[v10,0/8] KVM: allow mapping non-refcounted pages

Message

David Stevens Feb. 21, 2024, 7:25 a.m. UTC

From: David Stevens <[e-mail]>

This patch series adds support for mapping VM_IO and VM_PFNMAP memory
that is backed by struct pages that aren't currently being refcounted
(e.g. tail pages of non-compound higher order allocations) into the
guest.

Our use case is virtio-gpu blob resources [1], which directly map host
graphics buffers into the guest as "vram" for the virtio-gpu device.
This feature currently does not work on systems using the amdgpu driver,
as that driver allocates non-compound higher order pages via
ttm_pool_alloc_page().

First, this series replaces the gfn_to_pfn_memslot() API with a more
extensible kvm_follow_pfn() API. The updated API rearranges
gfn_to_pfn_memslot()'s args into a struct and where possible packs the
bool arguments into a FOLL_ flags argument. The refactoring changes do
not change any behavior.

From there, this series extends the kvm_follow_pfn() API so that
non-refconuted pages can be safely handled. This invloves adding an
input parameter to indicate whether the caller can safely use
non-refcounted pfns and an output parameter to tell the caller whether
or not the returned page is refcounted. This change includes a breaking
change, by disallowing non-refcounted pfn mappings by default, as such
mappings are unsafe. To allow such systems to continue to function, an
opt-in module parameter is added to allow the unsafe behavior.

This series only adds support for non-refcounted pages to x86. Other
MMUs can likely be updated without too much difficulty, but it is not
needed at this point. Updating other parts of KVM (e.g. pfncache) is not
straightforward [2].

[1]
https://patchwork.kernel.org/project/dri-devel/cover/[e-mail]/
[2] https://lore.kernel.org/all/[e-mail]/

v9 -> v10:
 - Re-add FOLL_GET changes.
 - Split x86/mmu spte+non-refcount-page patch into two patches.
 - Rename 'foll' variables to 'kfp'.
 - Properly gate usage of refcount spte bit when it's not available.
 - Replace kfm_follow_pfn's is_refcounted_page output parameter with
   a struct page *refcounted_page pointing to the page in question.
 - Add patch downgrading BUG_ON to WARN_ON_ONCE.
v8 -> v9:
 - Make paying attention to is_refcounted_page mandatory. This means
   that FOLL_GET is no longer necessary. For compatibility with
   un-migrated callers, add a temporary parameter to sidestep
   ref-counting issues.
 - Add allow_unsafe_mappings, which is a breaking change.
 - Migrate kvm_vcpu_map and other callsites used by x86 to the new API.
 - Drop arm and ppc changes.
v7 -> v8:
 - Set access bits before releasing mmu_lock.
 - Pass FOLL_GET on 32-bit x86 or !tdp_enabled.
 - Refactor FOLL_GET handling, add kvm_follow_refcounted_pfn helper.
 - Set refcounted bit on >4k pages.
 - Add comments and apply formatting suggestions.
 - rebase on kvm next branch.
v6 -> v7:
 - Replace __gfn_to_pfn_memslot with a more flexible __kvm_faultin_pfn,
   and extend that API to support non-refcounted pages (complete
   rewrite).

David Stevens (7):
  KVM: Relax BUG_ON argument validation
  KVM: mmu: Introduce kvm_follow_pfn()
  KVM: mmu: Improve handling of non-refcounted pfns
  KVM: Migrate kvm_vcpu_map() to kvm_follow_pfn()
  KVM: x86: Migrate to kvm_follow_pfn()
  KVM: x86/mmu: Track if sptes refer to refcounted pages
  KVM: x86/mmu: Handle non-refcounted pages

Sean Christopherson (1):
  KVM: Assert that a page's refcount is elevated when marking
    accessed/dirty

 arch/x86/kvm/mmu/mmu.c          | 104 +++++++---
 arch/x86/kvm/mmu/mmu_internal.h |   2 +
 arch/x86/kvm/mmu/paging_tmpl.h  |   7 +-
 arch/x86/kvm/mmu/spte.c         |   4 +-
 arch/x86/kvm/mmu/spte.h         |  22 +-
 arch/x86/kvm/mmu/tdp_mmu.c      |  22 +-
 arch/x86/kvm/x86.c              |  11 +-
 include/linux/kvm_host.h        |  53 ++++-
 virt/kvm/guest_memfd.c          |   8 +-
 virt/kvm/kvm_main.c             | 349 +++++++++++++++++++-------------
 virt/kvm/kvm_mm.h               |   3 +-
 virt/kvm/pfncache.c             |  11 +-
 12 files changed, 399 insertions(+), 197 deletions(-)


base-commit: 54be6c6c5ae8e0d93a6c4641cb7528eb0b6ba478

Comments

Paolo Bonzini Feb. 23, 2024, 9:44 a.m. UTC | #1

On 2/21/24 08:25, David Stevens wrote:
> From: David Stevens <[e-mail]>
> 
> This patch series adds support for mapping VM_IO and VM_PFNMAP memory
> that is backed by struct pages that aren't currently being refcounted
> (e.g. tail pages of non-compound higher order allocations) into the
> guest.
> 
> Our use case is virtio-gpu blob resources [1], which directly map host
> graphics buffers into the guest as "vram" for the virtio-gpu device.
> This feature currently does not work on systems using the amdgpu driver,
> as that driver allocates non-compound higher order pages via
> ttm_pool_alloc_page().
> 
> First, this series replaces the gfn_to_pfn_memslot() API with a more
> extensible kvm_follow_pfn() API. The updated API rearranges
> gfn_to_pfn_memslot()'s args into a struct and where possible packs the
> bool arguments into a FOLL_ flags argument. The refactoring changes do
> not change any behavior.
> 
>  From there, this series extends the kvm_follow_pfn() API so that
> non-refconuted pages can be safely handled. This invloves adding an
> input parameter to indicate whether the caller can safely use
> non-refcounted pfns and an output parameter to tell the caller whether
> or not the returned page is refcounted. This change includes a breaking
> change, by disallowing non-refcounted pfn mappings by default, as such
> mappings are unsafe. To allow such systems to continue to function, an
> opt-in module parameter is added to allow the unsafe behavior.
> 
> This series only adds support for non-refcounted pages to x86. Other
> MMUs can likely be updated without too much difficulty, but it is not
> needed at this point. Updating other parts of KVM (e.g. pfncache) is not
> straightforward [2].

Looks good to me, apart that two patches were sent twice.  I only have a 
small comment on patch 4, to which I'll reply separately.

Paolo

> [1]
> https://patchwork.kernel.org/project/dri-devel/cover/[e-mail]/
> [2] https://lore.kernel.org/all/[e-mail]/
> 
> v9 -> v10:
>   - Re-add FOLL_GET changes.
>   - Split x86/mmu spte+non-refcount-page patch into two patches.
>   - Rename 'foll' variables to 'kfp'.
>   - Properly gate usage of refcount spte bit when it's not available.
>   - Replace kfm_follow_pfn's is_refcounted_page output parameter with
>     a struct page *refcounted_page pointing to the page in question.
>   - Add patch downgrading BUG_ON to WARN_ON_ONCE.
> v8 -> v9:
>   - Make paying attention to is_refcounted_page mandatory. This means
>     that FOLL_GET is no longer necessary. For compatibility with
>     un-migrated callers, add a temporary parameter to sidestep
>     ref-counting issues.
>   - Add allow_unsafe_mappings, which is a breaking change.
>   - Migrate kvm_vcpu_map and other callsites used by x86 to the new API.
>   - Drop arm and ppc changes.
> v7 -> v8:
>   - Set access bits before releasing mmu_lock.
>   - Pass FOLL_GET on 32-bit x86 or !tdp_enabled.
>   - Refactor FOLL_GET handling, add kvm_follow_refcounted_pfn helper.
>   - Set refcounted bit on >4k pages.
>   - Add comments and apply formatting suggestions.
>   - rebase on kvm next branch.
> v6 -> v7:
>   - Replace __gfn_to_pfn_memslot with a more flexible __kvm_faultin_pfn,
>     and extend that API to support non-refcounted pages (complete
>     rewrite).
> 
> David Stevens (7):
>    KVM: Relax BUG_ON argument validation
>    KVM: mmu: Introduce kvm_follow_pfn()
>    KVM: mmu: Improve handling of non-refcounted pfns
>    KVM: Migrate kvm_vcpu_map() to kvm_follow_pfn()
>    KVM: x86: Migrate to kvm_follow_pfn()
>    KVM: x86/mmu: Track if sptes refer to refcounted pages
>    KVM: x86/mmu: Handle non-refcounted pages
> 
> Sean Christopherson (1):
>    KVM: Assert that a page's refcount is elevated when marking
>      accessed/dirty
> 
>   arch/x86/kvm/mmu/mmu.c          | 104 +++++++---
>   arch/x86/kvm/mmu/mmu_internal.h |   2 +
>   arch/x86/kvm/mmu/paging_tmpl.h  |   7 +-
>   arch/x86/kvm/mmu/spte.c         |   4 +-
>   arch/x86/kvm/mmu/spte.h         |  22 +-
>   arch/x86/kvm/mmu/tdp_mmu.c      |  22 +-
>   arch/x86/kvm/x86.c              |  11 +-
>   include/linux/kvm_host.h        |  53 ++++-
>   virt/kvm/guest_memfd.c          |   8 +-
>   virt/kvm/kvm_main.c             | 349 +++++++++++++++++++-------------
>   virt/kvm/kvm_mm.h               |   3 +-
>   virt/kvm/pfncache.c             |  11 +-
>   12 files changed, 399 insertions(+), 197 deletions(-)
> 
> 
> base-commit: 54be6c6c5ae8e0d93a6c4641cb7528eb0b6ba478