[PATCH v3 01/16] KVM: Take vcpu->mutex outside vcpu_load

[Cornelia Huck]

On Mon,  4 Dec 2017 21:35:23 +0100
Christoffer Dall <cdall at kernel.org> wrote:

> From: Christoffer Dall <christoffer.dall at linaro.org>
> 
> As we're about to call vcpu_load() from architecture-specific
> implementations of the KVM vcpu ioctls, but yet we access data
> structures protected by the vcpu->mutex in the generic code, factor
> this logic out from vcpu_load().
> 
> x86 is the only architecture which calls vcpu_load() outside of the main
> vcpu ioctl function, and these calls will no longer take the vcpu mutex
> following this patch.  However, with the exception of
> kvm_arch_vcpu_postcreate (see below), the callers are either in the
> creation or destruction path of the VCPU, which means there cannot be
> any concurrent access to the data structure, because the file descriptor
> is not yet accessible, or is already gone.
> 
> kvm_arch_vcpu_postcreate makes the newly created vcpu potentially
> accessible by other in-kernel threads through the kvm->vcpus array, and
> we therefore take the vcpu mutex in this case directly.
> 
> Signed-off-by: Christoffer Dall <christoffer.dall at linaro.org>
> ---
>  arch/x86/kvm/vmx.c       |  4 +---
>  arch/x86/kvm/x86.c       | 20 +++++++-------------
>  include/linux/kvm_host.h |  2 +-
>  virt/kvm/kvm_main.c      | 17 ++++++-----------
>  4 files changed, 15 insertions(+), 28 deletions(-)

Reviewed-by: Cornelia Huck <cohuck at redhat.com>