Re: [PATCH v5 02/14] KVM: arm64: Stash observed pte value in visitor context

[Gavin Shan <gshan@xxxxxxxxxx>]

Hi Oliver,

On 11/8/22 5:56 AM, Oliver Upton wrote:
> Rather than reading the ptep all over the shop, read the ptep once from
> __kvm_pgtable_visit() and stick it in the visitor context. Reread the
> ptep after visiting a leaf in case the callback installed a new table
> underneath.
>
> No functional change intended.
>
> Signed-off-by: Oliver Upton <oliver.upton@xxxxxxxxx>
> ---
>   arch/arm64/include/asm/kvm_pgtable.h  |  1 +
>   arch/arm64/kvm/hyp/nvhe/mem_protect.c |  5 +-
>   arch/arm64/kvm/hyp/nvhe/setup.c       |  7 +--
>   arch/arm64/kvm/hyp/pgtable.c          | 86 +++++++++++++--------------
>   4 files changed, 48 insertions(+), 51 deletions(-)

Reviewed-by: Gavin Shan <gshan@xxxxxxxxxx>

A nit as below.

> diff --git a/arch/arm64/include/asm/kvm_pgtable.h b/arch/arm64/include/asm/kvm_pgtable.h
> index 607f9bb8aab4..14d4b68a1e92 100644
> --- a/arch/arm64/include/asm/kvm_pgtable.h
> +++ b/arch/arm64/include/asm/kvm_pgtable.h
> @@ -201,6 +201,7 @@ enum kvm_pgtable_walk_flags {
>   
>   struct kvm_pgtable_visit_ctx {
>   	kvm_pte_t				*ptep;
> +	kvm_pte_t				old;
>   	void					*arg;
>   	u64					addr;
>   	u64					end;
> diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> index 8f5b6a36a039..d21d1b08a055 100644
> --- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> +++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
> @@ -421,12 +421,11 @@ static int __check_page_state_visitor(const struct kvm_pgtable_visit_ctx *ctx,
>   				      enum kvm_pgtable_walk_flags visit)
>   {
>   	struct check_walk_data *d = ctx->arg;
> -	kvm_pte_t pte = *ctx->ptep;
>   
> -	if (kvm_pte_valid(pte) && !addr_is_memory(kvm_pte_to_phys(pte)))
> +	if (kvm_pte_valid(ctx->old) && !addr_is_memory(kvm_pte_to_phys(ctx->old)))
>   		return -EINVAL;
>   
> -	return d->get_page_state(pte) == d->desired ? 0 : -EPERM;
> +	return d->get_page_state(ctx->old) == d->desired ? 0 : -EPERM;
>   }
>   
>   static int check_page_state_range(struct kvm_pgtable *pgt, u64 addr, u64 size,
> diff --git a/arch/arm64/kvm/hyp/nvhe/setup.c b/arch/arm64/kvm/hyp/nvhe/setup.c
> index a293cf5eba1b..6af443c9d78e 100644
> --- a/arch/arm64/kvm/hyp/nvhe/setup.c
> +++ b/arch/arm64/kvm/hyp/nvhe/setup.c
> @@ -192,10 +192,9 @@ static int finalize_host_mappings_walker(const struct kvm_pgtable_visit_ctx *ctx
>   	struct kvm_pgtable_mm_ops *mm_ops = ctx->arg;
>   	enum kvm_pgtable_prot prot;
>   	enum pkvm_page_state state;
> -	kvm_pte_t pte = *ctx->ptep;
>   	phys_addr_t phys;
>   
> -	if (!kvm_pte_valid(pte))
> +	if (!kvm_pte_valid(ctx->old))
>   		return 0;
>   
>   	/*
> @@ -210,7 +209,7 @@ static int finalize_host_mappings_walker(const struct kvm_pgtable_visit_ctx *ctx
>   	if (ctx->level != (KVM_PGTABLE_MAX_LEVELS - 1))
>   		return -EINVAL;
>   
> -	phys = kvm_pte_to_phys(pte);
> +	phys = kvm_pte_to_phys(ctx->old);
>   	if (!addr_is_memory(phys))
>   		return -EINVAL;
>   
> @@ -218,7 +217,7 @@ static int finalize_host_mappings_walker(const struct kvm_pgtable_visit_ctx *ctx
>   	 * Adjust the host stage-2 mappings to match the ownership attributes
>   	 * configured in the hypervisor stage-1.
>   	 */
> -	state = pkvm_getstate(kvm_pgtable_hyp_pte_prot(pte));
> +	state = pkvm_getstate(kvm_pgtable_hyp_pte_prot(ctx->old));
>   	switch (state) {
>   	case PKVM_PAGE_OWNED:
>   		return host_stage2_set_owner_locked(phys, PAGE_SIZE, pkvm_hyp_id);
> diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
> index 900c8b9c0cfc..fb3696b3a997 100644
> --- a/arch/arm64/kvm/hyp/pgtable.c
> +++ b/arch/arm64/kvm/hyp/pgtable.c
> @@ -189,6 +189,7 @@ static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
>   	enum kvm_pgtable_walk_flags flags = data->walker->flags;
>   	struct kvm_pgtable_visit_ctx ctx = {
>   		.ptep	= ptep,
> +		.old	= READ_ONCE(*ptep),
>   		.arg	= data->walker->arg,
>   		.addr	= data->addr,
>   		.end	= data->end,
> @@ -196,16 +197,16 @@ static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
>   		.flags	= flags,
>   	};
>   	int ret = 0;
> -	kvm_pte_t *childp, pte = *ptep;
> -	bool table = kvm_pte_table(pte, level);
> +	kvm_pte_t *childp;
> +	bool table = kvm_pte_table(ctx.old, level);
>   
>   	if (table && (ctx.flags & KVM_PGTABLE_WALK_TABLE_PRE))
>   		ret = kvm_pgtable_visitor_cb(data, &ctx, KVM_PGTABLE_WALK_TABLE_PRE);
>   
>   	if (!table && (ctx.flags & KVM_PGTABLE_WALK_LEAF)) {
>   		ret = kvm_pgtable_visitor_cb(data, &ctx, KVM_PGTABLE_WALK_LEAF);
> -		pte = *ptep;
> -		table = kvm_pte_table(pte, level);
> +		ctx.old = READ_ONCE(*ptep);
> +		table = kvm_pte_table(ctx.old, level);
>   	}
>   

Since we're here, it may be worthy to have comments to explain why we need
to read the PTE again, due to page table splitting. For example, write-protection
because of dirty logging for one page, which has been mapped by a PMD entry.

>   	if (ret)
> @@ -217,7 +218,7 @@ static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
>   		goto out;
>   	}
>   
> -	childp = kvm_pte_follow(pte, data->pgt->mm_ops);
> +	childp = kvm_pte_follow(ctx.old, data->pgt->mm_ops);
>   	ret = __kvm_pgtable_walk(data, childp, level + 1);
>   	if (ret)
>   		goto out;
> @@ -299,7 +300,7 @@ static int leaf_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   {
>   	struct leaf_walk_data *data = ctx->arg;
>   
> -	data->pte   = *ctx->ptep;
> +	data->pte   = ctx->old;
>   	data->level = ctx->level;
>   
>   	return 0;
> @@ -388,7 +389,7 @@ enum kvm_pgtable_prot kvm_pgtable_hyp_pte_prot(kvm_pte_t pte)
>   static bool hyp_map_walker_try_leaf(const struct kvm_pgtable_visit_ctx *ctx,
>   				    struct hyp_map_data *data)
>   {
> -	kvm_pte_t new, old = *ctx->ptep;
> +	kvm_pte_t new;
>   	u64 granule = kvm_granule_size(ctx->level), phys = data->phys;
>   
>   	if (!kvm_block_mapping_supported(ctx, phys))
> @@ -396,11 +397,11 @@ static bool hyp_map_walker_try_leaf(const struct kvm_pgtable_visit_ctx *ctx,
>   
>   	data->phys += granule;
>   	new = kvm_init_valid_leaf_pte(phys, data->attr, ctx->level);
> -	if (old == new)
> +	if (ctx->old == new)
>   		return true;
> -	if (!kvm_pte_valid(old))
> +	if (!kvm_pte_valid(ctx->old))
>   		data->mm_ops->get_page(ctx->ptep);
> -	else if (WARN_ON((old ^ new) & ~KVM_PTE_LEAF_ATTR_HI_SW))
> +	else if (WARN_ON((ctx->old ^ new) & ~KVM_PTE_LEAF_ATTR_HI_SW))
>   		return false;
>   
>   	smp_store_release(ctx->ptep, new);
> @@ -461,16 +462,16 @@ struct hyp_unmap_data {
>   static int hyp_unmap_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   			    enum kvm_pgtable_walk_flags visit)
>   {
> -	kvm_pte_t pte = *ctx->ptep, *childp = NULL;
> +	kvm_pte_t *childp = NULL;
>   	u64 granule = kvm_granule_size(ctx->level);
>   	struct hyp_unmap_data *data = ctx->arg;
>   	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
>   
> -	if (!kvm_pte_valid(pte))
> +	if (!kvm_pte_valid(ctx->old))
>   		return -EINVAL;
>   
> -	if (kvm_pte_table(pte, ctx->level)) {
> -		childp = kvm_pte_follow(pte, mm_ops);
> +	if (kvm_pte_table(ctx->old, ctx->level)) {
> +		childp = kvm_pte_follow(ctx->old, mm_ops);
>   
>   		if (mm_ops->page_count(childp) != 1)
>   			return 0;
> @@ -538,15 +539,14 @@ static int hyp_free_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   			   enum kvm_pgtable_walk_flags visit)
>   {
>   	struct kvm_pgtable_mm_ops *mm_ops = ctx->arg;
> -	kvm_pte_t pte = *ctx->ptep;
>   
> -	if (!kvm_pte_valid(pte))
> +	if (!kvm_pte_valid(ctx->old))
>   		return 0;
>   
>   	mm_ops->put_page(ctx->ptep);
>   
> -	if (kvm_pte_table(pte, ctx->level))
> -		mm_ops->put_page(kvm_pte_follow(pte, mm_ops));
> +	if (kvm_pte_table(ctx->old, ctx->level))
> +		mm_ops->put_page(kvm_pte_follow(ctx->old, mm_ops));
>   
>   	return 0;
>   }
> @@ -691,7 +691,7 @@ static void stage2_put_pte(const struct kvm_pgtable_visit_ctx *ctx, struct kvm_s
>   	 * Clear the existing PTE, and perform break-before-make with
