1 files changed, 202 insertions, 36 deletions

diff --git a/arch/arm64/kvm/hyp/nvhe/mem_protect.c b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
index 7a0c4af186a0..bacd493a4eac 100644
--- a/arch/arm64/kvm/hyp/nvhe/mem_protect.c
+++ b/arch/arm64/kvm/hyp/nvhe/mem_protect.c
@@ -31,7 +31,7 @@ static struct hyp_pool host_s2_pool;
 u64 id_aa64mmfr0_el1_sys_val;
 u64 id_aa64mmfr1_el1_sys_val;
 
-static const u8 pkvm_hyp_id = 1;
+const u8 pkvm_hyp_id = 1;
 
 static void *host_s2_zalloc_pages_exact(size_t size)
 {
@@ -89,6 +89,8 @@ static void prepare_host_vtcr(void)
 					  id_aa64mmfr1_el1_sys_val, phys_shift);
 }
 
+static bool host_stage2_force_pte_cb(u64 addr, u64 end, enum kvm_pgtable_prot prot);
+
 int kvm_host_prepare_stage2(void *pgt_pool_base)
 {
 	struct kvm_s2_mmu *mmu = &host_kvm.arch.mmu;
@@ -101,8 +103,9 @@ int kvm_host_prepare_stage2(void *pgt_pool_base)
 	if (ret)
 		return ret;
 
-	ret = kvm_pgtable_stage2_init_flags(&host_kvm.pgt, &host_kvm.arch,
-					    &host_kvm.mm_ops, KVM_HOST_S2_FLAGS);
+	ret = __kvm_pgtable_stage2_init(&host_kvm.pgt, &host_kvm.arch,
+					&host_kvm.mm_ops, KVM_HOST_S2_FLAGS,
+					host_stage2_force_pte_cb);
 	if (ret)
 		return ret;
 
@@ -159,6 +162,11 @@ static int host_stage2_unmap_dev_all(void)
 	return kvm_pgtable_stage2_unmap(pgt, addr, BIT(pgt->ia_bits) - addr);
 }
 
+struct kvm_mem_range {
+	u64 start;
+	u64 end;
+};
+
 static bool find_mem_range(phys_addr_t addr, struct kvm_mem_range *range)
 {
 	int cur, left = 0, right = hyp_memblock_nr;
@@ -189,16 +197,26 @@ static bool find_mem_range(phys_addr_t addr, struct kvm_mem_range *range)
 	return false;
 }
 
+bool addr_is_memory(phys_addr_t phys)
+{
+	struct kvm_mem_range range;
+
+	return find_mem_range(phys, &range);
+}
+
+static bool is_in_mem_range(u64 addr, struct kvm_mem_range *range)
+{
+	return range->start <= addr && addr < range->end;
+}
+
 static bool range_is_memory(u64 start, u64 end)
 {
-	struct kvm_mem_range r1, r2;
+	struct kvm_mem_range r;
 
-	if (!find_mem_range(start, &r1) || !find_mem_range(end, &r2))
-		return false;
-	if (r1.start != r2.start)
+	if (!find_mem_range(start, &r))
 		return false;
 
-	return true;
+	return is_in_mem_range(end - 1, &r);
 }
 
 static inline int __host_stage2_idmap(u64 start, u64 end,
@@ -208,60 +226,208 @@ static inline int __host_stage2_idmap(u64 start, u64 end,
 				      prot, &host_s2_pool);
 }
 
+/*
+ * The pool has been provided with enough pages to cover all of memory with
+ * page granularity, but it is difficult to know how much of the MMIO range
+ * we will need to cover upfront, so we may need to 'recycle' the pages if we
+ * run out.
+ */
+#define host_stage2_try(fn, ...)					\
+	({								\
+		int __ret;						\
+		hyp_assert_lock_held(&host_kvm.lock);			\
+		__ret = fn(__VA_ARGS__);				\
+		if (__ret == -ENOMEM) {					\
+			__ret = host_stage2_unmap_dev_all();		\
+			if (!__ret)					\
+				__ret = fn(__VA_ARGS__);		\
+		}							\
+		__ret;							\
+	 })
+
+static inline bool range_included(struct kvm_mem_range *child,
+				  struct kvm_mem_range *parent)
+{
+	return parent->start <= child->start && child->end <= parent->end;
+}
+
+static int host_stage2_adjust_range(u64 addr, struct kvm_mem_range *range)
+{
+	struct kvm_mem_range cur;
+	kvm_pte_t pte;
+	u32 level;
+	int ret;
+
+	hyp_assert_lock_held(&host_kvm.lock);
+	ret = kvm_pgtable_get_leaf(&host_kvm.pgt, addr, &pte, &level);
+	if (ret)
+		return ret;
+
+	if (kvm_pte_valid(pte))
+		return -EAGAIN;
+
+	if (pte)
+		return -EPERM;
+
+	do {
+		u64 granule = kvm_granule_size(level);
+		cur.start = ALIGN_DOWN(addr, granule);
+		cur.end = cur.start + granule;
+		level++;
+	} while ((level < KVM_PGTABLE_MAX_LEVELS) &&
+			!(kvm_level_supports_block_mapping(level) &&
+			  range_included(&cur, range)));
+
+	*range = cur;
+
+	return 0;
+}
+
+int host_stage2_idmap_locked(phys_addr_t addr, u64 size,
+			     enum kvm_pgtable_prot prot)
+{
+	hyp_assert_lock_held(&host_kvm.lock);
+
+	return host_stage2_try(__host_stage2_idmap, addr, addr + size, prot);
+}
+
+int host_stage2_set_owner_locked(phys_addr_t addr, u64 size, u8 owner_id)
+{
+	hyp_assert_lock_held(&host_kvm.lock);
+
+	return host_stage2_try(kvm_pgtable_stage2_set_owner, &host_kvm.pgt,
+			       addr, size, &host_s2_pool, owner_id);
+}
+
+static bool host_stage2_force_pte_cb(u64 addr, u64 end, enum kvm_pgtable_prot prot)
+{
+	/*
+	 * Block mappings must be used with care in the host stage-2 as a
+	 * kvm_pgtable_stage2_map() operation targeting a page in the range of
+	 * an existing block will delete the block under the assumption that
+	 * mappings in the rest of the block range can always be rebuilt lazily.
+	 * That assumption is correct for the host stage-2 with RWX mappings
+	 * targeting memory or RW mappings targeting MMIO ranges (see
+	 * host_stage2_idmap() below which implements some of the host memory
+	 * abort logic). However, this is not safe for any other mappings where
+	 * the host stage-2 page-table is in fact the only place where this
+	 * state is stored. In all those cases, it is safer to use page-level
+	 * mappings, hence avoiding to lose the state because of side-effects in
+	 * kvm_pgtable_stage2_map().
+	 */
+	if (range_is_memory(addr, end))
+		return prot != PKVM_HOST_MEM_PROT;
+	else
+		return prot != PKVM_HOST_MMIO_PROT;
+}
+
 static int host_stage2_idmap(u64 addr)
 {
-	enum kvm_pgtable_prot prot = KVM_PGTABLE_PROT_R | KVM_PGTABLE_PROT_W;
 	struct kvm_mem_range range;
 	bool is_memory = find_mem_range(addr, &range);
+	enum kvm_pgtable_prot prot;
 	int ret;
 
-	if (is_memory)
-		prot |= KVM_PGTABLE_PROT_X;
+	prot = is_memory ? PKVM_HOST_MEM_PROT : PKVM_HOST_MMIO_PROT;
 
 	hyp_spin_lock(&host_kvm.lock);
-	ret = kvm_pgtable_stage2_find_range(&host_kvm.pgt, addr, prot, &range);
+	ret = host_stage2_adjust_range(addr, &range);
 	if (ret)
 		goto unlock;
 
