Xen devel - Dec 2010 - [PATCH] xen : Replace hard coded domain_id checks with a macro

Replace a number of checks for Dom0, that have been hard coded to check for
domain_id being zero with a macro is_dom0_domain().

Signed-off by: Mihir Nanavati <mihirn@cs.ubc.ca>



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Fri, 2010-12-10 at 07:07 +0000, Mihir Nanavati wrote:
> Replace a number of checks for Dom0, that have been hard coded to
> check for domain_id being zero with a macro is_dom0_domain().
Is the intention for this macro return true for some domid != 0 under
some future circumstance? In that case the macro name will turn out to
be badly chosen.

I''m not sure there is any benefit to hard coding a 0 in the function
name as opposed to hardcoding at the call site. I suppose it''s a little
easier to search and replace...

Is there a name which describes the actual semantics which the callers
want, as opposed to testing the dom0-ness? Or perhaps there is more than
one desired semantic, in which case multiple predicates would be ok
IMHO. Does the existing IS_PRIV cover some of the cases?

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Yes, the idea is to later have it, or another similar macro return true for
domids != 0. At the moment, I think it''s likely that there will be
other
separate predicates (maybe something like is_xenstore_domain,
is_control_domain, etc) for different disaggregated domains, and then have
the last bit continue to use this, even though it may no longer be domid 0.

You''re right about the name being ill-chosen, but the only other name I
could come up with was is_what_used_to_be_dom0 which was even worse ;)
I''m
open to suggestions. Perhaps, hardware domain or pci domain?

At the moment, IS_PRIV could be used, but it would lead to a coupling of the
privileges with functionality which could be problematic later on.

~M

On Fri, Dec 10, 2010 at 1:08 AM, Ian Campbell
<Ian.Campbell@citrix.com>wrote:
> On Fri, 2010-12-10 at 07:07 +0000, Mihir Nanavati wrote:
> > Replace a number of checks for Dom0, that have been hard coded to
> > check for domain_id being zero with a macro is_dom0_domain().
>
> Is the intention for this macro return true for some domid != 0 under
> some future circumstance? In that case the macro name will turn out to
> be badly chosen.
>
> I''m not sure there is any benefit to hard coding a 0 in the
function
> name as opposed to hardcoding at the call site. I suppose it''s a
little
> easier to search and replace...
>
> Is there a name which describes the actual semantics which the callers
> want, as opposed to testing the dom0-ness? Or perhaps there is more than
> one desired semantic, in which case multiple predicates would be ok
> IMHO. Does the existing IS_PRIV cover some of the cases?
>
> Ian.
>
>

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

I''m undecided. The patch by itself is kind of harmless but also kind of
pointless. Probably we should leave this until you have something more
substantial to propose. Trickling in trivial patches like this is a waste of
time.

 -- Keir

On 10/12/2010 10:13, "Mihir Nanavati" <mihirn@cs.ubc.ca> wrote:
> Yes, the idea is to later have it, or another similar macro return true for
> domids != 0. At the moment, I think it''s likely that there will be
other
> separate predicates (maybe something like is_xenstore_domain,
> is_control_domain, etc) for different disaggregated domains, and then have
the
> last bit continue to use this, even though it may no longer be domid 0.
> 
> You''re right about the name being ill-chosen, but the only other
name I could
> come up with was is_what_used_to_be_dom0 which was even worse ;)
I''m open to
> suggestions. Perhaps, hardware domain or pci domain?
> 
> At the moment, IS_PRIV could be used, but it would lead to a coupling of
the
> privileges with functionality which could be problematic later on.
> 
> ~M
> 
> On Fri, Dec 10, 2010 at 1:08 AM, Ian Campbell
<Ian.Campbell@citrix.com> wrote:
>> On Fri, 2010-12-10 at 07:07 +0000, Mihir Nanavati wrote:
>>> Replace a number of checks for Dom0, that have been hard coded to
>>> check for domain_id being zero with a macro is_dom0_domain().
>> 
>> Is the intention for this macro return true for some domid != 0 under
>> some future circumstance? In that case the macro name will turn out to
>> be badly chosen.
>> 
>> I''m not sure there is any benefit to hard coding a 0 in the
function
>> name as opposed to hardcoding at the call site. I suppose it''s
a little
>> easier to search and replace...
>> 
>> Is there a name which describes the actual semantics which the callers
>> want, as opposed to testing the dom0-ness? Or perhaps there is more
than
>> one desired semantic, in which case multiple predicates would be ok
>> IMHO. Does the existing IS_PRIV cover some of the cases?
>> 
>> Ian.
>> 
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Fri, 2010-12-10 at 10:13 +0000, Mihir Nanavati wrote:
> Yes, the idea is to later have it, or another similar macro return
> true for domids != 0. At the moment, I think it''s likely that
there
> will be other separate predicates (maybe something like
> is_xenstore_domain, is_control_domain, etc) for different
> disaggregated domains, and then have the last bit continue to use
> this, even though it may no longer be domid 0.
> 
> You''re right about the name being ill-chosen, but the only other
name
> I could come up with was is_what_used_to_be_dom0 which was even
> worse ;) I''m open to suggestions. Perhaps, hardware domain or pci
> domain?
I don''t have any better suggestions :-(