>   	 * TLB maintenance if it was valid.
>   	 */
> -	if (kvm_pte_valid(*ctx->ptep)) {
> +	if (kvm_pte_valid(ctx->old)) {
>   		kvm_clear_pte(ctx->ptep);
>   		kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, mmu, ctx->addr, ctx->level);
>   	}
> @@ -722,7 +722,7 @@ static bool stage2_leaf_mapping_allowed(const struct kvm_pgtable_visit_ctx *ctx,
>   static int stage2_map_walker_try_leaf(const struct kvm_pgtable_visit_ctx *ctx,
>   				      struct stage2_map_data *data)
>   {
> -	kvm_pte_t new, old = *ctx->ptep;
> +	kvm_pte_t new;
>   	u64 granule = kvm_granule_size(ctx->level), phys = data->phys;
>   	struct kvm_pgtable *pgt = data->mmu->pgt;
>   	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
> @@ -735,14 +735,14 @@ static int stage2_map_walker_try_leaf(const struct kvm_pgtable_visit_ctx *ctx,
>   	else
>   		new = kvm_init_invalid_leaf_owner(data->owner_id);
>   
> -	if (stage2_pte_is_counted(old)) {
> +	if (stage2_pte_is_counted(ctx->old)) {
>   		/*
>   		 * Skip updating the PTE if we are trying to recreate the exact
>   		 * same mapping or only change the access permissions. Instead,
>   		 * the vCPU will exit one more time from guest if still needed
>   		 * and then go through the path of relaxing permissions.
>   		 */
> -		if (!stage2_pte_needs_update(old, new))
> +		if (!stage2_pte_needs_update(ctx->old, new))
>   			return -EAGAIN;
>   
>   		stage2_put_pte(ctx, data->mmu, mm_ops);
> @@ -773,7 +773,7 @@ static int stage2_map_walk_table_pre(const struct kvm_pgtable_visit_ctx *ctx,
>   	if (!stage2_leaf_mapping_allowed(ctx, data))
>   		return 0;
>   
> -	data->childp = kvm_pte_follow(*ctx->ptep, data->mm_ops);
> +	data->childp = kvm_pte_follow(ctx->old, data->mm_ops);
>   	kvm_clear_pte(ctx->ptep);
>   
>   	/*
> @@ -790,11 +790,11 @@ static int stage2_map_walk_leaf(const struct kvm_pgtable_visit_ctx *ctx,
>   				struct stage2_map_data *data)
>   {
>   	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
> -	kvm_pte_t *childp, pte = *ctx->ptep;
> +	kvm_pte_t *childp;
>   	int ret;
>   
>   	if (data->anchor) {
> -		if (stage2_pte_is_counted(pte))
> +		if (stage2_pte_is_counted(ctx->old))
>   			mm_ops->put_page(ctx->ptep);
>   
>   		return 0;
> @@ -819,7 +819,7 @@ static int stage2_map_walk_leaf(const struct kvm_pgtable_visit_ctx *ctx,
>   	 * a table. Accesses beyond 'end' that fall within the new table
>   	 * will be mapped lazily.
>   	 */
> -	if (stage2_pte_is_counted(pte))
> +	if (stage2_pte_is_counted(ctx->old))
>   		stage2_put_pte(ctx, data->mmu, mm_ops);
>   
>   	kvm_set_table_pte(ctx->ptep, childp, mm_ops);
> @@ -844,7 +844,7 @@ static int stage2_map_walk_table_post(const struct kvm_pgtable_visit_ctx *ctx,
>   		data->childp = NULL;
>   		ret = stage2_map_walk_leaf(ctx, data);
>   	} else {
> -		childp = kvm_pte_follow(*ctx->ptep, mm_ops);
> +		childp = kvm_pte_follow(ctx->old, mm_ops);
>   	}
>   
>   	mm_ops->put_page(childp);
> @@ -954,23 +954,23 @@ static int stage2_unmap_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   	struct kvm_pgtable *pgt = ctx->arg;
>   	struct kvm_s2_mmu *mmu = pgt->mmu;
>   	struct kvm_pgtable_mm_ops *mm_ops = pgt->mm_ops;
> -	kvm_pte_t pte = *ctx->ptep, *childp = NULL;
> +	kvm_pte_t *childp = NULL;
>   	bool need_flush = false;
>   
> -	if (!kvm_pte_valid(pte)) {
