Date:   Fri, 26 Aug 2022 10:08:25 +0200
From:   Miquel Raynal <miquel.raynal@...tlin.com>
To:     Alexander Aring <aahringo@...hat.com>
Cc:     Alexander Aring <alex.aring@...il.com>,
        Stefan Schmidt <stefan@...enfreihafen.org>,
        linux-wpan - ML <linux-wpan@...r.kernel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Network Development <netdev@...r.kernel.org>,
        David Girault <david.girault@...vo.com>,
        Romuald Despres <romuald.despres@...vo.com>,
        Frederic Blain <frederic.blain@...vo.com>,
        Nicolas Schodet <nico@...fr.eu.org>,
        Thomas Petazzoni <thomas.petazzoni@...tlin.com>
Subject: Re: [PATCH wpan-next 01/20] net: mac802154: Allow the creation of
 coordinator interfaces

Hi Alexander,

aahringo@...hat.com wrote on Thu, 25 Aug 2022 21:35:05 -0400:

> Hi,
> 
> On Thu, Aug 25, 2022 at 8:51 PM Alexander Aring <aahringo@...hat.com> wrote:
> >
> > Hi,
> >
> > On Thu, Aug 25, 2022 at 4:41 AM Miquel Raynal <miquel.raynal@...tlin.com> wrote:  
> > >
> > > Hi Alexander,
> > >
> > > aahringo@...hat.com wrote on Wed, 24 Aug 2022 17:43:11 -0400:
> > >  
> > > > On Wed, Aug 24, 2022 at 3:35 AM Miquel Raynal <miquel.raynal@...tlin.com> wrote:  
> > > > >
> > > > > Hi Alexander,
> > > > >
> > > > > aahringo@...hat.com wrote on Tue, 23 Aug 2022 17:44:52 -0400:
> > > > >  
> > > > > > Hi,
> > > > > >
> > > > > > On Tue, Aug 23, 2022 at 12:29 PM Miquel Raynal
> > > > > > <miquel.raynal@...tlin.com> wrote:  
> > > > > > >
> > > > > > > Hi Alexander,
> > > > > > >
> > > > > > > aahringo@...hat.com wrote on Tue, 23 Aug 2022 08:33:30 -0400:
> > > > > > >  
> > > > > > > > Hi,
> > > > > > > >
> > > > > > > > On Fri, Aug 19, 2022 at 1:11 PM Miquel Raynal <miquel.raynal@...tlin.com> wrote:  
> > > > > > > > >
> > > > > > > > > Hi Alexander,
> > > > > > > > >
> > > > > > > > > aahringo@...hat.com wrote on Tue, 5 Jul 2022 21:51:02 -0400:
> > > > > > > > >  
> > > > > > > > > > Hi,
> > > > > > > > > >
> > > > > > > > > > On Fri, Jul 1, 2022 at 10:36 AM Miquel Raynal <miquel.raynal@...tlin.com> wrote:  
> > > > > > > > > > >
> > > > > > > > > > > As a first strep in introducing proper PAN management and association,
> > > > > > > > > > > we need to be able to create coordinator interfaces which might act as
> > > > > > > > > > > coordinator or PAN coordinator.
> > > > > > > > > > >
> > > > > > > > > > > Hence, let's add the minimum support to allow the creation of these
> > > > > > > > > > > interfaces. This might be restrained and improved later.
> > > > > > > > > > >
> > > > > > > > > > > Signed-off-by: Miquel Raynal <miquel.raynal@...tlin.com>
> > > > > > > > > > > ---
> > > > > > > > > > >  net/mac802154/iface.c | 14 ++++++++------
> > > > > > > > > > >  net/mac802154/rx.c    |  2 +-
> > > > > > > > > > >  2 files changed, 9 insertions(+), 7 deletions(-)
> > > > > > > > > > >
> > > > > > > > > > > diff --git a/net/mac802154/iface.c b/net/mac802154/iface.c
> > > > > > > > > > > index 500ed1b81250..7ac0c5685d3f 100644
> > > > > > > > > > > --- a/net/mac802154/iface.c
> > > > > > > > > > > +++ b/net/mac802154/iface.c
> > > > > > > > > > > @@ -273,13 +273,13 @@ ieee802154_check_concurrent_iface(struct ieee802154_sub_if_data *sdata,
> > > > > > > > > > >                 if (nsdata != sdata && ieee802154_sdata_running(nsdata)) {
> > > > > > > > > > >                         int ret;
> > > > > > > > > > >
> > > > > > > > > > > -                       /* TODO currently we don't support multiple node types
> > > > > > > > > > > -                        * we need to run skb_clone at rx path. Check if there
> > > > > > > > > > > -                        * exist really an use case if we need to support
> > > > > > > > > > > -                        * multiple node types at the same time.