I think it very much depends on what the call site is actually trying to
test and how it will be disaggregated in the future. If you are just
going to have to switch to another macro when you do the disaggregation
what''s the benefit of switching now?
> At the moment, IS_PRIV could be used, but it would lead to a coupling
> of the privileges with functionality which could be problematic later
> on.
> 
> ~M
> 
> On Fri, Dec 10, 2010 at 1:08 AM, Ian Campbell
> <Ian.Campbell@citrix.com> wrote:
>         On Fri, 2010-12-10 at 07:07 +0000, Mihir Nanavati wrote:
>         > Replace a number of checks for Dom0, that have been hard
>         coded to
>         > check for domain_id being zero with a macro
>         is_dom0_domain().
>         
>         
>         Is the intention for this macro return true for some domid !>   
0 under
>         some future circumstance? In that case the macro name will
>         turn out to
>         be badly chosen.
>         
>         I''m not sure there is any benefit to hard coding a 0 in
the
>         function
>         name as opposed to hardcoding at the call site. I suppose
it''s
>         a little
>         easier to search and replace...
>         
>         Is there a name which describes the actual semantics which the
>         callers
>         want, as opposed to testing the dom0-ness? Or perhaps there is
>         more than
>         one desired semantic, in which case multiple predicates would
>         be ok
>         IMHO. Does the existing IS_PRIV cover some of the cases?
>         
>         Ian.
>         
> 


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Fair enough, I guess it''s more useful for us here in testing to be able
to
switch to having another domid as dom0 from a single place then it would be
in a production system. Will keep it on hold till we''ve gotten some
more
pieces into place.

~M

On Fri, Dec 10, 2010 at 2:42 AM, Keir Fraser <keir@xen.org> wrote:
> I''m undecided. The patch by itself is kind of harmless but also
kind of
> pointless. Probably we should leave this until you have something more
> substantial to propose. Trickling in trivial patches like this is a waste
> of
> time.
>
>  -- Keir
>
> On 10/12/2010 10:13, "Mihir Nanavati" <mihirn@cs.ubc.ca>
wrote:
>
> > Yes, the idea is to later have it, or another similar macro return
true
> for
> > domids != 0. At the moment, I think it''s likely that there
will be other
> > separate predicates (maybe something like is_xenstore_domain,
> > is_control_domain, etc) for different disaggregated domains, and then
> have the
> > last bit continue to use this, even though it may no longer be domid
0.
> >
> > You''re right about the name being ill-chosen, but the only
other name I
> could
> > come up with was is_what_used_to_be_dom0 which was even worse ;)
I''m open
> to
> > suggestions. Perhaps, hardware domain or pci domain?
> >
> > At the moment, IS_PRIV could be used, but it would lead to a coupling
of
> the
> > privileges with functionality which could be problematic later on.
> >
> > ~M
> >
> > On Fri, Dec 10, 2010 at 1:08 AM, Ian Campbell
<Ian.Campbell@citrix.com>
> wrote:
> >> On Fri, 2010-12-10 at 07:07 +0000, Mihir Nanavati wrote:
> >>> Replace a number of checks for Dom0, that have been hard coded
to
> >>> check for domain_id being zero with a macro is_dom0_domain().
> >>
> >> Is the intention for this macro return true for some domid != 0
under
> >> some future circumstance? In that case the macro name will turn
out to
> >> be badly chosen.
> >>
> >> I''m not sure there is any benefit to hard coding a 0 in
the function
> >> name as opposed to hardcoding at the call site. I suppose
it''s a little
> >> easier to search and replace...
> >>
> >> Is there a name which describes the actual semantics which the
callers
> >> want, as opposed to testing the dom0-ness? Or perhaps there is
more than
> >> one desired semantic, in which case multiple predicates would be
ok
> >> IMHO. Does the existing IS_PRIV cover some of the cases?
> >>
> >> Ian.
> >>
> >
> >
> >
> > _______________________________________________
> > Xen-devel mailing list
> > Xen-devel@lists.xensource.com
> > http://lists.xensource.com/xen-devel
>
>
>

