Migrate non-migrated pages of a SVM.
Subsystems (3)
v3 cover Migrate non-migrated pages of a SVM.
The time taken to switch a VM to Secure-VM, increases by the size of the VM. A 100GB VM takes about 7minutes. This is unacceptable. This linear increase is caused by a suboptimal behavior by the Ultravisor and the Hypervisor. The Ultravisor unnecessarily migrates all the GFN of the VM from normal-memory to secure-memory. It has to just migrate the necessary and sufficient GFNs.
However when the optimization is incorporated in the Ultravisor, the Hypervisor starts misbehaving. The Hypervisor has a inbuilt assumption that the Ultravisor will explicitly request to migrate, each and every GFN of the VM. If only necessary and sufficient GFNs are requested for migration, the Hypervisor continues to manage the remaining GFNs as normal GFNs. This leads of memory corruption, manifested consistently when the SVM reboots.
The same is true, when a memory slot is hotplugged into a SVM. The Hypervisor expects the ultravisor to request migration of all GFNs to secure-GFN. But at the same time, the hypervisor is unable to handle any H_SVM_PAGE_IN requests from the Ultravisor, done in the context of UV_REGISTER_MEM_SLOT ucall. This problem manifests as random errors in the SVM, when a memory-slot is hotplugged.
This patch series automatically migrates the non-migrated pages of a SVM,
and thus solves the problem.
Testing: Passed rigorous SVM test using various sized SVMs.
Changelog:
v2: . fixed a bug observed by Laurent. The state of the GFN's associated
with Secure-VMs were not reset during memslot flush.
. Re-organized the code, for easier review.
. Better description of the patch series.
v1: fixed a bug observed by Bharata. Pages that where paged-in and later paged-out must also be skipped from migration during H_SVM_INIT_DONE.
Laurent Dufour (1):
KVM: PPC: Book3S HV: migrate hot plugged memory
Ram Pai (3):
KVM: PPC: Book3S HV: Fix function definition in book3s_hv_uvmem.c
KVM: PPC: Book3S HV: track the state GFNs associated with secure VMs
KVM: PPC: Book3S HV: migrate remaining normal-GFNs to secure-GFNs in
H_SVM_INIT_DONE
Documentation/powerpc/ultravisor.rst | 2 + arch/powerpc/include/asm/kvm_book3s_uvmem.h | 2 + arch/powerpc/kvm/book3s_hv.c | 10 +- arch/powerpc/kvm/book3s_hv_uvmem.c | 360 +++++++++++++++++++++++----- 4 files changed, 315 insertions(+), 59 deletions(-)
Signature · 1 line
1.8.3.1
PATCH v3 1/4 KVM: PPC: Book3S HV: Fix function definition in book3s_hv_uvmem.c
Without this fix, git is confused. It generates wrong function context for code changes in subsequent patches. Weird, but true.
--- arch/powerpc/kvm/book3s_hv_uvmem.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-)
arch/powerpc/kvm/book3s_hv_uvmem.c +10 -11
@@ -369,8 +369,7 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm)* 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, +static int kvmppc_svm_page_in (struct vm_area_struct *vma,unsigned long start,unsigned long end,unsigned long gpa,struct kvm *kvm,unsigned long page_shift,bool *downgrade) {
@@ -437,8 +436,8 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm)* 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) +static unsigned long kvmppc_share_page (struct kvm *kvm,unsigned long gpa, +unsigned long page_shift) {int ret =H_PARAMETER ;
@@ -487,9 +486,9 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm)* 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, -unsigned long flags,unsigned long page_shift) +unsigned long kvmppc_h_svm_page_in (struct kvm *kvm,unsigned long gpa, +unsigned long flags, +unsigned long page_shift) {bool downgrade = false;unsigned long start, end;
@@ -546,10 +545,10 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm)* 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 gpa) +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 gpa) {unsigned long src_pfn, dst_pfn =0 ;struct migrate_vma mig;
Signature · 1 line
1.8.3.1
PATCH v3 2/4 KVM: PPC: Book3S HV: track the state GFNs associated with secure VMs
In reply to Ram PaiDuring the life of SVM, its GFNs transition through normal, secure and shared states. Since the kernel does not track GFNs that are shared, it is not possible to disambiguate a shared GFN from a GFN whose PFN has not yet been migrated to a secure-PFN. Also it is not possible to disambiguate a secure-GFN from a GFN whose GFN has been pagedout from the ultravisor.
The ability to identify the state of a GFN is needed to skip migration of its PFN to secure-PFN during ESM transition.
The code is re-organized to track the states of a GFN as explained below.
************************************************************************
1. 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 accesible to the Hypervisor. Its content is never encrypted.
2. States of a VM.
---------------
(a) Normal VM: A VM whose contents are always accessible to the hypervisor. All its GFNs are normal-GFNs.
(b) 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.
(c) 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.
3. Memory slot State.
------------------
The state of a memory slot mirrors the state of the VM the memory slot is associated with.
4. 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.
5. 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.
