Re: [PATCH v3 4/6] KVM: Documentation: Add the missing description for tdp_mmu_root_count into kvm_mmu_page

[Sean Christopherson]

On Tue, Aug 01, 2023, Mingwei Zhang wrote:
> Add the description of tdp_mmu_root_count into kvm_mmu_page description and
> combine it with the description of root_count. tdp_mmu_root_count is an
> atomic counter used only in TDP MMU. Update the doc.
> 
> Signed-off-by: Mingwei Zhang <mizhang@xxxxxxxxxx>
> Reviewed-by: Kai Huang <kai.huang@xxxxxxxxx>
> ---
>  Documentation/virt/kvm/x86/mmu.rst | 12 ++++++++----
>  1 file changed, 8 insertions(+), 4 deletions(-)
> 
> diff --git a/Documentation/virt/kvm/x86/mmu.rst b/Documentation/virt/kvm/x86/mmu.rst
> index 17d90974204e..40daf8beb9b1 100644
> --- a/Documentation/virt/kvm/x86/mmu.rst
> +++ b/Documentation/virt/kvm/x86/mmu.rst
> @@ -229,10 +229,14 @@ Shadow pages contain the following information:
>      can be calculated from the gfn field when used.  In addition, when
>      role.direct is set, KVM does not track access permission for each of the
>      gfn. See role.direct and gfn.
> -  root_count:
> -    A counter keeping track of how many hardware registers (guest cr3 or
> -    pdptrs) are now pointing at the page.  While this counter is nonzero, the
> -    page cannot be destroyed.  See role.invalid.
> +  root_count / tdp_mmu_root_count:
> +     root_count is a reference counter for root shadow pages in Shadow MMU.
> +     vCPUs elevate the refcount when getting a shadow page that will be used as
> +     a root page, i.e. page that will be loaded into hardware directly (CR3,
> +     PDPTRs, nCR3 EPTP). Root pages cannot be destroyed while their refcount is
> +     non-zero. See role.invalid. tdp_mmu_root_count is similar but exclusively
> +     used in TDP MMU as an atomic refcount. When the value is non-zero, it
> +     allows vCPUs acquire references while holding mmu_lock for read.

That last sentence is wrong.  *vCPUs* can't acquire references while holding
mmu_lock for read.  And actually, they don't ever put references while holding
for read either.  vCPUs *must* hold mmu_lock for write to obtain a new root,
Not putting references while holding mmu_lock for read is mostly an implementation
quirk.

Maybe replace it with this?

    tdp_mmu_root_count is similar but exclusively used in the TDP MMU as an
    atomic refcount (select TDP MMU flows walk all roots while holding mmu_lock
    for read, e.g. when clearing dirty bits).