* semantics of GETDOMAININFO
@ 2005-05-08 21:39 Kip Macy
2005-05-08 22:10 ` Keir Fraser
0 siblings, 1 reply; 3+ messages in thread
From: Kip Macy @ 2005-05-08 21:39 UTC (permalink / raw)
To: xen-devel
One of the things that I've always thought was weird, but didn't pay
too close attention to is the fact that GETDOMAININFO will return a
valid result even if we give it a domid that is no longer valid.
Looking at the code we get back the first valid domain after the domid
we pass in.
What is the reason for this design choice? When I request the
attributes of a process I don't get the attributes for the next pid up
with a pid field set to the process id of the process I actually got
the attributes for.
-Kip
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: semantics of GETDOMAININFO
2005-05-08 21:39 semantics of GETDOMAININFO Kip Macy
@ 2005-05-08 22:10 ` Keir Fraser
2005-05-08 23:17 ` Kip Macy
0 siblings, 1 reply; 3+ messages in thread
From: Keir Fraser @ 2005-05-08 22:10 UTC (permalink / raw)
To: Kip Macy; +Cc: xen-devel
On 8 May 2005, at 22:39, Kip Macy wrote:
> One of the things that I've always thought was weird, but didn't pay
> too close attention to is the fact that GETDOMAININFO will return a
> valid result even if we give it a domid that is no longer valid.
> Looking at the code we get back the first valid domain after the domid
> we pass in.
>
> What is the reason for this design choice? When I request the
> attributes of a process I don't get the attributes for the next pid up
> with a pid field set to the process id of the process I actually got
> the attributes for.
Easy enumeration when we don't know which domain id's are in use. There
are other ways to allow for enumeration that might be neater.
-- Keir
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: semantics of GETDOMAININFO
2005-05-08 22:10 ` Keir Fraser
@ 2005-05-08 23:17 ` Kip Macy
0 siblings, 0 replies; 3+ messages in thread
From: Kip Macy @ 2005-05-08 23:17 UTC (permalink / raw)
To: Keir Fraser; +Cc: xen-devel
That seems reasonable - but I find its lack of uniformity a bit grating.
Maybe instead of having the current domain be first fit, we could add
a "next_domid" field. That way when "xm list" and the like want to
enumerate, provided they start zero, they can keep querying until
next_domid == -1. I can't think of a case when one wouldn't enumerate
the domains starting at zero. So, at least superficially, it seems
like a reasonable alternative.
I'm willing to accept the possiblity that I'm just nitpicking so I
won't push it.
-Kip
On 5/8/05, Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote:
>
> On 8 May 2005, at 22:39, Kip Macy wrote:
>
> > One of the things that I've always thought was weird, but didn't pay
> > too close attention to is the fact that GETDOMAININFO will return a
> > valid result even if we give it a domid that is no longer valid.
> > Looking at the code we get back the first valid domain after the domid
> > we pass in.
> >
> > What is the reason for this design choice? When I request the
> > attributes of a process I don't get the attributes for the next pid up
> > with a pid field set to the process id of the process I actually got
> > the attributes for.
>
> Easy enumeration when we don't know which domain id's are in use. There
> are other ways to allow for enumeration that might be neater.
>
> -- Keir
>
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2005-05-08 23:17 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-05-08 21:39 semantics of GETDOMAININFO Kip Macy
2005-05-08 22:10 ` Keir Fraser
2005-05-08 23:17 ` Kip Macy
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.