KVM: PPC: Book3S HV: Unify dirty page map between HPT and radix

Currently, the HPT code in HV KVM maintains a dirty bit per guest page
in the rmap array, whether or not dirty page tracking has been enabled
for the memory slot.  In contrast, the radix code maintains a dirty
bit per guest page in memslot->dirty_bitmap, and only does so when
dirty page tracking has been enabled.

This changes the HPT code to maintain the dirty bits in the memslot
dirty_bitmap like radix does.  This results in slightly less code
overall, and will mean that we do not lose the dirty bits when
transitioning between HPT and radix mode in future.

There is one minor change to behaviour as a result.  With HPT, when
dirty tracking was enabled for a memslot, we would previously clear
all the dirty bits at that point (both in the HPT entries and in the
rmap arrays), meaning that a KVM_GET_DIRTY_LOG ioctl immediately
following would show no pages as dirty (assuming no vcpus have run
in the meantime).  With this change, the dirty bits on HPT entries
are not cleared at the point where dirty tracking is enabled, so
KVM_GET_DIRTY_LOG would show as dirty any guest pages that are
resident in the HPT and dirty.  This is consistent with what happens
on radix.

This also fixes a bug in the mark_pages_dirty() function for radix
(in the sense that the function no longer exists).  In the case where
a large page of 64 normal pages or more is marked dirty, the
addressing of the dirty bitmap was incorrect and could write past
the end of the bitmap.  Fortunately this case was never hit in
practice because a 2MB large page is only 32 x 64kB pages, and we
don't support backing the guest with 1GB huge pages at this point.

Signed-off-by: Paul Mackerras <paulus@ozlabs.org>

4 files changed, 70 insertions, 109 deletions

diff --git a/arch/powerpc/kvm/book3s_64_mmu_hv.c b/arch/powerpc/kvm/book3s_64_mmu_hv.c
index 84728038e943..944f7a532879 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_hv.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_hv.c
@@ -776,6 +776,7 @@ static int kvm_handle_hva(struct kvm *kvm, unsigned long hva,
 
 /* Must be called with both HPTE and rmap locked */
 static void kvmppc_unmap_hpte(struct kvm *kvm, unsigned long i,
+			      struct kvm_memory_slot *memslot,
 			      unsigned long *rmapp, unsigned long gfn)
 {
 	__be64 *hptep = (__be64 *) (kvm->arch.hpt.virt + (i << 4));
@@ -807,8 +808,8 @@ static void kvmppc_unmap_hpte(struct kvm *kvm, unsigned long i,
 		/* Harvest R and C */
 		rcbits = be64_to_cpu(hptep[1]) & (HPTE_R_R | HPTE_R_C);
 		*rmapp |= rcbits << KVMPPC_RMAP_RC_SHIFT;
-		if (rcbits & HPTE_R_C)
-			kvmppc_update_rmap_change(rmapp, psize);
+		if ((rcbits & HPTE_R_C) && memslot->dirty_bitmap)
+			kvmppc_update_dirty_map(memslot, gfn, psize);
 		if (rcbits & ~rev[i].guest_rpte) {
 			rev[i].guest_rpte = ptel | rcbits;
 			note_hpte_modification(kvm, &rev[i]);
@@ -846,7 +847,7 @@ static int kvm_unmap_rmapp(struct kvm *kvm, struct kvm_memory_slot *memslot,
 			continue;
 		}
 
