ng_net: header building facilities by miri64 · Pull Request #2575 · RIOT-OS/RIOT

miri64 · 2015-03-10T22:29:52Z

This was taken out of #2553

The idea is that an upper-layer can build a header for the layer it wants to send data over without knowledge about the actual structure of the header. I also added a convinience function of similar format to ng_netif_hdr.

~~Depends on #2706~~ (merged)

OlegHahm · 2015-03-13T14:04:08Z

sys/include/net/ng_netif/hdr.h

s/network interface/network layer/ ? But what does "generic" mean in this context?

No this is correct. This module (ng_netif_hdr) defines the "generic network interface header". This name was introduced after @OlegHahm was not happy with the name "generic link layer header" ;-)

I'm confused. I thought this was meant to implement (in a IMO weird way) IPv6 upper layer checksum calculation. Why do one need to build a generic network interface header?

No that's in #2553. This just offers netif_hdr_init() and pktbuf_add() in one go.

OlegHahm · 2015-03-13T14:44:18Z

Sorry, I guess I asked this a 100 times before, but I'm confused again why an upper layer wants to build a lower layer's header?

miri64 · 2015-03-13T14:46:30Z

Use-case: IPv6 wants to send packet.

IPv6 checks next hop
IPv6 searches link-layer address for next hop
wants to tell link layer were to send it (and maybe additional information): uses link-layer header.

miri64 · 2015-03-13T14:48:06Z

Why not just the address? Because in the rest of the model we use the header for exactly the same reason as I sketched out previously: The upper-layer might want to set additional information (hop-limit, traffic class, etc. you name it).

OlegHahm · 2015-03-13T15:02:33Z

The upper-layer might want to set additional information (hop-limit, traffic class, etc. you name it).

But this would require an additional function, right? Why can I do this not just with a set_opt command?

miri64 · 2015-03-13T15:09:01Z

Because this requires additional context switches. And no it would not need additional functions. Once you've build the header, you know the type of the header anyways, so if you need to, you can just change the fields after the creation. This functions are primarily for service functions for layers that DON'T need to know the build-up of a lower-layer header.

miri64 · 2015-03-19T14:50:26Z

Rebased to current master

Lotterleben · 2015-03-24T11:08:02Z

sys/include/net/ng_netreg.h

nitpick: I find this sentence a bit hard to parse. Maybe the “for sending” could be left out?

No, since a receiving packet would store the payload differently.

Oh. Then maybe we could split it up into two sentences or something?

"Builds a header for sending and adds it to the packet buffer" would also work, I think. That the protocol is defined through type is apparent through the type parameter I think.

miri64 · 2015-03-24T18:15:27Z

Simplified doc a little.

miri64 · 2015-03-25T01:19:25Z

Needed to add #2706 as a depenency since Travis is taking forever -.-

Lotterleben · 2015-03-25T12:25:07Z

Awww come on Travis!

Kijewski · 2015-03-25T12:31:31Z

The last successful Travis run is two and a half days old. I think we've killed that poor fellow.

saurabh984 · 2015-03-25T12:44:45Z

I'll sponsor a dedicated server if anyone fancies setting up Janky (or Jenkins) on it! Travis is a nighmare

miri64 · 2015-03-25T12:47:21Z

@saurabh984 we're (or rather @phiros is) working on a solution. We had Jenkins in the past. This was also no solution, since someone(TM) needs to manage that machine and nobody got time for that.

miri64 · 2015-03-25T12:47:44Z

It's probably gonna be build bot.

miri64 · 2015-03-25T12:48:20Z

But thanks for the offer

miri64 · 2015-03-25T12:57:48Z

Fixed thingie

miri64 · 2015-03-26T05:47:09Z

Moved author tag.

miri64 · 2015-03-28T17:10:44Z

Apart from the squash and PR dependency warning and an aborted (and now restarted) build in the msp430 group travis seems to be fine. Squash?

Lotterleben · 2015-03-28T17:12:52Z

Yes please!

miri64 · 2015-03-28T17:18:24Z

Done

Lotterleben · 2015-03-28T17:20:09Z

👍 Okay... Let's do this! ACK & Go.

Lotterleben · 2015-03-28T17:20:44Z

Correction: ACK, wait for Travis & go. I'll keep this tab open.

miri64 · 2015-03-28T17:31:25Z

Ahhh sorry… I needed to change something for constistency: s/dest/dst/

Lotterleben · 2015-03-28T17:36:36Z

:D good catch! Tab is still open, will merge as soon as Travis is happy.

OlegHahm · 2015-03-28T20:22:10Z

I would like to put this topic on the list for the next NSTF meeting. I'm not nacking this, but am still not convinced of this approach.

miri64 · 2015-03-29T00:13:09Z

@OlegHahm can you point me to the problem with this approach?

ng_net: header building facilities

OlegHahm · 2015-03-30T14:31:07Z

No, I just still don't like upper layers building lower layers' headers. But it's so far more like a gut feeling.

miri64 added Area: network Area: Networking Type: new feature The issue requests / The PR implemements a new feature for RIOT Process: API change Integration Process: PR contains or issue proposes an API change. Should be handled with care. NSTF labels Mar 10, 2015

miri64 added this to the Network Stack Task Force milestone Mar 10, 2015

miri64 assigned haukepetersen Mar 10, 2015

This was referenced Mar 10, 2015

ng_net: introduce checksum calculation #2553

Merged

ng_ipv6: initial import #2454

Merged

OlegHahm reviewed Mar 13, 2015
View reviewed changes

miri64 force-pushed the ng_net/feat/hdr-build branch from 6881f2e to d6b7e38 Compare March 19, 2015 14:50

miri64 mentioned this pull request Mar 22, 2015

ng_slip: initial import #2688

Merged

miri64 assigned Lotterleben and unassigned haukepetersen Mar 23, 2015

Lotterleben reviewed Mar 24, 2015
View reviewed changes

miri64 force-pushed the ng_net/feat/hdr-build branch from 3dad5a1 to c547d93 Compare March 25, 2015 01:18

miri64 added CI: needs squashing Commits in this PR need to be squashed; If set, CI systems will mark this PR as unmergable State: waiting for other PR State: The PR requires another PR to be merged first labels Mar 25, 2015

miri64 force-pushed the ng_net/feat/hdr-build branch from c547d93 to 9674697 Compare March 25, 2015 02:06

This was referenced Mar 26, 2015

sys/shell: added UDP shell commands #2720

Closed

net: added ng_udp implementation #2430

Merged

miri64 removed the State: waiting for other PR State: The PR requires another PR to be merged first label Mar 28, 2015

miri64 force-pushed the ng_net/feat/hdr-build branch from 44e0eb1 to 5ae0325 Compare March 28, 2015 17:18

miri64 added 2 commits March 28, 2015 18:28

netreg: add multiplexer for header building

69c593b

netif: add convinience function for header building

1ea70be

miri64 force-pushed the ng_net/feat/hdr-build branch from 5ae0325 to 1ea70be Compare March 28, 2015 17:30

miri64 removed the CI: needs squashing Commits in this PR need to be squashed; If set, CI systems will mark this PR as unmergable label Mar 29, 2015

miri64 added a commit that referenced this pull request Mar 29, 2015

Merge pull request #2575 from authmillenon/ng_net/feat/hdr-build

fa689fe

ng_net: header building facilities

miri64 merged commit fa689fe into RIOT-OS:master Mar 29, 2015

miri64 deleted the ng_net/feat/hdr-build branch March 29, 2015 12:28

Conversation

miri64 commented Mar 10, 2015

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!

OlegHahm commented Mar 13, 2015

Uh oh!

miri64 commented Mar 13, 2015

Uh oh!

miri64 commented Mar 13, 2015

Uh oh!

OlegHahm commented Mar 13, 2015

Uh oh!

miri64 commented Mar 13, 2015

Uh oh!

miri64 commented Mar 19, 2015

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 Mar 24, 2015

Uh oh!

miri64 commented Mar 25, 2015

Uh oh!

Lotterleben commented Mar 25, 2015

Uh oh!

Kijewski commented Mar 25, 2015

Uh oh!

saurabh984 commented Mar 25, 2015

Uh oh!

miri64 commented Mar 25, 2015

Uh oh!

miri64 commented Mar 25, 2015

Uh oh!

miri64 commented Mar 25, 2015

Uh oh!

miri64 commented Mar 25, 2015

Uh oh!

miri64 commented Mar 26, 2015

Uh oh!

miri64 commented Mar 28, 2015

Uh oh!

Lotterleben commented Mar 28, 2015

Uh oh!

miri64 commented Mar 28, 2015

Uh oh!

Lotterleben commented Mar 28, 2015

Uh oh!

Lotterleben commented Mar 28, 2015

Uh oh!

miri64 commented Mar 28, 2015

Uh oh!

Lotterleben commented Mar 28, 2015

Uh oh!

OlegHahm commented Mar 28, 2015

Uh oh!

miri64 commented Mar 29, 2015

Uh oh!

OlegHahm commented Mar 30, 2015

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!