> > > > > > > > > > > +                       /* TODO currently we don't support multiple node/coord
> > > > > > > > > > > +                        * types we need to run skb_clone at rx path. Check if
> > > > > > > > > > > +                        * there exist really an use case if we need to support
> > > > > > > > > > > +                        * multiple node/coord types at the same time.
> > > > > > > > > > >                          */
> > > > > > > > > > > -                       if (wpan_dev->iftype == NL802154_IFTYPE_NODE &&
> > > > > > > > > > > -                           nsdata->wpan_dev.iftype == NL802154_IFTYPE_NODE)
> > > > > > > > > > > +                       if (wpan_dev->iftype != NL802154_IFTYPE_MONITOR &&
> > > > > > > > > > > +                           nsdata->wpan_dev.iftype != NL802154_IFTYPE_MONITOR)
> > > > > > > > > > >                                 return -EBUSY;
> > > > > > > > > > >
> > > > > > > > > > >                         /* check all phy mac sublayer settings are the same.
> > > > > > > > > > > @@ -577,6 +577,7 @@ ieee802154_setup_sdata(struct ieee802154_sub_if_data *sdata,
> > > > > > > > > > >         wpan_dev->short_addr = cpu_to_le16(IEEE802154_ADDR_BROADCAST);
> > > > > > > > > > >
> > > > > > > > > > >         switch (type) {
> > > > > > > > > > > +       case NL802154_IFTYPE_COORD:
> > > > > > > > > > >         case NL802154_IFTYPE_NODE:
> > > > > > > > > > >                 ieee802154_be64_to_le64(&wpan_dev->extended_addr,
> > > > > > > > > > >                                         sdata->dev->dev_addr);
> > > > > > > > > > > @@ -636,6 +637,7 @@ ieee802154_if_add(struct ieee802154_local *local, const char *name,
> > > > > > > > > > >         ieee802154_le64_to_be64(ndev->perm_addr,
> > > > > > > > > > >                                 &local->hw.phy->perm_extended_addr);
> > > > > > > > > > >         switch (type) {
> > > > > > > > > > > +       case NL802154_IFTYPE_COORD:
> > > > > > > > > > >         case NL802154_IFTYPE_NODE:
> > > > > > > > > > >                 ndev->type = ARPHRD_IEEE802154;
> > > > > > > > > > >                 if (ieee802154_is_valid_extended_unicast_addr(extended_addr)) {
> > > > > > > > > > > diff --git a/net/mac802154/rx.c b/net/mac802154/rx.c
> > > > > > > > > > > index b8ce84618a55..39459d8d787a 100644
> > > > > > > > > > > --- a/net/mac802154/rx.c
> > > > > > > > > > > +++ b/net/mac802154/rx.c
> > > > > > > > > > > @@ -203,7 +203,7 @@ __ieee802154_rx_handle_packet(struct ieee802154_local *local,
> > > > > > > > > > >         }
> > > > > > > > > > >
> > > > > > > > > > >         list_for_each_entry_rcu(sdata, &local->interfaces, list) {
> > > > > > > > > > > -               if (sdata->wpan_dev.iftype != NL802154_IFTYPE_NODE)
> > > > > > > > > > > +               if (sdata->wpan_dev.iftype == NL802154_IFTYPE_MONITOR)
> > > > > > > > > > >                         continue;  
> > > > > > > > > >
> > > > > > > > > > I probably get why you are doing that, but first the overall design is
> > > > > > > > > > working differently - means you should add an additional receive path
> > > > > > > > > > for the special interface type.
> > > > > > > > > >
> > > > > > > > > > Also we "discovered" before that the receive path of node vs
> > > > > > > > > > coordinator is different... Where is the different handling here? I
> > > > > > > > > > don't see it, I see that NODE and COORD are the same now (because that
> > > > > > > > > > is _currently_ everything else than monitor). This change is not
> > > > > > > > > > enough and does "something" to handle in some way coordinator receive
> > > > > > > > > > path but there are things missing.
> > > > > > > > > >
> > > > > > > > > > 1. Changing the address filters that it signals the transceiver it's