-		kvmppc_unmap_hpte(kvm, i, rmapp, gfn);
+		kvmppc_unmap_hpte(kvm, i, memslot, rmapp, gfn);
 		unlock_rmap(rmapp);
 		__unlock_hpte(hptep, be64_to_cpu(hptep[0]));
 	}
@@ -1029,14 +1030,6 @@ static int kvm_test_clear_dirty_npages(struct kvm *kvm, unsigned long *rmapp)
 
  retry:
 	lock_rmap(rmapp);
-	if (*rmapp & KVMPPC_RMAP_CHANGED) {
-		long change_order = (*rmapp & KVMPPC_RMAP_CHG_ORDER)
-			>> KVMPPC_RMAP_CHG_SHIFT;
-		*rmapp &= ~(KVMPPC_RMAP_CHANGED | KVMPPC_RMAP_CHG_ORDER);
-		npages_dirty = 1;
-		if (change_order > PAGE_SHIFT)
-			npages_dirty = 1ul << (change_order - PAGE_SHIFT);
-	}
 	if (!(*rmapp & KVMPPC_RMAP_PRESENT)) {
 		unlock_rmap(rmapp);
 		return npages_dirty;
@@ -1128,7 +1121,7 @@ void kvmppc_harvest_vpa_dirty(struct kvmppc_vpa *vpa,
 long kvmppc_hv_get_dirty_log_hpt(struct kvm *kvm,
 			struct kvm_memory_slot *memslot, unsigned long *map)
 {
-	unsigned long i, j;
+	unsigned long i;
 	unsigned long *rmapp;
 
 	preempt_disable();
@@ -1140,9 +1133,8 @@ long kvmppc_hv_get_dirty_log_hpt(struct kvm *kvm,
 		 * since we always put huge-page HPTEs in the rmap chain
 		 * corresponding to their page base address.
 		 */
-		if (npages && map)
-			for (j = i; npages; ++j, --npages)
-				__set_bit_le(j, map);
+		if (npages)
+			set_dirty_bits(map, i, npages);
 		++rmapp;
 	}
 	preempt_enable();
@@ -1186,7 +1178,6 @@ void kvmppc_unpin_guest_page(struct kvm *kvm, void *va, unsigned long gpa,
 	struct page *page = virt_to_page(va);
 	struct kvm_memory_slot *memslot;
 	unsigned long gfn;
-	unsigned long *rmap;
 	int srcu_idx;
 
 	put_page(page);
@@ -1194,20 +1185,12 @@ void kvmppc_unpin_guest_page(struct kvm *kvm, void *va, unsigned long gpa,
 	if (!dirty)
 		return;
 
-	/* We need to mark this page dirty in the rmap chain */
+	/* We need to mark this page dirty in the memslot dirty_bitmap, if any */
 	gfn = gpa >> PAGE_SHIFT;
 	srcu_idx = srcu_read_lock(&kvm->srcu);
 	memslot = gfn_to_memslot(kvm, gfn);
-	if (memslot) {
-		if (!kvm_is_radix(kvm)) {
-			rmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
-			lock_rmap(rmap);
-			*rmap |= KVMPPC_RMAP_CHANGED;
-			unlock_rmap(rmap);
-		} else if (memslot->dirty_bitmap) {
-			mark_page_dirty(kvm, gfn);
-		}
-	}
+	if (memslot && memslot->dirty_bitmap)
+		set_bit_le(gfn - memslot->base_gfn, memslot->dirty_bitmap);
 	srcu_read_unlock(&kvm->srcu, srcu_idx);
 }
 
@@ -1282,7 +1265,7 @@ static unsigned long resize_hpt_rehash_hpte(struct kvm_resize_hpt *resize,
 			rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
 
 			lock_rmap(rmapp);
-			kvmppc_unmap_hpte(kvm, idx, rmapp, gfn);
+			kvmppc_unmap_hpte(kvm, idx, memslot, rmapp, gfn);
 			unlock_rmap(rmapp);
 		}
 
diff --git a/arch/powerpc/kvm/book3s_64_mmu_radix.c b/arch/powerpc/kvm/book3s_64_mmu_radix.c
index c5d7435455f1..6336b13ed233 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_radix.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_radix.c
@@ -474,26 +474,6 @@ int kvmppc_book3s_radix_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 	return ret;
 }
 
