KVM: PPC: Book3S HV: Introduce rmap to track nested guest mappings

When a host (L0) page which is mapped into a (L1) guest is in turn
mapped through to a nested (L2) guest we keep a reverse mapping (rmap)
so that these mappings can be retrieved later.

Whenever we create an entry in a shadow_pgtable for a nested guest we
create a corresponding rmap entry and add it to the list for the
L1 guest memslot at the index of the L1 guest page it maps. This means
at the L1 guest memslot we end up with lists of rmaps.

When we are notified of a host page being invalidated which has been
mapped through to a (L1) guest, we can then walk the rmap list for that
guest page, and find and invalidate all of the corresponding
shadow_pgtable entries.

In order to reduce memory consumption, we compress the information for
each rmap entry down to 52 bits -- 12 bits for the LPID and 40 bits
for the guest real page frame number -- which will fit in a single
unsigned long.  To avoid a scenario where a guest can trigger
unbounded memory allocations, we scan the list when adding an entry to
see if there is already an entry with the contents we need.  This can
occur, because we don't ever remove entries from the middle of a list.

A struct nested guest rmap is a list pointer and an rmap entry;
----------------
| next pointer |
----------------
| rmap entry   |
----------------

Thus the rmap pointer for each guest frame number in the memslot can be
either NULL, a single entry, or a pointer to a list of nested rmap entries.

gfn	 memslot rmap array
 	-------------------------
 0	| NULL			|	(no rmap entry)
 	-------------------------
 1	| single rmap entry	|	(rmap entry with low bit set)
 	-------------------------
 2	| list head pointer	|	(list of rmap entries)
 	-------------------------

The final entry always has the lowest bit set and is stored in the next
pointer of the last list entry, or as a single rmap entry.
With a list of rmap entries looking like;

-----------------	-----------------	-------------------------
| list head ptr	| ----> | next pointer	| ---->	| single rmap entry	|
-----------------	-----------------	-------------------------
			| rmap entry	|	| rmap entry		|
			-----------------	-------------------------

Signed-off-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

3 files changed, 169 insertions, 14 deletions

diff --git a/arch/powerpc/kvm/book3s_64_mmu_radix.c b/arch/powerpc/kvm/book3s_64_mmu_radix.c
index c4b1a9e1e3ff..4c1eccb20190 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_radix.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_radix.c
@@ -256,27 +256,38 @@ static void kvmppc_pmd_free(pmd_t *pmdp)
 	kmem_cache_free(kvm_pmd_cache, pmdp);
 }
 
-void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte,
-		      unsigned long gpa, unsigned int shift,
-		      struct kvm_memory_slot *memslot,
+/* Called with kvm->mmu_lock held */
+void kvmppc_unmap_pte(struct kvm *kvm, pte_t *pte, unsigned long gpa,
+		      unsigned int shift, struct kvm_memory_slot *memslot,
 		      unsigned int lpid)
 
 {
 	unsigned long old;
+	unsigned long gfn = gpa >> PAGE_SHIFT;
+	unsigned long page_size = PAGE_SIZE;
+	unsigned long hpa;
 
 	old = kvmppc_radix_update_pte(kvm, pte, ~0UL, 0, gpa, shift);
 	kvmppc_radix_tlbie_page(kvm, gpa, shift, lpid);
-	if ((old & _PAGE_DIRTY) && (lpid == kvm->arch.lpid)) {
-		unsigned long gfn = gpa >> PAGE_SHIFT;
-		unsigned long page_size = PAGE_SIZE;
 
-		if (shift)
-			page_size = 1ul << shift;
+	/* The following only applies to L1 entries */
+	if (lpid != kvm->arch.lpid)
+		return;
+
+	if (!memslot) {
+		memslot = gfn_to_memslot(kvm, gfn);
 		if (!memslot)
-			memslot = gfn_to_memslot(kvm, gfn);
-		if (memslot && memslot->dirty_bitmap)
-			kvmppc_update_dirty_map(memslot, gfn, page_size);
+			return;
 	}
+	if (shift)
+		page_size = 1ul << shift;
+
+	gpa &= ~(page_size - 1);
+	hpa = old & PTE_RPN_MASK;
+	kvmhv_remove_nest_rmap_range(kvm, memslot, gpa, hpa, page_size);
+
+	if ((old & _PAGE_DIRTY) && memslot->dirty_bitmap)
+		kvmppc_update_dirty_map(memslot, gfn, page_size);
 }
 
