Friday, June 24, 2016< ^ >
mcr/credil has set the subject to: IETF96 ROLL WG meeting
Room Configuration
Room Occupants

[07:35:12] eintopf joins the room
[07:35:25] <eintopf> stefan will be there @IETF
[07:35:29] <eintopf> also for plugfest
[07:40:26] <eintopf> he need to approve some NDA document :-)
[08:34:10] <eintopf> acknowledge*
[14:24:25] <mcr/soho> cool.
[16:13:03] <eintopf> stefan told me isn't @IETF itself
[16:13:07] <eintopf> plugtest only
[16:13:25] <eintopf> and then I was thinking if I really get any benefits to join this IETF stuff
[16:23:35] <mcr/soho> the value to be there for a day (I suggest Monday) is that you would get a better insight into who is who.
[16:23:46] <mcr/soho> but, if we are going to come to the plugfest, then you would meet people anyway.
[16:56:08] <mcr/soho> er, "but if you are going", I mean.
[17:01:17] <eintopf> I am a very shy person
[17:01:26] <eintopf> and hotel is already booked...
[17:04:08] <mcr/soho> i am in an airbnb.
[17:07:37] <eintopf> do you suggest the tutorial for newcomers @ ietf?
[17:07:56] <mcr/soho> if you are around, it is worth going to.
[17:08:43] <eintopf> I will visit the full ietf now
[17:10:06] <eintopf> Attending Emerging Work in IEEE 802 (Pat Thaler, Juan-Carlos Zuniga) at 15:00 Sunday?
[17:10:09] <eintopf> what's that
[17:10:13] <eintopf> IEEE at IETF?
[17:12:05] <eintopf> :-)
[17:12:40] <eintopf> oh creat 6lo/6tisch register works again
[17:16:56] <mcr/soho> basically, the session is Pat and company telling us unwashed IETF types what the IEEE plans to screw up next.
[17:17:21] <mcr/soho> (so that we won't be surprised, and possibly so that some of us might pay attention and help them get it right)
[17:18:05] <mcr/soho> I guess I'm overwhelmed by Brexit stuff today.  (I'm dual Cdn/British. So I'll lose my EU citizenship)
[17:20:50] <eintopf> do you visit the IEEE sutff?
[17:20:54] <eintopf> oh :(
[17:23:48] <mcr/soho> Never been to an IEEE meeting; it costs more (they ALWAYS meet at resorts), requires a more "corporate" commitment, they have enfranchisement rules before you can "vote", and do almost all their work in Word.  
[17:26:28] <eintopf> with comic sans :-)
[17:27:12] <eintopf> do you are sad because loosing the EU citizien stuff?
[17:27:26] <eintopf> grml, I hate that busybox ping implementation
[17:28:21] <mcr/soho> Just gobsmacked.  I have lots of British friends (who live there) and Cdn friends who are British ex-pats on facebook...
[17:35:21] <eintopf> ah, ok
[17:35:35] <eintopf> I also didn't thought this would be the result
[17:35:38] <eintopf> but it is
[17:35:41] <eintopf> 64 bytes from fe80::8a77:6655:4433:2212: seq=1 ttl=64 time=1004.618 ms
[17:35:48] <eintopf> looong RTT
[17:35:57] <mcr/soho> 1s. Something living in buffers too long?
[17:36:11] <eintopf> that's not real hardware
[17:36:15] <mcr/soho> Are you old enough to remember the Wall coming down?
[17:36:23] <eintopf> no
[17:36:49] <eintopf> it's pinging the RIOT userspace stack (native) from linux 6lowpan fakelb driver
[17:36:50] <mcr/soho> It's much like that, I think. Of course, that was a major positive.
[17:37:14] <mcr/soho> oh, so from native through tunnel into RIOT. Do you have more than one CPU core available?
[17:37:28] <mcr/soho> not really tunnel...
[17:37:29] <eintopf> I think it's because the SIGIO stuff they used
[17:37:42] <eintopf> I hate this SIGIO for sockets
[17:37:48] <mcr/soho> yes, you probably have to wait for RIOT to schedule itself...
[17:38:06] <eintopf> I need to think about that
[17:38:18] <eintopf> they use the same mechanism for tap interface
[17:38:34] <eintopf> I need to tune it, (difficult receive handling etc.) but at first it works!
