Re: [Xen-devel] [PATCH 2/2] xen/evtchn: optimize evtchn_security_struct

[Daniel De Graaf <dgdegra@xxxxxxxxxxxxx>]

On 03/20/2014 06:23 AM, Jan Beulich wrote:
> > > > On 19.03.14 at 20:18, Daniel De Graaf <dgdegra@xxxxxxxxxxxxx> wrote:
> > Because FLASK is the only implementation of the XSM hooks in Xen, some
> > of the abstractions required to handle multiple XSM providers are
> > redundant and only produce unneeded overhead.  This patch reduces the
> > memory overhead of enabling XSM on event channels by replacing the
> > untyped ssid pointer from struct evtchn with the sid field that it would
> > point to when FLASK is enabled.  This avoids an additional heap
> > allocation for every event channel, and on 64-bit systems, reduces the
> > size of struct evtchn by 4 bytes.
> Considering the comments on patch 1 I think it goes without
> saying that this is going too far, locking out any eventual other
> XSM provider.
>
> Jan
The existing code was already locking out other XSM providers from using
this particular field (it was #ifdef FLASK_ENABLE not #ifdef XSM_ENABLE),
probably because ACM did not use it. I will fix that issue in addition to
making the next version more capable of supporting other XSM providers
that wish to mark event channels.

--
Daniel De Graaf
National Security Agency

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