Re: 6lowpan on a RPi using at86rf233

Re: 6lowpan on a RPi using at86rf233

[Alexander Aring]

On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> Hi,
> I'm trying to get 6lowpan to work using bluetooth-next on the above
> hardware[1]. I'm currently using bluetooth-next as of
> baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> 
>    iwpan dev wpan0 set pan_id 0xbeef
> 
> I get:
> 
>    command failed: Device or resource busy (-16)
> 

you can set the pan_id when you interface is down only.

> or in detail:
> 
> #  src/iwpan --debug dev wpan0 set pan_id 0xbeef
> -- Debug: Sent Message:
> --------------------------   BEGIN NETLINK MESSAGE ---------------------------
>   [HEADER] 16 octets
>     .nlmsg_len = 36
>     .nlmsg_type = 21 <0x15>
>     .nlmsg_flags = 5 <REQUEST,ACK>
>     .nlmsg_seq = 1428071645
>     .nlmsg_pid = 2760
>   [PAYLOAD] 20 octets
>     0a 00 00 00 08 00 03 00 05 00 00 00 06 00 09 00 ef be ..................
>     00 00                                                 ..
> ---------------------------  END NETLINK MESSAGE   ---------------------------
> -- Debug: Received Message:
> --------------------------   BEGIN NETLINK MESSAGE ---------------------------
>   [HEADER] 16 octets
>     .nlmsg_len = 56
>     .nlmsg_type = 2 <ERROR>
>     .nlmsg_flags = 0 <>
>     .nlmsg_seq = 1428071645
>     .nlmsg_pid = 2760
>   [ERRORMSG] 20 octets
>     .error = -16 "Device or resource busy"
>   [ORIGINAL MESSAGE] 16 octets
>     .nlmsg_len = 16
>     .nlmsg_type = 21 <0x15>
>     .nlmsg_flags = 5 <REQUEST,ACK>
>     .nlmsg_seq = 1428071645
>     .nlmsg_pid = 2760
> ---------------------------  END NETLINK MESSAGE   ---------------------------
> 
> This looks similar to
> 
>    http://www.spinics.net/lists/linux-wpan/msg01290.html
> 
> but I'm running wpan-tools master already.
> 
> Before digging deeper it'd be great to know if these version should work together or if
> another kernel tree should be used otherwise I'll sprink some pr_debugs().
> Cheers,
>  -- Guido
> 
> [1] https://honk.sigxcpu.org/piki/hw/rpi6lowpan/

- Alex

Re: 6lowpan on a RPi using at86rf233

[Guido Günther]

On Fri, Apr 03, 2015 at 04:45:02PM +0200, Alexander Aring wrote:
> On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> > Hi,
> > I'm trying to get 6lowpan to work using bluetooth-next on the above
> > hardware[1]. I'm currently using bluetooth-next as of
> > baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> > 
> >    iwpan dev wpan0 set pan_id 0xbeef
> > 
> > I get:
> > 
> >    command failed: Device or resource busy (-16)
> > 
> 
> you can set the pan_id when you interface is down only.

There was an ifplugd on the pi which continued to up the device -
damnit. Sorry for the noise. I can now ping the two devices but am
seeing massive packet loss:

    # ping6 -s 20   -I lowpan0  fe80::f836:9287:905c:a684 
    PING fe80::f836:9287:905c:a684(fe80::f836:9287:905c:a684) from fe80::cb23:b779:742d:4fd2 lowpan0: 20 data bytes
    28 bytes from fe80::f836:9287:905c:a684: icmp_seq=3 ttl=64 time=27.7 ms
    28 bytes from fe80::f836:9287:905c:a684: icmp_seq=4 ttl=64 time=12.7 ms
    28 bytes from fe80::f836:9287:905c:a684: icmp_seq=5 ttl=64 time=12.7 ms
    ^C
    --- fe80::f836:9287:905c:a684 ping statistics ---
    16 packets transmitted, 3 received, 81% packet loss, time 15017ms

Will need to dig into this again.

