OpenThread port #5552

jia200x · 2016-06-15T14:46:00Z

Hi.

This is the WIP PR for the port of OpenThread.
Best regards.

jia200x · 2016-06-15T14:47:40Z

pkg/openthread/patches/0001-fix-naming-conflict-mkversion.patch

+
+-ROOTDIR=${PREFIX}/
+#ROOTDIR=${PREFIX}/
+ROOTDIR=/


There's a conflict between mkversion's PREFIX and RIOT PREFIX variable. This is not the cleanest solution, but for now it's working.

jia200x · 2016-06-15T14:50:55Z

For now the pkg is being cloned, compiled and the openthread library is copied in BINDIR.
Static linking is working.

miri64 · 2016-06-15T14:51:38Z

pkg/openthread/Makefile

+.PHONY: all
+
+all: git-download
+		$($(PKG_BUILDIR)/bootstrap)


One tab should suffice here?

OlegHahm · 2016-07-12T07:46:38Z

Just checking: Is this WIP? (If yes, we should set the label, if not, we should update the title.)

jia200x · 2016-07-17T16:30:37Z

Not WIP anymore (now is in test phase).
There are some files that shouldn't be in the PR. I will remove them asap.
Also, I have to move the example to tests.

Cheers!

jia200x · 2016-07-18T07:51:53Z

2 known bugs after testing with IotLAB:

When a node disconnects and connects again to the same leader, sometimes the leader reject the packets. This produces 2 OpenThread leaders.
When there are a lot of nodes, sometimes they crash if you try to do an IEEE802.15.4 scan or pinging (it seems to be an assert of OpenThread's otPlatRadioTransmit that expects the driver to be ready).

Since the pkg version is outdated, I'm migrating to the latest to see if this fixes the problems.
Cheers

kaspar030 · 2016-07-18T11:30:39Z

pkg/openthread/Makefile

+PKG_NAME=openthread
+PKG_URL=git://github.com/openthread/openthread.git
+PKG_VERSION=4779fb20eb6f027b4ca3ff56f58ee789829d72f8
+PKG_BUILDDIR ?= $(BINDIRBASE)/pkg/$(BOARD)/$(PKG_NAME)


Isn't exactly this line part of pkg.mk?

jonas-rem · 2016-07-25T09:42:26Z

For testing with the phyNODE board I rebased the updated driver from #5469 on this PR. There are some minor adaptions necessary, which can be found at [1]. I tested with the branch at my Github [2].

Status:

Building works
RX and TX packets seems to work
when starting the Thread Network, device sends out multicast messages which I can see in Wireshark
The other device receives these packets and forwards them within the function recv_pkt

However, when starting the commissioning procedure with start at both boards, they fail pair to each other. Both devices state that they are leader. When I try the same with two SAMR21 devices, everything works. @jia200x any idea whats going wrong?

[1] - 92e5b91
[2] - https://github.com/jremmert-phytec-iot/RIOT/tree/kw2xrf

jia200x · 2016-07-25T10:01:57Z

Hi Jonas.

I sometimes had that problem when the packets where not being forwarded to OpenThread via otPlatRadioReceiveDone (in function recv_pkt.)

Also, I added a NETOPT_RX_LISTENING function in the driver for turning of the receiver. It's required by OpenThread, since does the rx/tx synchronization with otPlatRadioIdle, otPlatRadioReceive and otPlatRadioTransmit. If the driver triggers a RX_COMPLETE event while OT is in idle, unexpected results my occur.

Can you turn on OpenThread debug and post the logs? This is done in file openthread_example/bin/pkg/{BOARD}/openthread/src/core/openthread-core-config.h, line 144, changing OPENTHREAD_LOG_LEVEL_CRIT to OPENTHREAD_LOG_LEVEL_DEBG.

Cheers

adjih · 2016-08-01T16:20:15Z

tested last commit on two SAMR21-xpro, works for me 👍

aabadie · 2016-08-01T16:38:11Z

@jia200x, it seems this one needs some rebase ;)

jia200x · 2016-08-02T09:45:35Z

@aabadie Heavy rebase actually. I'm working on it :)

2016-08-01 18:38 GMT+02:00 Alexandre Abadie notifications@github.com:

@jia200x https://github.com/jia200x, it seems this one needs some
rebase ;)

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#5552 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/ABM8SIfB44xAPJ0NygopYAw66yNZrQcKks5qbiD1gaJpZM4I2b6I
.

jonas-rem · 2016-08-02T10:07:14Z

pkg/openthread/contrib/platform_radio.c

+{
+    netopt_enable_t enable = true;
+
+    _dev->driver->set(_dev, NETOPT_RX_LISTENING, &enable, sizeof(enable));


@jia200x what is the reason to not using set_state(NETOPT_STATE_IDLE)? This method bypasses the radio driver state handling and requires modification of each radio driver implementations.