-	ret = __host_stage2_idmap(range.start, range.end, prot);
-	if (ret != -ENOMEM)
+	ret = host_stage2_idmap_locked(range.start, range.end - range.start, prot);
+unlock:
+	hyp_spin_unlock(&host_kvm.lock);
+
+	return ret;
+}
+
+static inline bool check_prot(enum kvm_pgtable_prot prot,
+			      enum kvm_pgtable_prot required,
+			      enum kvm_pgtable_prot denied)
+{
+	return (prot & (required | denied)) == required;
+}
+
+int __pkvm_host_share_hyp(u64 pfn)
+{
+	phys_addr_t addr = hyp_pfn_to_phys(pfn);
+	enum kvm_pgtable_prot prot, cur;
+	void *virt = __hyp_va(addr);
+	enum pkvm_page_state state;
+	kvm_pte_t pte;
+	int ret;
+
+	if (!addr_is_memory(addr))
+		return -EINVAL;
+
+	hyp_spin_lock(&host_kvm.lock);
+	hyp_spin_lock(&pkvm_pgd_lock);
+
+	ret = kvm_pgtable_get_leaf(&host_kvm.pgt, addr, &pte, NULL);
+	if (ret)
 		goto unlock;
+	if (!pte)
+		goto map_shared;
 
 	/*
-	 * The pool has been provided with enough pages to cover all of memory
-	 * with page granularity, but it is difficult to know how much of the
-	 * MMIO range we will need to cover upfront, so we may need to 'recycle'
-	 * the pages if we run out.
+	 * Check attributes in the host stage-2 PTE. We need the page to be:
+	 *  - mapped RWX as we're sharing memory;
+	 *  - not borrowed, as that implies absence of ownership.
+	 * Otherwise, we can't let it got through
 	 */
-	ret = host_stage2_unmap_dev_all();
-	if (ret)
+	cur = kvm_pgtable_stage2_pte_prot(pte);
+	prot = pkvm_mkstate(0, PKVM_PAGE_SHARED_BORROWED);
+	if (!check_prot(cur, PKVM_HOST_MEM_PROT, prot)) {
+		ret = -EPERM;
 		goto unlock;
+	}
 
-	ret = __host_stage2_idmap(range.start, range.end, prot);
+	state = pkvm_getstate(cur);
+	if (state == PKVM_PAGE_OWNED)
+		goto map_shared;
 
-unlock:
-	hyp_spin_unlock(&host_kvm.lock);
+	/*
+	 * Tolerate double-sharing the same page, but this requires
+	 * cross-checking the hypervisor stage-1.
+	 */
+	if (state != PKVM_PAGE_SHARED_OWNED) {
+		ret = -EPERM;
+		goto unlock;
+	}
 
-	return ret;
-}
+	ret = kvm_pgtable_get_leaf(&pkvm_pgtable, (u64)virt, &pte, NULL);
+	if (ret)
+		goto unlock;
 
-int __pkvm_mark_hyp(phys_addr_t start, phys_addr_t end)
-{
-	int ret;
+	/*
+	 * If the page has been shared with the hypervisor, it must be
+	 * already mapped as SHARED_BORROWED in its stage-1.
+	 */
+	cur = kvm_pgtable_hyp_pte_prot(pte);
+	prot = pkvm_mkstate(PAGE_HYP, PKVM_PAGE_SHARED_BORROWED);
+	if (!check_prot(cur, prot, ~prot))
+		ret = -EPERM;
+	goto unlock;
 
+map_shared:
 	/*
-	 * host_stage2_unmap_dev_all() currently relies on MMIO mappings being
-	 * non-persistent, so don't allow changing page ownership in MMIO range.
+	 * If the page is not yet shared, adjust mappings in both page-tables
+	 * while both locks are held.
 	 */
-	if (!range_is_memory(start, end))
-		return -EINVAL;
+	prot = pkvm_mkstate(PAGE_HYP, PKVM_PAGE_SHARED_BORROWED);
+	ret = pkvm_create_mappings_locked(virt, virt + PAGE_SIZE, prot);
+	BUG_ON(ret);
 
-	hyp_spin_lock(&host_kvm.lock);
-	ret = kvm_pgtable_stage2_set_owner(&host_kvm.pgt, start, end - start,
-					   &host_s2_pool, pkvm_hyp_id);
+	prot = pkvm_mkstate(PKVM_HOST_MEM_PROT, PKVM_PAGE_SHARED_OWNED);
+	ret = host_stage2_idmap_locked(addr, PAGE_SIZE, prot);
+	BUG_ON(ret);
+
+unlock:
+	hyp_spin_unlock(&pkvm_pgd_lock);
 	hyp_spin_unlock(&host_kvm.lock);
 
-	return ret != -EAGAIN ? ret : 0;
+	return ret;
 }
 
 void handle_host_mem_abort(struct kvm_cpu_context *host_ctxt)