gmane.linux.network.zigbee.devel

Results of a small experiment: improve serialdriver latency by increasing the baud rate (with econotags)

Tony Cheneau — 2012-05-25T18:13:46

Good afternoon,

By default, the Econotag and izattach (part of the linux-zigbee 
userspace tools) expects the communications to run at 115200 bps. I 
wanted to know of it was affecting the performances, so I tried 
different values (by recompiling the firmware and the tools 
accordingly). I was mainly interested in reducing latency, this is why I 
exclusively used ping6 for my tests, but I guess the throughput should 
improve as well.

Here is the average round-trip time reported by ping6 between two 
nodes:
- at 115200 bps: ~ 400ms
- at 921600 bps: ~ 20ms (yes , 20 times faster)
- at 1843200 bps (maximum baudrate supported by the econotag): ~ 20 ms 
(same number as above)

As a side note, 1843200 bps is not a "standard" baud rate supported by 
the Linux kernel, which make it more complex to set on the software 
side. Given that there is not performance gain compared to 921600 bps, 
there seem to be no benefit in using this baud rate (but I might be 
wrong).

I do not believe my changes should make it to th

Re: [PATCH] BeagleBone: Patch to register SPIDEV on one of the SPI ports.

Alan Ott — 2012-05-24T14:07:40

Yes, that's why I clarified in the second email. I sent the wrong patch,
but the idea is the same, substituting cc120 for spidev. I thought your
previous email said cc2420. If that's the case, then there's already of
course a driver for it and you just need to attach to it (similar to
what I did in the patch I sent). If cc120 is a new device, then you'll
need to write a driver for it.


The first thing you'll want to see You'll want to see
Documentation/spi/spi-summary for information about SPI, assuming that
you're using an SPI device like the cc2420.

Also see Documentation/networking/ieee802154.txt .

For general information about making drivers, see Linux Device Drivers,
3rd Edition at http://lwn.net/Kernel/LDD3/ .

Beyond that you'll need to read the code. Look at drivers which are
similar to your device to see how they work.

Alan.


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's secur

Re: [PATCH] BeagleBone: Patch to register SPIDEV on one of the SPI ports.

Pierre-Emmanuel Goudet — 2012-05-24T07:24:41

Hi Alan,

Thanks for this.
We are already trying such a modification but not for "spidev" but for "cc120" that we are trying to connect to beaglebone, through "mac802154/iee802154" proto drivers.
Not yet successful but still working on. Will soon send our state on the mailing list.

We are a bit "new" to kernel and hardware/protocols drivers in Linux. Why the way is so tricky for us. Do you know a good "tutorial" showing global schemes of interfaces between Network stack / Protocols drivers / hardware drivers ?

Best regards

Pierre-emmanuel

-----Message d'origine-----
De : Alan Ott [mailto:alan-yzvJWuRpmD1zbRFIqnYvSA< at >public.gmane.org] 
Envoyé : mercredi 23 mai 2012 20:18
À : Pierre-Emmanuel Goudet; linux-zigbee-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org
Cc : Alan Ott
Objet : [PATCH] BeagleBone: Patch to register SPIDEV on one of the SPI ports.

---
 arch/arm/mach-omap2/board-am335xevm.c |   25 +++++++++++++++++++++++++
 1 files changed, 25 insertions(+), 0 deletions(-)

diff --git a/arc

Re: [PATCH] BeagleBone: Patch to register SPIDEV on one of the SPI ports.

Alan Ott — 2012-05-23T18:23:39

Actaully, that's the wrong patch, sorry. That one attaches the spidev
driver to the SPI device. The idea is the same though. Instead of using
.modalias = "spidev", you'll use the name of the driver you want to use.




------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

[PATCH] BeagleBone: Patch to register SPIDEVon one of the SPI ports.

Alan Ott — 2012-05-23T18:17:30

---
 arch/arm/mach-omap2/board-am335xevm.c |   25 +++++++++++++++++++++++++
 1 files changed, 25 insertions(+), 0 deletions(-)

diff --git a/arch/arm/mach-omap2/board-am335xevm.c b/arch/arm/mach-omap2/board-am335xevm.c
index 36efca1..84daf62 100644
--- a/arch/arm/mach-omap2/board-am335xevm.c
+++ b/arch/arm/mach-omap2/board-am335xevm.c
< at >< at > -1836,6 +1836,20 < at >< at > static void tt3201_init(int evm_id, int profile)
 
 am33xx_d_can_init(1);
 }
+
+static struct spi_board_info am335x_spidev1_info[] = {
+{
+.modalias      = "spidev",
+.platform_data = NULL,
+.irq           = 0,
+.max_speed_hz  = 12000000,
+.bus_num       = 2,
+.mode          = SPI_MODE_1,
+.chip_select   = 0,
+},
+};
+
+
 static void beaglebone_cape_setup(struct memory_accessor *mem_acc, void *context)
 {
 capecount++;
< at >< at > -2143,6 +2157,16 < at >< at > static void spi1_init(int evm_id, int profile)
 return;
 }
 
+/* setup spidev */
+static void spi1_spidev_init(int evm_id, int profile)
+{
+pr_info("Setting up spidev on bus 2\n");
+
+setup_pin_mu

Re: Using a device driver for Linux-Zigbee

Alan Ott — 2012-05-23T18:15:38

I'll reply with a patch that I use to attach my mrf24j40 to the
BeagleBone. It of course won't work as-is for your device, but you'll
likely want to do something very similar.

Alan.

On 05/23/2012 12:18 PM, Pierre-Emmanuel Goudet wrote:

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/_______________________________________________
Linux-zigbee-devel mailing list
Linux-zigbee-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org
https://lists.sourceforge.net/lists/listinfo/linux-zigbee-devel

Re: Using a device driver for Linux-Zigbee

Alan Ott — 2012-05-23T14:48:39

Hi Pierre-Emmanuel,

For SPI devices, you have to add support for them in the board file,
basically telling the SPI subsystem what kind of device you have
attached to which port (since SPI is not a self-describing bus).

Check out:
    arch/arm/mach-pxa/stargate2.c
for an example of how to do this. (search for spi_board_info).

I hope this is the question you were asking :)

Alan.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/_______________________________________________
Linux-zigbee-devel mailing list
Linux-zigbee-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org
https://lists.sourceforge.net/lists/listinfo/linux-zigbee-devel

Using a device driver for Linux-Zigbee

Pierre-Emmanuel Goudet — 2012-05-23T14:05:18

Hi all,

I'm an early tester of linux-zigbee implementation. I was, until now, playing with "Serial" driver and having quite success to use it.

Is there any tutorial in the way that a driver like CC2420 or RF231 would be used on the user side.

I mean once the modules are loaded, what should be the next action from user side to really create, test, use the corresponding interface ... ? As with serial we could use izattach , iz, ip ... to manage the wpanx interface.

For now I've compiled the modules I could "modprobe" them, but except seeing them correctly with "lsmod", I do not see any new wpan-phy interface or other ? Event more after adding some debug "printk" I saw that the "probe" function was never called after registering ?

In advance thanks for your tips

-----------------------------------------------------------------------------------------
Pierre-emmanuel

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will co

Is 6lowpan fragmentation working?

Tony Cheneau — 2012-05-22T21:42:56

Hello,

I'm currently playing with the 6lowpan code and was wondering if the 
fragmentation support is supposed to be operational. The reason I ask is 
because when my packets grow too big (when fragmentation should kick 
in), my communications stop working.
I sniffed the traffic with Wireshark, and found out the first fragment 
(FRAG1) is always 25 bytes long and does not contain any payload. A 
quick look at the code seems to confirm that the first fragment carries 
no payload (see lowpan_skb_fragmentation()). I fail to see where RFC 
4944 indicates that the first fragment should be empty (and I fail to 
see what would be the benefit of this behavior anyway). So, is that the 
intended behavior? Did I miss something?

Either way, I'll look more into it, maybe this is only specific to my 
system (I ported the serial driver and applied some patches to it, so 
the packets might be mangled somewhere else).

Thank you in advance.