Hi Jonas.

In the current pkg version of OpenThread there's a race condition between Idle state and packet arrival.
When the driver is in NETOPT_STATE_IDLE, is listening to packets.

When OpenThread sends packets, it calls otPlatRadioIdle first and then otPlatRadioTransmit. From the first call it expects the driver to be Idle and with receiver off. If there's a packet arrival between otPlatRadioIdle and otPlatRadioTransmit, otPlatRadioTransmit will throw kThreadError_Busy. This case shouldn't be possible in the OT implementation, so it executes an assert(false) and crashes. Although this could be fixed by simulating driver states (and dropping packets), I added this option for easily enabling/disabling the receiver and making it work properly.

The new implementation of OpenThread doesn't have this race condition (otPlatRadioIdle is not needed anymore), so it shouldn't be necessary to implement this RX_LISTENING option (I don't know if it's useful for other applications). I'm currently working in the pkg update to deal with this.

Best.

Hi Jose, ah ok, thx for your explanation! Then I'll wait for the pkg update and give it an other try then.

From the first call it expects the driver to be Idle and with receiver off.

But isn't the "Idle" state (NETOPT_STATE_IDLE) in the current RIOT netdev2 interface to put the transceiver in RX mode? Idle state with receiver off would be called NETOPT_STATE_SLEEP.

Hi again.

I will try to get the new pkg working asap.

Concerning the SLEEP state, is not exactly the same as IDLE without RX. Although the result is very similar, in SLEEP the device is in a low power mode with the transceiver off. It can't send packets unless it goes again to an awake state (SLEEP -> IDLE -> TX).
In the enable/disable rx option you are still able to send packets. OpenThread puts the device to Idle (rx off), then sends a packet in that state.

Also OpenThread expects to have a SLEEP state as well, that matches with NETOPT_STATE_SLEEP.

Anyway, in the new version of OpenThread this RX_LISTENING option becomes unnecessary since there's no race condition in packet sending/arrival.

Best.

Would it maybe possible to wrap this behavior device-agnostic in the glue-code layer between OT and netdev2?

biboc · 2017-05-16T12:11:17Z

@jia200x @aabadie
I created a PR on @jia200x repo (jia200x#1):

Add support for OpenThread FTD (Full Thread Device), MTD (Minimal Thread Device) and NCP (Network Co-Processor) device
Apply correction from aabadie review

Known bug:

After a while a FTD catches an assert
A MTD does not manage to connect to a FTD on SAMR21

aabadie · 2017-05-16T12:33:37Z

I created a PR on @jia200x repo (jia200x#1)

Thanks this helps a lot. I'm looking forward to see this merged in this PR. I still think that it would be a good thing to split this PR in several PRs:

one for the basic OT port
one for sock adaption
one for each other tests: ncp, ftd, mtd

Enable debug to show bug fram too long

jia200x · 2017-05-18T18:56:30Z

@aabadie Roger. I proposed Baptiste to push his PR directly, so I can use this one for the core part. I will post another one with the sock adoption.

Cheers

aabadie · 2017-05-18T19:07:03Z

I proposed Baptiste to push his PR directly, so I can use this one for the core part. I will post another one with the sock adoption

Do you mean that @biboc will open a new PR here with the core part ? That would be a good since GitHub has some lag with this huge thread ;)

jia200x · 2017-05-18T19:53:52Z

I meant, the FTD and MTD part. But anyway, If it's easier, I can open another PR for the core and coordinate with @biboc.
Cheers!

Openthread FTD, MTD and NCP

…nthread

jia200x · 2017-05-18T23:48:50Z

@aabadie I think I addressed all issues in #7082. I was having inconsistencies in the Github viewer, and seemed to point to older files.

Cheers

miri64 · 2017-06-01T15:35:44Z

With #7082 next to being merged: is this PR still needed?

aabadie · 2017-06-01T15:36:49Z

No I think we can safely close

miri64 · 2017-06-01T15:41:28Z

The line count indicates, that there is some code here, that isn't in #7082. So I close as memo.

jia200x reviewed Jun 15, 2016
View reviewed changes

miri64 reviewed Jun 15, 2016
View reviewed changes

pkg/openthread/Makefile Outdated

.PHONY: all

all: git-download

$($(PKG_BUILDIR)/bootstrap)

Copy link

Member

miri64 Jun 15, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

One tab should suffice here?