 /*
@@ -430,7 +441,8 @@ static void kvmppc_unmap_free_pud_entry_table(struct kvm *kvm, pud_t *pud,
 
 int kvmppc_create_pte(struct kvm *kvm, pgd_t *pgtable, pte_t pte,
 		      unsigned long gpa, unsigned int level,
-		      unsigned long mmu_seq, unsigned int lpid)
+		      unsigned long mmu_seq, unsigned int lpid,
+		      unsigned long *rmapp, struct rmap_nested **n_rmap)
 {
 	pgd_t *pgd;
 	pud_t *pud, *new_pud = NULL;
@@ -509,6 +521,8 @@ int kvmppc_create_pte(struct kvm *kvm, pgd_t *pgtable, pte_t pte,
 			kvmppc_unmap_free_pud_entry_table(kvm, pud, gpa, lpid);
 		}
 		kvmppc_radix_set_pte_at(kvm, gpa, (pte_t *)pud, pte);
+		if (rmapp && n_rmap)
+			kvmhv_insert_nest_rmap(kvm, rmapp, n_rmap);
 		ret = 0;
 		goto out_unlock;
 	}
@@ -559,6 +573,8 @@ int kvmppc_create_pte(struct kvm *kvm, pgd_t *pgtable, pte_t pte,
 			kvmppc_unmap_free_pmd_entry_table(kvm, pmd, gpa, lpid);
 		}
 		kvmppc_radix_set_pte_at(kvm, gpa, pmdp_ptep(pmd), pte);
+		if (rmapp && n_rmap)
+			kvmhv_insert_nest_rmap(kvm, rmapp, n_rmap);
 		ret = 0;
 		goto out_unlock;
 	}
@@ -583,6 +599,8 @@ int kvmppc_create_pte(struct kvm *kvm, pgd_t *pgtable, pte_t pte,
 		goto out_unlock;
 	}
 	kvmppc_radix_set_pte_at(kvm, gpa, ptep, pte);
+	if (rmapp && n_rmap)
+		kvmhv_insert_nest_rmap(kvm, rmapp, n_rmap);
 	ret = 0;
 
  out_unlock:
@@ -710,7 +728,7 @@ int kvmppc_book3s_instantiate_page(struct kvm_vcpu *vcpu,
 
 	/* Allocate space in the tree and write the PTE */
 	ret = kvmppc_create_pte(kvm, kvm->arch.pgtable, pte, gpa, level,
-				mmu_seq, kvm->arch.lpid);
+				mmu_seq, kvm->arch.lpid, NULL, NULL);
 	if (inserted_pte)
 		*inserted_pte = pte;
 	if (levelp)
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index a87912508f63..6e69b4de9a9a 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -4486,6 +4486,7 @@ int kvmppc_switch_mmu_to_radix(struct kvm *kvm)
 	kvmppc_free_hpt(&kvm->arch.hpt);
 	kvmppc_update_lpcr(kvm, LPCR_UPRT | LPCR_GTSE | LPCR_HR,
 			   LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR);
+	kvmppc_rmap_reset(kvm);
 	kvm->arch.radix = 1;
 	return 0;
 }
diff --git a/arch/powerpc/kvm/book3s_hv_nested.c b/arch/powerpc/kvm/book3s_hv_nested.c
index 21a210c134af..3fa676b2acd9 100644
--- a/arch/powerpc/kvm/book3s_hv_nested.c
+++ b/arch/powerpc/kvm/book3s_hv_nested.c
@@ -10,6 +10,7 @@
 
 #include <linux/kernel.h>
 #include <linux/kvm_host.h>
+#include <linux/llist.h>
 
 #include <asm/kvm_ppc.h>
 #include <asm/kvm_book3s.h>
@@ -22,6 +23,7 @@
 static struct patb_entry *pseries_partition_tb;
 
 static void kvmhv_update_ptbl_cache(struct kvm_nested_guest *gp);
+static void kvmhv_free_memslot_nest_rmap(struct kvm_memory_slot *free);
 
 void kvmhv_save_hv_regs(struct kvm_vcpu *vcpu, struct hv_guest_state *hr)
 {
@@ -456,6 +458,8 @@ void kvmhv_release_all_nested(struct kvm *kvm)
 	int i;
 	struct kvm_nested_guest *gp;
 	struct kvm_nested_guest *freelist = NULL;
+	struct kvm_memory_slot *memslot;
+	int srcu_idx;
 
 	spin_lock(&kvm->mmu_lock);
 	for (i = 0; i <= kvm->arch.max_nested_lpid; i++) {
@@ -474,6 +478,11 @@ void kvmhv_release_all_nested(struct kvm *kvm)
 		freelist = gp->next;
 		kvmhv_release_nested(gp);
 	}
+
+	srcu_idx = srcu_read_lock(&kvm->srcu);
+	kvm_for_each_memslot(memslot, kvm_memslots(kvm))
+		kvmhv_free_memslot_nest_rmap(memslot);
+	srcu_read_unlock(&kvm->srcu, srcu_idx);
 }
 
 /* caller must hold gp->tlb_lock */
@@ -544,6 +553,123 @@ void kvmhv_put_nested(struct kvm_nested_guest *gp)
 		kvmhv_release_nested(gp);
 }
 
+static struct kvm_nested_guest *kvmhv_find_nested(struct kvm *kvm, int lpid)
+{
+	if (lpid > kvm->arch.max_nested_lpid)
+		return NULL;
+	return kvm->arch.nested_guests[lpid];
+}
+
+static inline bool kvmhv_n_rmap_is_equal(u64 rmap_1, u64 rmap_2)
+{
+	return !((rmap_1 ^ rmap_2) & (RMAP_NESTED_LPID_MASK |
+				       RMAP_NESTED_GPA_MASK));
+}
+
+void kvmhv_insert_nest_rmap(struct kvm *kvm, unsigned long *rmapp,
+			    struct rmap_nested **n_rmap)
+{
+	struct llist_node *entry = ((struct llist_head *) rmapp)->first;
+	struct rmap_nested *cursor;
+	u64 rmap, new_rmap = (*n_rmap)->rmap;
+
+	/* Are there any existing entries? */
+	if (!(*rmapp)) {
+		/* No -> use the rmap as a single entry */
+		*rmapp = new_rmap | RMAP_NESTED_IS_SINGLE_ENTRY;
+		return;
+	}
+
+	/* Do any entries match what we're trying to insert? */
+	for_each_nest_rmap_safe(cursor, entry, &rmap) {
+		if (kvmhv_n_rmap_is_equal(rmap, new_rmap))
+			return;
+	}
+
+	/* Do we need to create a list or just add the new entry? */
+	rmap = *rmapp;
+	if (rmap & RMAP_NESTED_IS_SINGLE_ENTRY) /* Not previously a list */
+		*rmapp = 0UL;
+	llist_add(&((*n_rmap)->list), (struct llist_head *) rmapp);
+	if (rmap & RMAP_NESTED_IS_SINGLE_ENTRY) /* Not previously a list */
+		(*n_rmap)->list.next = (struct llist_node *) rmap;
+
+	/* Set NULL so not freed by caller */
+	*n_rmap = NULL;
+}
+
+static void kvmhv_remove_nest_rmap(struct kvm *kvm, u64 n_rmap,
+				   unsigned long hpa, unsigned long mask)
+{
+	struct kvm_nested_guest *gp;
+	unsigned long gpa;
+	unsigned int shift, lpid;
+	pte_t *ptep;
+
+	gpa = n_rmap & RMAP_NESTED_GPA_MASK;
+	lpid = (n_rmap & RMAP_NESTED_LPID_MASK) >> RMAP_NESTED_LPID_SHIFT;
+	gp = kvmhv_find_nested(kvm, lpid);
+	if (!gp)
+		return;
+
+	/* Find and invalidate the pte */
+	ptep = __find_linux_pte(gp->shadow_pgtable, gpa, NULL, &shift);
+	/* Don't spuriously invalidate ptes if the pfn has changed */
+	if (ptep && pte_present(*ptep) && ((pte_val(*ptep) & mask) == hpa))
+		kvmppc_unmap_pte(kvm, ptep, gpa, shift, NULL, gp->shadow_lpid);
+}
+
+static void kvmhv_remove_nest_rmap_list(struct kvm *kvm, unsigned long *rmapp,
+					unsigned long hpa, unsigned long mask)
+{
+	struct llist_node *entry = llist_del_all((struct llist_head *) rmapp);
+	struct rmap_nested *cursor;
+	unsigned long rmap;
+
+	for_each_nest_rmap_safe(cursor, entry, &rmap) {
+		kvmhv_remove_nest_rmap(kvm, rmap, hpa, mask);
+		kfree(cursor);
+	}
+}
+
+/* called with kvm->mmu_lock held */
+void kvmhv_remove_nest_rmap_range(struct kvm *kvm,
+				  struct kvm_memory_slot *memslot,
+				  unsigned long gpa, unsigned long hpa,
+				  unsigned long nbytes)
+{
+	unsigned long gfn, end_gfn;
+	unsigned long addr_mask;
+
+	if (!memslot)
+		return;
+	gfn = (gpa >> PAGE_SHIFT) - memslot->base_gfn;
+	end_gfn = gfn + (nbytes >> PAGE_SHIFT);
+
+	addr_mask = PTE_RPN_MASK & ~(nbytes - 1);
+	hpa &= addr_mask;
+
+	for (; gfn < end_gfn; gfn++) {
+		unsigned long *rmap = &memslot->arch.rmap[gfn];
+		kvmhv_remove_nest_rmap_list(kvm, rmap, hpa, addr_mask);
+	}
+}
+
+static void kvmhv_free_memslot_nest_rmap(struct kvm_memory_slot *free)
+{
+	unsigned long page;
+
+	for (page = 0; page < free->npages; page++) {
+		unsigned long rmap, *rmapp = &free->arch.rmap[page];
+		struct rmap_nested *cursor;
+		struct llist_node *entry;
+
+		entry = llist_del_all((struct llist_head *) rmapp);
+		for_each_nest_rmap_safe(cursor, entry, &rmap)
+			kfree(cursor);
+	}
+}
+
 static bool kvmhv_invalidate_shadow_pte(struct kvm_vcpu *vcpu,
 					struct kvm_nested_guest *gp,
 					long gpa, int *shift_ret)
@@ -695,11 +821,13 @@ static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu,
 {
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_memory_slot *memslot;
+	struct rmap_nested *n_rmap;
 	struct kvmppc_pte gpte;
 	pte_t pte, *pte_p;
 	unsigned long mmu_seq;
 	unsigned long dsisr = vcpu->arch.fault_dsisr;
 	unsigned long ea = vcpu->arch.fault_dar;
+	unsigned long *rmapp;
 	unsigned long n_gpa, gpa, gfn, perm = 0UL;
 	unsigned int shift, l1_shift, level;
 	bool writing = !!(dsisr & DSISR_ISSTORE);
@@ -833,8 +961,16 @@ static long int __kvmhv_nested_page_fault(struct kvm_vcpu *vcpu,
 
 	/* 4. Insert the pte into our shadow_pgtable */
 
+	n_rmap = kzalloc(sizeof(*n_rmap), GFP_KERNEL);
+	if (!n_rmap)
+		return RESUME_GUEST; /* Let the guest try again */
+	n_rmap->rmap = (n_gpa & RMAP_NESTED_GPA_MASK) |
+		(((unsigned long) gp->l1_lpid) << RMAP_NESTED_LPID_SHIFT);
+	rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
 	ret = kvmppc_create_pte(kvm, gp->shadow_pgtable, pte, n_gpa, level,
-				mmu_seq, gp->shadow_lpid);
+				mmu_seq, gp->shadow_lpid, rmapp, &n_rmap);
+	if (n_rmap)
+		kfree(n_rmap);
 	if (ret == -EAGAIN)
 		ret = RESUME_GUEST;	/* Let the guest try again */