTT given that the state of things after this patch series
> > > is that we're using exactly 1 bit.
> > 
> > Right, and that was the reason for the PROT_NONE approach in the
> > previous version, but we agreed it'd be worth generalizing to allow for
> > future use-cases :-)
> 
> Yeah, just generalising to 32 bits feels like going too far! I dunno,
> make it a u8 for now, or define the hypervisor owner ID as 1 and reject
> owners greater than that? We can easily extend it later.

Alrighty I'll do _both_

> > > > > > @@ -517,28 +543,36 @@ static int stage2_map_walker_try_leaf(u64 addr, u64 end, u32 level,
> > > > > >  	if (!kvm_block_mapping_supported(addr, end, phys, level))
> > > > > >  		return -E2BIG;
> > > > > >  
> > > > > > -	new = kvm_init_valid_leaf_pte(phys, data->attr, level);
> > > > > > -	if (kvm_pte_valid(old)) {
> > > > > > +	if (kvm_pte_valid(data->attr))
> > > > > 
> > > > > This feels like a bit of a hack to me: the 'attr' field in stage2_map_data
> > > > > is intended to correspond directly to the lower/upper attributes of the
> > > > > descriptor as per the architecture, so tagging the valid bit in there is
> > > > > pretty grotty. However, I can see the significant advantage in being able
> > > > > to re-use the stage2_map_walker functionality, so about instead of nobbling
> > > > > attr, you set phys to something invalid instead, e.g.:
> > > > > 
> > > > > 	#define KVM_PHYS_SET_OWNER	(-1ULL)
> > > > 
> > > > That'll confuse kvm_block_mapping_supported() and friends I think, at
> > > > least in their current form. If you _really_ don't like this, maybe we
> > > > could have an extra 'flags' field in stage2_map_data?
> > > 
> > > I was pondering this last night and I thought of two ways to do it:
> > > 
> > > 1. Add a 'bool valid' and then stick the owner and the phys in a union.
> > >    (yes, you'll need to update the block mapping checks to look at the
> > >     valid flag)
> > 
> > Right, though that is also used for the hyp s1 which doesn't use any of
> > this ATM. That shouldn't be too bad to change, I'll have a look.
> 
> Oh, I meant stick the bool in the stage2_map_data so that should be limited
> to the stage2 path.

I mean I still want to use kvm_block_mapping_supported() but ignore the
phys check when it's not valid. I find it ugly to add a 'valid'
parameter to the function itself, so maybe we're better off with just
special casing phys == -1ULL as you first suggested. How much do you hate
the below (totally untested)?

diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
index 8e4599256969..9ec937462fd6 100644
--- a/arch/arm64/kvm/hyp/pgtable.c
+++ b/arch/arm64/kvm/hyp/pgtable.c
@@ -71,6 +71,13 @@ static u64 kvm_granule_size(u32 level)
        return BIT(kvm_granule_shift(level));
 }

+#define KVM_PHYS_INVALID (-1ULL)
+
+static bool kvm_phys_is_valid(u64 phys)
+{
+       return phys != KVM_PHYS_INVALID;
+}
+
 static bool kvm_level_support_block_mappings(u32 level)
 {
        /*
@@ -90,7 +97,10 @@ static bool kvm_block_mapping_supported(u64 addr, u64 end, u64 phys, u32 level)
        if (granule > (end - addr))
                return false;

-       return IS_ALIGNED(addr, granule) && IS_ALIGNED(phys, granule);
+       if (kvm_phys_is_valid(phys) && !IS_ALIGNED(phys, granule))
+               return false;
+
+       return IS_ALIGNED(addr, granule);
 }

 static u32 kvm_pgtable_idx(struct kvm_pgtable_walk_data *data, u32 level)
@@ -550,7 +560,7 @@ static int stage2_map_walker_try_leaf(u64 addr, u64 end, u32 level,
        if (!kvm_block_mapping_supported(addr, end, phys, level))
                return -E2BIG;

-       if (kvm_pte_valid(data->attr))
+       if (kvm_phys_is_valid(phys))
                new = kvm_init_valid_leaf_pte(phys, data->attr, level);
        else
                new = kvm_init_invalid_leaf_owner(data->owner_id);
@@ -580,7 +590,8 @@ static int stage2_map_walker_try_leaf(u64 addr, u64 end, u32 level,
        smp_store_release(ptep, new);
        if (stage2_pte_is_counted(new))
                mm_ops->get_page(ptep);
-       data->phys += granule;
+       if (kvm_phys_is_valid(phys))
+               data->phys += granule;
        return 0;
 }

@@ -739,9 +750,6 @@ int kvm_pgtable_stage2_map(struct kvm_pgtable *pgt, u64 addr, u64 size,
        if (ret)
                return ret;

-       /* Set the valid flag to distinguish with the set_owner() path. */
-       map_data.attr |= KVM_PTE_VALID;
-
        ret = kvm_pgtable_walk(pgt, addr, size, &walker);
        dsb(ishst);
        return ret;
@@ -752,6 +760,7 @@ int kvm_pgtable_stage2_set_owner(struct kvm_pgtable *pgt, u64 addr, u64 size,
 {
        int ret;
        struct stage2_map_data map_data = {
+               .phys           = KVM_PHYS_INVALID,
                .mmu            = pgt->mmu,
                .memcache       = mc,
                .mm_ops         = pgt->mm_ops,



_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm