babeld: add new package by BKPepe · Pull Request #30162 · openwrt/packages

BKPepe · 2026-08-06T08:10:00Z

Adds babeld from the openwrt/routing feed — the routing packages are being moved into openwrt/packages one by one, as discussed in openwrt/routing#184.

Includes the small MAKE_FLAGS cleanup pending in openwrt/routing#1197.

The content matches the current routing feed master. Once this is merged, the package will be removed from the routing feed (a coordinated removal PR is prepared there).

Maintainer: @PolynomialDivision

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

BKPepe · 2026-08-06T09:52:06Z

+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Nick Hainke <vincent@systemli.org>
+Date: Thu, 17 Dec 2020 12:41:32 +0100
+Subject: [PATCH] add ubus bindings


@PolynomialDivision This is your patch, please take a look if it can be upstreamed or what to do with this one

it is ubus and so openwrt specific. I think it is not upstreamable.

I know it’s ubus and pretty OpenWrt-specific, but our goal should simply be to get some eyes on the patch and have it reviewed, ideally by an upstream developer. I honestly think you could put in a bit of effort and try to upstream it, because otherwise this patch will just sit in our repo forever and we'll be stuck maintaining it. If upstream rejects it, so be it, but right now we're just acting on the assumption that someone, somewhere, thinks upstream won't take it.

If you look at the Babel repository, you'll see that I tried to upstream multiple PRs, but they're still open. So I don't know if that is worth the effort.

Come on, I see you have two open pull requests there. One is a WIP, and the other is just waiting to be merged and looks good. I really don't think it's that hard to just take the patch and put it out there. This patch has been sitting here since 2020, and with this kind of mindset, we're really not going to get anywhere. Meanwhile, babeld development is pretty active, so in my view, we're mostly just making excuses instead of putting in the effort to make things better around here.

openwrt-ai

Reviewed 1 new commit; the commit message matches the change. Packaging looks sound — conffiles layout, the procd init script, the 2-space/tab indentation split in the Makefile, and the literal BuildPackage,babeld all match the feed conventions, and src/ubus.c/ubus.h land in the build dir via the default Build/Prepare, so no explicit prepare step is needed.

The only finding I'd call a real defect is the uninitialized metric in babeld_ubus_add_filter() — one line, and worth fixing while the code is being moved rather than after. The unvalidated filter type next to it is a question, not a blocker. The rest are marked nit:, plus one non-blocking heads-up about this feed's generic runtime version check, which the routing feed does not run.

I left the patch's missing Signed-off-by/upstream reference alone since you've already raised the upstreaming question with @PolynomialDivision on that file.

Generated by Claude Code

openwrt-ai · 2026-08-06T09:58:55Z

+  struct blob_buf b = {0};
+  struct filter *filter = NULL;
+  char *ifname;
+  int metric, type;


metric is only assigned when tb[FILTER_METRIC] is present, but it is read unconditionally at filter->action.add_metric = metric; (ubus.c:109). metric is optional in filter_policy — only ifname and type are rejected when missing — so ubus call babeld add_filter '{"ifname":"eth0","type":0}' installs a filter whose metric is whatever was on the stack. In babeld a filter's add_metric of INFINITY means "deny", so a garbage value can silently turn an allow-filter into a deny-filter.

Suggested change

int metric, type;

int metric = 0, type;

Generated by Claude Code

fixed, thanks

Generated by Claude Code

openwrt-ai · 2026-08-06T09:59:04Z

+  if (!tb[FILTER_TYPE])
+    return UBUS_STATUS_INVALID_ARGUMENT;
+
+  type = blobmsg_get_u32(tb[FILTER_TYPE]);


type is taken straight from the ubus message and handed to add_filter(filter, type) without a range check, even though ubus.h:5-10 documents only 0..3 (FILTER_TYPE_INPUT … FILTER_TYPE_INSTALL) as valid.

I can't see babeld's add_filter() from this diff to confirm what it does with an out-of-range type — if it ends its if/else chain in abort() (as several babeld dispatchers do), an unexpected value from a ubus caller would take the daemon down rather than return an error. Would it be worth rejecting out-of-range values with UBUS_STATUS_INVALID_ARGUMENT here, so the validation doesn't depend on upstream's internals?