> > > > > > > > > > acting as coordinator
> > > > > > > > > > 2. We _should_ also have additional handling for whatever the
> > > > > > > > > > additional handling what address filters are doing in mac802154
> > > > > > > > > > _because_ there is hardware which doesn't have address filtering e.g.
> > > > > > > > > > hwsim which depend that this is working in software like other
> > > > > > > > > > transceiver hardware address filters.
> > > > > > > > > >
> > > > > > > > > > For the 2. one, I don't know if we do that even for NODE right or we
> > > > > > > > > > just have the bare minimal support there... I don't assume that
> > > > > > > > > > everything is working correctly here but what I want to see is a
> > > > > > > > > > separate receive path for coordinators that people can send patches to
> > > > > > > > > > fix it.  
> > > > > > > > >
> > > > > > > > > Yes, we do very little differently between the two modes, that's why I
> > > > > > > > > took the easy way: just changing the condition. I really don't see what
> > > > > > > > > I can currently add here, but I am fine changing the style to easily
> > > > > > > > > show people where to add filters for such or such interface, but right
> > > > > > > > > now both path will look very "identical", do we agree on that?  
> > > > > > > >
> > > > > > > > mostly yes, but there exists a difference and we should at least check
> > > > > > > > if the node receive path violates the coordinator receive path and
> > > > > > > > vice versa.
> > > > > > > > Put it in a receive_path() function and then coord_receive_path(),
> > > > > > > > node_receive_path() that calls the receive_path() and do the
> > > > > > > > additional filtering for coordinators, etc.
> > > > > > > >
> > > > > > > > There should be a part in the standard about "third level filter rule
> > > > > > > > if it's a coordinator".
> > > > > > > > btw: this is because the address filter on the transceiver needs to
> > > > > > > > have the "i am a coordinator" boolean set which is missing in this
> > > > > > > > series. However it depends on the transceiver filtering level and the
> > > > > > > > mac802154 receive path if we actually need to run such filtering or
> > > > > > > > not.  
> > > > > > >
> > > > > > > I must be missing some information because I can't find any places
> > > > > > > where what you suggest is described in the spec.
> > > > > > >
> > > > > > > I agree there are multiple filtering level so let's go through them one
> > > > > > > by one (6.7.2 Reception and rejection):
> > > > > > > - first level: is the checksum (FCS) valid?
> > > > > > >         yes -> goto second level
> > > > > > >         no -> drop
> > > > > > > - second level: are we in promiscuous mode?
> > > > > > >         yes -> forward to upper layers
> > > > > > >         no -> goto second level (bis)
> > > > > > > - second level (bis): are we scanning?
> > > > > > >         yes -> goto scan filtering
> > > > > > >         no -> goto third level
> > > > > > > - scan filtering: is it a beacon?
> > > > > > >         yes -> process the beacon
> > > > > > >         no -> drop
> > > > > > > - third level: is the frame valid? (type, source, destination, pan id,
> > > > > > >   etc)
> > > > > > >         yes -> forward to upper layers
> > > > > > >         no -> drop
> > > > > > >
> > > > > > > But none of them, as you said, is dependent on the interface type.
> > > > > > > There is no mention of a specific filtering operation to do in all
> > > > > > > those cases when running in COORD mode. So I still don't get what
> > > > > > > should be included in either node_receive_path() which should be
> > > > > > > different than in coord_receive_path() for now.
> > > > > > >
> > > > > > > There are, however, two situations where the interface type has its
