gnrc_sixlowpan_frag: initial import of minimal forwarding by miri64 · Pull Request #11068 · RIOT-OS/RIOT

miri64 · 2019-02-26T14:30:12Z

Contribution description

This implements minimal fragment forwarding for GNRC.

Testing procedure

There is a test application tests/gnrc_sixlowpan_frag_minfwd. Having a set-up with >2 nodes and forward packets between them should also work, but for some reason (I suspect the radio) it doesn't work that well at the moment. Once I find out why, I will update the PR and the testing procedures accordingly (see #11256 for what I found out).

examples/gnrc_networking should work as normal on a 6LoWPAN compatible board when not using gnrc_sixlowpan_frag_minfwd and when using the module both with fragmentation (packet size >100) or without.

Issues/PRs references

Depends on ~~#11000, #11061, #11063, #11164, #11182, #11263, #11264, and #12629~~ (all merged):

miri64 · 2019-03-08T18:40:43Z

Rebased to current master and #11063 to resolve conflict caused by merging #11135.

miri64 · 2019-03-12T16:03:58Z

With the latest addition of #11164 as a dependency and the other fixes I am able to forward, at least when debug output is activated ... I'm not able to forward without it yet though :-(. I will for now first add some test-cases for the bugs I found in the real-world application, and then try to find the bug.

miri64 · 2019-03-12T16:05:36Z

OK, I got one successful send (out of 15 so far) :(

miri64 · 2019-03-23T17:40:41Z

In #11068 (comment) I wrote:

With the latest addition of #11164 as a dependency and the other fixes I am able to forward, at least when debug output is activated ... I'm not able to forward without it yet though :-(. I will for now first add some test-cases for the bugs I found in the real-world application, and then try to find the bug.

Some intensive days of hardware debugging later (thank you @PeterKietzmann for all the help and patience) we figured out what happens: appearently most of the time the framebuffer of the driver is overridden with the send data of the forwarded fragment, after the next fragment is received. I'll open a separate issue to explain this in a bit more detail.

miri64 · 2019-03-28T08:03:36Z

Rebased and adapted for current master and dependencies (including new dependencies #11263 and #11264)

miri64 · 2019-04-09T09:07:01Z

Rebased to include current RC (so I can refer to the 2019.04 release for my experiments)

miri64 · 2019-04-09T09:51:27Z

And rebased once again to include fix in gnrc_netif

miri64 · 2019-04-18T13:37:58Z

Set to WIP because I added some stuff that does not belong with this PR and need to be factored out.

miri64 · 2020-12-01T11:35:55Z

how many reassembly buffers can there?

Until the RAM is full, I'd say. The number is capped of by CONFIG_GNRC_SIXLOWPAN_FRAG_RBUF_SIZE.

Will we wait a bit before sending the next fragment, so not to immediately cause a collision with the retransmission from the next hop?

That's basically what is done in SFR (#12648). One could add this as a feature generally to 6LoWPAN, but I rather would do it in a separate PR and when we go into congestion control.

I didn't check yet, but I guess there is no effect on code size / behavior if the module is not used.

Should be, yes.

miri64 · 2020-12-01T12:54:07Z

Pushed everything. Please check, if everything was addressed.

miri64 · 2020-12-01T12:58:06Z

Should have at least compile tested before pushing ^^"

sys/net/gnrc/network_layer/sixlowpan/frag/rb/gnrc_sixlowpan_frag_rb.c

benpicco · 2020-12-01T13:41:34Z

Please squash!

See https://tools.ietf.org/html/draft-ietf-6lo-minimal-fragment-04

miri64 · 2020-12-01T14:10:30Z

Squashed!

benpicco · 2020-12-01T14:13:22Z

Squashed!

almost 😉

miri64 · 2020-12-01T14:15:42Z

Oh, there I seemed to have referenced a commit foreign to this PR (used git blame to identify it) Will see where I put it :D

miri64 · 2020-12-01T14:18:20Z

Fixed!

benpicco

I didn't test multi-hop fragmenting since I don't have a test setup for that (already thought of extending the zep_dispatcher), but I rely on your extensive testing documented in your paper.

I can confirm that there are no regressions when the module is not used, so ACK.

miri64 · 2020-12-01T14:43:48Z

Fixed and squashed the one doxygen error caused by a legacy name.

benpicco · 2020-12-01T15:16:14Z

ah looks like the test will need a Makefile.ci.
Do we have a script yet to generate that automatically, like add_insufficient_memory_board.sh?

When a VRB entry exists use minfwd to forward. When a route exist for the first fragment received in reassembly create a virtual reassembly buffer entry.

