// 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. 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.
*/
/* * 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. * * 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. kvm->srcu - Protects KVM memslots * 2. kvm->mm->mmap_lock - find_vma, migrate_vma_pages and helpers, ksm_madvise * 3. kvm->arch.uvmem_lock - protects read/writes to uvmem slots thus acting * as sync-points for page-in/out
*/
/* * Notes on page size * * Currently UV uses 2MB mappings internally, but will issue H_SVM_PAGE_IN * and H_SVM_PAGE_OUT hcalls in PAGE_SIZE(64K) granularity. HV tracks * secure GPAs at 64K page size and maintains one device PFN for each * 64K secure GPA. UV_PAGE_IN and UV_PAGE_OUT calls by HV are also issued * for 64K page at a time. * * HV faulting on secure pages: When HV touches any secure page, it * faults and issues a UV_PAGE_OUT request with 64K page size. Currently * UV splits and remaps the 2MB page if necessary and copies out the * required 64K page contents. * * Shared pages: Whenever guest shares a secure page, UV will split and * remap the 2MB page if required and issue H_SVM_PAGE_IN with 64K page size. * * HV invalidating a page: When a regular page belonging to secure * guest gets unmapped, HV informs UV with UV_PAGE_INVAL of 64K * page size. Using 64K page size is correct here because any non-secure * page will essentially be of 64K page size. Splitting by UV during sharing * and page-out ensures this. * * Page fault handling: When HV handles page fault of a page belonging * to secure guest, it sends that to UV with a 64K UV_PAGE_IN request. * Using 64K size is correct here too as UV would have split the 2MB page * into 64k mappings and would have done page-outs earlier. * * In summary, the current secure pages handling code in HV assumes * 64K page size and in fact fails any page-in/page-out requests of * non-64K size upfront. If and when UV starts supporting multiple * page-sizes, we need to break this assumption.
*/
/* * States of a GFN * --------------- * The GFN can be in one of the following states. * * (a) Secure - The GFN is secure. The GFN is associated with * a Secure VM, the contents of the GFN is not accessible * to the Hypervisor. This GFN can be backed by a secure-PFN, * or can be backed by a normal-PFN with contents encrypted. * The former is true when the GFN is paged-in into the * ultravisor. The latter is true when the GFN is paged-out * of the ultravisor. * * (b) Shared - The GFN is shared. The GFN is associated with a * a secure VM. The contents of the GFN is accessible to * Hypervisor. This GFN is backed by a normal-PFN and its * content is un-encrypted. * * (c) Normal - The GFN is a normal. The GFN is associated with * a normal VM. The contents of the GFN is accessible to * the Hypervisor. Its content is never encrypted. * * States of a VM. * --------------- * * Normal VM: A VM whose contents are always accessible to * the hypervisor. All its GFNs are normal-GFNs. * * Secure VM: A VM whose contents are not accessible to the * hypervisor without the VM's consent. Its GFNs are * either Shared-GFN or Secure-GFNs. * * Transient VM: A Normal VM that is transitioning to secure VM. * The transition starts on successful return of * H_SVM_INIT_START, and ends on successful return * of H_SVM_INIT_DONE. This transient VM, can have GFNs * in any of the three states; i.e Secure-GFN, Shared-GFN, * and Normal-GFN. The VM never executes in this state * in supervisor-mode. * * Memory slot State. * ----------------------------- * The state of a memory slot mirrors the state of the * VM the memory slot is associated with. * * VM State transition. * -------------------- * * A VM always starts in Normal Mode. * * H_SVM_INIT_START moves the VM into transient state. During this * time the Ultravisor may request some of its GFNs to be shared or * secured. So its GFNs can be in one of the three GFN states. * * H_SVM_INIT_DONE moves the VM entirely from transient state to * secure-state. At this point any left-over normal-GFNs are * transitioned to Secure-GFN. * * H_SVM_INIT_ABORT moves the transient VM back to normal VM. * All its GFNs are moved to