-static void mark_pages_dirty(struct kvm *kvm, struct kvm_memory_slot *memslot,
-			     unsigned long gfn, unsigned int order)
-{
-	unsigned long i, limit;
-	unsigned long *dp;
-
-	if (!memslot->dirty_bitmap)
-		return;
-	limit = 1ul << order;
-	if (limit < BITS_PER_LONG) {
-		for (i = 0; i < limit; ++i)
-			mark_page_dirty(kvm, gfn + i);
-		return;
-	}
-	dp = memslot->dirty_bitmap + (gfn - memslot->base_gfn);
-	limit /= BITS_PER_LONG;
-	for (i = 0; i < limit; ++i)
-		*dp++ = ~0ul;
-}
-
 /* Called with kvm->lock held */
 int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
 		    unsigned long gfn)
@@ -508,12 +488,11 @@ int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
 		old = kvmppc_radix_update_pte(kvm, ptep, _PAGE_PRESENT, 0,
 					      gpa, shift);
 		kvmppc_radix_tlbie_page(kvm, gpa, shift);
-		if (old & _PAGE_DIRTY) {
-			if (!shift)
-				mark_page_dirty(kvm, gfn);
-			else
-				mark_pages_dirty(kvm, memslot,
-						 gfn, shift - PAGE_SHIFT);
+		if ((old & _PAGE_DIRTY) && memslot->dirty_bitmap) {
+			unsigned long npages = 1;
+			if (shift)
+				npages = 1ul << (shift - PAGE_SHIFT);
+			kvmppc_update_dirty_map(memslot, gfn, npages);
 		}
 	}
 	return 0;				
@@ -579,20 +558,8 @@ long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm,
 			struct kvm_memory_slot *memslot, unsigned long *map)
 {
 	unsigned long i, j;
-	unsigned long n, *p;
 	int npages;
 
-	/*
-	 * Radix accumulates dirty bits in the first half of the
-	 * memslot's dirty_bitmap area, for when pages are paged
-	 * out or modified by the host directly.  Pick up these
-	 * bits and add them to the map.
-	 */
-	n = kvm_dirty_bitmap_bytes(memslot) / sizeof(long);
-	p = memslot->dirty_bitmap;
-	for (i = 0; i < n; ++i)
-		map[i] |= xchg(&p[i], 0);
-
 	for (i = 0; i < memslot->npages; i = j) {
 		npages = kvm_radix_test_clear_dirty(kvm, memslot, i);
 
@@ -604,9 +571,10 @@ long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm,
 		 * real address, if npages > 1 we can skip to i + npages.
 		 */
 		j = i + 1;
-		if (npages)
-			for (j = i; npages; ++j, --npages)
-				__set_bit_le(j, map);
+		if (npages) {
+			set_dirty_bits(map, i, npages);
+			i = j + npages;
+		}
 	}
 	return 0;
 }
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index f6522075edbe..96b92d4adb87 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3376,7 +3376,7 @@ static int kvm_vm_ioctl_get_dirty_log_hv(struct kvm *kvm,
 	struct kvm_memory_slot *memslot;
 	int i, r;
 	unsigned long n;
-	unsigned long *buf;
+	unsigned long *buf, *p;
 	struct kvm_vcpu *vcpu;
 
 	mutex_lock(&kvm->slots_lock);
@@ -3392,8 +3392,8 @@ static int kvm_vm_ioctl_get_dirty_log_hv(struct kvm *kvm,
 		goto out;
 
 	/*
-	 * Use second half of bitmap area because radix accumulates
-	 * bits in the first half.
+	 * Use second half of bitmap area because both HPT and radix
+	 * accumulate bits in the first half.
 	 */
 	n = kvm_dirty_bitmap_bytes(memslot);
 	buf = memslot->dirty_bitmap + n / sizeof(long);
@@ -3406,6 +3406,16 @@ static int kvm_vm_ioctl_get_dirty_log_hv(struct kvm *kvm,
 	if (r)
 		goto out;
 