Thaks a lot for the hint!
Cheers,
 -- Guido

Re: 6lowpan on a RPi using at86rf233

[Alexander Aring]

On Fri, Apr 03, 2015 at 08:27:44PM +0200, Guido Günther wrote:
> On Fri, Apr 03, 2015 at 04:45:02PM +0200, Alexander Aring wrote:
> > On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> > > Hi,
> > > I'm trying to get 6lowpan to work using bluetooth-next on the above
> > > hardware[1]. I'm currently using bluetooth-next as of
> > > baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> > > 
> > >    iwpan dev wpan0 set pan_id 0xbeef
> > > 
> > > I get:
> > > 
> > >    command failed: Device or resource busy (-16)
> > > 
> > 
> > you can set the pan_id when you interface is down only.
> 
> There was an ifplugd on the pi which continued to up the device -
> damnit. Sorry for the noise. I can now ping the two devices but am
> seeing massive packet loss:

This "ifplugd" thing is a common issue, most people which using raspbian
reported similar issues.

> 
>     # ping6 -s 20   -I lowpan0  fe80::f836:9287:905c:a684 
>     PING fe80::f836:9287:905c:a684(fe80::f836:9287:905c:a684) from fe80::cb23:b779:742d:4fd2 lowpan0: 20 data bytes
>     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=3 ttl=64 time=27.7 ms
>     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=4 ttl=64 time=12.7 ms
>     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=5 ttl=64 time=12.7 ms
>     ^C
>     --- fe80::f836:9287:905c:a684 ping statistics ---
>     16 packets transmitted, 3 received, 81% packet loss, time 15017ms
> 
> Will need to dig into this again.
>

mhh, is this a linux to linux connection? Maybe try to activate ARET
mode, you can do that with:

iwpan dev wpan0 set max_frame_retries 3

or something differs with 3, -1 will turn off the ARET mode. Value 0
will do csma handling only. All above 0 will do retransmission tries.
You should do that on both nodes and all nodes in your network need to
support AACK handling. (Automatck ACK).

Or maybe switch channel, I am running out of ideas now. :-)

- Alex

Re: 6lowpan on a RPi using at86rf233

[Alexander Aring]

On Fri, Apr 03, 2015 at 08:27:44PM +0200, Guido Günther wrote:
> On Fri, Apr 03, 2015 at 04:45:02PM +0200, Alexander Aring wrote:
> > On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> > > Hi,
> > > I'm trying to get 6lowpan to work using bluetooth-next on the above
> > > hardware[1]. I'm currently using bluetooth-next as of
> > > baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> > > 
> > >    iwpan dev wpan0 set pan_id 0xbeef
> > > 
> > > I get:
> > > 
> > >    command failed: Device or resource busy (-16)
> > > 
> > 
> > you can set the pan_id when you interface is down only.
> 
> There was an ifplugd on the pi which continued to up the device -
> damnit. Sorry for the noise. I can now ping the two devices but am
> seeing massive packet loss:
> 
>     # ping6 -s 20   -I lowpan0  fe80::f836:9287:905c:a684 
>     PING fe80::f836:9287:905c:a684(fe80::f836:9287:905c:a684) from fe80::cb23:b779:742d:4fd2 lowpan0: 20 data bytes
>     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=3 ttl=64 time=27.7 ms
>     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=4 ttl=64 time=12.7 ms
>     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=5 ttl=64 time=12.7 ms
>     ^C
>     --- fe80::f836:9287:905c:a684 ping statistics ---
>     16 packets transmitted, 3 received, 81% packet loss, time 15017ms
> 
> Will need to dig into this again.
> 

ah, maybe I know why. Did you use the openlabs dts at86rf233 entry?

Try this:

&spi {
	status = "okay";
	at86rf233@0 {
		compatible = "atmel,at86rf233";
		spi-max-frequency = <7500000>;
		reg = <0>;
		interrupts = <23 4>;
		interrupt-parent = <&gpio>;
		reset-gpio = <&gpio 24 1>;
		sleep-gpio = <&gpio 25 1>;
		xtal-trim = /bits/ 8 <0x0F>;
	};
};


The openlabs entry uses a clock ~ 500kHz I experience similar issues with
such a low clock. Also the xtal-trim should be "somewhat correct" for the
openlabs transceivers. I need HF equipment to really test this value  I
supposed datasheet values.

- Alex

Re: 6lowpan on a RPi using at86rf233

[Guido Günther]

On Fri, Apr 03, 2015 at 08:38:06PM +0200, Alexander Aring wrote:
> On Fri, Apr 03, 2015 at 08:27:44PM +0200, Guido Günther wrote:
> > On Fri, Apr 03, 2015 at 04:45:02PM +0200, Alexander Aring wrote:
> > > On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> > > > Hi,
> > > > I'm trying to get 6lowpan to work using bluetooth-next on the above
> > > > hardware[1]. I'm currently using bluetooth-next as of
> > > > baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> > > > 
> > > >    iwpan dev wpan0 set pan_id 0xbeef
> > > > 
> > > > I get:
> > > > 
> > > >    command failed: Device or resource busy (-16)
> > > > 
> > > 
> > > you can set the pan_id when you interface is down only.
> > 
> > There was an ifplugd on the pi which continued to up the device -
> > damnit. Sorry for the noise. I can now ping the two devices but am
> > seeing massive packet loss:
> 
> This "ifplugd" thing is a common issue, most people which using raspbian
> reported similar issues.

