Re: [Xen-devel] [PATCH v3 0/11] Rename/remove IS_PRIV

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 12.04.13 at 23:04, Daniel De Graaf <dgdegra@xxxxxxxxxxxxx> wrote:
> Changes since v2:
>  - Handle XEN_SYSCTL_CPUPOOL_OP_MOVEDOMAIN separately
>  - Use is_control_domain for CPUID
>  - Use is_{control,hardware}_domain for TSC
>  - MAINTAINERS patch included in series
> 
> ---
> 
> Following the conversion of most IS_PRIV hooks to XSM, the remaining
> references to this function generally deal with direct hardware access
> and not with the type of privilege checks that are best controlled by
> XSM. To reflect this, the IS_PRIV check is renamed to is_hardware_domain
> and is used only when dealing with accesses that are both required by
> dom0 and where it does not make sense to grant access to a domain other
> than dom0.
> 
> There are a number of existing places in the hypervisor that check
> domain_id for equality to zero to make some distinction on dom0; this
> series replaces these checks with is_hardware_domain to be consistent in
> how the hypervisor checks a domain's access.
> 
> Independent changes related to this series:
>       [PATCH 01/11] MAINTAINERS: Add myself as XSM maintainer
>       [PATCH 08/11] xen/cpupool: prevent a domain from moving itself
> 
> Cleanup of IS_PRIV checks that should not be is_hardware_domain:
>       [PATCH 02/11] xen/arch/x86: remove IS_PRIV access check bypasses
>       [PATCH 03/11] xen/xsm: add hooks for claim
>       [PATCH 04/11] hvm: convert access check for nested HVM to XSM
>       [PATCH 05/11] xen/arch/x86: remove IS_PRIV_FOR references
>       [PATCH 06/11] xen/arch/arm: remove rcu_lock_target_domain_by_id
> 
> Replace remaining calls to IS_PRIV:
>       [PATCH 07/11] xen: rename IS_PRIV to is_hardware_domain
> 
> Use is_hardware_domain locations where (domid == 0) was used:
>       [PATCH 09/11] xen: use domid check in is_hardware_domain
>       [PATCH 10/11] xen/arch/x86: clarify domid == 0 checks
>       [PATCH 11/11] IOMMU: use is_hardware_domain instead of domid == 0

While patch 1 went in a few days ago, patch 2 was held up just by
XSA-46, which went public today. Consequently I also committed
patch 2 a few minutes ago.

For the rest of the series, however, I would want you two to work
out the release related aspects, and I'd look into committing parts
that I'm permitted to commit once I saw George's ack.

Patch 10 may, according to the reply I just sent, need another small
tweak before getting applied - as long as you agree, I could certainly
do this while committing.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel