[PATCH v9 0/8] KVM: PPC: Driver to manage pages of secure guest

@@ -232,11 +232,31 @@ struct revmap_entry {
 };
 
 /*
- * We use the top bit of each memslot->arch.rmap entry as a lock bit,
- * and bit 32 as a present flag.  The bottom 32 bits are the
- * index in the guest HPT of a HPTE that points to the page.
+ * The rmap array of size number of guest pages is allocated for each memslot.
+ * This array is used to store usage specific information about the guest page.
+ * Below are the encodings of the various possible usage types.
  */
-#define KVMPPC_RMAP_LOCK_BIT	63
+/* Free bits which can be used to define a new usage */
+#define KVMPPC_RMAP_TYPE_MASK	0xff00000000000000
+#define KVMPPC_RMAP_NESTED	0xc000000000000000	/* Nested rmap array */
+#define KVMPPC_RMAP_HPT		0x0100000000000000	/* HPT guest */
+#define KVMPPC_RMAP_UVMEM_PFN	0x0200000000000000	/* Secure GPA */
+
+static inline unsigned long kvmppc_rmap_type(unsigned long *rmap)
+{
+	return (*rmap & KVMPPC_RMAP_TYPE_MASK);
+}
+
+/*
+ * rmap usage definition for a hash page table (hpt) guest:
+ * 0x0000080000000000	Lock bit
+ * 0x0000018000000000	RC bits
+ * 0x0000000100000000	Present bit
+ * 0x00000000ffffffff	HPT index bits
+ * The bottom 32 bits are the index in the guest HPT of a HPTE that points to
+ * the page.
+ */
+#define KVMPPC_RMAP_LOCK_BIT	43
 #define KVMPPC_RMAP_RC_SHIFT	32
 #define KVMPPC_RMAP_REFERENCED	(HPTE_R_R << KVMPPC_RMAP_RC_SHIFT)
 #define KVMPPC_RMAP_PRESENT	0x100000000ul

@@ -99,7 +99,7 @@ void kvmppc_add_revmap_chain(struct kvm *kvm, struct revmap_entry *rev,
 	} else {
 		rev->forw = rev->back = pte_index;
 		*rmap = (*rmap & ~KVMPPC_RMAP_INDEX) |
-			pte_index | KVMPPC_RMAP_PRESENT;
+			pte_index | KVMPPC_RMAP_PRESENT | KVMPPC_RMAP_HPT;
 	}
 	unlock_rmap(rmap);
 }

@@ -342,6 +342,10 @@
 #define H_TLB_INVALIDATE	0xF808
 #define H_COPY_TOFROM_GUEST	0xF80C
 
+/* Platform-specific hcalls used by the Ultravisor */
+#define H_SVM_PAGE_IN		0xEF00
+#define H_SVM_PAGE_OUT		0xEF04
+
 /* Values for 2nd argument to H_SET_MODE */
 #define H_SET_MODE_RESOURCE_SET_CIABR		1
 #define H_SET_MODE_RESOURCE_SET_DAWR		2

@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __POWERPC_KVM_PPC_HMM_H__
+#define __POWERPC_KVM_PPC_HMM_H__
+
+#ifdef CONFIG_PPC_UV
+unsigned long kvmppc_h_svm_page_in(struct kvm *kvm,
+				   unsigned long gra,
+				   unsigned long flags,
+				   unsigned long page_shift);
+unsigned long kvmppc_h_svm_page_out(struct kvm *kvm,
+				    unsigned long gra,
+				    unsigned long flags,
+				    unsigned long page_shift);
+#else
+static inline unsigned long
+kvmppc_h_svm_page_in(struct kvm *kvm, unsigned long gra,
+		     unsigned long flags, unsigned long page_shift)
+{
+	return H_UNSUPPORTED;
+}
+
+static inline unsigned long
+kvmppc_h_svm_page_out(struct kvm *kvm, unsigned long gra,
+		      unsigned long flags, unsigned long page_shift)
+{
+	return H_UNSUPPORTED;
+}
+#endif /* CONFIG_PPC_UV */
+#endif /* __POWERPC_KVM_PPC_HMM_H__ */

@@ -336,6 +336,7 @@ struct kvm_arch {
 #endif
 	struct kvmppc_ops *kvm_ops;
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+	struct mutex uvmem_lock;
 	struct mutex mmu_setup_lock;	/* nests inside vcpu mutexes */
 	u64 l1_ptcr;
 	int max_nested_lpid;

@@ -869,4 +870,16 @@ static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu) {}
 static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu *vcpu) {}
 
+#ifdef CONFIG_PPC_UV
+int kvmppc_uvmem_init(void);
+void kvmppc_uvmem_free(void);
+#else
+static inline int kvmppc_uvmem_init(void)
+{
+	return 0;
+}
+
+static inline void kvmppc_uvmem_free(void) {}
+#endif /* CONFIG_PPC_UV */
+
 #endif /* __POWERPC_KVM_HOST_H__ */

@@ -25,5 +25,7 @@
 /* opcodes */
 #define UV_WRITE_PATE			0xF104
 #define UV_RETURN			0xF11C
+#define UV_PAGE_IN			0xF128
+#define UV_PAGE_OUT			0xF12C
 
 #endif /* _ASM_POWERPC_ULTRAVISOR_API_H */

@@ -31,4 +31,18 @@ static inline int uv_register_pate(u64 lpid, u64 dw0, u64 dw1)
 	return ucall_norets(UV_WRITE_PATE, lpid, dw0, dw1);
 }
 
+static inline int uv_page_in(u64 lpid, u64 src_ra, u64 dst_gpa, u64 flags,
+			     u64 page_shift)
+{
+	return ucall_norets(UV_PAGE_IN, lpid, src_ra, dst_gpa, flags,
+			    page_shift);
+}
+
+static inline int uv_page_out(u64 lpid, u64 dst_ra, u64 src_gpa, u64 flags,
+			      u64 page_shift)
+{
+	return ucall_norets(UV_PAGE_OUT, lpid, dst_ra, src_gpa, flags,
+			    page_shift);
+}
+
 #endif	/* _ASM_POWERPC_ULTRAVISOR_H */

@@ -71,6 +71,9 @@ kvm-hv-y += \
 	book3s_64_mmu_radix.o \
 	book3s_hv_nested.o
 
+kvm-hv-$(CONFIG_PPC_UV) += \
+	book3s_hv_uvmem.o
+
 kvm-hv-$(CONFIG_PPC_TRANSACTIONAL_MEM) += \
 	book3s_hv_tm.o
 

@@ -72,6 +72,8 @@
 #include <asm/xics.h>
 #include <asm/xive.h>
 #include <asm/hw_breakpoint.h>
+#include <asm/kvm_host.h>
+#include <asm/kvm_book3s_uvmem.h>
 
 #include "book3s.h"
 

@@ -1075,6 +1077,18 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
 					 kvmppc_get_gpr(vcpu, 5),
 					 kvmppc_get_gpr(vcpu, 6));
 		break;
+	case H_SVM_PAGE_IN:
+		ret = kvmppc_h_svm_page_in(vcpu->kvm,
+					   kvmppc_get_gpr(vcpu, 4),
+					   kvmppc_get_gpr(vcpu, 5),
+					   kvmppc_get_gpr(vcpu, 6));
+		break;
+	case H_SVM_PAGE_OUT:
+		ret = kvmppc_h_svm_page_out(vcpu->kvm,
+					    kvmppc_get_gpr(vcpu, 4),
+					    kvmppc_get_gpr(vcpu, 5),
+					    kvmppc_get_gpr(vcpu, 6));
+		break;
 	default:
 		return RESUME_HOST;
 	}

@@ -4769,6 +4783,7 @@ static int kvmppc_core_init_vm_hv(struct kvm *kvm)
 	char buf[32];
 	int ret;
 
+	mutex_init(&kvm->arch.uvmem_lock);
 	mutex_init(&kvm->arch.mmu_setup_lock);
 
 	/* Allocate the guest's logical partition ID */

@@ -5523,11 +5538,16 @@ static int kvmppc_book3s_init_hv(void)
 			no_mixing_hpt_and_radix = true;
 	}
 
+	r = kvmppc_uvmem_init();
+	if (r < 0)
+		pr_err("KVM-HV: kvmppc_uvmem_init failed %d\n", r);
+
 	return r;
 }
 
 static void kvmppc_book3s_exit_hv(void)
 {
+	kvmppc_uvmem_free();
 	kvmppc_free_host_rm_ops();
 	if (kvmppc_radix_possible())
 		kvmppc_radix_exit();

@@ -0,0 +1,481 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Secure pages management: Migration of pages between normal and secure
+ * memory of KVM guests.
+ *
+ * Copyright 2018 Bharata B Rao, IBM Corp. <bharata@linux.ibm.com>
+ */
+
+/*
+ * A pseries guest can be run as secure guest on Ultravisor-enabled
+ * POWER platforms. On such platforms, this driver will be used to manage
+ * the movement of guest pages between the normal memory managed by
+ * hypervisor (HV) and secure memory managed by Ultravisor (UV).
+ *
+ * The page-in or page-out requests from UV will come to HV as hcalls and
+ * HV will call back into UV via ultracalls to satisfy these page requests.
+ *
+ * Private ZONE_DEVICE memory equal to the amount of secure memory
+ * available in the platform for running secure guests is hotplugged.
+ * Whenever a page belonging to the guest becomes secure, a page from this
+ * private device memory is used to represent and track that secure page
+ * on the HV side.
+ *
+ * For each page that gets moved into secure memory, a device PFN is used
+ * on the HV side and migration PTE corresponding to that PFN would be
+ * populated in the QEMU page tables. Device PFNs are stored in the rmap
+ * array. Whenever a guest page becomes secure, device PFN allocated for
+ * the same will be populated in the corresponding slot in the rmap
+ * array. The overloading of rmap array's usage which otherwise is
+ * used primarily by HPT guests means that this feature (secure
+ * guest on PEF platforms) is available only for Radix MMU guests.
+ * Also since the same rmap array is used differently by nested
+ * guests, a secure guest can't have further nested guests.
+ */
+
+/*
+ * Notes on locking
+ *
+ * kvm->arch.uvmem_lock is a per-guest lock that prevents concurrent
+ * page-in and page-out requests for the same GPA. Concurrent accesses
+ * can either come via UV (guest vCPUs requesting for same page)
+ * or when HV and guest simultaneously access the same page.
+ * This mutex serializes the migration of page from HV(normal) to
+ * UV(secure) and vice versa. So the serialization points are around
+ * migrate_vma routines and page-in/out routines.
+ *
+ * It would have been desirable to use lock_rmap() which essentially
+ * is spinlock at guest PFN granularity. However migrate_vma_pages() and