> > > > > > > importance:
> > > > > > > - Enhanced beacon requests with Enhanced beacon filter IE, which asks
> > > > > > >   the receiving device to process/drop the request upon certain
> > > > > > >   conditions (minimum LQI and/or randomness), as detailed in
> > > > > > >   7.4.4.6 Enhanced Beacon Filter IE. But, as mentioned in
> > > > > > >   7.5.9 Enhanced Beacon Request command: "The Enhanced Beacon Request
> > > > > > >   command is optional for an FFD and an RFD", so this series was only
> > > > > > >   targeting basic beaconing for now.
> > > > > > > - In relaying mode, the destination address must not be validated
> > > > > > >   because the message needs to be re-emitted. Indeed, a receiver in
> > > > > > >   relaying mode may not be the recipient. This is also optional and out
> > > > > > >   of the scope of this series.
> > > > > > >
> > > > > > > Right now I have the below diff, which clarifies the two path, without
> > > > > > > too much changes in the current code because I don't really see why it
> > > > > > > would be necessary. Unless you convince me otherwise or read the spec
> > > > > > > differently than I do :) What do you think?
> > > > > > >  
> > > > > >
> > > > > > "Reception and rejection"
> > > > > >
> > > > > > third-level filtering regarding "destination address" and if the
> > > > > > device is "PAN coordinator".
> > > > > > This is, in my opinion, what the coordinator boolean tells the
> > > > > > transceiver to do on hardware when doing address filter there. You can
> > > > > > also read that up in datasheets of transceivers as atf86rf233, search
> > > > > > for I_AM_COORD.  
> > > > >
> > > > > Oh right, I now see what you mean!
> > > > >  
> > > > > > Whereas they use the word "PAN coordinator" not "coordinator", if they
> > > > > > really make a difference there at this point..., if so then the kernel
> > > > > > must know if the coordinator is a pan coordinator or coordinator
> > > > > > because we need to set the address filter in kernel.  
> > > > >
> > > > > Yes we need to make a difference, you can have several coordinators but
> > > > > a single PAN coordinator in a PAN. I think we can assume that the PAN
> > > > > coordinator is the coordinator with no parent (association-wise). With
> > > > > the addition of the association series, I can handle that, so I will
> > > > > create the two path as you advise, add a comment about this additional
> > > > > filter rule that we don't yet support, and finally after the
> > > > > association series add another commit to make this filtering rule real.
> > > > >  
> > > > > >  
> > > > > > > Thanks,
> > > > > > > Miquèl
> > > > > > >
> > > > > > > ---
> > > > > > >
> > > > > > > --- a/net/mac802154/rx.c
> > > > > > > +++ b/net/mac802154/rx.c
> > > > > > > @@ -194,6 +194,7 @@ __ieee802154_rx_handle_packet(struct ieee802154_local *local,
> > > > > > >         int ret;
> > > > > > >         struct ieee802154_sub_if_data *sdata;
> > > > > > >         struct ieee802154_hdr hdr;
> > > > > > > +       bool iface_found = false;
> > > > > > >
> > > > > > >         ret = ieee802154_parse_frame_start(skb, &hdr);
> > > > > > >         if (ret) {
> > > > > > > @@ -203,18 +204,31 @@ __ieee802154_rx_handle_packet(struct ieee802154_local *local,
> > > > > > >         }
> > > > > > >
> > > > > > >         list_for_each_entry_rcu(sdata, &local->interfaces, list) {
> > > > > > > -               if (sdata->wpan_dev.iftype != NL802154_IFTYPE_NODE)
> > > > > > > +               if (sdata->wpan_dev.iftype == NL802154_IFTYPE_MONITOR)
> > > > > > >                         continue;
> > > > > > >
> > > > > > >                 if (!ieee802154_sdata_running(sdata))
> > > > > > >                         continue;
> > > > > > >