I've documented this at [1] so hopefully Google helps to avoid this a bit.

> 
> > 
> >     # ping6 -s 20   -I lowpan0  fe80::f836:9287:905c:a684 
> >     PING fe80::f836:9287:905c:a684(fe80::f836:9287:905c:a684) from fe80::cb23:b779:742d:4fd2 lowpan0: 20 data bytes
> >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=3 ttl=64 time=27.7 ms
> >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=4 ttl=64 time=12.7 ms
> >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=5 ttl=64 time=12.7 ms
> >     ^C
> >     --- fe80::f836:9287:905c:a684 ping statistics ---
> >     16 packets transmitted, 3 received, 81% packet loss, time 15017ms
> > 
> > Will need to dig into this again.
> >
> 
> mhh, is this a linux to linux connection? Maybe try to activate ARET
> mode, you can do that with:

Linux to Linux, yes. Thought that this is simpler before trying to talk
to Contiki.

> 
> iwpan dev wpan0 set max_frame_retries 3
> 
> or something differs with 3, -1 will turn off the ARET mode. Value 0
> will do csma handling only. All above 0 will do retransmission tries.
> You should do that on both nodes and all nodes in your network need to
> support AACK handling. (Automatck ACK).
> 
> Or maybe switch channel, I am running out of ideas now. :-)

I turned out that the 1.5 meters with some metal in the walls was far
too much of a faraday cage for that module (altough it isn't a problem
for 802.11). Moving things closer together made the packet loss mostly
go away.

Cheers,
 -- Guido




[1] https://honk.sigxcpu.org/piki/hw/rpi6lowpan/

Re: 6lowpan on a RPi using at86rf233

[Alexander Aring]

On Tue, Apr 07, 2015 at 01:49:21PM +0200, Guido Günther wrote:
> On Fri, Apr 03, 2015 at 08:38:06PM +0200, Alexander Aring wrote:
> > On Fri, Apr 03, 2015 at 08:27:44PM +0200, Guido Günther wrote:
> > > On Fri, Apr 03, 2015 at 04:45:02PM +0200, Alexander Aring wrote:
> > > > On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> > > > > Hi,
> > > > > I'm trying to get 6lowpan to work using bluetooth-next on the above
> > > > > hardware[1]. I'm currently using bluetooth-next as of
> > > > > baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> > > > > 
> > > > >    iwpan dev wpan0 set pan_id 0xbeef
> > > > > 
> > > > > I get:
> > > > > 
> > > > >    command failed: Device or resource busy (-16)
> > > > > 
> > > > 
> > > > you can set the pan_id when you interface is down only.
> > > 
> > > There was an ifplugd on the pi which continued to up the device -
> > > damnit. Sorry for the noise. I can now ping the two devices but am
> > > seeing massive packet loss:
> > 
> > This "ifplugd" thing is a common issue, most people which using raspbian
> > reported similar issues.
> 
> I've documented this at [1] so hopefully Google helps to avoid this a bit.
> 

thanks.

> > 
> > > 
> > >     # ping6 -s 20   -I lowpan0  fe80::f836:9287:905c:a684 
> > >     PING fe80::f836:9287:905c:a684(fe80::f836:9287:905c:a684) from fe80::cb23:b779:742d:4fd2 lowpan0: 20 data bytes
> > >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=3 ttl=64 time=27.7 ms
> > >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=4 ttl=64 time=12.7 ms
> > >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=5 ttl=64 time=12.7 ms
> > >     ^C
> > >     --- fe80::f836:9287:905c:a684 ping statistics ---
> > >     16 packets transmitted, 3 received, 81% packet loss, time 15017ms
> > > 
> > > Will need to dig into this again.
> > >
> > 
> > mhh, is this a linux to linux connection? Maybe try to activate ARET
> > mode, you can do that with:
> 
> Linux to Linux, yes. Thought that this is simpler before trying to talk
> to Contiki.
> 