This is only the case when minfwd is compiled in.

miri64 · 2020-12-01T15:26:04Z

And added boards that are to small to Makefile.ci to make Murdock happy.

miri64 · 2020-12-01T15:29:33Z

Do we have a script yet to generate that automatically, like add_insufficient_memory_board.sh?

Murdock reported 35 boards. Most GNRC applications list 35 boards in their Makefile.ci, so I just copied it from tests/gnrc_dhcpv6_client. Chances are, those are the same boards ;-).

miri64 · 2020-12-01T17:59:28Z

Finally 🎉

miri64 added Area: network Area: Networking Type: new feature The issue requests / The PR implemements a new feature for RIOT State: waiting for other PR State: The PR requires another PR to be merged first labels Feb 26, 2019

miri64 requested review from cgundogan and waehlisch February 26, 2019 14:30

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 93d3b6c to 6542506 Compare March 8, 2019 18:40

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 2e4c0ad to e64b92b Compare March 12, 2019 16:00

miri64 mentioned this pull request Mar 14, 2019

gnrc_ipv6_nib: check if pkt is NULL on error #11182

Merged

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 66e2538 to e19efcd Compare March 20, 2019 15:45

This was referenced Mar 24, 2019

tests: add tests for netdev flooding race-condition #11256

Open

gnrc_netif: add packet to queue when device is busy #11263

Merged

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from e19efcd to f9326d4 Compare March 28, 2019 08:02

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 1596f20 to 9ef5f1d Compare March 29, 2019 11:38

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 91032ad to 189ff58 Compare April 9, 2019 08:56

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch 2 times, most recently from 496d05c to 9d835f2 Compare April 9, 2019 11:18

miri64 added the State: WIP State: The PR is still work-in-progress and its code is not in its final presentable form yet label Apr 18, 2019

benpicco reviewed Dec 1, 2020

View reviewed changes

sys/net/gnrc/network_layer/sixlowpan/frag/rb/gnrc_sixlowpan_frag_rb.c Show resolved Hide resolved

gnrc_sixlowpan_frag: initial import of minimal forwarding

4ce2d01

See https://tools.ietf.org/html/draft-ietf-6lo-minimal-fragment-04

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 229459b to c8ef052 Compare December 1, 2020 14:10

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from c8ef052 to 2c35465 Compare December 1, 2020 14:18

benpicco approved these changes Dec 1, 2020

View reviewed changes

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 2c35465 to 3a90dc2 Compare December 1, 2020 14:42

miri64 and others added 8 commits December 1, 2020 16:25

tests: introduce tests for gnrc_sixlowpan_frag_minfwd

8564a86

gnrc_sixlowpan_frag_rb: make rbuf pointer generic

df1171a

gnrc_sixlowpan_frag_rb: add VRB and use minfwd handling

7fe07e3

When a VRB entry exists use minfwd to forward. When a route exist for the first fragment received in reassembly create a virtual reassembly buffer entry.

tests: add gnrc_sixlowpan_frag_minfwd tests for rbuf integration

383790a

gnrc_sixlowpan_iphc/minfwd: put only IPHC dispatch in first fragment

d9be79b

This is only the case when minfwd is compiled in.

tests: add gnrc_sixlowpan_frag_minfwd tests for IPHC sending

d8c5a22

gnrc_sixlowpan_iphc: forward received fragments

27e1236

tests: add gnrc_sixlowpan_frag_minfwd tests for IPHC reception

e70a42c

miri64 force-pushed the gnrc_sixlowpan_frag/new/minfwd branch from 3a90dc2 to e70a42c Compare December 1, 2020 15:25

benpicco merged commit b18cbd8 into RIOT-OS:master Dec 1, 2020

miri64 deleted the gnrc_sixlowpan_frag/new/minfwd branch December 1, 2020 17:56

Conversation

miri64 commented Feb 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Contribution description

Testing procedure

Issues/PRs references

Uh oh!

miri64 commented Mar 8, 2019

Uh oh!

miri64 commented Mar 12, 2019

Uh oh!

miri64 commented Mar 12, 2019

Uh oh!

miri64 commented Mar 23, 2019

Uh oh!

miri64 commented Mar 28, 2019

Uh oh!

miri64 commented Apr 9, 2019

Uh oh!

miri64 commented Apr 9, 2019

Uh oh!

miri64 commented Apr 18, 2019

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

Uh oh!

benpicco commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

benpicco commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

benpicco left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

benpicco commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

miri64 commented Dec 1, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

miri64 commented Feb 26, 2019 •

edited

Loading

benpicco left a comment •

edited

Loading