Generated by Claude Code

openwrt-ai · 2026-08-06T09:59:22Z

+include $(TOPDIR)/rules.mk
+
+PKG_NAME:=babeld
+PKG_VERSION:=1.13.1


Heads-up rather than a defect: this feed's CI runs a generic runtime check that executes each installed binary with version/help flags and expects PKG_VERSION in the output — the routing feed doesn't run that, so it's new surface for this package.

Two things make it worth a look before merge: babeld's own option parser is short-option based, and 600-add-ubus.patch rewrites generate-version.sh to emit "$version-ubus-mod". The suffix itself is harmless (1.13.1-ubus-mod still contains 1.13.1), but if $version resolves to unknown when building from the release tarball rather than a git checkout, the printed string won't contain 1.13.1 at all. If the check doesn't pass, the fix is a net/babeld/test-version.sh that returns 0 for this package. The multi-arch jobs were still queued when I looked, so I couldn't confirm either way.

Generated by Claude Code

openwrt-ai

Reviewed 1 new commit; no new issues found.

fc5110b does exactly what its message says — int metric = 0, type; in babeld_ubus_add_filter() — and closes the uninitialized-read path I flagged. Nothing else changed.

Two follow-ups from the previous round, for the record:

The generic runtime version check I warned about is answered: all ten Feeds Package Test Build jobs are green on fc5110b, so no net/babeld/test-version.sh is needed and the -ubus-mod version suffix is not a problem in practice.
Still open, both non-blocking and unchanged since they live in the moved-as-is code: the unvalidated filter type handed to add_filter(), the /tmp/babel.d/ vs /tmp/babeld.d/ typo in files/babeld.config, and the tab+spaces indentation in 600-add-ubus.patch. None of these need to hold up the move from the routing feed — they can be follow-ups here just as easily.

Generated by Claude Code

The package has been moved to the openwrt/packages feed, as discussed in openwrt#184. See openwrt/packages#30162. Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

BKPepe · 2026-09-07T06:50:14Z

ping @PolynomialDivision

BKPepe · 2026-09-09T08:00:27Z

@PolynomialDivision Gentle ping this is sitting here for a month.

openwrt · 2026-09-11T18:50:10Z

Formality Check: Failed

We checked this pull request against the contribution guidelines. Here is what needs your attention:

🛑 CRITICAL ERRORS

Commit f78c2c8 - Update net/babeld/files/babeld.config:

Commit subject must start with <package name or prefix>:

Commit cab46c4 - Update net/babeld/patches/600-add-ubus.patch:

Commit subject must start with <package name or prefix>:

Commit 0169cb7 - Update net/babeld/files/babeld.config:

Commit subject must start with <package name or prefix>:

Commit 9b7d496 - Update net/babeld/patches/600-add-ubus.patch:

Commit subject must start with <package name or prefix>:

Tip

Do not close this pull request to make corrections. Instead, modify your existing commits (e.g. git commit --amend) and update the branch using git push --force-with-lease --force-if-includes. The checks will re-run automatically.

_{Something broken? Consider reporting an issue.}
_{Running version 26ec621 deployed on 2026-09-20 09:46:35 CEST}

openwrt-ai

Commit checks

3517621 "Update net/babeld/files/babeld.config" — GitHub-web template subject, no body, and it only corrects a typo in the file added by 93f3f3d in this same unmerged series; squash it into 93f3f3d.
476205a "Update net/babeld/patches/600-add-ubus.patch" — same shape, same series; squash into 93f3f3d so the patch is added in its final form.

Generated by Claude Code

openwrt-ai

Commit checks

f02c5ef, 7316b6e — template subjects with no body, both fixing up 93f3f3d inside this same unmerged series; squash all four fixups into 93f3f3d so the package is added in its final form.

Generated by Claude Code

openwrt-ai · 2026-09-13T05:27:16Z

+             local_notify_route_1(&local_sockets[i], route, kind);
+     }
+    if(ubus_bindings)
+	    ubus_notify_route(route, kind);


nit: last remaining tab+four-spaces line in this patch — the other two if(ubus_bindings) bodies are now four spaces, and surrounding babeld code is four spaces per level.

