pkg/openwsn: re-integrate the network stack as a package #8570

PeterKietzmann · 2018-02-15T16:52:02Z

Contribution description

This PR re-integrates the OpenWSN network stack as a pkg to RIOT. It is marked as WIP because it does not yet work stable and I'm asking for your help in terms of debugging, testing and review.

An example application is provided, including documentation in the README next to it. Furthermore, I added documentation about the port in general, the status of this implementation as well as additional information about debugging capabilities and currently open issues/TODOs. The two commits covering Atmel radio driver changes and iotlab-m3 timer configuration are not intended to be merged but rather to be included as a patch later.

I have a collection of additional test applications for different wrappers lying around, which I would provide in a separate PR if we agree on how to integrate them into the repository.

Implementation Issues

~~- Nodes reset after some time (here?). Without channel hopping this effect does not show up that often. With channel hopping it gets worse~~
~~- After a node restarts, it won't join the topology any time soon, even though neighbored nodes still send DIOs~~

People that might be interested and in charge to support

(alphabetical order of last name)
@emmanuelsearch, @kb2ma, @changtengfei, @OlegHahm, @sieben, @fjmolinas, @Lotterleben, @twatteyne

Issues/PRs references

Relates to #3039 and #7893

PeterKietzmann · 2018-02-15T16:56:42Z

Useful hint no 1: For easier debugging I'm using a local copy of the adapted openwsn-fw repository. To incorporate the local repo I replace the git-dowlnload command in the pkg Makefile by something like:

 OPENWSN_PATH := ../../../openwsn-fw
 all:
 	rm -rf $(PKG_BUILDDIR) && mkdir -p $(PKG_BUILDDIR)
 	rsync -a $(OPENWSN_PATH) $(dir $(PKG_BUILDDIR)) --exclude .git
 	$(MAKE) -C $(PKG_BUILDDIR)

PeterKietzmann · 2018-02-15T17:01:44Z

Useful hint no 2: RIOT provides some build commands to easily submit experiments to the FIT IoT.LAB testbed. For documentation see here. TL;DR: From the test application directory use this command to build, submit an experiment with three nodes and start a tmux session to the aggregated serial output of that experiment:
BOARD=iotlab-m3 make clean all iotlab-exp iotlab-term IOTLAB_NODES=3

miri64

Just some preliminary comments.

miri64 · 2018-02-16T18:15:41Z

pkg/openwsn/contrib/eui64.c

+
+void eui64_get(uint8_t *addressToWrite)
+{
+    luid_get((void *)addressToWrite, IEEE802154_LONG_ADDRESS_LEN);


This will get you a different EUI-64 everytime the function is called or is this intended (couldn't find any documentation, on this but I saw it is called twice here)?

If not, why not get the long-address from the device somehow? E.g. have

netdev_t *openwsn_netdev; void eui64_get(uint8_t *addressToWrite) { eui64_t eui64; if (openwsn_netdev->driver->get(dev, NETOPT_ADDRESS_LONG, &eui64, sizeof(eui64_t)) == sizeof(eui64_t)) { memcpy(addressToWrite, eui64.u8, sizeof(eui64.u8)); } else { /* Use your code as a fallback or print an error message here */ luid_get((void *)addressToWrite, IEEE802154_LONG_ADDRESS_LEN); } }

And set netdev in radio_init():

extern netdev_t *openwsn_netdev; void radio_init(void) { /* ... */ #ifdef MODULE_AT86RF2XX openwsn_netdev = (netdev_t *)&at86rf2xx_dev.netdev.netdev; radio_vars.dev = openwsn_netdev; at86rf2xx_setup(&at86rf2xx_dev, &at86rf2xx_params[0]); #endif

If you fear memory waste, take netdev out of radio_vars_t and use openwsn_netdev instead of radio_vars.dev.