Regards,
Tony


------------------------------------------------------------------

Bug in 6lowpan when uncompressing UDP datagram

Tony Cheneau — 2012-05-21T18:50:13

Hello again,

When running some connectivity tests, I discovered that UDP 
communication would not work between my nodes. Looking at the various 
printk() in the kernel, I found out that the IPv6 next header field is 
not set (set to 0) upon decompression and thus prevented correct 
processing of the decompressed packet at the upper layer (IPv6 layer).
When the next header is UDP, IPHC next header compression applies (as 
defined in Section 4.1 of RFC6282). The compression code in 6lowpan.c 
seems to behave correctly. However, upon decompression code (in 
lowpan_process_data()) fails to properly restore/set the next header 
field.

I attached a (very small) patch for this issue to this mail. It works 
in my configuration and enables the two nodes to communicate. I do not 
believe it will introduce any side-effects.

Regards,
Tony------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
thr

Re: Bug in 6lowpan 16-bit short address detection code

Alexander Smirnov — 2012-05-21T15:52:49

Hi Tony,

thanks for your next finding!

Unfortunately currently I'm sick, please give me a couple of days to
get well and I'll test/check your patch.

With best regards,
Alexander

2012/5/21 Tony Cheneau <tony.cheneau+zigbeedev-jNfjcPZKvDhg9hUCZPvPmw< at >public.gmane.org>:

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Bug in 6lowpan 16-bit short address detectioncode

Tony Cheneau — 2012-05-21T15:27:30

Hello,

Last week, I was analyzing some 6lowpan packet with Wireshark and got 
notified by the tool that the checksum (TCP, UDP or ICMPv6) was wrong. 
Some investigations showed me that this issue has been discussed and 
solved in Contiki, as pointed out in the following thread:
http://comments.gmane.org/gmane.os.contiki.devel/7053

In my case, the issue was triggered by the format of the address I 
assigned to my nodes. They were assigned with addresses fe80::1 and 
fe80::2 and the 6lowpan code wrongly interpreted them as being 16-bit 
short address mapped an EUI-64 address (although the ff:fe part was 
missing).

It turns out that the lowpan_is_iid_16_bit_compressable macro in 
net/ieee802154/6lowpan.h is not correct and does not behave according to 
RFC6282. I attached a patch that intend to correct this behavior.

Regards,
Tony
From 84a884346b175d2c2890249d122e2f34735d433a Mon Sep 17 00:00:00 2001
From: Tony Cheneau <tony.cheneau< at >amnesiak.org>
Date: Thu, 17 May 2012 17:13:06 -0400
Subject: [PATCH 1/1] lo

Re: 6lowpan Contiki fixes

2012-05-17T14:56:21

On Mon, May 14, 2012 at 8:21 AM, Alexander Smirnov
<alex.bluesman.smirnov-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

git diff is probably more useful.

First I would clone a Conitiki git tree make yourself a tag and push
it to git hub. That will let you remember what you have checked.

Next note that 'git diff' take commit ids and the file name as
parameters.  Do a 'git log' on just the 6lowpan files and figure out a
commit id that is old enough. Then 'git diff' between that id and head
just to get the delta for that file.

Another way I do it is to go back to a commit id I know is old enough
in the project and then do a full project diff for that id. I load
that diff into an edit and just delete out the diffs that don't
matter. Then study what is left.

Next time this will be easier since you will have labeled your check
points in that tree at gitub. Each time you check you changes just
make a new label and push it to git hub.

Completely different way - I think there is a commit mailing list for
Cont

Fwd: [PATCH net-next v5 0/13] basicieee802.15.4 mac support

Alexander Smirnov — 2012-05-17T12:24:56

Hi all,

another part of the project was merged to mainline.

---------- Forwarded message ----------
From: David Miller <davem< at >davemloft.net>
Date: 2012/5/16
Subject: Re: [PATCH net-next v5 0/13] basic ieee802.15.4 mac support
To: alex.bluesman.smirnov< at >gmail.com
Копия: netdev< at >vger.kernel.org, dbaryshkov< at >gmail.com


From: Alexander Smirnov <alex.bluesman.smirnov< at >gmail.com>
Date: Wed, 16 May 2012 10:50:18 +0400


All applied, thanks.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Linux-zigbee-devel mailing list
Linux-zigbee-devel< at >lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-zigbee-devel

Re: 6lowpan Contiki fixes

Mariano Alvira — 2012-05-14T14:27:26

Unfortunately, Contiki doesn't have a tracking system or a list of
bugs. As you guessed, git is going to be the best way to figure out a
history.