> > > > > > > +               iface_found = true;
> > > > > > > +               break;
> > > > > > > +       }
> > > > > > > +
> > > > > > > +       if (!iface_found) {
> > > > > > > +               kfree_skb(skb);
> > > > > > > +               return;
> > > > > > > +       }
> > > > > > > +
> > > > > > > +       /* TBD: Additional filtering is possible on NODEs and/or COORDINATORs */
> > > > > > > +       switch (sdata->wpan_dev.iftype) {
> > > > > > > +       case NL802154_IFTYPE_COORD:
> > > > > > > +       case NL802154_IFTYPE_NODE:
> > > > > > >                 ieee802154_subif_frame(sdata, skb, &hdr);
> > > > > > > -               skb = NULL;
> > > > > > > +               break;
> > > > > > > +       default:
> > > > > > > +               kfree_skb(skb);
> > > > > > >                 break;
> > > > > > >         }  
> > > > > >
> > > > > > Why do you remove the whole interface looping above and make it only
> > > > > > run for one ?first found? ?  
> > > > >
> > > > > To reduce the indentation level.
> > > > >  
> > > > > > That code changes this behaviour and I do
> > > > > > not know why.  
> > > > >
> > > > > The precedent code did:
> > > > > for_each_iface() {
> > > > >         if (not a node)
> > > > >                 continue;
> > > > >         if (not running)
> > > > >                 continue;
> > > > >
> > > > >         subif_frame();
> > > > >         break;
> > > > > }
> > > > >
> > > > > That final break also elected only the first running node iface.
> > > > > Otherwise it would mean that we allow the same skb to be consumed
> > > > > twice, which is wrong IMHO?  
> > > >
> > > > no? Why is that wrong? There is a real use-case to have multiple
> > > > interfaces on one phy (or to do it in near future, I said that
> > > > multiple times). This patch does a step backwards to this.  
> > >
> > > So we need to duplicate the skb because it automatically gets freed in
> > > the "forward to upper layer" path. Am I right? I'm fine doing so if  
> >
> > What is the definition of "duplicate the skb" here.
> >  
> > > this is the way to go, but I am interested if you can give me a real
> > > use case where having NODE+COORDINATOR on the same PHY is useful?
> > >  
> >
> > Testing.  
> 
> I need to say that I really used multiple monitors at the same time on
> one phy only and I did that with hwsim to run multiple user space
> stacks. It was working and I was happy and didn't need to do a lot of
> phy creations in hwsim.

Indeed, looking at the code, you could use as many MONITOR interfaces
you needed, but only a single NODE. I've changed that to use as many
NODE and COORD that we wish.

> Most hardware can probably not run multiple
> nodes and coordinators at the same time ?yet?, _but_ there is a
> candidate which can do that and this is atusb. On atusb we have a
> co-processor that can deal with multiple address filters. People
> already asked to do something like a node which can operate on two
> pans as I remember, that would be a candidate for such a feature.

Oh nice! Yes this makes sense.

> I
> really don't want to move step backwards here and delete this thing
> which probably can be useful later. I don't know how wireless history
> dealt with it and how complicated it was to bring such a feature in to
> e.g. run multiple access points on one phy. I also see it in ethernet
> with macvlan, which is a similar feature.
> 
> We don't need to support it, make it so that on an ifup it returns
> -EBUSY if something doesn't fit together as it currently is. We can
> later add support for it after playing around with hwsim a little bit
> more. We should at least take care that I can still run my multiple
> monitors at the same time (which is currently allowed).
> 
> - Alex
> 


Thanks,
Miquèl

Powered by blists - more mailing lists