6. 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 |
--------------------------------------------------------------
7. 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 |
--------------------------------------------------------------------
************************************************************************
--- arch/powerpc/kvm/book3s_hv_uvmem.c | 187 +++++++++++++++++++++++++++++++++---- 1 file changed, 168 insertions(+), 19 deletions(-)
arch/powerpc/kvm/book3s_hv_uvmem.c +168 -19
@@ -98,7 +98,127 @@static unsigned long *kvmppc_uvmem_bitmap;static DEFINE_SPINLOCK (kvmppc_uvmem_bitmap_lock); -#define KVMPPC_UVMEM_PFN (1UL << 63) +/* +* 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 accesible 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 | +* -------------------------------------------------------------------- +*/ + +#define KVMPPC_GFN_UVMEM_PFN (1UL << 63) +#define KVMPPC_GFN_MEM_PFN (1UL << 62) +#define KVMPPC_GFN_SHARED (1UL << 61) +#define KVMPPC_GFN_SECURE (KVMPPC_GFN_UVMEM_PFN | KVMPPC_GFN_MEM_PFN) +#define KVMPPC_GFN_FLAG_MASK (KVMPPC_GFN_SECURE | KVMPPC_GFN_SHARED) +#define KVMPPC_GFN_PFN_MASK (~KVMPPC_GFN_FLAG_MASK)struct kvmppc_uvmem_slot {struct list_head list;
@@ -106,11 +226,11 @@ struct kvmppc_uvmem_slot {unsigned long base_pfn;unsigned long *pfns; }; -struct kvmppc_uvmem_page_pvt {struct kvm *kvm;unsigned long gpa;bool skip_page_out; +bool remove_gfn; };int kvmppc_uvmem_slot_init (struct kvm *kvm,const struct kvm_memory_slot *slot)
@@ -154,8 +274,8 @@ void kvmppc_uvmem_slot_free(struct kvm *kvm, const struct kvm_memory_slot *slot)mutex_unlock (&kvm->arch.uvmem_lock); } -static void kvmppc_uvmem_pfn_insert (unsigned long gfn,unsigned long uvmem_pfn, -struct kvm *kvm) +static void kvmppc_mark_gfn (unsigned long gfn,struct kvm *kvm, +unsigned long flag,unsigned long uvmem_pfn) {struct kvmppc_uvmem_slot *p;
@@ -163,24 +283,41 @@ static void kvmppc_uvmem_pfn_insert(unsigned long gfn, unsigned long uvmem_pfn,if (gfn >= p->base_pfn && gfn < p->base_pfn + p->nr_pfns) {unsigned long index = gfn - p->base_pfn; - p->pfns[index] = uvmem_pfn |KVMPPC_UVMEM_PFN ; +if (flag =KVMPPC_GFN_UVMEM_PFN ) + p->pfns[index] = uvmem_pfn | flag; +else + p->pfns[index] = flag;return ; } } } -static void kvmppc_uvmem_pfn_remove (unsigned long gfn,struct kvm *kvm) +/* mark the GFN as secure-GFN associated with @uvmem pfn device-PFN. */ +static void kvmppc_gfn_secure_uvmem_pfn (unsigned long gfn, +unsigned long uvmem_pfn,struct kvm *kvm) { -struct kvmppc_uvmem_slot *p; +kvmppc_mark_gfn (gfn, kvm,KVMPPC_GFN_UVMEM_PFN , uvmem_pfn); +} -list_for_each_entry (p, &kvm->arch.uvmem_pfns, list) { -if (gfn >= p->base_pfn && gfn < p->base_pfn + p->nr_pfns) { - p->pfns[gfn - p->base_pfn] =0 ; -return ; - } - } +/* mark the GFN as secure-GFN associated with a memory-PFN. */ +static void kvmppc_gfn_secure_mem_pfn (unsigned long gfn,struct kvm *kvm) +{ +kvmppc_mark_gfn (gfn, kvm,KVMPPC_GFN_MEM_PFN ,0 ); +} + +/* mark the GFN as a shared GFN. */ +static void kvmppc_gfn_shared (unsigned long gfn,struct kvm *kvm) +{ +kvmppc_mark_gfn (gfn, kvm,KVMPPC_GFN_SHARED ,0 ); +} + +/* mark the GFN as a non-existent GFN. */ +static void kvmppc_gfn_remove (unsigned long 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 */ static bool kvmppc_gfn_is_uvmem_pfn (unsigned long gfn,struct kvm *kvm,unsigned long *uvmem_pfn) {
