ng_rpl: port to the new network stack #3050

cgundogan · 2015-05-22T20:38:29Z

~~Depends on: #2783 and #2818~~

This PR introduces RPL for the new network stack. I used the old RPL implementation as a blueprint and copied code blocks, which could be reused.

For a better internal dodag management I made use of utlist.h (linked list).
Thus, it was quite easy to adopt the code to support multiple dodags and/or instances.

Furthermore, I changed the parsing of control packets. The old implementation did use a mediary data structure to parse all the information and then assign them to the internal dodag structure. I completely removed this mediary step and operate directly on the internal data structures.

While adopting the code, I found small discrepancies betwen the implementation and the RFC. I tried to correct them to the best of my knowledge.

This port is far from being complete, as I only included the Storing mode of operation
Missing things:

Non-Storing Mode of Operation
- Source Routing Header missing RFC6554 (see also ng_ipv6_ext: initial import of extension header handling (including RPL SRH) #2898)
ETX/MrHof is not included, because ETX needs to be ported to the new network stack separately
some message options are missing, like the Prefix Information Object (in DIOs) ng_rpl: auto address configuration #3489
RPL Information Option missing RFC6553

I did all my testing with the ng_nativenet implementation by rebasing this to the branch of @kaspar030 #2776 and then use the test application in ./tests/driver_netdev_eth/.
There I modifid the Makefile to include ng_ipv6_default and ng_rpl.

When starting two to four nodes (tap0..tap3), I first have to add an IPv6 address to the nodes and can then initialize RPL on the given interface.

ifconfig 5 add 2001:db8::x/64
rpl init 5

Please make sure, that you use the same prefix part for all nodes.

On one node I create a DODAG by issuing: rpl root 1 2001:db8::x, where 1 is the instance id by choice and 2001:db8::x the dodag id. Note that you should not use the prefix notation (*/xx) like I did for ifconfig here.

rpl shows some basic information about the dodags and rpl help will display some more commands, like deleting/adding dodags, starting/stoping trickle timers, sending DIS etc.

I would like to test this on some iotlab m3s, but I don't know how. Could somebody point me to a test/example application, which I can use as a baseline?

@OlegHahm @BytesGalore @authmillenon @gebart @Lotterleben @haukepetersen @emmanuelsearch and basically everyone else: please help me by reviewing this code of a Kraken...

@authmillenon you should especially look into the communication with the network layer part, because I do not know if I implemented it the right way - it works though.

miri64 · 2015-05-22T22:29:05Z

I did all my testing with the ng_nativenet implementation by rebasing this to the branch of @kaspar030 #2776 and then use the test application in ./tests/driver_netdev_eth/.
There I modifid the Makefile to include ng_ipv6_default and ng_rpl.

I advice to use ng_ipv6_router_default instead. This way to FIB is included automatically, as well as the router behavior for NDP/6LOWPAN-ND (later on). Also you compile it in the configuration that RPL will be used most likely in.

miri64 · 2015-05-22T22:41:27Z

(for the router behavior of NDP you need #3049 of course)

OlegHahm · 2015-05-26T18:44:08Z

sys/include/net/ng_nettype.h

@@ -58,6 +58,9 @@ typedef enum {
 #ifdef MODULE_NG_ICMPV6
    NG_NETTYPE_ICMPV6,          /**< Protocol is ICMPv6 */
 #endif
+#ifdef MODULE_NG_RPL
+    NG_NETTYPE_RPL_CTRL,        /**< Protocol is ICMPv6, RPL sub-type */


I think I would prefer something like NG_NETTYPE_ICMP_RPL.

It shouldn't need a NG_NETTYPE at all oO let me review ths

It's not even used ;-)

Demux is done in https://github.com/RIOT-OS/RIOT/pull/3050/files#diff-a263cf517da28750259527ba0aab62dfR122 so everything is as it should be :-)

Right, seems like a left-over from my previous attempt.. will address all your comments soon

OlegHahm · 2015-05-26T20:27:08Z

sys/net/network_layer/fib/universal_address.c

+        /* if the bytes-1 equals we check the bits of the lowest byte */
+        uint8_t bitmask = 0x00;
+        /* get a bitmask for the trailing 0b */
+        for( uint8_t j = 0; j < 8; ++j ) {


again wrong prefix incrementor?