+ * helpers can sleep and hence using spinlock across these routines is not
+ * possible. This first attempt uses a per-guest mutex though per-guest-PFN
+ * mutex would have been ideal and preferrable.
+ *
+ * Per-guest mutex comes with a cost though. Mainly it serializes the
+ * fault path as page-out can occur when HV faults on accessing secure
+ * guest pages. Currently UV issues page-in requests for all the guest
+ * PFNs one at a time during early boot (UV_ESM uvcall), so this is
+ * not a cause for concern. Also currently the number of page-outs caused
+ * by HV touching secure pages is very very low. If an when UV supports
+ * overcommitting, then we might see concurrent guest driven page-outs.
+ *
+ * Locking order
+ *
+ * 1. srcu_read_lock(&kvm->srcu) - Protects KVM memslots
+ * 2. down_read(&kvm->mm->mmap_sem) - find_vma, migrate_vma_pages and helpers
+ * 3. mutex_lock(&kvm->arch.uvmem_lock) - protects read/writes to rmap[] slots
+ *					thus acting as sync-points for
+ *					page-in/out
+ */
+
+#include <linux/pagemap.h>
+#include <linux/migrate.h>
+#include <linux/kvm_host.h>
+#include <asm/ultravisor.h>
+
+static struct dev_pagemap kvmppc_uvmem_pgmap;
+static unsigned long *kvmppc_uvmem_pfn_bitmap;
+static DEFINE_SPINLOCK(kvmppc_uvmem_pfn_lock);
+
+struct kvmppc_uvmem_page_pvt {
+	unsigned long *rmap;
+	struct kvm *kvm;
+	unsigned long gpa;
+};
+
+/*
+ * Get a free device PFN from the pool
+ *
+ * Called when a normal page is moved to secure memory (UV_PAGE_IN). Device
+ * PFN will be used to keep track of the secure page on HV side.
+ *
+ * @rmap here is the slot in the rmap array that corresponds to @gpa.
+ * An rmap entry of type KVMPPC_RMAP_UVMEM_PFN indicates that the
+ * corresponding guest page has become secure, and is not mapped on
+ * the HV side.
+ *
+ * Called with kvm->arch.uvmem_lock held
+ */
+static struct page *kvmppc_uvmem_get_page(unsigned long *rmap,
+					  unsigned long gpa, struct kvm *kvm)
+{
+	struct page *dpage = NULL;
+	unsigned long bit, uvmem_pfn;
+	struct kvmppc_uvmem_page_pvt *pvt;
+	unsigned long pfn_last, pfn_first;
+
+	pfn_first = kvmppc_uvmem_pgmap.res.start >> PAGE_SHIFT;
+	pfn_last = pfn_first +
+		   (resource_size(&kvmppc_uvmem_pgmap.res) >> PAGE_SHIFT);
+
+	spin_lock(&kvmppc_uvmem_pfn_lock);
+	bit = find_first_zero_bit(kvmppc_uvmem_pfn_bitmap,
+				  pfn_last - pfn_first);
+	if (bit >= (pfn_last - pfn_first))
+		goto out;
+	bitmap_set(kvmppc_uvmem_pfn_bitmap, bit, 1);
+	spin_unlock(&kvmppc_uvmem_pfn_lock);
+
+	pvt = kzalloc(sizeof(*pvt), GFP_KERNEL);
+	if (!pvt)
+		goto out_clear;
+
+	uvmem_pfn = bit + pfn_first;
+	dpage = pfn_to_page(uvmem_pfn);
+	*rmap = uvmem_pfn | KVMPPC_RMAP_UVMEM_PFN;
+	pvt->rmap = rmap;
+	pvt->gpa = gpa;
+	pvt->kvm = kvm;
+	dpage->zone_device_data = pvt;
+
+	get_page(dpage);
+	lock_page(dpage);
+	return dpage;
+out_clear:
+	spin_lock(&kvmppc_uvmem_pfn_lock);
+	bitmap_clear(kvmppc_uvmem_pfn_bitmap, bit, 1);
+out:
+	spin_unlock(&kvmppc_uvmem_pfn_lock);
+	return NULL;
+}
+
+/*
+ * Alloc a PFN from private device memory pool and copy page from normal
+ * memory to secure memory using UV_PAGE_IN uvcall.
+ */
+static int
+kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start,
+		   unsigned long end, unsigned long *rmap,
+		   unsigned long gpa, struct kvm *kvm,
+		   unsigned long page_shift)
+{
+	unsigned long src_pfn, dst_pfn = 0;
+	struct migrate_vma mig;
+	struct page *spage;
+	unsigned long pfn;
+	struct page *dpage;
+	int ret = 0;
+
+	memset(&mig, 0, sizeof(mig));
+	mig.vma = vma;
+	mig.start = start;
+	mig.end = end;
+	mig.src = &src_pfn;
+	mig.dst = &dst_pfn;
+
+	ret = migrate_vma_setup(&mig);
+	if (ret)
+		return ret;
+
+	if (!(*mig.src & MIGRATE_PFN_MIGRATE)) {
+		ret = -1;
+		goto out_finalize;
+	}
+
+	dpage = kvmppc_uvmem_get_page(rmap, gpa, kvm);
+	if (!dpage) {
+		ret = -1;
+		goto out_finalize;
+	}
+
+	pfn = *mig.src >> MIGRATE_PFN_SHIFT;
+	spage = migrate_pfn_to_page(*mig.src);
+	if (spage)
+		uv_page_in(kvm->arch.lpid, pfn << page_shift, gpa, 0,
+			   page_shift);
+
+	*mig.dst = migrate_pfn(page_to_pfn(dpage)) | MIGRATE_PFN_LOCKED;
+	migrate_vma_pages(&mig);
+out_finalize:
+	migrate_vma_finalize(&mig);
+	return ret;
+}
+
+/*
+ * H_SVM_PAGE_IN: Move page from normal memory to secure memory.
+ */
+unsigned long
+kvmppc_h_svm_page_in(struct kvm *kvm, unsigned long gpa,
+		     unsigned long flags, unsigned long page_shift)
+{
+	unsigned long start, end;
+	struct vm_area_struct *vma;
+	int srcu_idx;
+	unsigned long gfn = gpa >> page_shift;
+	struct kvm_memory_slot *slot;
+	unsigned long *rmap;
+	int ret;
+
+	if (page_shift != PAGE_SHIFT)
+		return H_P3;
+
+	if (flags)
+		return H_P2;
+
+	ret = H_PARAMETER;
+	srcu_idx = srcu_read_lock(&kvm->srcu);
+	down_read(&kvm->mm->mmap_sem);
+	slot = gfn_to_memslot(kvm, gfn);
+	if (!slot)
+		goto out;
+
+	start = gfn_to_hva(kvm, gfn);
+	if (kvm_is_error_hva(start))
+		goto out;
+
+	rmap = &slot->arch.rmap[gfn - slot->base_gfn];
+	mutex_lock(&kvm->arch.uvmem_lock);
+	/* Fail the page-in request of an already paged-in page */
+	if (kvmppc_rmap_type(rmap) == KVMPPC_RMAP_UVMEM_PFN)
+		goto out_unlock;
+
+	end = start + (1UL << page_shift);
+	vma = find_vma_intersection(kvm->mm, start, end);
+	if (!vma || vma->vm_start > start || vma->vm_end < end)
+		goto out_unlock;
+
+	if (!kvmppc_svm_page_in(vma, start, end, rmap, gpa, kvm, page_shift))
+		ret = H_SUCCESS;
+out_unlock:
+	mutex_unlock(&kvm->arch.uvmem_lock);
+out:
+	up_read(&kvm->mm->mmap_sem);
+	srcu_read_unlock(&kvm->srcu, srcu_idx);
+	return ret;
+}
+
+/*
+ * Provision a new page on HV side and copy over the contents
+ * from secure memory using UV_PAGE_OUT uvcall.
+ */
+static int
+kvmppc_svm_page_out(struct vm_area_struct *vma, unsigned long start,
