03:29 <philipp64> I recently rebased my tree to master and it's been a minute... was last running 5.15 if I remember. Now when I build I'm having issues with linkage related to CONFIG_CRYPTO_ACOMP2 wanting to be a module and not built-in. Did I miss a step refreshing my env/kernel-config file? Anyone else having issues building 6.6 for x86_64 hardware?

03:29 <philipp64> https://paste.centos.org/view/c661003a

06:37 <KanjiMonster> philipp64: I presume the culprit is https://github.com/openwrt/openwrt/blob/main/target/linux/generic/hack-6.6/260-crypto_test_dependencies.patch - try disabling CRYPTO_MANAGER_DISABLE_TESTS

07:35 <f00b4r0> hauke: ok i'll prep a patch

07:37 <f00b4r0> I'll just update_git_source_package.sh in 23.05 branch, cherry picking all the intermediary patches is just too much hassle

07:41 <f00b4r0> hauke: in the meantime you're welcome to deal with 22.03, I won't be touching that :)

09:03 <f00b4r0> hauke: FWIW been test-building ath79 with 23.05 buildbot config to confirm ucode builds, I'm getting image file too big errors for a couple tp-link devices (re355-v1, re450-v{1,2}). Dunno if that's expected.

09:04 <f00b4r0> i'll submit the patch now

09:38 <romany> hi all! anybody is having problems with ethernet link detection (always stays UP even if there is no cable) on ath79 target?

09:39 <romany> I'm having this problem on ar9331 but curious maybe it's also a problem with other SoCs

10:23 <schmars[m]> hauke: mt76 802.11s regressions but they're already "older" and not since last 23.05.x release

10:23 <schmars[m]> more like a known problem

14:42 <tmn505> romany: You might be interested in https://github.com/openwrt/openwrt/pull/9971

14:56 <owrt-images-builds> Build [#302](https://buildbot.openwrt.org/images/#/builders/205/builds/302) of `master_mpc85xx/p1020` completed successfully.

15:02 <philipp64> KanjiMonster: that looks right. I also think that CRYPTO_ACOMP needs to be selected when CRYPTO_ACOMP2. so what's the solution? back out this patch for now?

15:05 <KanjiMonster> philipp64: just disable CRYPTO_MANAGER_DISABLE_TESTS in your kernel config, that should fix it for you

15:24 <philipp64> building...

15:25 <philipp64> being prompted for settings for CRASH_HOTPLUG and CRASH_MAX_MEMORY_RANGES ...

16:13 <philipp64> KanjiMonster: sigh. still fails exactly the same.

16:15 <philipp64> nbd: any ideas?

16:33 <romany> tmn505, that looks similar, yeah... but I already tried adding phy-handle to eth node (referencing correct internal switch port), it only starts detecting link if I remove fixed-link, but then there is no data flow, looks like phy-handle and fixed-link are in conflict somehow :/

16:40 <tmn505> the link is up but probably no speed is negotiated, that is why Leo added builtin-switch property in the mentioned PR

16:53 <romany> tmn505, ah, let me try

17:20 <romany> tmn505, that worked!

17:20 <romany> I wonder why that PR was never merged

17:22 <tmn505> beats me, don't know either, maybe with enough reports it'll get in

17:30 <tmn505> .names

17:30 <tmn505> oops

18:13 <hauke> tmn505: Is it fine to switch tegra to kernel 6.6 by default?

18:14 <tmn505> yes

18:14 <tmn505> want to send me the patch?

18:14 <hauke> schmars[m]: I was hoping that one of the ~10 fixes related to 802.11s would fix it, but this will not block the release

18:15 <hauke> tmn505: yes it would be nice if you could send a patch

18:17 <hauke> f00b4r0: The ucode change is pretty late. I am not sure if we should take it

18:17 <f00b4r0> hauke: I mentioned many times now that there are bug fixes in there

18:17 <f00b4r0> and we came to the apparent consensus a week ago that we should just bump instead of cherrypick

18:18 <f00b4r0> so, ¯\_(ツ)_/¯

18:19 <tmn505> hauke: will do

18:21 <f00b4r0> hauke: fwiw I made sure to use the exact same version as what has been in master since June 21.

18:21 <hauke> f00b4r0: I will have a look at this

18:59 <owrt-images-builds> Build [#303](https://buildbot.openwrt.org/images/#/builders/231/builds/303) of `master_mpc85xx/p1010` failed.

19:46 <philipp64> KanjiMonster: okay, unsetting CONFIG_CRYPTO_MANAGER_DISABLE_TESTS and CONFIG_CRYPTO_MANAGER_EXTRA_TESTS works.

19:57 <Mangix> this dante gcc 14 compile failure is pissing me off

21:13 <Mangix> hrm fixed in an awful way