@@ -190,10 +327,10 @@ static bool kvmppc_gfn_is_uvmem_pfn(unsigned long gfn, struct kvm *kvm,if (gfn >= p->base_pfn && gfn < p->base_pfn + p->nr_pfns) {unsigned long index = gfn - p->base_pfn; -if (p->pfns[index] &KVMPPC_UVMEM_PFN ) { +if (p->pfns[index] &KVMPPC_GFN_UVMEM_PFN ) {if (uvmem_pfn) *uvmem_pfn = p->pfns[index] & - ~KVMPPC_UVMEM_PFN ; +KVMPPC_GFN_PFN_MASK ;return true; }else return false;
@@ -271,6 +408,7 @@ void kvmppc_uvmem_drop_pages(const struct kvm_memory_slot *free,mutex_lock (&kvm->arch.uvmem_lock);if (!kvmppc_gfn_is_uvmem_pfn (gfn, kvm, &uvmem_pfn)) { +kvmppc_gfn_remove (gfn, kvm);mutex_unlock (&kvm->arch.uvmem_lock);continue ; }
@@ -278,6 +416,7 @@ void kvmppc_uvmem_drop_pages(const struct kvm_memory_slot *free, uvmem_page =pfn_to_page (uvmem_pfn); pvt = uvmem_page->zone_device_data; pvt->skip_page_out = skip_page_out; + pvt->remove_gfn = true;mutex_unlock (&kvm->arch.uvmem_lock); pfn =gfn_to_pfn (kvm, gfn);
@@ -347,7 +486,7 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm) goto out_clear; uvmem_pfn = bit + pfn_first; -kvmppc_uvmem_pfn_insert (gpa >>PAGE_SHIFT , uvmem_pfn, kvm); +kvmppc_gfn_secure_uvmem_pfn (gpa >>PAGE_SHIFT , uvmem_pfn, kvm); pvt->gpa = gpa; pvt->kvm = kvm;
@@ -454,6 +593,7 @@ static unsigned long kvmppc_share_page(struct kvm *kvm, unsigned long gpa, uvmem_page =pfn_to_page (uvmem_pfn); pvt = uvmem_page->zone_device_data; pvt->skip_page_out = true; + pvt->remove_gfn = false; } retry:
@@ -467,12 +607,16 @@ static unsigned long kvmppc_share_page(struct kvm *kvm, unsigned long gpa, uvmem_page =pfn_to_page (uvmem_pfn); pvt = uvmem_page->zone_device_data; pvt->skip_page_out = true; + pvt->remove_gfn = false;kvm_release_pfn_clean (pfn); goto retry; } -if (!uv_page_in (kvm->arch.lpid, pfn << page_shift, gpa,0 , page_shift)) +if (!uv_page_in (kvm->arch.lpid, pfn << page_shift, gpa,0 , + page_shift)) { +kvmppc_gfn_shared (gfn, kvm); ret =H_SUCCESS ; + }kvm_release_pfn_clean (pfn);mutex_unlock (&kvm->arch.uvmem_lock); out:
@@ -530,6 +674,7 @@ unsigned long kvmppc_h_svm_page_in(struct kvm *kvm, unsigned long gpa,if (!kvmppc_svm_page_in (vma, start, end, gpa, kvm, page_shift, &downgrade)) ret =H_SUCCESS ; + out_unlock:mutex_unlock (&kvm->arch.uvmem_lock); out:
@@ -640,7 +785,8 @@ static vm_fault_t kvmppc_uvmem_migrate_to_ram(struct vm_fault *vmf)/* * Release the device PFN back to the pool * -* Gets called when secure page becomes a normal page during H_SVM_PAGE_OUT. +* 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. */ static void kvmppc_uvmem_page_free (struct page *page)
@@ -655,7 +801,10 @@ static void kvmppc_uvmem_page_free(struct page *page) pvt = page->zone_device_data; page->zone_device_data =NULL ; -kvmppc_uvmem_pfn_remove (pvt->gpa >>PAGE_SHIFT , pvt->kvm); +if (pvt->remove_gfn) +kvmppc_gfn_remove (pvt->gpa >>PAGE_SHIFT , pvt->kvm); +else +kvmppc_gfn_secure_mem_pfn (pvt->gpa >>PAGE_SHIFT , pvt->kvm);kfree (pvt); }
Signature · 1 line
1.8.3.1
PATCH v3 3/4 KVM: PPC: Book3S HV: migrate remaining normal-GFNs to secure-GFNs in H_SVM_INIT_DONE
In reply to Ram PaiH_SVM_INIT_DONE incorrectly assumes that the Ultravisor has explicitly called H_SVM_PAGE_IN for all secure pages. These GFNs continue to be normal GFNs associated with normal PFNs; when infact, these GFNs should have been secure GFNs, associated with device PFNs.
Move all the PFNs associated with the SVM's GFNs, to secure-PFNs, in H_SVM_INIT_DONE. Skip the GFNs that are already Paged-in or Shared through H_SVM_PAGE_IN, or Paged-in followed by a Paged-out through UV_PAGE_OUT.
--- Documentation/powerpc/ultravisor.rst | 2 + arch/powerpc/include/asm/kvm_book3s_uvmem.h | 2 + arch/powerpc/kvm/book3s_hv_uvmem.c | 154 +++++++++++++++++++++++----- 3 files changed, 132 insertions(+), 26 deletions(-)
Documentation/powerpc/ultravisor.rst +2 -0
@@ -933,6 +933,8 @@ Return values * H_UNSUPPORTED if called from the wrong context (e.g. from an SVM or before an H_SVM_INIT_START hypercall). + * H_STATE if the hypervisor could not successfully + transition the VM to Secure VM. Description ~~~~~~~~~~~
arch/powerpc/include/asm/kvm_book3s_uvmem.h +2 -0
@@ -22,6 +22,8 @@ unsigned long kvmppc_h_svm_page_out(struct kvm *kvm,unsigned long kvmppc_h_svm_init_abort (struct kvm *kvm);void kvmppc_uvmem_drop_pages (const struct kvm_memory_slot *free,struct kvm *kvm,bool skip_page_out); +int kvmppc_uv_migrate_mem_slot (struct kvm *kvm, +const struct kvm_memory_slot *memslot);#else static inline int kvmppc_uvmem_init (void ) {
arch/powerpc/kvm/book3s_hv_uvmem.c +128 -26
208 lines not on this page · raw message
Signature · 1 line
1.8.3.1
Re: [PATCH v3 3/4] KVM: PPC: Book3S HV: migrate remaining normal-GFNs to secure-GFNs in H_SVM_INIT_D
Quoted 1 line · Ram Pai
H_SVM_INIT_DONEincorrectly assumes that the Ultravisor has explicitly
As noted in the last iteration, can you reword the above please? I don't see it as an incorrect assumption, but see it as extension of scope now :-)
Quoted 142 lines
called
H_SVM_PAGE_INfor all secure pages. These GFNs continue to be normal GFNs associated with normal PFNs; when infact, these GFNs should have been secure GFNs, associated with device PFNs.Move all the PFNs associated with the SVM's GFNs, to secure-PFNs, in
H_SVM_INIT_DONE. Skip the GFNs that are already Paged-in or Shared throughH_SVM_PAGE_IN, or Paged-in followed by a Paged-out throughUV_PAGE_OUT.Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Cc Bharata B Rao <[email protected]>Cc Aneesh Kumar K.V <[email protected]>Cc Sukadev Bhattiprolu <[email protected]>Cc Laurent Dufour <[email protected]>Cc Thiago Jung Bauermann <[email protected]>Cc David Gibson <[email protected]>Cc Claudio Carvalho <[email protected]>Signed-off-by Ram Pai <[email protected]>--- Documentation/powerpc/ultravisor.rst | 2 + arch/powerpc/include/asm/kvm_book3s_uvmem.h | 2 + arch/powerpc/kvm/book3s_hv_uvmem.c | 154 +++++++++++++++++++++++----- 3 files changed, 132 insertions(+), 26 deletions(-)Documentation/powerpc/ultravisor.rst +2
-0@@ -933,6 +933,8 @@ Return values * H_UNSUPPORTED if called from the wrong context (e.g. from an SVM or before an H_SVM_INIT_START hypercall). + * H_STATE if the hypervisor could not successfully + transition the VM to Secure VM. Description ~~~~~~~~~~~arch/powerpc/include/asm/kvm_book3s_uvmem.h +2
-0@@ -22,6 +22,8 @@ unsigned long kvmppc_h_svm_page_out(struct kvm *kvm,unsigned long kvmppc_h_svm_init_abort (struct kvm *kvm);void kvmppc_uvmem_drop_pages (const struct kvm_memory_slot *free,struct kvm *kvm,bool skip_page_out); +int kvmppc_uv_migrate_mem_slot (struct kvm *kvm, +const struct kvm_memory_slot *memslot);#else static inline int kvmppc_uvmem_init (void ) {arch/powerpc/kvm/book3s_hv_uvmem.c +43
-6@@ -93,6 +93,7 @@#include <asm/ultravisor.h> #include <asm/mman.h> #include <asm/kvm_ppc.h> +#include <asm/kvm_book3s_uvmem.h> static struct dev_pagemap kvmppc_uvmem_pgmap;static unsigned long *kvmppc_uvmem_bitmap;@@ -339,6 +340,21 @@ static bool kvmppc_gfn_is_uvmem_pfn(unsigned long gfn, struct kvm *kvm,return false; } +/* return true, if the GFN is a shared-GFN, or a secure-GFN */ +bool kvmppc_gfn_has_transitioned (unsigned long gfn,struct kvm *kvm) +{ +struct kvmppc_uvmem_slot *p; + +list_for_each_entry (p, &kvm->arch.uvmem_pfns, list) { +if (gfn >= p->base_pfn && gfn < p->base_pfn + p->nr_pfns) { +unsigned long index = gfn - p->base_pfn; + +return (p->pfns[index] &KVMPPC_GFN_FLAG_MASK ); + } + } +return false; +} +unsigned long kvmppc_h_svm_init_start (struct kvm *kvm) {struct kvm_memslots *slots;@@ -379,12 +395,31 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)unsigned long kvmppc_h_svm_init_done (struct kvm *kvm) { +struct kvm_memslots *slots; +struct kvm_memory_slot *memslot; +int srcu_idx; +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, slots) { + ret =kvmppc_uv_migrate_mem_slot (kvm, memslot); +if (ret) { + ret =H_STATE ; + goto out; + } + } + kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_DONE ;pr_info ("LPID %d went secure\n" , kvm->arch.lpid); -return H_SUCCESS ; + +out: +srcu_read_unlock (&kvm->srcu, srcu_idx); +return ret; }/* @@ -505,12 +540,14 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm) }/* -* Alloc a PFN from private device memory pool and copy page from normal -* memory to secure memory using UV_PAGE_IN uvcall. +* Alloc a PFN from private device memory pool. If @pagein is true, +* 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 gpa,struct kvm *kvm, -unsigned long page_shift,bool *downgrade) +static int kvmppc_svm_migrate_page (struct vm_area_struct *vma, +unsigned long start, +unsigned long end,unsigned long gpa,struct kvm *kvm, +unsigned long page_shift, +bool pagein) {unsigned long src_pfn, dst_pfn =0 ;struct migrate_vma mig;
arch/powerpc/kvm/book3s_hv_uvmem.c static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, · line 533
Quoted 8 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-5@@ -526,18 +563,6 @@ static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, mig.src = &src_pfn; mig.dst = &dst_pfn; -/* -* We come here with mmap_sem write lock held just for -* ksm_madvise(), otherwise we only need read mmap_sem. -* Hence downgrade to read lock once ksm_madvise() is done. -*/
Can you please retain this comment? This explains why we take write lock and then downgrade.
arch/powerpc/kvm/book3s_hv_uvmem.c static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, · line 537
Quoted 4 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-4- ret =ksm_madvise (vma, vma->vm_start, vma->vm_end, -MADV_UNMERGEABLE , &vma->vm_flags); -downgrade_write (&kvm->mm->mmap_sem); - *downgrade = true;
When I introduced this variable, there was a suggestion to rename it to "downgraded", but we were a bit late then. When you are touching this now, can you rename it appropriately?