Suggested change

+ ubus_notify_route(route, kind);

+ ubus_notify_route(route, kind);

Generated by Claude Code

openwrt-ai

Reviewed 6 new commits; no new issues found.

Generated by Claude Code

IvanTheGeek · 2026-09-23T19:21:55Z

A data point from downstream, from someone waiting on this one. Since routing#1198 landed on 2026-08-11, SNAPSHOT has had no babeld in any feed. luci-app-babeld is still in luci master (LUCI_DEPENDS:=+luci-base +babeld), so every SNAPSHOT build now prints

WARNING: Makefile 'package/feeds/luci/luci-app-babeld/Makefile' has a dependency on 'babeld', which does not exist

and neither package can be selected. openwrt-25.12 is fine: its routing feed still has 1.13.1-r3, including the add_filter metric fix.

As far as I can tell from this thread, the only thing left is the four Update net/babeld/... fixup commits that the Formality Check flags. They are the maintainer's own review fixes and carry his Signed-off-by, so they only need squashing into the first commit, as openwrt-ai suggested. Until then the Feeds Package Test Build is skipped, so CI has not built the current head.

We plan to run babeld on snapshot-based builds for mediatek/filogic (Cudy WR3000S and WR3000H).

Edit: I first listed "an ack from the maintainer" as outstanding, but those fixups are already his.

AI assistance: researched and drafted with help from Claude Opus 5.5 (Anthropic); I reviewed it before posting.

Copilot

Copilot review overview

🟡 Changes recommended

The ubus API has malformed notification payloads, missing recovery and validation paths, and the init script can generate invalid configuration.

Review effort: Balanced
Findings: 2 · 4

Open (6)

Exclude the OpenWrt ignore option from babeld configuration · New
Handle route node allocation failures safely · New
Validate filter type and handle add_filter failure · New
Recover the ubus connection after ubusd restarts · New
Serialize notification route fields at the message root · New
Emit documented neighbour event names · New

+			# "option ifname" is a special option, don't actually
+			# generate configuration for it.
+			[ "$option" = "ifname" ] && return
+			[ -n "$interface" ] && _interface="interface $interface" || _interface="default"
+			cfg_append "$_interface ${option//_/-} $value"


+      struct xroute_list_entry *xr =
+          calloc(1, sizeof(struct xroute_list_entry));
+      xr->xroute = xroute;


+  if (!tb[FILTER_TYPE])
+    return UBUS_STATUS_INVALID_ARGUMENT;
+
+  type = blobmsg_get_u32(tb[FILTER_TYPE]);


+  shared_ctx = ubus_connect(NULL);
+  if (!shared_ctx)
+    return false;


+  blob_buf_init(&b, 0);
+  babeld_add_route_buf(route, &b);
+  snprintf(method, sizeof(method), "route.%s", local_kind(kind));
+  ubus_notify(shared_ctx, &babeld_object, method, b.head, -1);


+
+  blob_buf_init(&b, 0);
+  babeld_add_neighbour_buf(neigh, &b);
+  snprintf(method, sizeof(method), "neigh.%s", local_kind(kind));


Babel is a loop-avoiding distance-vector routing protocol for IPv6 and IPv4 with fast convergence properties (RFC 8966). Moved from the openwrt/routing feed, as discussed in openwrt/routing#184. Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

The metric is optional in filter_policy: only ifname and type are rejected when missing, so `ubus call babeld add_filter '{"ifname":"eth0","type":0}'` reaches filter->action.add_metric = metric; with metric never assigned, i.e. whatever was left on the stack. In babeld an add_metric of INFINITY means "deny", so an uninitialized read can silently turn an allow filter into a deny filter. Initialize it to 0, which is the neutral value the filter would have had if the caller had passed it explicitly. Reported-by: openwrt-ai[bot] Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Co-authored-by: OpenWrt AI review account <openwrt-ai@hauke-m.de> Signed-off-by: Nick Hainke <vincent@systemli.org>