/cc @fjmolinas

miri64 · 2018-02-16T18:18:30Z

boards/common/iotlab/include/periph_conf_common.h

-#define RTT_FREQUENCY       (1)             /* in Hz */
-#define RTT_PRESCALER       (0x7fff)        /* run with 1 Hz */
+#define RTT_FREQUENCY       (16384U)     /* in Hz */
+#define RTT_PRESCALER       (0x1)        /* run with ~16 kHz */


Probably still a TODO, but these should be configured at compile-time, when OpenWSN is present.

miri64 · 2018-02-16T18:19:37Z

drivers/at86rf2xx/at86rf2xx.c

@@ -147,6 +147,9 @@ void at86rf2xx_tx_prepare(at86rf2xx_t *dev)
    if (state != AT86RF2XX_STATE_TX_ARET_ON) {
        dev->idle_state = state;
    }
+
+    at86rf2xx_set_state(dev, AT86RF2XX_STATE_PLL_ON);


Probably also a TODO, but hard-wiring states required for OpenWSN seems like a bad idea to me.

miri64 · 2018-02-16T18:19:44Z

drivers/at86rf2xx/at86rf2xx_netdev.c

@@ -213,7 +213,7 @@ static int _set_state(at86rf2xx_t *dev, netopt_state_t state)
                     * know when to switch back to the idle state. */
                    ++dev->pending_tx;
                }
-                at86rf2xx_set_state(dev, AT86RF2XX_STATE_TX_ARET_ON);
+                at86rf2xx_set_state(dev, AT86RF2XX_STATE_PLL_ON);


miri64 · 2018-02-16T18:20:01Z

drivers/at86rf2xx/at86rf2xx_netdev.c

+                }  else {
+                    DEBUG("[at86rf2xx] Unhandled TRAC_STATUS: %d\n",trac_status >> 5);
+                }
+            }


Move to separate PR?

miri64 · 2018-02-16T18:22:13Z

pkg/openwsn/contrib/radio.c

+    netdev_t                 *dev;
+} radio_vars_t;
+
+radio_vars_t radio_vars;


Isn't exposed, so static?

PeterKietzmann · 2018-04-03T16:18:29Z

The above mentioned issues have been solved. Furthermore, I squashed and rebased both RIOT as well as openwsn. Help in form of reviewing and testing is happily welcome. There is still work TODO which can be taken from the doc.txt. Most importantly:

find a generic way to enable the basic mode of the Atmel radio
make the sctimer wrapper hardware independant
run on different platforms with different radios

miri64 · 2018-10-08T13:15:40Z

find a generic way to enable the basic mode of the Atmel radio

Can #9592 help with that?

kb2ma · 2018-11-07T19:00:55Z

Able to send a simple UDP packet on IoT-LAB nodes using the test application. It's a start! FWIW, unable to compile with CFLAGS=-DOW_RIOT_SINGLE_CHANNEL=26. Lots of errors. But if channel hopping works, that's fine with me.

Next I'll see if I can use OpenWSN as a sock-like provider for nanocoap.

udp send bbbb::b85:fb65:106b:1115 61617 1234
udp send bbbb::b85:fb65:106b:1115 61617 1234
Send 4 byte over UDP to [bbbb::b85:fb65:106b:1115]:61617
Send success

udp server start 61617
Set UDP server port to 61617

Received 13 bytes on port 61617
00000000 45 81 7A 73 00 00 00 00 00 31 32 33 34 E.zs.....1234

miri64

Oops I had these review comments never commited and they seem to be in here for a while now ...

miri64 · 2018-12-16T12:28:17Z

pkg/openwsn/contrib/sctimer.c

+void sctimer_setCompare(uint32_t val)
+{
+    DEBUG("sctimer_setCompare\n");
+    // enable the compare interrupt


Comment style

miri64 · 2018-12-16T12:28:29Z

pkg/openwsn/contrib/sctimer.c

+    DEBUG("sctimer_setCompare\n");
+    // enable the compare interrupt
+    /*
+       if (current Timer counter - val < TIMERLOOP_THRESHOLD){


Can be removed, I guess

miri64 · 2018-12-16T12:28:37Z

pkg/openwsn/contrib/sctimer.c

+}
+
+/**
+   \brief set compare interrupt


Doxygen style

miri64 · 2018-12-16T12:30:04Z

pkg/openwsn/patches/0001-Add-Makefiles.patch

+From 0c967954f444044cbc5c51d25230a5ca6cfa1383 Mon Sep 17 00:00:00 2001
+From: PeterKietzmann <peter.kietzmann@haw-hamburg.de>
+Date: Thu, 8 Feb 2018 10:19:38 +0100
+Subject: [PATCH 1/6] Add Makefiles