biboc assigned miri64 Jul 5, 2016

biboc added Type: new feature The issue requests / The PR implemements a new feature for RIOT Area: pkg Area: External package ports labels Jul 5, 2016

jia200x changed the title ~~[WIP] OpenThread port~~ OpenThread port Jul 17, 2016

OlegHahm added the Area: network Area: Networking label Jul 18, 2016

kaspar030 reviewed Jul 18, 2016
View reviewed changes

OlegHahm added the PR-award-nominee Deprecated. Will be removed soon. label Jul 20, 2016

kYc0o mentioned this pull request Jul 26, 2016

pkg: RELIC uses build system of the original source #4511

Closed

kaspar030 added the State: WIP State: The PR is still work-in-progress and its code is not in its final presentable form yet label Jul 29, 2016

jia200x force-pushed the openthread branch 2 times, most recently from e0be1fa to d525f20 Compare August 1, 2016 15:06

jonas-rem reviewed Aug 2, 2016
View reviewed changes

biboc and others added 2 commits May 16, 2017 14:14

Fix openthread NCP readme

0ac7846

Fix OpenThread MTD and FTD README

37392a1

Refactor Makefile with new openthread library

e9bbfc4

Enable debug to show bug fram too long

jia200x added 14 commits May 18, 2017 16:00

Merge pull request #1 from biboc/openthread-NCP-FTD-MTD

584e417

Openthread FTD, MTD and NCP

fix merge

6a8651c

uncrustify files

e5512c3

sorted openthread in pseudomodules

d88e121

revert debug in at86rf2xx

d4c0e08

remove newline

583658b

changed 1000U with US_PER_MS

459eb71

fix space around operators

a039dbf

remove error variable

593ce1e

Merge branch 'openthread' of https://github.com/jia200x/RIOT into ope…

fe02b9a

…nthread

revert unrelated change

1b253de

fix ascii chars in copyright

3acc830

rewrite README

a5e2bed

revert ot.h file

56ef435

jia200x mentioned this pull request May 18, 2017

pkg/openthread: core and tests #7082

Merged

miri64 closed this Jun 1, 2017

miri64 added the State: archived State: The PR has been archived for possible future re-adaptation label Jun 1, 2017

Hyungsin mentioned this pull request Aug 26, 2017

[OpenThread] report RSSI for openthread #7518

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

OpenThread port #5552

OpenThread port #5552

jia200x commented Jun 15, 2016

jia200x Jun 15, 2016

jia200x commented Jun 15, 2016

miri64 Jun 15, 2016

OlegHahm commented Jul 12, 2016

jia200x commented Jul 17, 2016 •

edited

Loading

jia200x commented Jul 18, 2016

kaspar030 Jul 18, 2016

jonas-rem commented Jul 25, 2016

jia200x commented Jul 25, 2016

adjih commented Aug 1, 2016

aabadie commented Aug 1, 2016

jia200x commented Aug 2, 2016

jonas-rem Aug 2, 2016

jia200x Aug 2, 2016 •

edited

Loading

jonas-rem Aug 2, 2016

jia200x Aug 2, 2016

miri64 Oct 19, 2016

biboc commented May 16, 2017 •

edited

Loading

aabadie commented May 16, 2017

jia200x commented May 18, 2017

aabadie commented May 18, 2017

jia200x commented May 18, 2017

jia200x commented May 18, 2017

miri64 commented Jun 1, 2017

aabadie commented Jun 1, 2017

miri64 commented Jun 1, 2017

OpenThread port #5552

OpenThread port #5552

Conversation

jia200x commented Jun 15, 2016

jia200x Jun 15, 2016

Choose a reason for hiding this comment

jia200x commented Jun 15, 2016

miri64 Jun 15, 2016

Choose a reason for hiding this comment

OlegHahm commented Jul 12, 2016

jia200x commented Jul 17, 2016 • edited Loading

jia200x commented Jul 18, 2016

kaspar030 Jul 18, 2016

Choose a reason for hiding this comment

jonas-rem commented Jul 25, 2016

jia200x commented Jul 25, 2016

adjih commented Aug 1, 2016

aabadie commented Aug 1, 2016

jia200x commented Aug 2, 2016

jonas-rem Aug 2, 2016

Choose a reason for hiding this comment

jia200x Aug 2, 2016 • edited Loading

Choose a reason for hiding this comment

jonas-rem Aug 2, 2016

Choose a reason for hiding this comment

jia200x Aug 2, 2016

Choose a reason for hiding this comment

miri64 Oct 19, 2016

Choose a reason for hiding this comment

biboc commented May 16, 2017 • edited Loading

aabadie commented May 16, 2017

jia200x commented May 18, 2017

aabadie commented May 18, 2017

jia200x commented May 18, 2017

jia200x commented May 18, 2017

miri64 commented Jun 1, 2017

aabadie commented Jun 1, 2017

miri64 commented Jun 1, 2017

jia200x commented Jul 17, 2016 •

edited

Loading

jia200x Aug 2, 2016 •

edited

Loading

biboc commented May 16, 2017 •

edited

Loading