BKPepe · 2026-10-03T18:30:38Z

Those commits should be squashed before merging and someone e.g. @PolynomialDivision should take a look at Copilot's review

robimarko · 2026-10-03T18:36:19Z

Well, I am getting quite annoyed with the dropping of the package for it to just sit in this state here for months

Copilot AI lite review requested due to automatic review settings August 6, 2026 08:10

openwrt Bot added Add package not following guidelines labels Aug 6, 2026

Copilot AI reviewed Aug 6, 2026

BKPepe mentioned this pull request Aug 6, 2026

treewide: remove packages moved to openwrt/packages openwrt/routing#1198

Merged

BKPepe force-pushed the add-babeld branch from d44f7c6 to 93f3f3d Compare August 6, 2026 09:36

openwrt Bot removed the not following guidelines label Aug 6, 2026

BKPepe commented Aug 6, 2026

View reviewed changes

openwrt-ai reviewed Aug 6, 2026

View reviewed changes

BKPepe assigned PolynomialDivision Aug 18, 2026

openwrt Bot added the not following guidelines label Sep 11, 2026

openwrt-ai reviewed Sep 11, 2026

View reviewed changes

Comment thread net/babeld/files/babeld.config Outdated

Comment thread net/babeld/patches/600-add-ubus.patch Outdated

openwrt-ai reviewed Sep 13, 2026

View reviewed changes

BKPepe force-pushed the add-babeld branch from 7316b6e to bc68bc3 Compare September 21, 2026 20:15

openwrt-ai reviewed Sep 22, 2026

View reviewed changes

BKPepe requested a balanced review from Copilot October 2, 2026 08:05

Copilot started reviewing on behalf of BKPepe October 2, 2026 08:07 View session

Copilot AI reviewed Oct 2, 2026

View reviewed changes

BKPepe and others added 6 commits October 3, 2026 20:28

Update net/babeld/files/babeld.config

f78c2c8

Co-authored-by: OpenWrt AI review account <openwrt-ai@hauke-m.de> Signed-off-by: Nick Hainke <vincent@systemli.org>

Update net/babeld/patches/600-add-ubus.patch

cab46c4

Co-authored-by: OpenWrt AI review account <openwrt-ai@hauke-m.de> Signed-off-by: Nick Hainke <vincent@systemli.org>

Update net/babeld/files/babeld.config

0169cb7

Co-authored-by: OpenWrt AI review account <openwrt-ai@hauke-m.de> Signed-off-by: Nick Hainke <vincent@systemli.org>

Update net/babeld/patches/600-add-ubus.patch

9b7d496

Co-authored-by: OpenWrt AI review account <openwrt-ai@hauke-m.de> Signed-off-by: Nick Hainke <vincent@systemli.org>

robimarko force-pushed the add-babeld branch from bc68bc3 to 9b7d496 Compare October 3, 2026 18:28

	+ ubus_notify_route(route, kind);
	+ ubus_notify_route(route, kind);

Conversation

BKPepe commented Aug 6, 2026

Uh oh!

Copilot AI 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!

openwrt-ai 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!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

openwrt-ai left a comment

Choose a reason for hiding this comment

Uh oh!

BKPepe commented Sep 7, 2026

Uh oh!

BKPepe commented Sep 9, 2026

Uh oh!

openwrt Bot commented Sep 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Formality Check: Failed

🛑 CRITICAL ERRORS

Uh oh!

openwrt-ai left a comment

Choose a reason for hiding this comment

Commit checks

Uh oh!

Uh oh!

Uh oh!

openwrt-ai left a comment

Choose a reason for hiding this comment

Commit checks

Uh oh!

Choose a reason for hiding this comment

Uh oh!

openwrt-ai left a comment

Choose a reason for hiding this comment

Uh oh!

IvanTheGeek commented Sep 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

BKPepe commented Oct 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

robimarko commented Oct 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

openwrt Bot commented Sep 11, 2026 •

edited

Loading

IvanTheGeek commented Sep 23, 2026 •

edited

Loading

BKPepe commented Oct 3, 2026 •

edited

Loading