Have a look at the current lwIP package. Adding makefiles is now possible without a patch ;-).

miri64 · 2018-12-16T12:31:07Z

pkg/openwsn/patches/0003-Comment-out-problematic-includes.patch

+From 587609720250906a6ae5b6d5bd5c1daca2dd3349 Mon Sep 17 00:00:00 2001
+From: PeterKietzmann <peter.kietzmann@haw-hamburg.de>
+Date: Thu, 8 Feb 2018 11:25:18 +0100
+Subject: [PATCH 3/6] Comment out problematic includes


Why are they problematic? A short reasoning in the description might be beneficial

miri64 · 2019-02-26T18:45:09Z

Ping @PeterKietzmann?

PeterKietzmann · 2019-03-24T13:33:20Z

Thanks for the feedback, I'll address what I can. There was some "major" reconstruction and added features to this work which is why I'm closing the PR for now. Will re-open once ready.

PeterKietzmann added Area: network Area: Networking State: WIP State: The PR is still work-in-progress and its code is not in its final presentable form yet Type: new feature The issue requests / The PR implemements a new feature for RIOT Area: pkg Area: External package ports labels Feb 15, 2018

PeterKietzmann assigned jia200x Feb 15, 2018

PeterKietzmann changed the title ~~pkg/openwsn: rei-ntegrate the network stack as a package~~ pkg/openwsn: re-integrate the network stack as a package Feb 15, 2018

miri64 reviewed Feb 16, 2018

View reviewed changes

cgundogan mentioned this pull request Mar 1, 2018

How to port ccn-lite to Contiki-OS? cn-uofbasel/ccn-lite#205

Closed

fjmolinas and others added 3 commits April 3, 2018 17:54

Add OpenWSN as a package

0f448ad

Changes to enable AT86RF2XX basic mode

f7d1e7d

Change iotlab RTT config

1030668

PeterKietzmann force-pushed the OpenWSN_port branch from 396db24 to 1030668 Compare April 3, 2018 16:09

miri64 mentioned this pull request Jun 8, 2018

netif: use where appropriate #9287

Open

8 tasks

PeterKietzmann mentioned this pull request Jun 22, 2018

openwsn: remove deprecated application RIOT-OS/applications#41

Merged

PeterKietzmann mentioned this pull request Jul 19, 2018

pkg/lwip: assert on samr21-xpro #9573

Closed

miri64 self-assigned this Oct 8, 2018

kb2ma mentioned this pull request Dec 10, 2018

sock UDP for OpenWSN 6TiSCH pkg #10579

Closed

miri64 reviewed Jan 23, 2019

View reviewed changes

PeterKietzmann closed this Mar 24, 2019

miri64 mentioned this pull request Apr 18, 2019

GoMacH on nRF52840 #11421

Closed

fjmolinas mentioned this pull request Mar 6, 2020

6TiSCH not supported in RIOT #13259

Closed

4 tasks

fjmolinas mentioned this pull request Apr 6, 2020

pkg/openwsn: re-integrate the network stack as a package #13824

Merged

11 tasks

pkg/openwsn: re-integrate the network stack as a package #8570

pkg/openwsn: re-integrate the network stack as a package #8570

Uh oh!

Conversation

PeterKietzmann commented Feb 15, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Contribution description

Implementation Issues

People that might be interested and in charge to support

Issues/PRs references

Uh oh!

PeterKietzmann commented Feb 15, 2018

Uh oh!

PeterKietzmann commented Feb 15, 2018

Uh oh!

miri64 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

PeterKietzmann commented Apr 3, 2018

Uh oh!

miri64 commented Oct 8, 2018

Uh oh!

kb2ma commented Nov 7, 2018

Uh oh!

miri64 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

miri64 commented Feb 26, 2019

Uh oh!

PeterKietzmann commented Mar 24, 2019

Uh oh!

Uh oh!

PeterKietzmann commented Feb 15, 2018 •

edited

Loading