Normal-GFNs. * * UV_TERMINATE transitions the secure-VM back to normal-VM. All * the secure-GFN and shared-GFNs are tranistioned to normal-GFN * Note: The contents of the normal-GFN is undefined at this point. * * GFN state implementation: * ------------------------- * * Secure GFN is associated with a secure-PFN; also called uvmem_pfn, * when the GFN is paged-in. Its pfn[] has KVMPPC_GFN_UVMEM_PFN flag * set, and contains the value of the secure-PFN. * It is associated with a normal-PFN; also called mem_pfn, when * the GFN is pagedout. Its pfn[] has KVMPPC_GFN_MEM_PFN flag set. * The value of the normal-PFN is not tracked. * * Shared GFN is associated with a normal-PFN. Its pfn[] has * KVMPPC_UVMEM_SHARED_PFN flag set. The value of the normal-PFN * is not tracked. * * Normal GFN is associated with normal-PFN. Its pfn[] has * no flag set. The value of the normal-PFN is not tracked. * * Life cycle of a GFN * -------------------- * * -------------------------------------------------------------- * | | Share | Unshare | SVM |H_SVM_INIT_DONE| * | |operation |operation | abort/ | | * | | | | terminate | | * ------------------------------------------------------------- * | | | | | | * | Secure | Shared | Secure |Normal |Secure | * | | | | | | * | Shared | Shared | Secure |Normal |Shared | * | | | | | | * | Normal | Shared | Secure |Normal |Secure | * -------------------------------------------------------------- * * Life cycle of a VM * -------------------- * * -------------------------------------------------------------------- * | | start | H_SVM_ |H_SVM_ |H_SVM_ |UV_SVM_ | * | | VM |INIT_START|INIT_DONE|INIT_ABORT |TERMINATE | * | | | | | | | * --------- ---------------------------------------------------------- * | | | | | | | * | Normal | Normal | Transient|Error |Error |Normal | * | | | | | | | * | Secure | Error | Error |Error |Error |Normal | * | | | | | | | * |Transient| N/A | Error |Secure |Normal |Normal | * --------------------------------------------------------------------
*/
bool kvmppc_uvmem_available(void)
{ /* * If kvmppc_uvmem_bitmap != NULL, then there is an ultravisor * and our data structures have been initialized successfully.
*/ return !!kvmppc_uvmem_bitmap;
}
/* * All device PFNs are already released by the time we come here.
*/ void kvmppc_uvmem_slot_free(struct kvm *kvm, conststruct kvm_memory_slot *slot)
{ struct kvmppc_uvmem_slot *p, *next;
/* mark the GFN as secure-GFN associated with @uvmem pfn device-PFN. */ staticvoid kvmppc_gfn_secure_uvmem_pfn(unsignedlong gfn, unsignedlong uvmem_pfn, struct kvm *kvm)
{
kvmppc_mark_gfn(gfn, kvm, KVMPPC_GFN_UVMEM_PFN, uvmem_pfn);
}
/* mark the GFN as secure-GFN associated with a memory-PFN. */ staticvoid kvmppc_gfn_secure_mem_pfn(unsignedlong gfn, struct kvm *kvm)
{
kvmppc_mark_gfn(gfn, kvm, KVMPPC_GFN_MEM_PFN, 0);
}
/* mark the GFN as a shared GFN. */ staticvoid kvmppc_gfn_shared(unsignedlong gfn, struct kvm *kvm)
{
kvmppc_mark_gfn(gfn, kvm, KVMPPC_GFN_SHARED, 0);
}
/* mark the GFN as a non-existent GFN. */ staticvoid kvmppc_gfn_remove(unsignedlong gfn, struct kvm *kvm)
{
kvmppc_mark_gfn(gfn, kvm, 0, 0);
}
/* return true, if the GFN is a secure-GFN backed by a secure-PFN */ staticbool kvmppc_gfn_is_uvmem_pfn(unsignedlong gfn, struct kvm *kvm, unsignedlong *uvmem_pfn)
{ struct kvmppc_uvmem_slot *p;
if (p->pfns[index] & KVMPPC_GFN_UVMEM_PFN) { if (uvmem_pfn)
*uvmem_pfn = p->pfns[index] &
KVMPPC_GFN_PFN_MASK; returntrue;
} else returnfalse;
}
} returnfalse;
}
/* * starting from *gfn search for the next available GFN that is not yet * transitioned to a secure GFN. return the value of that GFN in *gfn. If a * GFN is found, return true, else return false * * Must be called with kvm->arch.uvmem_lock held.
*/ staticbool kvmppc_next_nontransitioned_gfn(conststruct kvm_memory_slot *memslot, struct kvm *kvm, unsignedlong *gfn)
{ struct kvmppc_uvmem_slot *p = NULL, *iter; bool ret = false; unsignedlong i;
list_for_each_entry(iter, &kvm->arch.uvmem_pfns, list) if (*gfn >= iter->base_pfn && *gfn < iter->base_pfn + iter->nr_pfns) {
p = iter; break;
} if (!p) return ret; /* * The code below assumes, one to one correspondence between * kvmppc_uvmem_slot and memslot.