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Yeah, that''s the pain of an out-of-tree patch queue I''m
afraid. You have to
suck it up. :-)

On 10/12/2010 11:06, "Mihir Nanavati" <mihirn@cs.ubc.ca> wrote:
> Fair enough, I guess it''s more useful for us here in testing to be
able to
> switch to having another domid as dom0 from a single place then it would be
in
> a production system. Will keep it on hold till we''ve gotten some
more pieces
> into place.
> 
> ~M
> 
> On Fri, Dec 10, 2010 at 2:42 AM, Keir Fraser <keir@xen.org> wrote:
>> I''m undecided. The patch by itself is kind of harmless but
also kind of
>> pointless. Probably we should leave this until you have something more
>> substantial to propose. Trickling in trivial patches like this is a
waste of
>> time.
>> 
>>  -- Keir
>> 
>> On 10/12/2010 10:13, "Mihir Nanavati"
<mihirn@cs.ubc.ca> wrote:
>> 
>>> Yes, the idea is to later have it, or another similar macro return
true for
>>> domids != 0. At the moment, I think it''s likely that there
will be other
>>> separate predicates (maybe something like is_xenstore_domain,
>>> is_control_domain, etc) for different disaggregated domains, and
then have
>>> the
>>> last bit continue to use this, even though it may no longer be
domid 0.
>>> 
>>> You''re right about the name being ill-chosen, but the only
other name I
>>> could
>>> come up with was is_what_used_to_be_dom0 which was even worse ;)
I''m open to
>>> suggestions. Perhaps, hardware domain or pci domain?
>>> 
>>> At the moment, IS_PRIV could be used, but it would lead to a
coupling of the
>>> privileges with functionality which could be problematic later on.
>>> 
>>> ~M
>>> 
>>> On Fri, Dec 10, 2010 at 1:08 AM, Ian Campbell
<Ian.Campbell@citrix.com>
>>> wrote:
>>>> On Fri, 2010-12-10 at 07:07 +0000, Mihir Nanavati wrote:
>>>>> Replace a number of checks for Dom0, that have been hard
coded to
>>>>> check for domain_id being zero with a macro
is_dom0_domain().
>>>> 
>>>> Is the intention for this macro return true for some domid != 0
under
>>>> some future circumstance? In that case the macro name will turn
out to
>>>> be badly chosen.
>>>> 
>>>> I''m not sure there is any benefit to hard coding a 0
in the function
>>>> name as opposed to hardcoding at the call site. I suppose
it''s a little
>>>> easier to search and replace...
>>>> 
>>>> Is there a name which describes the actual semantics which the
callers
>>>> want, as opposed to testing the dom0-ness? Or perhaps there is
more than
>>>> one desired semantic, in which case multiple predicates would
be ok
>>>> IMHO. Does the existing IS_PRIV cover some of the cases?
>>>> 
>>>> Ian.
>>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> Xen-devel mailing list
>>> Xen-devel@lists.xensource.com
>>> http://lists.xensource.com/xen-devel
>> 
>> 
> 
> 


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel