Re: [Xen-devel] [PATCH v3] xsm: add a default policy to .init.data

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 30.06.16 at 17:13, <konrad.wilk@xxxxxxxxxx> wrote:
> On Thu, Jun 30, 2016 at 10:01:18AM -0400, Daniel De Graaf wrote:
>> On 06/30/2016 09:45 AM, Konrad Rzeszutek Wilk wrote:
>> > On Wed, Jun 29, 2016 at 11:09:01AM -0400, Daniel De Graaf wrote:
>> > > --- a/xen/xsm/xsm_core.c
>> > > +++ b/xen/xsm/xsm_core.c
>> > > @@ -36,6 +36,24 @@ static inline int verify(struct xsm_operations *ops)
>> > >      return 0;
>> > >  }
>> > > 
>> > > +#ifdef CONFIG_XSM_POLICY
>> > > +extern char xsm_init_policy[];
>> > 
>> > > +extern int xsm_init_policy_size;
>> > > +#else
>> > > +#define xsm_init_policy 0
>> > > +#endif
>> > > +
>> > > +static void __init xsm_policy_init(void)
>> > > +{
>> > > +#ifdef CONFIG_XSM_POLICY
>> > > +    if ( policy_size == 0 )
>> > > +    {
>> > > +        policy_buffer = xsm_init_policy;
>> > > +        policy_size = xsm_init_policy_size;
>> > > +    }
>> > > +#endif
>> > > +}
>> > > +
>> > 
>> > This all looks like it could go in a header file?
>> 
>> Sure, I could move it to xsm.h.  I thought it might be clearer to have the
>> assignment and the xfree() check in the same file.
> 
> You are the maintainer of that code, so it is your call.
> 
> Having externs in C code is not that great (from my perspective),
> so if you move all of this to a header file that should get easier?

I agree, fwiw, for the extern-s to be moved to a header. That
header should then also be included by the generated source
file producing the symbols. I don't, however, see why the
function should go to a header.

Jan


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