Quoted 69 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +47
-8-if (ret) -return ret; - ret =migrate_vma_setup (&mig);if (ret)return ret;@@ -553,11 +578,16 @@ static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, 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); +if (pagein) { + pfn = *mig.src >>MIGRATE_PFN_SHIFT ; + spage =migrate_pfn_to_page (*mig.src); +if (spage) { + ret =uv_page_in (kvm->arch.lpid, pfn << page_shift, + gpa,0 , page_shift); +if (ret) + goto out_finalize; + } + } *mig.dst =migrate_pfn (page_to_pfn (dpage)) |MIGRATE_PFN_LOCKED ;migrate_vma_pages (&mig);@@ -566,6 +596,66 @@ static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start,return ret; } +int kvmppc_uv_migrate_mem_slot (struct kvm *kvm, +const struct kvm_memory_slot *memslot) +{ +unsigned long gfn = memslot->base_gfn; +unsigned long end; +bool downgrade = false; +struct vm_area_struct *vma; +int i, ret =0 ; +unsigned long start =gfn_to_hva (kvm, gfn); + +if (kvm_is_error_hva (start)) +return H_STATE ; + + end = start + (memslot->npages <<PAGE_SHIFT ); + +down_write (&kvm->mm->mmap_sem); + +mutex_lock (&kvm->arch.uvmem_lock); + vma =find_vma_intersection (kvm->mm, start, end); +if (!vma || vma->vm_start > start || vma->vm_end < end) { + ret =H_STATE ; + goto out_unlock; + } + + ret =ksm_madvise (vma, vma->vm_start, vma->vm_end, +MADV_UNMERGEABLE , &vma->vm_flags); +downgrade_write (&kvm->mm->mmap_sem); + downgrade = true; +if (ret) { + ret =H_STATE ; + goto out_unlock; + } + +for (i =0 ; i < memslot->npages; i++, ++gfn) { +/* +* skip GFNs that have already tranistioned. +* paged-in GFNs, shared GFNs, paged-in GFNs
arch/powerpc/kvm/book3s_hv_uvmem.c static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, · line 645
Quoted 10 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +10
-0+* that were later paged-out. +*/ +if (kvmppc_gfn_has_transitioned (gfn, kvm)) +continue ; + + start =gfn_to_hva (kvm, gfn); + end = start + (1UL <<PAGE_SHIFT ); + ret =kvmppc_svm_migrate_page (vma, start, end, + (gfn <<PAGE_SHIFT ), kvm,PAGE_SHIFT , false); +
As I said last time, you are assuming that the vma that you obtained in the beginning actually spans the entire memslot range. This might be true as you haven't found any issues during testing, but I feel it is better if there is no such implicit assumption in the code here.
Regards,
Bharata.
Re: [PATCH v3 3/4] KVM: PPC: Book3S HV: migrate remaining normal-GFNs to secure-GFNs in H_SVM_INIT_D
Le 28/06/2020 à 18:20, Bharata B Rao a écrit :
Quoted 246 lines · Ram Pai
Quoted 1 line
H_SVM_INIT_DONEincorrectly assumes that the Ultravisor has explicitlyAs noted in the last iteration, can you reword the above please? I don't see it as an incorrect assumption, but see it as extension of scope now :-)
Quoted 142 lines
called
H_SVM_PAGE_INfor all secure pages. These GFNs continue to be normal GFNs associated with normal PFNs; when infact, these GFNs should have been secure GFNs, associated with device PFNs.Move all the PFNs associated with the SVM's GFNs, to secure-PFNs, in
H_SVM_INIT_DONE. Skip the GFNs that are already Paged-in or Shared throughH_SVM_PAGE_IN, or Paged-in followed by a Paged-out throughUV_PAGE_OUT.Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Cc Bharata B Rao <[email protected]>Cc Aneesh Kumar K.V <[email protected]>Cc Sukadev Bhattiprolu <[email protected]>Cc Laurent Dufour <[email protected]>Cc Thiago Jung Bauermann <[email protected]>Cc David Gibson <[email protected]>Cc Claudio Carvalho <[email protected]>Signed-off-by Ram Pai <[email protected]>--- Documentation/powerpc/ultravisor.rst | 2 + arch/powerpc/include/asm/kvm_book3s_uvmem.h | 2 + arch/powerpc/kvm/book3s_hv_uvmem.c | 154 +++++++++++++++++++++++----- 3 files changed, 132 insertions(+), 26 deletions(-)Documentation/powerpc/ultravisor.rst +2
-0@@ -933,6 +933,8 @@ Return values * H_UNSUPPORTED if called from the wrong context (e.g. from an SVM or before an H_SVM_INIT_START hypercall). + * H_STATE if the hypervisor could not successfully + transition the VM to Secure VM. Description ~~~~~~~~~~~arch/powerpc/include/asm/kvm_book3s_uvmem.h +2
-0@@ -22,6 +22,8 @@ unsigned long kvmppc_h_svm_page_out(struct kvm *kvm,unsigned long kvmppc_h_svm_init_abort (struct kvm *kvm);void kvmppc_uvmem_drop_pages (const struct kvm_memory_slot *free,struct kvm *kvm,bool skip_page_out); +int kvmppc_uv_migrate_mem_slot (struct kvm *kvm, +const struct kvm_memory_slot *memslot);#else static inline int kvmppc_uvmem_init (void ) {arch/powerpc/kvm/book3s_hv_uvmem.c +43