*/ for (i = *gfn; i < p->base_pfn + p->nr_pfns; i++) { unsignedlong index = i - p->base_pfn;
if (!(p->pfns[index] & KVMPPC_GFN_FLAG_MASK)) {
*gfn = i;
ret = true; break;
}
} return ret;
}
/* * This function is used in two cases: * - When HV touches a secure page, for which we do UV_PAGE_OUT * - When a secure page is converted to shared page, we *get* * the page to essentially unmap the device page. In this * case we skip page-out.
*/ if (!pvt->skip_page_out)
ret = uv_page_out(kvm->arch.lpid, pfn << page_shift,
gpa, 0, page_shift);
/* * Drop device pages that we maintain for the secure guest * * We first mark the pages to be skipped from UV_PAGE_OUT when there * is HV side fault on these pages. Next we *get* these pages, forcing * fault on them, do fault time migration to replace the device PTEs in * QEMU page table with normal PTEs from newly allocated pages.
*/ void kvmppc_uvmem_drop_pages(conststruct kvm_memory_slot *slot, struct kvm *kvm, bool skip_page_out)
{ int i; struct kvmppc_uvmem_page_pvt *pvt; struct page *uvmem_page; struct vm_area_struct *vma = NULL; unsignedlong uvmem_pfn, gfn; unsignedlong addr;
mmap_read_lock(kvm->mm);
addr = slot->userspace_addr;
gfn = slot->base_gfn; for (i = slot->npages; i; --i, ++gfn, addr += PAGE_SIZE) {
/* Fetch the VMA if addr is not in the latest fetched one */ if (!vma || addr >= vma->vm_end) {
vma = vma_lookup(kvm->mm, addr); if (!vma) {
pr_err("Can't find VMA for gfn:0x%lx\n", gfn); break;
}
}
/* * Expect to be called only after INIT_START and before INIT_DONE. * If INIT_DONE was completed, use normal VM termination sequence.
*/ if (!(kvm->arch.secure_guest & KVMPPC_SECURE_INIT_START)) return H_UNSUPPORTED;
if (kvm->arch.secure_guest & KVMPPC_SECURE_INIT_DONE) return H_STATE;
/* * 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. * * Called with kvm->arch.uvmem_lock held
*/ staticstruct page *kvmppc_uvmem_get_page(unsignedlong gpa, struct kvm *kvm)
{ struct page *dpage = NULL; unsignedlong bit, uvmem_pfn; struct kvmppc_uvmem_page_pvt *pvt; unsignedlong pfn_last, pfn_first;
ret = kvmppc_svm_page_in(vma, start, end,
(gfn << PAGE_SHIFT), kvm, PAGE_SHIFT, false); if (ret) {
ret = H_STATE; break;
}
/* relinquish the cpu if needed */
cond_resched();
}
mutex_unlock(&kvm->arch.uvmem_lock);
mmap_read_unlock(kvm->mm); return ret;
}
unsignedlong kvmppc_h_svm_init_done(struct kvm *kvm)
{ struct kvm_memslots *slots; struct kvm_memory_slot *memslot; int srcu_idx, bkt; long ret = H_SUCCESS;
if (!(kvm->arch.secure_guest & KVMPPC_SECURE_INIT_START)) return H_UNSUPPORTED;
/* migrate any unmoved normal pfn to device pfns*/
srcu_idx = srcu_read_lock(&kvm->srcu);
slots = kvm_memslots(kvm);
kvm_for_each_memslot(memslot, bkt, slots) {
ret = kvmppc_uv_migrate_mem_slot(kvm, memslot); if (ret) { /* * The pages will remain transitioned. * Its the callers responsibility to * terminate the VM, which will undo * all state of the VM. Till then * this VM is in a erroneous state. * Its KVMPPC_SECURE_INIT_DONE will * remain unset.
*/
ret = H_STATE; goto out;
}
}
kvm->arch.secure_guest |= KVMPPC_SECURE_INIT_DONE;
pr_info("LPID %lld went secure\n", kvm->arch.lpid);
/* * 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.
*/ staticunsignedlong kvmppc_share_page(struct kvm *kvm, unsignedlong gpa, unsignedlong page_shift)
{
int ret = H_PARAMETER; struct page *page, *uvmem_page; struct kvmppc_uvmem_page_pvt *pvt; unsignedlong gfn = gpa >> page_shift; int srcu_idx; unsignedlong uvmem_pfn;
srcu_idx = srcu_read_lock(&kvm->srcu);
mutex_lock(&kvm->arch.uvmem_lock); if (kvmppc_gfn_is_uvmem_pfn(gfn, kvm, &uvmem_pfn)) {
uvmem_page = pfn_to_page(uvmem_pfn);
pvt = uvmem_page->zone_device_data;
pvt->skip_page_out = true; /* * do not drop the GFN. It is a valid GFN * that is transitioned to a shared GFN.
*/
pvt->remove_gfn = false;
}
retry:
mutex_unlock(&kvm->arch.uvmem_lock);
page = gfn_to_page(kvm, gfn); if (!page) goto out;
mutex_lock(&kvm->arch.uvmem_lock); if (kvmppc_gfn_is_uvmem_pfn(gfn, kvm, &uvmem_pfn)) {
uvmem_page = pfn_to_page(uvmem_pfn);
pvt = uvmem_page->zone_device_data;
pvt->skip_page_out = true;
pvt->remove_gfn = false; /* it continues to be a valid GFN */
kvm_release_page_unused(page); goto retry;
}
/* * 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.
*/ unsignedlong kvmppc_h_svm_page_in(struct kvm *kvm, unsignedlong gpa, unsignedlong flags, unsignedlong page_shift)
{ unsignedlong start, end; struct vm_area_struct *vma; int srcu_idx; unsignedlong gfn = gpa >> page_shift; int ret;
if (!(kvm->arch.secure_guest & KVMPPC_SECURE_INIT_START)) return H_UNSUPPORTED;
if (page_shift != PAGE_SHIFT) return H_P3;
if (flags & ~H_PAGE_IN_SHARED) return H_P2;
if (flags & H_PAGE_IN_SHARED) return kvmppc_share_page(kvm, gpa, page_shift);
ret = H_PARAMETER;
srcu_idx = srcu_read_lock(&kvm->srcu);
mmap_read_lock(kvm->mm);
start = gfn_to_hva(kvm, gfn); if (kvm_is_error_hva(start)) goto out;
mutex_lock(&kvm->arch.uvmem_lock); /* Fail the page-in request of an already paged-in page */ if (kvmppc_gfn_is_uvmem_pfn(gfn, kvm, NULL)) goto out_unlock;
/* * 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;
/* * Release the device PFN back to the pool * * Gets called when secure GFN tranistions from a secure-PFN * to a normal PFN during H_SVM_PAGE_OUT. * Gets called with kvm->arch.uvmem_lock held.
*/ staticvoid kvmppc_uvmem_page_free(struct page *page)
{ unsignedlong pfn = page_to_pfn(page) -
(kvmppc_uvmem_pgmap.range.start >> PAGE_SHIFT); struct kvmppc_uvmem_page_pvt *pvt;
static u64 kvmppc_get_secmem_size(void)
{ struct device_node *np; int i, len; const __be32 *prop;
u64 size = 0;
/* * First try the new ibm,secure-memory nodes which supersede the * secure-memory-ranges property. * If we found some, no need to read the deprecated ones.
*/
for_each_compatible_node(np, NULL, "ibm,secure-memory") {
prop = of_get_property(np, "reg", &len); if (!prop) continue;
size += of_read_number(prop + 2, 2);
} if (size) return size;
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; unsignedlong size; struct resource *res; void *addr; unsignedlong pfn_last, pfn_first;
size = kvmppc_get_secmem_size(); if (!size) { /* * Don't fail the initialization of kvm-hv module if * the platform doesn't export ibm,uv-firmware node. * Let normal guests run on such PEF-disabled platform.
*/
pr_info("KVMPPC-UVMEM: No support for secure guests\n"); 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.range.start = res->start;
kvmppc_uvmem_pgmap.range.end = res->end;
kvmppc_uvmem_pgmap.nr_range = 1;
kvmppc_uvmem_pgmap.ops = &kvmppc_uvmem_ops; /* just one global instance: */
kvmppc_uvmem_pgmap.owner = &kvmppc_uvmem_pgmap;
addr = memremap_pages(&kvmppc_uvmem_pgmap, NUMA_NO_NODE); if (IS_ERR(addr)) {
ret = PTR_ERR(addr); goto out_free_region;
}
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.