[24.10] kernel: bump 6.6 to 6.6.137

[tmacholda]

Bumping from 6.6.135 to 6.6.137

[tmacholda]

pinging @hauke since you're bumping other stuff at the moment as well

[tmacholda]

nevermind, didn't think of refreshing non-generic patches...

[tmacholda]

i wrote a script to do it for all targets, but that seems a bit erroneous... is there a canonical way of refreshing patches for kernel version bumps?

[GoetzGoerisch]

i wrote a script to do it for all targets, but that seems a bit erroneous... is there a canonical way of refreshing patches for kernel version bumps?

https://github.com/openwrt/maintainer-tools/blob/master/update_kernel.sh

[hnyman]

Would be good to get 6.6.137 in, as it should solve the recent CVE-2026-31431 ("CopyFail", a serious local privilege escalation.

We will also need 6.18.12 and 6.12.85 for the newer branches.

[GoetzGoerisch]

Thank you @tmacholda

git diff 690dcc4a403c7b9314fc73db23ffff7e8dd67320 8474a31017540a84d905d44419bfb875081c1408
diff --git a/target/linux/generic/hack-6.6/902-debloat_proc.patch b/target/linux/generic/hack-6.6/902-debloat_proc.patch
index 0e66453df2..77b6856bda 100644
--- a/target/linux/generic/hack-6.6/902-debloat_proc.patch
+++ b/target/linux/generic/hack-6.6/902-debloat_proc.patch
@@ -291,14 +291,13 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
                        sizeof(struct seq_net_private)))
                goto out;
 -      if (!proc_create_seq("softnet_stat", 0444, net->proc_net,
--                       &softnet_seq_ops))
 +      if (!IS_ENABLED(CONFIG_PROC_STRIPPED) &&
 +                      !proc_create_seq("softnet_stat", 0444, net->proc_net,
-+                      &softnet_seq_ops))
+                        &softnet_seq_ops))
                goto out_dev;
 -      if (!proc_create_net("ptype", 0444, net->proc_net, &ptype_seq_ops,
 +      if (!IS_ENABLED(CONFIG_PROC_STRIPPED) &&
-+                      proc_create_net("ptype", 0444, net->proc_net, &ptype_seq_ops,
++                      !proc_create_net("ptype", 0444, net->proc_net, &ptype_seq_ops,
                        sizeof(struct ptype_iter_state)))
                goto out_softnet;

I think the refresh of generic/hack-6.6/902-debloat_proc.patch should look like above.

a is your commit .136 and b is my commit GoetzGoerisch@8474a31

As we are working on the same.

[tmacholda]

I think the refresh of generic/hack-6.6/902-debloat_proc.patch should look like above.

That looks good, thanks. I made a fixup commit in case you'd want to squash and merge this request still.

[tmacholda]

Though it's now failing on formalities, even though it's a temporary commit... I hope that this is the proper procedure and maintainers can autosquash before merge.

[tmacholda]

Though it's now failing on formalities, even though it's a temporary commit... I hope that this is the proper procedure and maintainers can autosquash before merge.

ready for review despite failing checks (formalities), jfyi @hauke

(please tell me if pinging under these circumstances is unnecessary)

[hnyman]

You need to squash the fixup commit into the commit that it fixes.
(It would be pointless to first upgrade to .136, then to 137 and only then fix the 136 commit.)

Pretty easy with the git rebase -i, and then force pushing the PR branch.

[tmacholda]

okay. i thought the changes would be more transparent like this so that's why i did it this way, leaving it up to the maintainers to squash.

[openwrt-ai]

Reviewed 2 new commits; one nit on commit-message hygiene noted inline. The Goetz-suggested fixup to 902-debloat_proc.patch is correctly incorporated into b36ccfd ("kernel: bump 6.6 to 6.6.136").

---

Generated by Claude Code

[openwrt-ai]

Reviewed 1 new commit (7341b4bd45 "kernel: bump 6.6 to 6.6.137") and the squash of the previously-reviewed .136 fixup; no new issues found.

Verified the .137 tarball hash 92f1e90d8be49f9d7f989e563c97a4f39cd7731025d7b81152814c68b5bd56c4 against kernel.org's published sha256sums.asc. The patch refresh in this commit is purely a -1 line-offset shift in crypto/Kconfig hunk headers across the four 906-*-jitter* backports, consistent with All patches autorefreshed. in the commit body. The squashed .136 commit correctly applies the 902-debloat_proc.patch fixup suggested by @GoetzGoerisch (the !proc_create_net !-prefix and sizeof(struct ptype_iter_state) are both correct), and the [2] footnote is now cited inline in the message body, resolving my previous nit.

---

Generated by Claude Code