Re: Xen 4.18 release: Reminder about code freeze

[Juergen Gross <jgross@xxxxxxxx>]

On 13.10.23 16:51, Julien Grall wrote:
> Hi Juergen,
>
> On 13/10/2023 15:47, Juergen Gross wrote:
> > On 13.10.23 13:22, Julien Grall wrote:
> > > Hi George,
> > >
> > > On 13/10/2023 11:16, George Dunlap wrote:
> > > > On Thu, Oct 12, 2023 at 11:36 PM Stefano Stabellini
> > > > <sstabellini@xxxxxxxxxx> wrote:
> > > > > On Thu, 12 Oct 2023, George Dunlap wrote:
> > > > > > > > Stop tinkering in the hope that it hides the problem.  You're only
> > > > > > > > making it harder to fix properly.
> > > > > > > Making it harder to fix properly would be a valid reason not to commit
> > > > > > > the (maybe partial) fix. But looking at the fix again:
> > > > > > >
> > > > > > > diff --git a/tools/xenstored/domain.c b/tools/xenstored/domain.c
> > > > > > > index a6cd199fdc..9cd6678015 100644
> > > > > > > --- a/tools/xenstored/domain.c
> > > > > > > +++ b/tools/xenstored/domain.c
> > > > > > > @@ -989,6 +989,7 @@ static struct domain *introduce_domain(const void *ctx,
> > > > > > >                  talloc_steal(domain->conn, domain);
> > > > > > >
> > > > > > >                  if (!restore) {
> > > > > > > +                       domain_conn_reset(domain);
> > > > > > >                          /* Notify the domain that xenstore is available */
> > > > > > >                          interface->connection = XENSTORE_CONNECTED;
> > > > > > >                          xenevtchn_notify(xce_handle, domain->port);
> > > > > > > @@ -1031,8 +1032,6 @@ int do_introduce(const void *ctx, struct connection 
> > > > > > > *conn,
> > > > > > >          if (!domain)
> > > > > > >                  return errno;
> > > > > > >
> > > > > > > -       domain_conn_reset(domain);
> > > > > > > -
> > > > > > >          send_ack(conn, XS_INTRODUCE);
> > > > > > >
> > > > > > > It is a 1-line movement. Textually small. Easy to understand and to
> > > > > > > revert. It doesn't seem to be making things harder to fix? We could
> > > > > > > revert it any time if a better fix is offered.
> > > > > > >
> > > > > > > Maybe we could have a XXX note in the commit message or in-code
> > > > > > > comment?
> > > > > > It moves a line from one function (do_domain_introduce()) into a
> > > > > > completely different function (introduce_domain()), nested inside two
> > > > > > if() statements; with no analysis on how the change will impact
> > > > > > things.
> > > > > I am not the original author of the patch, and I am not the maintainer
> > > > > of the code, so I don't feel I have the qualifications to give you the
> > > > > answers you are seeking. Julien as author of the patch and xenstore
> > > > > reviewer might be in a better position to answer. Or Juergen as xenstore
> > > > > maintainer.
> > > > I understand that; my main point is that the change is more complex
> > > > than you're characterizing it.  This is information necessary to
> > > > understand whether the patch is correct, but it's not in the patch
> > > > description, nor in the subsequent thread back in May.
> > > >
> > > > > > Are there any paths through do_domain_introduce() that now *won't* get
> > > > > > a domain_conn_reset() call?  Is that OK?
> > > > > Yes, the already-introduced and the restore code paths. The operations in
> > > > > the already-introduced or the restore code paths seem simple enough not
> > > > > to require a domain_conn_reset. Julien and Juergen should confirm.
> > > > There is no "restore" codepath through do_domain_introduce(); it
> > > > passes "false" for the "restore" argument.  So we  only have two paths
> > > > to consider through do_domain_introduce(): The "not introduced and not
> > > > restoring" path, and the "already-introduced" path.
> > > >
> > > > I'm not sure what the "simple" elements on the branch in
> > > > introduce_domain() have to do with whether the content of the page
> > > > needs to be cleaned up.  As I said, I don't 100% understand this code,
> > > > but it seems like if anything, the reset would be *more* important to
> > > > have in the "reintroduce" case than in the "initial introduction"
> > > > case, since I'd expect the "initial introduction" case to be empty
> > > > already.
> > > Indeed, there should be no watches/transactions/buffered I/O for the initial 
> > > introduction. However, the function is also clear part of the interface 
> > > because we can't guaranteed it was zeroed.
> > > The latter matter for the initial introduction. I believe the rest is just 
> > > called for simplicity.
> > > > > Doesn't it seem weird to you that we set a connection to CONNECTED,
> > > > > notify the domain that it is ready to go, and only *after* that we reset
> > > > > the connection to zero?
> > > > >
> > > > > What happens if a domain starts using the connection as soon as it
> > > > > receives the event channel notification and before domain_conn_reset is
> > > > > called?
> > > > Yes, it does seem weird, which is why I said the following. :-)
> > > >
> > > > > > I mean, it certainly seems strange to set the state to CONNECTED, send
> > > > > > off an event channel, and then after that delete all watches /
> > > > > > transactions / buffered data and so on;
> > > > But just because the current code is probably wrong, doesn't mean that
> > > > the modified code is probably correct.
> > > >
> > > > If the problem is the delay between the xenevtchn_notify() in
> > > > introduce_domain() and the domain_conn_reset() afterwards in
> > > > do_domain(), would it make sense instead to move the notification into
> > > > do_introduce(), after the domain_conn_reset()?  It is, after all, in
> > > > response to XS_INTRODUCE that we want to send the notification, not in
> > > > dom0_init() or read_state_connection() (which seems to be more about
> > > > restoring a domain).
> > > I understand that the event channel notification was specifically added for 
> > > dom0less. But I don't see why we don't want to send it to dom0 as well.
> > > Technically, dom0 has exactly the same problem as dom0less domains it boots 
> > > before Xenstored is running and therefore it may need to know when it is 
> > > ready to receive commands.
> > Umm, no, not really.
> >
> > The main difference between dom0 and a dom0less domU is, that xenstored
> > introduces dom0 by itself via a call of dom0_init(), while the dom0less
> > domUs get introduced by Xen tools in case a dom0 is coming up later. And
> > that XS_INTRODUCE will clobber any ring page contents, while a call of
> > dom0_init() won't do that.
> >
> > Dom0 (especially the kernel) is fine to start filling the ring page with
> > requests even before xenstored is running. It just shouldn't expect to
> > receive any responses right away.
> I am not sure what you mean by fine. You will see hang notifications if 
> Xenstored is not started in time. Isn't why we decided to go with a different 
> way for dom0less?
The main difference is that dom0 tells xenstored the connection parameters for
itself, so dom0 _knows_ that the ring page is setup correctly when xenstored
starts looking at it (it is dom0 which needs to do the ring page init).

A dom0less domU doesn't have that negotiation with xenstored, as xenstored just
uses the pre-defined grant for looking at the ring page. For the domU there is
no way to tell that xenstored has initialized the ring page (it is not the domU
to do the initialization, as the XS_INTRODUCE might be sent before the domU
even starts running), other than the "connected" indicator in the page itself.


Juergen

Attachment: OpenPGP_0xB0DE9DD628BF132F.asc
Description: OpenPGP public key

Attachment: OpenPGP_signature.asc
Description: OpenPGP digital signature