KVM: PPC: Book3S HV: Report VPA and DTL modifications in dirty map

At present, the KVM_GET_DIRTY_LOG ioctl doesn't report modifications
done by the host to the virtual processor areas (VPAs) and dispatch
trace logs (DTLs) registered by the guest.  This is because those
modifications are done either in real mode or in the host kernel
context, and in neither case does the access go through the guest's
HPT, and thus no change (C) bit gets set in the guest's HPT.

However, the changes done by the host do need to be tracked so that
the modified pages get transferred when doing live migration.  In
order to track these modifications, this adds a dirty flag to the
struct representing the VPA/DTL areas, and arranges to set the flag
when the VPA/DTL gets modified by the host.  Then, when we are
collecting the dirty log, we also check the dirty flags for the
VPA and DTL for each vcpu and set the relevant bit in the dirty log
if necessary.  Doing this also means we now need to keep track of
the guest physical address of the VPA/DTL areas.

So as not to lose track of modifications to a VPA/DTL area when it gets
unregistered, or when a new area gets registered in its place, we need
to transfer the dirty state to the rmap chain.  This adds code to
kvmppc_unpin_guest_page() to do that if the area was dirty.  To simplify
that code, we now require that all VPA, DTL and SLB shadow buffer areas
fit within a single host page.  Guests already comply with this
requirement because pHyp requires that these areas not cross a 4k
boundary.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>

6 files changed, 80 insertions, 21 deletions

diff --git a/arch/powerpc/include/asm/kvm_book3s.h b/arch/powerpc/include/asm/kvm_book3s.h
index bc81842ea25..c55f7e6affa 100644
--- a/arch/powerpc/include/asm/kvm_book3s.h
+++ b/arch/powerpc/include/asm/kvm_book3s.h
@@ -156,7 +156,8 @@ void kvmppc_clear_ref_hpte(struct kvm *kvm, unsigned long *hptep,
 			unsigned long pte_index);
 extern void *kvmppc_pin_guest_page(struct kvm *kvm, unsigned long addr,
 			unsigned long *nb_ret);
-extern void kvmppc_unpin_guest_page(struct kvm *kvm, void *addr);
+extern void kvmppc_unpin_guest_page(struct kvm *kvm, void *addr,
+			unsigned long gpa, bool dirty);
 extern long kvmppc_virtmode_h_enter(struct kvm_vcpu *vcpu, unsigned long flags,
 			long pte_index, unsigned long pteh, unsigned long ptel);
 extern long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index 8a48e686a75..1443768a658 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -301,11 +301,13 @@ struct kvmppc_vcore {
  * that a guest can register.
  */
 struct kvmppc_vpa {
+	unsigned long gpa;	/* Current guest phys addr */
 	void *pinned_addr;	/* Address in kernel linear mapping */
 	void *pinned_end;	/* End of region */
 	unsigned long next_gpa;	/* Guest phys addr for update */
 	unsigned long len;	/* Number of bytes required */
 	u8 update_pending;	/* 1 => update pinned_addr from next_gpa */
+	bool dirty;		/* true => area has been modified by kernel */
 };
 
 struct kvmppc_pte {
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index d87c90886c7..dbfd5498f44 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -477,6 +477,7 @@ int main(void)
 	DEFINE(VCPU_DSISR, offsetof(struct kvm_vcpu, arch.shregs.dsisr));
 	DEFINE(VCPU_DAR, offsetof(struct kvm_vcpu, arch.shregs.dar));
 	DEFINE(VCPU_VPA, offsetof(struct kvm_vcpu, arch.vpa.pinned_addr));
+	DEFINE(VCPU_VPA_DIRTY, offsetof(struct kvm_vcpu, arch.vpa.dirty));
 #endif
 #ifdef CONFIG_PPC_BOOK3S
 	DEFINE(VCPU_VCPUID, offsetof(struct kvm_vcpu, vcpu_id));
diff --git a/arch/powerpc/kvm/book3s_64_mmu_hv.c b/arch/powerpc/kvm/book3s_64_mmu_hv.c
index d641a6634b0..69efe0d6ced 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_hv.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_hv.c
@@ -1099,11 +1099,30 @@ static int kvm_test_clear_dirty(struct kvm *kvm, unsigned long *rmapp)
 	return ret;
 }
 