-6@@ -93,6 +93,7 @@#include <asm/ultravisor.h> #include <asm/mman.h> #include <asm/kvm_ppc.h> +#include <asm/kvm_book3s_uvmem.h> static struct dev_pagemap kvmppc_uvmem_pgmap;static unsigned long *kvmppc_uvmem_bitmap;@@ -339,6 +340,21 @@ static bool kvmppc_gfn_is_uvmem_pfn(unsigned long gfn, struct kvm *kvm,return false; } +/* return true, if the GFN is a shared-GFN, or a secure-GFN */ +bool kvmppc_gfn_has_transitioned (unsigned long gfn,struct kvm *kvm) +{ +struct kvmppc_uvmem_slot *p; + +list_for_each_entry (p, &kvm->arch.uvmem_pfns, list) { +if (gfn >= p->base_pfn && gfn < p->base_pfn + p->nr_pfns) { +unsigned long index = gfn - p->base_pfn; + +return (p->pfns[index] &KVMPPC_GFN_FLAG_MASK ); + } + } +return false; +} +unsigned long kvmppc_h_svm_init_start (struct kvm *kvm) {struct kvm_memslots *slots;@@ -379,12 +395,31 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)unsigned long kvmppc_h_svm_init_done (struct kvm *kvm) { +struct kvm_memslots *slots; +struct kvm_memory_slot *memslot; +int srcu_idx; +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, slots) { + ret =kvmppc_uv_migrate_mem_slot (kvm, memslot); +if (ret) { + ret =H_STATE ; + goto out; + } + } + kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_DONE ;pr_info ("LPID %d went secure\n" , kvm->arch.lpid); -return H_SUCCESS ; + +out: +srcu_read_unlock (&kvm->srcu, srcu_idx); +return ret; }/* @@ -505,12 +540,14 @@ static struct page *kvmppc_uvmem_get_page(unsigned long gpa, struct kvm *kvm) }/* -* Alloc a PFN from private device memory pool and copy page from normal -* memory to secure memory using UV_PAGE_IN uvcall. +* Alloc a PFN from private device memory pool. If @pagein is true, +* 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 gpa,struct kvm *kvm, -unsigned long page_shift,bool *downgrade) +static int kvmppc_svm_migrate_page (struct vm_area_struct *vma, +unsigned long start, +unsigned long end,unsigned long gpa,struct kvm *kvm, +unsigned long page_shift, +bool pagein) {unsigned long src_pfn, dst_pfn =0 ;struct migrate_vma mig;arch/powerpc/kvm/book3s_hv_uvmem.c static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, · line 533
Quoted 8 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-5@@ -526,18 +563,6 @@ static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, mig.src = &src_pfn; mig.dst = &dst_pfn; -/* -* We come here with mmap_sem write lock held just for -* ksm_madvise(), otherwise we only need read mmap_sem. -* Hence downgrade to read lock once ksm_madvise() is done. -*/ Can you please retain this comment? This explains why we take write lock and then downgrade.
arch/powerpc/kvm/book3s_hv_uvmem.c static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, · line 537
Quoted 4 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-4- ret =ksm_madvise (vma, vma->vm_start, vma->vm_end, -MADV_UNMERGEABLE , &vma->vm_flags); -downgrade_write (&kvm->mm->mmap_sem); - *downgrade = true;When I introduced this variable, there was a suggestion to rename it to "downgraded", but we were a bit late then. When you are touching this now, can you rename it appropriately?
Quoted 76 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +54
-8-if (ret) -return ret; - ret =migrate_vma_setup (&mig);if (ret)return ret;@@ -553,11 +578,16 @@ static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, 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); +if (pagein) { + pfn = *mig.src >>MIGRATE_PFN_SHIFT ; + spage =migrate_pfn_to_page (*mig.src); +if (spage) { + ret =uv_page_in (kvm->arch.lpid, pfn << page_shift, + gpa,0 , page_shift); +if (ret) + goto out_finalize; + } + } *mig.dst =migrate_pfn (page_to_pfn (dpage)) |MIGRATE_PFN_LOCKED ;migrate_vma_pages (&mig);@@ -566,6 +596,66 @@ static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start,return ret; } +int kvmppc_uv_migrate_mem_slot (struct kvm *kvm, +const struct kvm_memory_slot *memslot) +{ +unsigned long gfn = memslot->base_gfn; +unsigned long end; +bool downgrade = false; +struct vm_area_struct *vma; +int i, ret =0 ; +unsigned long start =gfn_to_hva (kvm, gfn); + +if (kvm_is_error_hva (start)) +return H_STATE ; + + end = start + (memslot->npages <<PAGE_SHIFT ); + +down_write (&kvm->mm->mmap_sem); + +mutex_lock (&kvm->arch.uvmem_lock); + vma =find_vma_intersection (kvm->mm, start, end); +if (!vma || vma->vm_start > start || vma->vm_end < end) { + ret =H_STATE ; + goto out_unlock; + } + + ret =ksm_madvise (vma, vma->vm_start, vma->vm_end, +MADV_UNMERGEABLE , &vma->vm_flags); +downgrade_write (&kvm->mm->mmap_sem); + downgrade = true; +if (ret) { + ret =H_STATE ; + goto out_unlock; + } + +for (i =0 ; i < memslot->npages; i++, ++gfn) { +/* +* skip GFNs that have already tranistioned. +* paged-in GFNs, shared GFNs, paged-in GFNs +* that were later paged-out. +*/ +if (kvmppc_gfn_has_transitioned (gfn, kvm)) +continue ; + + start =gfn_to_hva (kvm, gfn); + end = start + (1UL <<PAGE_SHIFT );
Quoted 8 lines
arch/powerpc/kvm/book3s_hv_uvmem.c static int kvmppc_svm_page_in(struct vm_area_struct *vma, unsigned long start, · line 645
Quoted 3 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +3
-0+ ret =kvmppc_svm_migrate_page (vma, start, end, + (gfn <<PAGE_SHIFT ), kvm,PAGE_SHIFT , false); +As I said last time, you are assuming that the vma that you obtained in the beginning actually spans the entire memslot range. This might be true as you haven't found any issues during testing, but I feel it is better if there is no such implicit assumption in the code here.
I agree that assumptions are sometimes not good for future work, but here the mmap_sem is held, and the VMA's boundaries have already been checked, so how could the VMA not spans over the memslot's range?
Am I missing something?
Cheers,
Laurent.
PATCH v3 4/4 KVM: PPC: Book3S HV: migrate hot plugged memory
In reply to Ram PaiFrom: Laurent Dufour <[email protected]>
When a memory slot is hot plugged to a SVM, PFNs associated with the GFNs in that slot must be migrated to the secure-PFNs, aka device-PFNs.
kvmppc_uv_migrate_mem_slot() is called to accomplish this. UV_PAGE_IN ucall is skipped, since the ultravisor does not trust the content of those pages and hence ignores it.
--- arch/powerpc/kvm/book3s_hv.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
arch/powerpc/kvm/book3s_hv.c +6 -4
@@ -4531,10 +4531,12 @@ static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,case KVM_MR_CREATE :if (kvmppc_uvmem_slot_init (kvm, new))return ; -uv_register_mem_slot (kvm->arch.lpid, - new->base_gfn <<PAGE_SHIFT , - new->npages *PAGE_SIZE , -0 , new->id); +if (uv_register_mem_slot (kvm->arch.lpid, + new->base_gfn <<PAGE_SHIFT , + new->npages *PAGE_SIZE , +0 , new->id)) +return ; +kvmppc_uv_migrate_mem_slot (kvm, new);break ;case KVM_MR_DELETE :uv_unregister_mem_slot (kvm->arch.lpid, old->id);
Signature · 1 line
1.8.3.1
Quoted 20 lines · Ram Pai
The time taken to switch a VM to Secure-VM, increases by the size of the VM. A 100GB VM takes about 7minutes. This is unacceptable. This linear increase is caused by a suboptimal behavior by the Ultravisor and the Hypervisor. The Ultravisor unnecessarily migrates all the GFN of the VM from normal-memory to secure-memory. It has to just migrate the necessary and sufficient GFNs.
However when the optimization is incorporated in the Ultravisor, the Hypervisor starts misbehaving. The Hypervisor has a inbuilt assumption that the Ultravisor will explicitly request to migrate, each and every GFN of the VM. If only necessary and sufficient GFNs are requested for migration, the Hypervisor continues to manage the remaining GFNs as normal GFNs. This leads of memory corruption, manifested consistently when the SVM reboots.
The same is true, when a memory slot is hotplugged into a SVM. The Hypervisor expects the ultravisor to request migration of all GFNs to secure-GFN. But at the same time, the hypervisor is unable to handle any
H_SVM_PAGE_INrequests from the Ultravisor, done in the context ofUV_REGISTER_MEM_SLOTucall. This problem manifests as random errors in the SVM, when a memory-slot is hotplugged.
Quoted 2 lines
This patch series automatically migrates the non-migrated pages of a SVM,
and thus solves the problem.
So this is what I understand as the objective of this patchset:
1. Getting all the pages into the secure memory right when the guest transitions into secure mode is expensive. Ultravisor wants to just get the necessary and sufficient pages in and put the onus on the Hypervisor to mark the remaining pages (w/o actual page-in) as secure during H_SVM_INIT_DONE.
2. During H_SVM_INIT_DONE, you want a way to differentiate the pages that are already secure from the pages that are shared and that are paged-out. For this you are introducing all these new states in HV.
UV knows about the shared GFNs and maintains the state of the same. Hence let HV send all the pages (minus already secured pages) via H_SVM_PAGE_IN and if UV finds any shared pages in them, let it fail the uv-page-in call. Then HV can fail the migration for it and the page continues to remain shared. With this, you don't need to maintain a state for secured GFN in HV.
In the unlikely case of sending a paged-out page to UV during H_SVM_INIT_DONE, let the page-in succeed and HV will fault on it again if required. With this, you don't need a state in HV to identify a paged-out-but-encrypted state.
Doesn't the above work? If so, we can avoid all those extra states in HV. That way HV can continue to differentiate only between two types of pages: secure and not-secure. The rest of the states (shared, paged-out-encrypted) actually belong to SVM/UV and let UV take care of them.
Or did I miss something?
Regards,
Bharata.
Quoted 41 lines · Bharata B Rao
Quoted 20 lines · Ram Pai
The time taken to switch a VM to Secure-VM, increases by the size of the VM. A 100GB VM takes about 7minutes. This is unacceptable. This linear increase is caused by a suboptimal behavior by the Ultravisor and the Hypervisor. The Ultravisor unnecessarily migrates all the GFN of the VM from normal-memory to secure-memory. It has to just migrate the necessary and sufficient GFNs.
However when the optimization is incorporated in the Ultravisor, the Hypervisor starts misbehaving. The Hypervisor has a inbuilt assumption that the Ultravisor will explicitly request to migrate, each and every GFN of the VM. If only necessary and sufficient GFNs are requested for migration, the Hypervisor continues to manage the remaining GFNs as normal GFNs. This leads of memory corruption, manifested consistently when the SVM reboots.
The same is true, when a memory slot is hotplugged into a SVM. The Hypervisor expects the ultravisor to request migration of all GFNs to secure-GFN. But at the same time, the hypervisor is unable to handle any
H_SVM_PAGE_INrequests from the Ultravisor, done in the context ofUV_REGISTER_MEM_SLOTucall. This problem manifests as random errors in the SVM, when a memory-slot is hotplugged.Quoted 2 lines
This patch series automatically migrates the non-migrated pages of a SVM,
and thus solves the problem.So this is what I understand as the objective of this patchset:
1. Getting all the pages into the secure memory right when the guest transitions into secure mode is expensive. Ultravisor wants to just get the necessary and sufficient pages in and put the onus on the Hypervisor to mark the remaining pages (w/o actual page-in) as secure during
H_SVM_INIT_DONE.
2. DuringH_SVM_INIT_DONE, you want a way to differentiate the pages that are already secure from the pages that are shared and that are paged-out. For this you are introducing all these new states in HV.UV knows about the shared GFNs and maintains the state of the same. Hence let HV send all the pages (minus already secured pages) via
H_SVM_PAGE_INand if UV finds any shared pages in them, let it fail the uv-page-in call. Then HV can fail the migration for it and the page continues to remain shared. With this, you don't need to maintain a state for secured GFN in HV.
Quoted 6 lines
In the unlikely case of sending a paged-out page to UV during
H_SVM_INIT_DONE, let the page-in succeed and HV will fault on it again if required. With this, you don't need a state in HV to identify a paged-out-but-encrypted state.Doesn't the above work?
I see that you want to infact skip the uv-page-in calls from H_SVM_INIT_DONE. So that would need the extra states in HV which you are proposing here.
Regards,
Bharata.
Quoted 47 lines · Bharata B Rao
Quoted 46 lines · Bharata B Rao
Quoted 20 lines · Ram Pai
The time taken to switch a VM to Secure-VM, increases by the size of the VM. A 100GB VM takes about 7minutes. This is unacceptable. This linear increase is caused by a suboptimal behavior by the Ultravisor and the Hypervisor. The Ultravisor unnecessarily migrates all the GFN of the VM from normal-memory to secure-memory. It has to just migrate the necessary and sufficient GFNs.
However when the optimization is incorporated in the Ultravisor, the Hypervisor starts misbehaving. The Hypervisor has a inbuilt assumption that the Ultravisor will explicitly request to migrate, each and every GFN of the VM. If only necessary and sufficient GFNs are requested for migration, the Hypervisor continues to manage the remaining GFNs as normal GFNs. This leads of memory corruption, manifested consistently when the SVM reboots.
The same is true, when a memory slot is hotplugged into a SVM. The Hypervisor expects the ultravisor to request migration of all GFNs to secure-GFN. But at the same time, the hypervisor is unable to handle any
H_SVM_PAGE_INrequests from the Ultravisor, done in the context ofUV_REGISTER_MEM_SLOTucall. This problem manifests as random errors in the SVM, when a memory-slot is hotplugged.Quoted 2 lines
This patch series automatically migrates the non-migrated pages of a SVM,
and thus solves the problem.So this is what I understand as the objective of this patchset:
1. Getting all the pages into the secure memory right when the guest transitions into secure mode is expensive. Ultravisor wants to just get the necessary and sufficient pages in and put the onus on the Hypervisor to mark the remaining pages (w/o actual page-in) as secure during
H_SVM_INIT_DONE.
2. DuringH_SVM_INIT_DONE, you want a way to differentiate the pages that are already secure from the pages that are shared and that are paged-out. For this you are introducing all these new states in HV.UV knows about the shared GFNs and maintains the state of the same. Hence let HV send all the pages (minus already secured pages) via
H_SVM_PAGE_INand if UV finds any shared pages in them, let it fail the uv-page-in call. Then HV can fail the migration for it and the page continues to remain shared. With this, you don't need to maintain a state for secured GFN in HV.In the unlikely case of sending a paged-out page to UV during
H_SVM_INIT_DONE, let the page-in succeed and HV will fault on it again if required. With this, you don't need a state in HV to identify a paged-out-but-encrypted state.
Quoted 4 lines
Quoted 1 line
Doesn't the above work?
I see that you want to infact skip the uv-page-in calls from
H_SVM_INIT_DONE. So that would need the extra states in HV which you are proposing here.
Yes. I want to skip to speed up the overall ESM switch.
RP