+		    unsigned long end, unsigned long page_shift,
+		    struct kvm *kvm, unsigned long *rmap)
+{
+	unsigned long src_pfn, dst_pfn = 0;
+	struct migrate_vma mig;
+	struct page *dpage, *spage;
+	struct kvmppc_uvmem_page_pvt *pvt;
+	unsigned long pfn;
+	int ret = U_SUCCESS;
+
+	memset(&mig, 0, sizeof(mig));
+	mig.vma = vma;
+	mig.start = start;
+	mig.end = end;
+	mig.src = &src_pfn;
+	mig.dst = &dst_pfn;
+
+	mutex_lock(&kvm->arch.uvmem_lock);
+	/* The requested page is already paged-out, nothing to do */
+	if (kvmppc_rmap_type(rmap) != KVMPPC_RMAP_UVMEM_PFN)
+		goto out;
+
+	ret = migrate_vma_setup(&mig);
+	if (ret)
+		return ret;
+
+	spage = migrate_pfn_to_page(*mig.src);
+	if (!spage || !(*mig.src & MIGRATE_PFN_MIGRATE))
+		goto out_finalize;
+
+	if (!is_zone_device_page(spage))
+		goto out_finalize;
+
+	dpage = alloc_page_vma(GFP_HIGHUSER, vma, start);
+	if (!dpage) {
+		ret = -1;
+		goto out_finalize;
+	}
+
+	lock_page(dpage);
+	pvt = spage->zone_device_data;
+	pfn = page_to_pfn(dpage);
+
+	ret = uv_page_out(pvt->kvm->arch.lpid, pfn << page_shift,
+			  pvt->gpa, 0, page_shift);
+	if (ret == U_SUCCESS)
+		*mig.dst = migrate_pfn(pfn) | MIGRATE_PFN_LOCKED;
+	else {
+		unlock_page(dpage);
+		__free_page(dpage);
+		goto out_finalize;
+	}
+
+	migrate_vma_pages(&mig);
+out_finalize:
+	migrate_vma_finalize(&mig);
+out:
+	mutex_unlock(&kvm->arch.uvmem_lock);
+	return ret;
+}
+
+/*
+ * Fault handler callback that gets called when HV touches any page that
+ * has been moved to secure memory, we ask UV to give back the page by
+ * issuing UV_PAGE_OUT uvcall.
+ *
+ * This eventually results in dropping of device PFN and the newly
+ * provisioned page/PFN gets populated in QEMU page tables.
+ */
+static vm_fault_t kvmppc_uvmem_migrate_to_ram(struct vm_fault *vmf)
+{
+	struct kvmppc_uvmem_page_pvt *pvt = vmf->page->zone_device_data;
+
+	if (kvmppc_svm_page_out(vmf->vma, vmf->address,
+				vmf->address + PAGE_SIZE, PAGE_SHIFT,
+				pvt->kvm, pvt->rmap))
+		return VM_FAULT_SIGBUS;
+	else
+		return 0;
+}
+
+/*
+ * Release the device PFN back to the pool
+ *
+ * Gets called when secure page becomes a normal page during H_SVM_PAGE_OUT.
+ * Gets called with kvm->arch.uvmem_lock held.
+ */
+static void kvmppc_uvmem_page_free(struct page *page)
+{
+	unsigned long pfn = page_to_pfn(page) -
+			(kvmppc_uvmem_pgmap.res.start >> PAGE_SHIFT);
+	struct kvmppc_uvmem_page_pvt *pvt;
+
+	spin_lock(&kvmppc_uvmem_pfn_lock);
+	bitmap_clear(kvmppc_uvmem_pfn_bitmap, pfn, 1);
+	spin_unlock(&kvmppc_uvmem_pfn_lock);
+
+	pvt = page->zone_device_data;
+	page->zone_device_data = NULL;
+	*pvt->rmap = 0;
+	kfree(pvt);
+}
+
+static const struct dev_pagemap_ops kvmppc_uvmem_ops = {
+	.page_free = kvmppc_uvmem_page_free,
+	.migrate_to_ram	= kvmppc_uvmem_migrate_to_ram,
+};
+
+/*
+ * H_SVM_PAGE_OUT: Move page from secure memory to normal memory.
+ */
+unsigned long
+kvmppc_h_svm_page_out(struct kvm *kvm, unsigned long gpa,
+		      unsigned long flags, unsigned long page_shift)
+{
+	unsigned long gfn = gpa >> page_shift;
+	struct kvm_memory_slot *slot;
+	unsigned long *rmap;
+	unsigned long start, end;
+	struct vm_area_struct *vma;
+	int srcu_idx;
+	int ret;
+
+	if (page_shift != PAGE_SHIFT)
+		return H_P3;
+
+	if (flags)
+		return H_P2;
+
+	ret = H_PARAMETER;
+	srcu_idx = srcu_read_lock(&kvm->srcu);
+	down_read(&kvm->mm->mmap_sem);
+	start = gfn_to_hva(kvm, gfn);
+	if (kvm_is_error_hva(start))
+		goto out;
+
+	end = start + (1UL << page_shift);
+	vma = find_vma_intersection(kvm->mm, start, end);
+	if (!vma || vma->vm_start > start || vma->vm_end < end)
+		goto out;
+
+	slot = gfn_to_memslot(kvm, gfn);
+	rmap = &slot->arch.rmap[gfn - slot->base_gfn];
+	if (!kvmppc_svm_page_out(vma, start, end, page_shift, kvm, rmap))
+		ret = H_SUCCESS;
+out:
+	up_read(&kvm->mm->mmap_sem);
+	srcu_read_unlock(&kvm->srcu, srcu_idx);
+	return ret;
+}
+
+static u64 kvmppc_get_secmem_size(void)
+{
+	struct device_node *np;
+	int i, len;
+	const __be32 *prop;
+	u64 size = 0;
+
+	np = of_find_compatible_node(NULL, NULL, "ibm,uv-firmware");
+	if (!np)
+		goto out;
+
+	prop = of_get_property(np, "secure-memory-ranges", &len);
+	if (!prop)
+		goto out_put;
+
+	for (i = 0; i < len / (sizeof(*prop) * 4); i++)
+		size += of_read_number(prop + (i * 4) + 2, 2);
+
+out_put:
+	of_node_put(np);
+out:
+	return size;
+}
+
+int kvmppc_uvmem_init(void)
+{
+	int ret = 0;
+	unsigned long size;
+	struct resource *res;
+	void *addr;
+	unsigned long pfn_last, pfn_first;
+
+	size = kvmppc_get_secmem_size();
+	if (!size) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	res = request_free_mem_region(&iomem_resource, size, "kvmppc_uvmem");
+	if (IS_ERR(res)) {
+		ret = PTR_ERR(res);
+		goto out;
+	}
+
+	kvmppc_uvmem_pgmap.type = MEMORY_DEVICE_PRIVATE;
+	kvmppc_uvmem_pgmap.res = *res;
+	kvmppc_uvmem_pgmap.ops = &kvmppc_uvmem_ops;
+	addr = memremap_pages(&kvmppc_uvmem_pgmap, NUMA_NO_NODE);
+	if (IS_ERR(addr)) {
+		ret = PTR_ERR(addr);
+		goto out_free_region;
+	}
+
+	pfn_first = res->start >> PAGE_SHIFT;
+	pfn_last = pfn_first + (resource_size(res) >> PAGE_SHIFT);
+	kvmppc_uvmem_pfn_bitmap = kcalloc(BITS_TO_LONGS(pfn_last - pfn_first),
+					  sizeof(unsigned long), GFP_KERNEL);
+	if (!kvmppc_uvmem_pfn_bitmap) {
+		ret = -ENOMEM;
+		goto out_unmap;
+	}
+
+	pr_info("KVMPPC-UVMEM: Secure Memory size 0x%lx\n", size);
+	return ret;
+out_unmap:
+	memunmap_pages(&kvmppc_uvmem_pgmap);
+out_free_region:
+	release_mem_region(res->start, size);
+out:
+	return ret;
+}
+
+void kvmppc_uvmem_free(void)
+{
+	memunmap_pages(&kvmppc_uvmem_pgmap);
+	release_mem_region(kvmppc_uvmem_pgmap.res.start,
+			   resource_size(&kvmppc_uvmem_pgmap.res));
+	kfree(kvmppc_uvmem_pfn_bitmap);
+}

@@ -342,6 +342,9 @@
 #define H_TLB_INVALIDATE	0xF808
 #define H_COPY_TOFROM_GUEST	0xF80C
 
+/* Flags for H_SVM_PAGE_IN */
+#define H_PAGE_IN_SHARED        0x1
+
 /* Platform-specific hcalls used by the Ultravisor */
 #define H_SVM_PAGE_IN		0xEF00
 #define H_SVM_PAGE_OUT		0xEF04

@@ -19,7 +19,10 @@
  * available in the platform for running secure guests is hotplugged.
  * Whenever a page belonging to the guest becomes secure, a page from this
  * private device memory is used to represent and track that secure page
- * on the HV side.
+ * on the HV side. Some pages (like virtio buffers, VPA pages etc) are
+ * shared between UV and HV. However such pages aren't represented by
+ * device private memory and mappings to shared memory exist in both
+ * UV and HV page tables.
  *
  * For each page that gets moved into secure memory, a device PFN is used
  * on the HV side and migration PTE corresponding to that PFN would be

@@ -80,6 +83,7 @@ struct kvmppc_uvmem_page_pvt {
 	unsigned long *rmap;
 	struct kvm *kvm;
 	unsigned long gpa;
+	bool skip_page_out;
 };
 
 /*

@@ -190,8 +194,70 @@ kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start,
 	return ret;
 }
 
+/*
+ * Shares the page with HV, thus making it a normal page.
+ *
+ * - If the page is already secure, then provision a new page and share
+ * - If the page is a normal page, share the existing page
+ *
+ * In the former case, uses dev_pagemap_ops.migrate_to_ram handler
+ * to unmap the device page from QEMU's page tables.
+ */
+static unsigned long
+kvmppc_share_page(struct kvm *kvm, unsigned long gpa, unsigned long page_shift)
+{
+
+	int ret = H_PARAMETER;
+	struct page *uvmem_page;
+	struct kvmppc_uvmem_page_pvt *pvt;
+	unsigned long pfn;
+	unsigned long *rmap;
+	struct kvm_memory_slot *slot;
+	unsigned long gfn = gpa >> page_shift;
+	int srcu_idx;
+
+	srcu_idx = srcu_read_lock(&kvm->srcu);
+	slot = gfn_to_memslot(kvm, gfn);
+	if (!slot)
+		goto out;
+
+	rmap = &slot->arch.rmap[gfn - slot->base_gfn];
+	mutex_lock(&kvm->arch.uvmem_lock);
+	if (kvmppc_rmap_type(rmap) == KVMPPC_RMAP_UVMEM_PFN) {
+		uvmem_page = pfn_to_page(*rmap & ~KVMPPC_RMAP_UVMEM_PFN);
+		pvt = uvmem_page->zone_device_data;
+		pvt->skip_page_out = true;
+	}
+
+retry:
+	mutex_unlock(&kvm->arch.uvmem_lock);
+	pfn = gfn_to_pfn(kvm, gfn);
+	if (is_error_noslot_pfn(pfn))
+		goto out;
+
+	mutex_lock(&kvm->arch.uvmem_lock);
+	if (kvmppc_rmap_type(rmap) == KVMPPC_RMAP_UVMEM_PFN) {
+		uvmem_page = pfn_to_page(*rmap & ~KVMPPC_RMAP_UVMEM_PFN);
+		pvt = uvmem_page->zone_device_data;
+		pvt->skip_page_out = true;
+		kvm_release_pfn_clean(pfn);
+		goto retry;
+	}
+
+	if (!uv_page_in(kvm->arch.lpid, pfn << page_shift, gpa, 0, page_shift))
+		ret = H_SUCCESS;
+	kvm_release_pfn_clean(pfn);
+	mutex_unlock(&kvm->arch.uvmem_lock);
+out:
+	srcu_read_unlock(&kvm->srcu, srcu_idx);
+	return ret;
+}
+
 /*
  * H_SVM_PAGE_IN: Move page from normal memory to secure memory.
+ *
+ * H_PAGE_IN_SHARED flag makes the page shared which means that the same
+ * memory in is visible from both UV and HV.
  */
 unsigned long
 kvmppc_h_svm_page_in(struct kvm *kvm, unsigned long gpa,