+static void harvest_vpa_dirty(struct kvmppc_vpa *vpa,
+			      struct kvm_memory_slot *memslot,
+			      unsigned long *map)
+{
+	unsigned long gfn;
+
+	if (!vpa->dirty || !vpa->pinned_addr)
+		return;
+	gfn = vpa->gpa >> PAGE_SHIFT;
+	if (gfn < memslot->base_gfn ||
+	    gfn >= memslot->base_gfn + memslot->npages)
+		return;
+
+	vpa->dirty = false;
+	if (map)
+		__set_bit_le(gfn - memslot->base_gfn, map);
+}
+
 long kvmppc_hv_get_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot,
 			     unsigned long *map)
 {
 	unsigned long i;
 	unsigned long *rmapp;
+	struct kvm_vcpu *vcpu;
 
 	preempt_disable();
 	rmapp = memslot->arch.rmap;
@@ -1112,6 +1131,15 @@ long kvmppc_hv_get_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot,
 			__set_bit_le(i, map);
 		++rmapp;
 	}
+
+	/* Harvest dirty bits from VPA and DTL updates */
+	/* Note: we never modify the SLB shadow buffer areas */
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		spin_lock(&vcpu->arch.vpa_update_lock);
+		harvest_vpa_dirty(&vcpu->arch.vpa, memslot, map);
+		harvest_vpa_dirty(&vcpu->arch.dtl, memslot, map);
+		spin_unlock(&vcpu->arch.vpa_update_lock);
+	}
 	preempt_enable();
 	return 0;
 }
@@ -1123,7 +1151,7 @@ void *kvmppc_pin_guest_page(struct kvm *kvm, unsigned long gpa,
 	unsigned long gfn = gpa >> PAGE_SHIFT;
 	struct page *page, *pages[1];
 	int npages;
-	unsigned long hva, psize, offset;
+	unsigned long hva, offset;
 	unsigned long pa;
 	unsigned long *physp;
 	int srcu_idx;
@@ -1155,14 +1183,9 @@ void *kvmppc_pin_guest_page(struct kvm *kvm, unsigned long gpa,
 	}
 	srcu_read_unlock(&kvm->srcu, srcu_idx);
 
-	psize = PAGE_SIZE;
-	if (PageHuge(page)) {
-		page = compound_head(page);
-		psize <<= compound_order(page);
-	}
-	offset = gpa & (psize - 1);
+	offset = gpa & (PAGE_SIZE - 1);
 	if (nb_ret)
-		*nb_ret = psize - offset;
+		*nb_ret = PAGE_SIZE - offset;
 	return page_address(page) + offset;
 
  err:
@@ -1170,11 +1193,31 @@ void *kvmppc_pin_guest_page(struct kvm *kvm, unsigned long gpa,
 	return NULL;
 }
 
-void kvmppc_unpin_guest_page(struct kvm *kvm, void *va)
+void kvmppc_unpin_guest_page(struct kvm *kvm, void *va, unsigned long gpa,
+			     bool dirty)
 {
 	struct page *page = virt_to_page(va);
+	struct kvm_memory_slot *memslot;
+	unsigned long gfn;
+	unsigned long *rmap;
+	int srcu_idx;
 
 	put_page(page);
+
+	if (!dirty || !kvm->arch.using_mmu_notifiers)
+		return;
+
+	/* We need to mark this page dirty in the rmap chain */
+	gfn = gpa >> PAGE_SHIFT;
+	srcu_idx = srcu_read_lock(&kvm->srcu);
+	memslot = gfn_to_memslot(kvm, gfn);
+	if (memslot) {
+		rmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
+		lock_rmap(rmap);
+		*rmap |= KVMPPC_RMAP_CHANGED;
+		unlock_rmap(rmap);
+	}
+	srcu_read_unlock(&kvm->srcu, srcu_idx);
 }
 
 /*
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 1e521baf9a7..5af0f297983 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -259,7 +259,7 @@ static unsigned long do_h_register_vpa(struct kvm_vcpu *vcpu,
 			len = ((struct reg_vpa *)va)->length.hword;
 		else
 			len = ((struct reg_vpa *)va)->length.word;
-		kvmppc_unpin_guest_page(kvm, va);
+		kvmppc_unpin_guest_page(kvm, va, vpa, false);
 
 		/* Check length */
 		if (len > nb || len < sizeof(struct reg_vpa))
@@ -359,13 +359,13 @@ static void kvmppc_update_vpa(struct kvm_vcpu *vcpu, struct kvmppc_vpa *vpap)
 		va = NULL;
 		nb = 0;
 		if (gpa)
-			va = kvmppc_pin_guest_page(kvm, vpap->next_gpa, &nb);
+			va = kvmppc_pin_guest_page(kvm, gpa, &nb);
 		spin_lock(&vcpu->arch.vpa_update_lock);
 		if (gpa == vpap->next_gpa)
 			break;
 		/* sigh... unpin that one and try again */
 		if (va)
-			kvmppc_unpin_guest_page(kvm, va);
+			kvmppc_unpin_guest_page(kvm, va, gpa, false);
 	}
 
 	vpap->update_pending = 0;
@@ -375,12 +375,15 @@ static void kvmppc_update_vpa(struct kvm_vcpu *vcpu, struct kvmppc_vpa *vpap)
 		 * has changed the mappings underlying guest memory,
 		 * so unregister the region.
 		 */
-		kvmppc_unpin_guest_page(kvm, va);
+		kvmppc_unpin_guest_page(kvm, va, gpa, false);
 		va = NULL;
 	}
 	if (vpap->pinned_addr)
-		kvmppc_unpin_guest_page(kvm, vpap->pinned_addr);
+		kvmppc_unpin_guest_page(kvm, vpap->pinned_addr, vpap->gpa,
+					vpap->dirty);
+	vpap->gpa = gpa;
 	vpap->pinned_addr = va;
+	vpap->dirty = false;
 	if (va)
 		vpap->pinned_end = va + vpap->len;
 }
@@ -472,6 +475,7 @@ static void kvmppc_create_dtl_entry(struct kvm_vcpu *vcpu,
 	/* order writing *dt vs. writing vpa->dtl_idx */
 	smp_wmb();
 	vpa->dtl_idx = ++vcpu->arch.dtl_index;
+	vcpu->arch.dtl.dirty = true;
 }
 
 int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
@@ -913,15 +917,19 @@ out:
 	return ERR_PTR(err);
 }
 
+static void unpin_vpa(struct kvm *kvm, struct kvmppc_vpa *vpa)
+{
+	if (vpa->pinned_addr)
+		kvmppc_unpin_guest_page(kvm, vpa->pinned_addr, vpa->gpa,
+					vpa->dirty);
+}
+
 void kvmppc_core_vcpu_free(struct kvm_vcpu *vcpu)
 {
 	spin_lock(&vcpu->arch.vpa_update_lock);
-	if (vcpu->arch.dtl.pinned_addr)
-		kvmppc_unpin_guest_page(vcpu->kvm, vcpu->arch.dtl.pinned_addr);
-	if (vcpu->arch.slb_shadow.pinned_addr)
-		kvmppc_unpin_guest_page(vcpu->kvm, vcpu->arch.slb_shadow.pinned_addr);
-	if (vcpu->arch.vpa.pinned_addr)
-		kvmppc_unpin_guest_page(vcpu->kvm, vcpu->arch.vpa.pinned_addr);
+	unpin_vpa(vcpu->kvm, &vcpu->arch.dtl);
+	unpin_vpa(vcpu->kvm, &vcpu->arch.slb_shadow);
+	unpin_vpa(vcpu->kvm, &vcpu->arch.vpa);
 	spin_unlock(&vcpu->arch.vpa_update_lock);
 	kvm_vcpu_uninit(vcpu);
 	kmem_cache_free(kvm_vcpu_cache, vcpu);
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index e33d11f1b97..0f23bb85171 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -260,6 +260,8 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206)
 	lwz	r5, LPPACA_YIELDCOUNT(r3)
 	addi	r5, r5, 1
 	stw	r5, LPPACA_YIELDCOUNT(r3)
+	li	r6, 1
+	stb	r6, VCPU_VPA_DIRTY(r4)
 25:
 	/* Load up DAR and DSISR */
 	ld	r5, VCPU_DAR(r4)
@@ -1018,6 +1020,8 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_206)
 	lwz	r3, LPPACA_YIELDCOUNT(r8)
 	addi	r3, r3, 1
 	stw	r3, LPPACA_YIELDCOUNT(r8)
+	li	r3, 1
+	stb	r3, VCPU_VPA_DIRTY(r9)
 25:
 	/* Save PMU registers if requested */
 	/* r8 and cr0.eq are live here */