> -		if (stage2_pte_is_counted(pte)) {
> +	if (!kvm_pte_valid(ctx->old)) {
> +		if (stage2_pte_is_counted(ctx->old)) {
>   			kvm_clear_pte(ctx->ptep);
>   			mm_ops->put_page(ctx->ptep);
>   		}
>   		return 0;
>   	}
>   
> -	if (kvm_pte_table(pte, ctx->level)) {
> -		childp = kvm_pte_follow(pte, mm_ops);
> +	if (kvm_pte_table(ctx->old, ctx->level)) {
> +		childp = kvm_pte_follow(ctx->old, mm_ops);
>   
>   		if (mm_ops->page_count(childp) != 1)
>   			return 0;
> -	} else if (stage2_pte_cacheable(pgt, pte)) {
> +	} else if (stage2_pte_cacheable(pgt, ctx->old)) {
>   		need_flush = !stage2_has_fwb(pgt);
>   	}
>   
> @@ -982,7 +982,7 @@ static int stage2_unmap_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   	stage2_put_pte(ctx, mmu, mm_ops);
>   
>   	if (need_flush && mm_ops->dcache_clean_inval_poc)
> -		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
> +		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(ctx->old, mm_ops),
>   					       kvm_granule_size(ctx->level));
>   
>   	if (childp)
> @@ -1013,11 +1013,11 @@ struct stage2_attr_data {
>   static int stage2_attr_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   			      enum kvm_pgtable_walk_flags visit)
>   {
> -	kvm_pte_t pte = *ctx->ptep;
> +	kvm_pte_t pte = ctx->old;
>   	struct stage2_attr_data *data = ctx->arg;
>   	struct kvm_pgtable_mm_ops *mm_ops = data->mm_ops;
>   
> -	if (!kvm_pte_valid(pte))
> +	if (!kvm_pte_valid(ctx->old))
>   		return 0;
>   
>   	data->level = ctx->level;
> @@ -1036,7 +1036,7 @@ static int stage2_attr_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   		 * stage-2 PTE if we are going to add executable permission.
>   		 */
>   		if (mm_ops->icache_inval_pou &&
> -		    stage2_pte_executable(pte) && !stage2_pte_executable(*ctx->ptep))
> +		    stage2_pte_executable(pte) && !stage2_pte_executable(ctx->old))
>   			mm_ops->icache_inval_pou(kvm_pte_follow(pte, mm_ops),
>   						  kvm_granule_size(ctx->level));
>   		WRITE_ONCE(*ctx->ptep, pte);
> @@ -1142,13 +1142,12 @@ static int stage2_flush_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   {
>   	struct kvm_pgtable *pgt = ctx->arg;
>   	struct kvm_pgtable_mm_ops *mm_ops = pgt->mm_ops;
> -	kvm_pte_t pte = *ctx->ptep;
>   
> -	if (!kvm_pte_valid(pte) || !stage2_pte_cacheable(pgt, pte))
> +	if (!kvm_pte_valid(ctx->old) || !stage2_pte_cacheable(pgt, ctx->old))
>   		return 0;
>   
>   	if (mm_ops->dcache_clean_inval_poc)
> -		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(pte, mm_ops),
> +		mm_ops->dcache_clean_inval_poc(kvm_pte_follow(ctx->old, mm_ops),
>   					       kvm_granule_size(ctx->level));
>   	return 0;
>   }
> @@ -1200,15 +1199,14 @@ static int stage2_free_walker(const struct kvm_pgtable_visit_ctx *ctx,
>   			      enum kvm_pgtable_walk_flags visit)
>   {
>   	struct kvm_pgtable_mm_ops *mm_ops = ctx->arg;
> -	kvm_pte_t pte = *ctx->ptep;
>   
> -	if (!stage2_pte_is_counted(pte))
> +	if (!stage2_pte_is_counted(ctx->old))
>   		return 0;
>   
>   	mm_ops->put_page(ctx->ptep);
>   
> -	if (kvm_pte_table(pte, ctx->level))
> -		mm_ops->put_page(kvm_pte_follow(pte, mm_ops));
> +	if (kvm_pte_table(ctx->old, ctx->level))
> +		mm_ops->put_page(kvm_pte_follow(ctx->old, mm_ops));
>   
>   	return 0;
>   }

Thanks,
Gavin