There shouldn't have been too many changes to 6lowpan though... most
of the churn has been in RPL and Contiki's ipv6 stack.

-Mar.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Re: 6lowpan Contiki fixes

2012-05-14T13:18:01

Mar should know what is going on. He tracks this more closely than I do.

On Mon, May 14, 2012 at 8:21 AM, Alexander Smirnov
<alex.bluesman.smirnov-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

6lowpan Contiki fixes

Alexander Smirnov — 2012-05-14T12:21:06

Hi Jon,

I want to sync Linux 6lowpan implementation with Contiki one, so how
can I derive latest 6lowpan fixes?
Do you have any tracking system or may be just a list with bugs were fixed?

Or 'git log' must be my favorite tool? :-)

Thanks,
Alex

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Re: [PATCH 2/2] net/ieee802154/6lowpan.c:reuse eth_mac_addr()

Alexander Smirnov — 2012-05-12T08:38:01

Hi Tony,

2012/4/28 Tony Cheneau <tony.cheneau+zigbeedev-jNfjcPZKvDhg9hUCZPvPmw< at >public.gmane.org>:

Yeah, you are definitely right. I don't exactly know where did this
patch arrive from, looks like I wasn't in Cc. I'm going to create
reverted patch.

Thanks


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Re: [PATCH 2/2] net/ieee802154/6lowpan.c: reuse eth_mac_addr()

Tony Cheneau — 2012-04-27T20:21:05

Hello,

I believe the following patch should not have been applied and should 
now be reverted.
The reason is because eth_mac_addr() is not functionally equivalent to 
lowpan_set_address() that it replaces:
- lowpan_set_address() copies dev->addr_len bytes, where dev->addr_len 
is set to 8 bytes for IEEE802154 devices (this is an IEEE EUI-64 
address). Hence, the function copies the full 8 bytes of the address.
- eth_mac_addr() copies ETH_ALEN bytes, where ETH_ALEN is set to 6. 
Hence, the function copies only 6 bytes of the 8 bytes address. So, 2 
bytes of the address are never copied.

Does this sound reasonable?

Regards,
Tony


Le 22.02.2012 14:36, Danny Kukawka a écrit :


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threa

Re: Bug when setting the hop limit in 6lowpan.c

Tony Cheneau — 2012-04-27T14:40:36

Hi,

Here is the patch. I am not too familiar with git, so I do not know if 
the patch is formatted correctly or not, if not, you can produce your 
own patch and cc me.
The patch was created against the latest version of the net-next 
(82981930125abfd39d7c8378a9cfdf5e1be2002b at the moment I write these 
lines).

Hope it helps.

Regards,
Tony


Le 27.04.2012 09:18, Alexander Smirnov a écrit :
From 40b6fa4495a1775fa1825b37fc19cf51e991cc13 Mon Sep 17 00:00:00 2001
From: Tony Cheneau <tony.cheneau< at >amnesiak.org>
Date: Fri, 27 Apr 2012 10:26:08 -0400
Subject: [PATCH] 6lowpan: fixes a bug in lowpan_header_create(). If the hop
 limit value was not 1, 64 or 255, the value would not have
 been set correctly.

---
 net/ieee802154/6lowpan.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/net/ieee802154/6lowpan.c b/net/ieee802154/6lowpan.c
index 4b5701c..9f6d6d7 100644
--- a/net/ieee802154/6lowpan.c
+++ b/net/ieee802154/6lowpan.c
< at >< at > -490,6 +490,7 < at >< at > static int lowpan_header_create(struct sk_bu

Re: Bug when setting the hop limit in 6lowpan.c

Alexander Smirnov — 2012-04-27T07:18:02

Hi Tony,

27 апреля 2012 г. 1:46 пользователь Tony Cheneau
<tony.cheneau+zigbeedev< at >amnesiak.org> написал:

yeah, it seems you are right! thanks for the finding..
BTW: are you going to make a patch? Or I can do it (with you in cc)
and sent with my next set.

With best regards,
Alex


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Linux-zigbee-devel mailing list
Linux-zigbee-devel< at >lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-zigbee-devel