[17:38:42] <eintopf> but I saw bugs in monitor interface handling
[17:40:04] <eintopf> wireshark thinks there is a FCS on monitor while transmit but there is none because other reasons :-)
[17:40:36] <eintopf> wireless has monitor flags to allow FCS from userspace or kernel
[17:40:47] <eintopf> and then wireshark will switch parsing
[17:40:56] <eintopf> (I think it's also for transmit)
[17:42:08] <eintopf> ahhh
[17:42:15] <eintopf> I type reboot in RIOT native shell
[17:42:19] <eintopf> now it works
[17:42:30] <eintopf> 64 bytes from fe80::8a77:6655:4433:2212: seq=33 ttl=64 time=0.368 ms
64 bytes from fe80::8a77:6655:4433:2212: seq=34 ttl=64 time=0.361 ms
64 bytes from fe80::8a77:6655:4433:2212: seq=35 ttl=64 time=0.375 ms
64 bytes from fe80::8a77:6655:4433:2212: seq=36 ttl=64 time=0.364 ms
64 bytes from fe80::8a77:6655:4433:2212: seq=37 ttl=64 time=0.369 ms
[17:42:40] <eintopf> I need to reset some queues there
[17:43:51] <eintopf> buffers, readout stucking frames in socketfd
[17:43:56] <eintopf> something like that
[17:44:03] <eintopf> I think when I fix that it's okay
[17:44:29] <eintopf> so, now I need to register me @ IETF
[17:44:58] <eintopf> Attending Sunday Welcome Reception? - is beer available there?
[17:50:35] <mcr/soho> beer available.
[18:02:22] <eintopf> ok
[18:02:36] <eintopf> there coming some heavy rain in my town
[18:04:14] <eintopf> Meeting participation: yes @ plugtest register?
[18:04:17] <eintopf> I think yes
[18:11:48] <eintopf> I thought the 6lo/6TiSCH plugtest is free
[18:11:53] <eintopf> isn't it?
[18:14:32] <mcr/soho> it is, but you have to register.
[18:14:37] <mcr/soho> and do an NDA thing.
[18:26:11] <eintopf> done
[18:26:23] <eintopf> so at last I need a ticket for the train
[18:29:03] <eintopf> and pinging RIOT userspace works nice now
[18:29:11] <eintopf> with some magic behind
[18:31:14] <eintopf> no login information yet...
[18:31:17] <eintopf> for plugtest
[18:34:00] <eintopf> 5000 packets transmitted, 5000 received, 0% packet loss, time 55016ms
rtt min/avg/max/mdev = 0.249/0.502/2.053/0.251 ms
[18:34:07] <eintopf> ping6 fe80::8a77:6655:4433:2212%lowpan0 -s 555 -i 0.01 -c 5000
[18:34:10] <eintopf> that's ok
[18:46:33] <eintopf> AdvOnLink off; in PIO radvd.conf
[18:46:51] <eintopf> I always get this issue, but it's no issue it's because I am too stupid and use some default configuration
[18:47:05] <eintopf> and RFC6775 says drop RA PIO if this flag is set
[18:47:14] <eintopf> so I need to set it off
[19:12:15] <eintopf> mhh, can't ping global address linux <-> global address RIOT
[19:12:21] <eintopf> and global address was assigned by RA PIO
[19:12:23] <eintopf> in RIOT
[19:12:30] <eintopf> I think that should be possible
[19:13:35] <eintopf> but RIOT looks funny with short addr activated in linux, then RIOT use the lastest address option (which is short address)
[19:13:47] <eintopf> and could be that they have some out of buffer access there
[21:37:43] <eintopf> mhh
[21:38:08] <eintopf> if you getting the fragments from FRAGN to FRAG1, then the sequence num in 802.15.4 frames is also vice versa
[21:38:28] <eintopf> we should think about a "good" default qdsic
[21:39:03] <eintopf> qdisc, if we have the possibility to change that. Because fancy init systems e.g. systemd change them directly to fq_codel or similar
[21:42:22] eintopf leaves the room: Disconnected: closed
Powered by ejabberd - robust, scalable and extensible XMPP server Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!