+	/*
+	 * We accumulate dirty bits in the first half of the
+	 * memslot's dirty_bitmap area, for when pages are paged
+	 * out or modified by the host directly.  Pick up these
+	 * bits and add them to the map.
+	 */
+	p = memslot->dirty_bitmap;
+	for (i = 0; i < n / sizeof(long); ++i)
+		buf[i] |= xchg(&p[i], 0);
+
 	/* Harvest dirty bits from VPA and DTL updates */
 	/* Note: we never modify the SLB shadow buffer areas */
 	kvm_for_each_vcpu(i, vcpu, kvm) {
@@ -3466,8 +3476,6 @@ static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
 				const struct kvm_memory_slot *new)
 {
 	unsigned long npages = mem->memory_size >> PAGE_SHIFT;
-	struct kvm_memslots *slots;
-	struct kvm_memory_slot *memslot;
 
 	/*
 	 * If we are making a new memslot, it might make
@@ -3477,18 +3485,6 @@ static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
 	 */
 	if (npages)
 		atomic64_inc(&kvm->arch.mmio_update);
-
-	if (npages && old->npages && !kvm_is_radix(kvm)) {
-		/*
-		 * If modifying a memslot, reset all the rmap dirty bits.
-		 * If this is a new memslot, we don't need to do anything
-		 * since the rmap array starts out as all zeroes,
-		 * i.e. no pages are dirty.
-		 */
-		slots = kvm_memslots(kvm);
-		memslot = id_to_memslot(slots, mem->slot);
-		kvmppc_hv_get_dirty_log_hpt(kvm, memslot, NULL);
-	}
 }
 
 /*
diff --git a/arch/powerpc/kvm/book3s_hv_rm_mmu.c b/arch/powerpc/kvm/book3s_hv_rm_mmu.c
index cf98f17c1aa6..26c11f678fbf 100644
--- a/arch/powerpc/kvm/book3s_hv_rm_mmu.c
+++ b/arch/powerpc/kvm/book3s_hv_rm_mmu.c
@@ -107,23 +107,39 @@ void kvmppc_add_revmap_chain(struct kvm *kvm, struct revmap_entry *rev,
 }
 EXPORT_SYMBOL_GPL(kvmppc_add_revmap_chain);
 
-/* Update the changed page order field of an rmap entry */
-void kvmppc_update_rmap_change(unsigned long *rmap, unsigned long psize)
+/* Update the dirty bitmap of a memslot */
+void kvmppc_update_dirty_map(struct kvm_memory_slot *memslot,
+			     unsigned long gfn, unsigned long psize)
 {
-	unsigned long order;
+	unsigned long npages;
 
-	if (!psize)
+	if (!psize || !memslot->dirty_bitmap)
 		return;
-	order = ilog2(psize);
-	order <<= KVMPPC_RMAP_CHG_SHIFT;
-	if (order > (*rmap & KVMPPC_RMAP_CHG_ORDER))
-		*rmap = (*rmap & ~KVMPPC_RMAP_CHG_ORDER) | order;
+	npages = (psize + PAGE_SIZE - 1) / PAGE_SIZE;
+	gfn -= memslot->base_gfn;
+	set_dirty_bits_atomic(memslot->dirty_bitmap, gfn, npages);
+}
+EXPORT_SYMBOL_GPL(kvmppc_update_dirty_map);
+
+static void kvmppc_set_dirty_from_hpte(struct kvm *kvm,
+				unsigned long hpte_v, unsigned long hpte_gr)
+{
+	struct kvm_memory_slot *memslot;
+	unsigned long gfn;
+	unsigned long psize;
+
+	psize = kvmppc_actual_pgsz(hpte_v, hpte_gr);
+	gfn = hpte_rpn(hpte_gr, psize);
+	memslot = __gfn_to_memslot(kvm_memslots_raw(kvm), gfn);
+	if (memslot && memslot->dirty_bitmap)
+		kvmppc_update_dirty_map(memslot, gfn, psize);
 }
-EXPORT_SYMBOL_GPL(kvmppc_update_rmap_change);
 
 /* Returns a pointer to the revmap entry for the page mapped by a HPTE */
 static unsigned long *revmap_for_hpte(struct kvm *kvm, unsigned long hpte_v,
-				      unsigned long hpte_gr)
+				      unsigned long hpte_gr,
+				      struct kvm_memory_slot **memslotp,
+				      unsigned long *gfnp)
 {
 	struct kvm_memory_slot *memslot;
 	unsigned long *rmap;
@@ -131,6 +147,10 @@ static unsigned long *revmap_for_hpte(struct kvm *kvm, unsigned long hpte_v,
 
 	gfn = hpte_rpn(hpte_gr, kvmppc_actual_pgsz(hpte_v, hpte_gr));
 	memslot = __gfn_to_memslot(kvm_memslots_raw(kvm), gfn);
+	if (memslotp)
+		*memslotp = memslot;
+	if (gfnp)
+		*gfnp = gfn;
 	if (!memslot)
 		return NULL;
 
@@ -147,10 +167,12 @@ static void remove_revmap_chain(struct kvm *kvm, long pte_index,
 	unsigned long ptel, head;
 	unsigned long *rmap;
 	unsigned long rcbits;
+	struct kvm_memory_slot *memslot;
+	unsigned long gfn;
 
 	rcbits = hpte_r & (HPTE_R_R | HPTE_R_C);
 	ptel = rev->guest_rpte |= rcbits;
-	rmap = revmap_for_hpte(kvm, hpte_v, ptel);
+	rmap = revmap_for_hpte(kvm, hpte_v, ptel, &memslot, &gfn);
 	if (!rmap)
 		return;
 	lock_rmap(rmap);
@@ -169,8 +191,8 @@ static void remove_revmap_chain(struct kvm *kvm, long pte_index,
 	}
 	*rmap |= rcbits << KVMPPC_RMAP_RC_SHIFT;
 	if (rcbits & HPTE_R_C)
-		kvmppc_update_rmap_change(rmap,
-					  kvmppc_actual_pgsz(hpte_v, hpte_r));
+		kvmppc_update_dirty_map(memslot, gfn,
+					kvmppc_actual_pgsz(hpte_v, hpte_r));
 	unlock_rmap(rmap);
 }
 
@@ -798,7 +820,7 @@ long kvmppc_h_clear_ref(struct kvm_vcpu *vcpu, unsigned long flags,
 		gr |= r & (HPTE_R_R | HPTE_R_C);
 		if (r & HPTE_R_R) {
 			kvmppc_clear_ref_hpte(kvm, hpte, pte_index);
-			rmap = revmap_for_hpte(kvm, v, gr);
+			rmap = revmap_for_hpte(kvm, v, gr, NULL, NULL);
 			if (rmap) {
 				lock_rmap(rmap);
 				*rmap |= KVMPPC_RMAP_REFERENCED;
@@ -820,7 +842,6 @@ long kvmppc_h_clear_mod(struct kvm_vcpu *vcpu, unsigned long flags,
 	__be64 *hpte;
 	unsigned long v, r, gr;
 	struct revmap_entry *rev;
-	unsigned long *rmap;
 	long ret = H_NOT_FOUND;
 
 	if (kvm_is_radix(kvm))
@@ -849,16 +870,9 @@ long kvmppc_h_clear_mod(struct kvm_vcpu *vcpu, unsigned long flags,
 		r = be64_to_cpu(hpte[1]);
 		gr |= r & (HPTE_R_R | HPTE_R_C);
 		if (r & HPTE_R_C) {
-			unsigned long psize = kvmppc_actual_pgsz(v, r);
 			hpte[1] = cpu_to_be64(r & ~HPTE_R_C);
 			eieio();
-			rmap = revmap_for_hpte(kvm, v, gr);
-			if (rmap) {
-				lock_rmap(rmap);
-				*rmap |= KVMPPC_RMAP_CHANGED;
-				kvmppc_update_rmap_change(rmap, psize);
-				unlock_rmap(rmap);
-			}
+			kvmppc_set_dirty_from_hpte(kvm, v, gr);
 		}
 	}
 	vcpu->arch.gpr[4] = gr;