yep.
> > 
> > iwpan dev wpan0 set max_frame_retries 3
> > 
> > or something differs with 3, -1 will turn off the ARET mode. Value 0
> > will do csma handling only. All above 0 will do retransmission tries.
> > You should do that on both nodes and all nodes in your network need to
> > support AACK handling. (Automatck ACK).
> > 
> > Or maybe switch channel, I am running out of ideas now. :-)
> 
> I turned out that the 1.5 meters with some metal in the walls was far
> too much of a faraday cage for that module (altough it isn't a problem
> for 802.11). Moving things closer together made the packet loss mostly
> go away.
> 

are you sure? The 12.7 ms for 28 bytes with a link-local (I suppose
SLAAC) smells like a small spi clock and I don't believe that. Can you
try a "ping -s 500 ..." then you can test fragmenation handling, I had
big issues with fragmentation (means it doesn't worked) and a slow spi
clock ~500Khz. Try my dts at86rf233 entry for the RPi in my other mail.

- Alex

Re: 6lowpan on a RPi using at86rf233

[Guido Günther]

On Tue, Apr 07, 2015 at 09:07:52PM +0200, Alexander Aring wrote:
> On Tue, Apr 07, 2015 at 01:49:21PM +0200, Guido Günther wrote:
> > On Fri, Apr 03, 2015 at 08:38:06PM +0200, Alexander Aring wrote:
> > > On Fri, Apr 03, 2015 at 08:27:44PM +0200, Guido Günther wrote:
> > > > On Fri, Apr 03, 2015 at 04:45:02PM +0200, Alexander Aring wrote:
> > > > > On Fri, Apr 03, 2015 at 04:36:47PM +0200, Guido Günther wrote:
> > > > > > Hi,
> > > > > > I'm trying to get 6lowpan to work using bluetooth-next on the above
n> > > > > > hardware[1]. I'm currently using bluetooth-next as of
> > > > > > baf880a96859cca79208122e555e7efeabd16e4d . When I do a:
> > > > > > 
> > > > > >    iwpan dev wpan0 set pan_id 0xbeef
> > > > > > 
> > > > > > I get:
> > > > > > 
> > > > > >    command failed: Device or resource busy (-16)
> > > > > > 
> > > > > 
> > > > > you can set the pan_id when you interface is down only.
> > > > 
> > > > There was an ifplugd on the pi which continued to up the device -
> > > > damnit. Sorry for the noise. I can now ping the two devices but am
> > > > seeing massive packet loss:
> > > 
> > > This "ifplugd" thing is a common issue, most people which using raspbian
> > > reported similar issues.
> > 
> > I've documented this at [1] so hopefully Google helps to avoid this a bit.
> > 
> 
> thanks.
> 
> > > 
> > > > 
> > > >     # ping6 -s 20   -I lowpan0  fe80::f836:9287:905c:a684 
> > > >     PING fe80::f836:9287:905c:a684(fe80::f836:9287:905c:a684) from fe80::cb23:b779:742d:4fd2 lowpan0: 20 data bytes
> > > >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=3 ttl=64 time=27.7 ms
> > > >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=4 ttl=64 time=12.7 ms
> > > >     28 bytes from fe80::f836:9287:905c:a684: icmp_seq=5 ttl=64 time=12.7 ms
> > > >     ^C
> > > >     --- fe80::f836:9287:905c:a684 ping statistics ---
> > > >     16 packets transmitted, 3 received, 81% packet loss, time 15017ms
> > > > 
> > > > Will need to dig into this again.
> > > >
> > > 
> > > mhh, is this a linux to linux connection? Maybe try to activate ARET
> > > mode, you can do that with:
> > 
> > Linux to Linux, yes. Thought that this is simpler before trying to talk
> > to Contiki.
> > 
> 
> yep.
> > > 
> > > iwpan dev wpan0 set max_frame_retries 3
> > > 
> > > or something differs with 3, -1 will turn off the ARET mode. Value 0
> > > will do csma handling only. All above 0 will do retransmission tries.
> > > You should do that on both nodes and all nodes in your network need to
> > > support AACK handling. (Automatck ACK).
> > > 
> > > Or maybe switch channel, I am running out of ideas now. :-)
> > 
> > I turned out that the 1.5 meters with some metal in the walls was far
> > too much of a faraday cage for that module (altough it isn't a problem
> > for 802.11). Moving things closer together made the packet loss mostly
> > go away.
> > 
> 
> are you sure? The 12.7 ms for 28 bytes with a link-local (I suppose
> SLAAC) smells like a small spi clock and I don't believe that. Can you
> try a "ping -s 500 ..." then you can test fragmenation handling, I had
> big issues with fragmentation (means it doesn't worked) and a slow spi
> clock ~500Khz. Try my dts at86rf233 entry for the RPi in my other mail.

# ping6  -c 500    fe80::51c6:6a47:dcf2:72e5%lowpan0
PING fe80::51c6:6a47:dcf2:72e5%lowpan0(fe80::51c6:6a47:dcf2:72e5) 56 data bytes
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=1 ttl=64 time=16.7 ms
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=2 ttl=64 time=16.3 ms
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=3 ttl=64 time=16.3 ms
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=4 ttl=64 time=16.3 ms
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=5 ttl=64 time=16.2 ms
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=8 ttl=64 time=16.9 ms
64 bytes from fe80::51c6:6a47:dcf2:72e5: icmp_seq=9 ttl=64 time=30.4 ms
^C
--- fe80::51c6:6a47:dcf2:72e5%lowpan0 ping statistics ---
9 packets transmitted, 7 received, 22% packet loss, time 8017ms
rtt min/avg/max/mdev = 16.257/18.495/30.421/4.877 ms


After adjusting the spi-max-frequence to 7500000 and switching to
edge-triggered interrupts [1] this now looks like:

# ping6 -s 500  fe80::18a4:6afa:7f78:56f6%lowpan0
PING fe80::18a4:6afa:7f78:56f6%lowpan0(fe80::18a4:6afa:7f78:56f6) 500 data bytes
508 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=1 ttl=64 time=62.6 ms
508 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=5 ttl=64 time=62.3 ms
508 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=6 ttl=64 time=62.3 ms
508 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=10 ttl=64 time=62.3 ms
508 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=14 ttl=64 time=62.3 ms
508 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=15 ttl=64 time=62.3 ms
^C
--- fe80::18a4:6afa:7f78:56f6%lowpan0 ping statistics ---
16 packets transmitted, 6 received, 62% packet loss, time 15031ms
rtt min/avg/max/mdev = 62.304/62.387/62.628/0.340 ms

so quiet some packet loss with fragmentation.

ping6 -s 20  fe80::18a4:6afa:7f78:56f6%lowpan0
PING fe80::18a4:6afa:7f78:56f6%lowpan0(fe80::18a4:6afa:7f78:56f6) 20 data bytes
28 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=1 ttl=64 time=6.43 ms
28 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=2 ttl=64 time=6.17 ms
28 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=3 ttl=64 time=6.22 ms
28 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=4 ttl=64 time=6.17 ms
28 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=5 ttl=64 time=6.25 ms
28 bytes from fe80::18a4:6afa:7f78:56f6: icmp_seq=8 ttl=64 time=6.21 ms
^C
--- fe80::18a4:6afa:7f78:56f6%lowpan0 ping statistics ---
8 packets transmitted, 6 received, 25% packet loss, time 7016ms
rtt min/avg/max/mdev = 6.171/6.246/6.439/0.120 ms

SSHing between the machines is ok. Nice.

Cheers,
 -- Guido

[1] as described in http://www.spinics.net/lists/linux-wpan/msg01468.html

Re: 6lowpan on a RPi using at86rf233

[Alexander Aring]

Hi,

On Thu, Apr 09, 2015 at 11:06:52AM +0200, Guido Günther wrote:
> 
> SSHing between the machines is ok. Nice.

For tune-up tcp connection I set also set tcp window size;

#reduce tcp receive buf, will improve tcp connections
echo "127 512 1024" > /proc/sys/net/ipv4/tcp_rmem
echo "127 512 1024" > /proc/sys/net/ipv4/tcp_wmem

This will do this for all tcp connection. You can also try to set MSS
with ip6tables.

to have much fragments isn't good on a 6LoWPAN connection. Upper layers
like CoAP, etc. will handle this. Also a general TCP connection isn't
good, because the ACK handling stuff. Maybe want to try out "mosh with
ipv6 support [1]" - currently only available on dev branch.


High fragments and fq_codel qdisc will also occur trouble, because the
skb's are too long inside the queue and fq_codel will drop them then.
This occurs on my setup at payload of ~2000 bytes, but we should never
reach such payload, also 1024 is too high. If you use with contiki I
don't think that contiki implement ipv6 fragmentation, so you can set
the highest payload to 1280. Does somebody knows if contiki supports
ipv6 fragmentation?

- Alex

[1] https://github.com/keithw/mosh