OlegHahm · 2015-08-17T09:21:16Z

I think we're finally ready to go: ACK and go as soon as Travis is green.

cgundogan · 2015-08-17T09:51:26Z

travis is not happy, because I modified netif.h. And netif.h has some doxygen warnings. (see #3641)

jnohlgard · 2015-08-17T09:58:10Z

I don't like merging this without any Travis-enabled build with USEMODULE+=ng_rpl. Currently the RPL implementation is not even being compile-tested by Travis.

cgundogan · 2015-08-17T10:02:11Z

@gebart good catch. I will add ng_rpl to the ng_networking example now. When the future of ng_networking is decided, this can be changed later in follow-up PRs.

jnohlgard · 2015-08-17T11:15:24Z

ACK, let's wait for Travis

OlegHahm · 2015-08-17T11:47:12Z

Spark-Core needs black-listing.

OlegHahm · 2015-08-17T11:47:50Z

cppcheck complains:

sys/net/routing/ng_rpl/ng_rpl_control_messages.c:398: warning (nullPointer): Possible null pointer dereference: parent - otherwise it is redundant to check it against null.

sys/net/routing/ng_rpl/ng_rpl_control_messages.c:38: style (variableScope): The scope of the variable 'tmp' can be reduced.

sys/net/routing/ng_rpl/ng_rpl_control_messages.c:95: style (variableScope): The scope of the variable 'dodag_conf' can be reduced.

sys/shell/commands/sc_ng_rpl.c:31: performance (redundantAssignment): Variable 'iface_pid' is reassigned a value before the old one has been used.

sys/shell/commands/sc_ng_rpl.c:48: style (variableScope): The scope of the variable 'addr_str' can be reduced.

OlegHahm · 2015-08-17T20:52:32Z

Needs a rebase, too.

cgundogan · 2015-08-18T10:18:24Z

rebased and eliminated cppcheck warnings

cgundogan · 2015-08-18T14:11:22Z

travis approves

OlegHahm · 2015-08-18T14:28:23Z

Let me run one final test.

OlegHahm · 2015-08-18T14:42:51Z

Test runs fine, ACK holds - wanna hit the button yourself?

cgundogan · 2015-08-18T14:47:53Z

I am sitting in the bus right now. Merge at will. Otherwise I merge when I
get home
Am 18.08.2015 16:42 schrieb "Oleg Hahm" [email protected]:

Test runs fine, ACK holds - wanna hit the button yourself?

—
Reply to this email directly or view it on GitHub
#3050 (comment).

ng_rpl: port to the new network stack

emmanuelsearch · 2015-08-18T14:50:33Z

Congrats @cgundogan !

cgundogan · 2015-08-18T14:52:03Z

Eureka!
Am 18.08.2015 16:50 schrieb "Emmanuel Baccelli" [email protected]:

Congrats @cgundogan https://github.com/cgundogan !

—
Reply to this email directly or view it on GitHub
#3050 (comment).

miri64 · 2015-08-18T15:01:35Z

✨ \o/ ✨

kaspar030 · 2015-08-18T15:03:49Z

nice!

BytesGalore · 2015-08-19T04:21:15Z

bäääm, you did it sir 🎆

cgundogan added Area: network Area: Networking Type: new feature The issue requests / The PR implemements a new feature for RIOT NSTF labels May 22, 2015

cgundogan assigned OlegHahm May 22, 2015

OlegHahm reviewed May 26, 2015
View reviewed changes

OlegHahm added the CI: needs squashing Commits in this PR need to be squashed; If set, CI systems will mark this PR as unmergable label May 26, 2015

OlegHahm reviewed May 26, 2015
View reviewed changes

cgundogan force-pushed the ng_rpl_fib branch from 782df1d to 7d4da48 Compare August 17, 2015 10:40

cgundogan force-pushed the ng_rpl_fib branch from 7d4da48 to efbf3bf Compare August 18, 2015 09:35

cgundogan added 2 commits August 18, 2015 12:03

ng_rpl: RPL for the gnrc network stack

7f6966a

ng_rpl: shell commands for RPL

acd07de

cgundogan force-pushed the ng_rpl_fib branch from efbf3bf to 01a322f Compare August 18, 2015 10:03

gnrc_networking: add ng_rpl to example

d1b1949

cgundogan force-pushed the ng_rpl_fib branch from 01a322f to d1b1949 Compare August 18, 2015 10:35

OlegHahm added a commit that referenced this pull request Aug 18, 2015

Merge pull request #3050 from cgundogan/ng_rpl_fib

e3edf34

ng_rpl: port to the new network stack

OlegHahm merged commit e3edf34 into RIOT-OS:master Aug 18, 2015

miri64 mentioned this pull request Aug 18, 2015

gnrc_rpl: fix dependencies #3653

Merged

cgundogan deleted the ng_rpl_fib branch August 18, 2015 17:13

cgundogan mentioned this pull request Aug 18, 2015

rpl: omit dodag id in DAO(-ACK) for global instances #3659

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ng_rpl: port to the new network stack #3050

ng_rpl: port to the new network stack #3050

cgundogan commented May 22, 2015

miri64 commented May 22, 2015

miri64 commented May 22, 2015

OlegHahm May 26, 2015

miri64 May 26, 2015

miri64 May 26, 2015

miri64 May 26, 2015

cgundogan May 26, 2015

OlegHahm May 26, 2015

OlegHahm commented Aug 17, 2015

cgundogan commented Aug 17, 2015

jnohlgard commented Aug 17, 2015

cgundogan commented Aug 17, 2015

jnohlgard commented Aug 17, 2015

OlegHahm commented Aug 17, 2015

OlegHahm commented Aug 17, 2015

OlegHahm commented Aug 17, 2015

cgundogan commented Aug 18, 2015

cgundogan commented Aug 18, 2015

OlegHahm commented Aug 18, 2015

OlegHahm commented Aug 18, 2015

cgundogan commented Aug 18, 2015

emmanuelsearch commented Aug 18, 2015

cgundogan commented Aug 18, 2015

miri64 commented Aug 18, 2015

kaspar030 commented Aug 18, 2015

BytesGalore commented Aug 19, 2015

ng_rpl: port to the new network stack #3050

ng_rpl: port to the new network stack #3050

Conversation

cgundogan commented May 22, 2015

miri64 commented May 22, 2015

miri64 commented May 22, 2015

OlegHahm May 26, 2015

Choose a reason for hiding this comment

miri64 May 26, 2015

Choose a reason for hiding this comment

miri64 May 26, 2015

Choose a reason for hiding this comment

miri64 May 26, 2015

Choose a reason for hiding this comment

cgundogan May 26, 2015

Choose a reason for hiding this comment

OlegHahm May 26, 2015

Choose a reason for hiding this comment

OlegHahm commented Aug 17, 2015

cgundogan commented Aug 17, 2015

jnohlgard commented Aug 17, 2015

cgundogan commented Aug 17, 2015

jnohlgard commented Aug 17, 2015

OlegHahm commented Aug 17, 2015

OlegHahm commented Aug 17, 2015

OlegHahm commented Aug 17, 2015

cgundogan commented Aug 18, 2015

cgundogan commented Aug 18, 2015

OlegHahm commented Aug 18, 2015

OlegHahm commented Aug 18, 2015

cgundogan commented Aug 18, 2015

emmanuelsearch commented Aug 18, 2015

cgundogan commented Aug 18, 2015

miri64 commented Aug 18, 2015

kaspar030 commented Aug 18, 2015

BytesGalore commented Aug 19, 2015