**** BEGIN LOGGING AT Thu Dec 24 03:00:14 2020 Dec 24 04:26:36 Hauke: something is wrong with the buildbots. They're still building ARC targets with uClibc-ng, even though it's gone. Dec 24 05:03:49 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_mediatek.html has been updated. (100.0% images and 97.3% packages reproducible in our current test framework.) Dec 24 06:27:24 yay, got some feedback on my patches! Dec 24 09:19:24 we have no plans for musl support. joy... Dec 24 09:20:19 let me guess. some redhat project? Dec 24 09:23:16 see mailing list Dec 24 10:08:48 mangix: working on testing 5.10. i already patched in all of nbd's patches, bit weird to see the ramips pr duplicate them. Dec 24 10:13:32 he does it for consistency I think Dec 24 10:19:32 I'm tempted to post an mvebu kernel bump but I don't I'll be testing it. Dec 24 10:19:47 *think Dec 24 10:35:33 :P Dec 24 11:07:15 cryptodev-linux is breaking here on 5.10 Dec 24 11:07:20 * Borromini tries an upstream patch Dec 24 11:16:17 lovely. three upstream patches :D Dec 24 11:23:12 lol you run cryptodev on mt7621? Dec 24 11:24:50 mangix: the archs38 target was build with glibc Dec 24 11:25:05 mangix: https://downloads.openwrt.org/snapshots/faillogs/arc_archs/ Dec 24 11:25:26 the busybox is linked against glibc Dec 24 11:49:24 mangix Dec 24 11:49:24 I will test the mvebu kernel bump for you. You will have to tell me how to build as I onley know how to build master. Dec 24 11:52:06 Hauke: breakage is about what I expected. I think this is one of the reasons the Turris people abandoned glibc when they had their OpenWrt fork. Dec 24 11:52:41 Tapper: dengqf6 will probably handle that :) Dec 24 11:52:50 OK no probs Dec 24 12:05:00 ldir ping Dec 24 12:12:47 If any one is around that knows about the bug were the wrt3200acm will not sysupgrade there is some one in the forums asking for help. Dec 24 12:12:50 https://forum.openwrt.org/t/linksys-wrt3200acm-sysupgrade-not-working/83041/10 Dec 24 12:13:03 mangix: i don't but openssl pulls it in. Dec 24 12:13:19 I hade the same thing go rong for me but when trying to fix it I bricked mine and ldir fixt it for me. Dec 24 12:16:51 Borromini: I think that dependency is adjustable in menuconfig Dec 24 12:17:12 mangix: no, somehow not. Dec 24 12:17:28 at least not with all the options i've ticked. i checked Dec 24 12:17:57 Hauke: I briefly looked at the build failures. Seems pkgconf is not being compiled properly. Dec 24 12:20:25 mangix: i don't mind to much, helping bring up 5.10 as well that way, so :) Dec 24 12:22:57 mangix: for me this looks like a problem in the SDK build Dec 24 12:23:15 libubox is compiling against glibc without SDK Dec 24 14:07:38 * russell-- working on updating iproute2, discovers complaints about missing dependencies libmnl and libelf despite them being selected as 'm', something to do with the condition dependency? Dec 24 14:08:05 exists on current 5.9.0 as well Dec 24 14:09:48 Package tc is missing dependencies for the following libraries: Dec 24 14:09:48 libelf.so.1 Dec 24 14:09:48 libmnl.so.0 Dec 24 14:09:54 for example Dec 24 14:10:48 tc's DEPENDS is: Dec 24 14:10:50 DEPENDS:=+kmod-sched-core +libxtables +libelf +(PACKAGE_devlink||PACKAGE_rdma):libmnl Dec 24 14:53:33 Hauke: seema like it. Dec 24 14:54:25 mangix: so... does original pkgconfig work? ;) Dec 24 14:54:49 ik Dec 24 14:55:00 whoops Dec 24 14:55:26 karlp: not related Dec 24 14:57:03 jsut teasing :) Dec 24 15:59:12 mangix: 5.10 shows same behaviour as 5.4 on the DIR-860L, 19.07 works as expected. Dec 24 16:59:01 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_kirkwood.html has been updated. (100.0% images and 97.3% packages reproducible in our current test framework.) Dec 24 17:40:21 anyone seen Rosen lately? Dec 24 17:43:33 grift: you’re a Strongswan user, aren’t you? Dec 24 18:04:57 philipp64: if you're talking about rosen penev, his nick is mangix Dec 24 18:04:59 on irc Dec 24 18:06:19 okay, thought it was neheb… Dec 24 18:06:42 odd that there’s a “people” page with nicks for “openwrt” on the website, but not for “packages” maintainers… Dec 24 18:07:51 mangix: ping Dec 24 18:09:37 philipp64: not on irc :) Dec 24 18:12:10 Borromini: who’s not on irc? Dec 24 18:13:31 his nick on irc is not neheb, i meant. Dec 24 19:55:33 ath79 broken on master, bisecting here, not sure what's the annoying warning from: Makefile:86: warning: ignoring old recipe for target openwrt-ath79-generic-dlink_dap-2660-a1-squashfs-sysupgrade.bin" Dec 24 20:04:42 rr123: nebster in #openwrt bisected it back to https://github.com/openwrt/openwrt/commit/5fc28ef47959c1256f2553d507224bb481b08494 Dec 24 20:04:59 he's on ipq806x though i think Dec 24 20:05:04 see https://gist.github.com/SerialVelocity/d590d86a90c4fff625cb3b3fd67e3c5f Dec 24 20:09:33 I may be missing the obvious, but that commit does appear to be harmless Dec 24 20:12:09 I would believe some freak interaction for sysupgrading from that commit onwards, maybe also problems with uci-defaults network preconfiguration (does not apply to upgrades with keeping settings - but I don't see any issues in those hunks), but nothing that would explain boot problems after a sysupgrade Dec 24 20:28:05 pkgadd: that's what i thought too (and i said as much) Dec 24 20:28:57 looked harmless to me as well Dec 24 20:29:07 what it does however, is touching the timestamp of the Makefile, which might make update the dependency chain Dec 24 20:29:44 *might make GNU-make Dec 24 20:52:43 would that make a build break in such a way it produced defective firmware Dec 24 20:55:21 https://bugs.openwrt.org/index.php?do=details&task_id=3531 the IMAGE/sysupgrade.bin/squashfs definition from 5fc28ef4 is leaking to other devices, result in sysupgrade.bin images being sysupgrade tarballs, which do not boot Dec 24 21:11:25 ah, hat makes sense Dec 24 21:11:30 that* Dec 24 21:13:28 oh lovely. Dec 24 21:13:33 thanks swegener Dec 24 21:13:47 * Borromini looks at his bricked 1043nd v2 :( Dec 24 21:23:17 philipp64: pong Dec 24 21:26:37 Borromini: we wish you a merry christmas and a happy new brick :) Dec 24 21:26:43 Borromini: yeah, i had to debrick my device with an spi programmer, because it has no uboot console or tftp recovery Dec 24 21:31:16 mangix: and a working recovery to you too :P Dec 24 21:31:50 swegener: yeah i have no soldering skills and the header is not populated, so this one's a write-off. such are the dangers of running master eh :P Dec 24 21:32:41 Borromini: some pins and a spare thumb to maintain contact is enough Dec 24 21:33:48 russell--: i have wires that just fit. i can tell you it spewed out garbage first, then i switched tx/rx (vcc not connected), and after that it went silent. Dec 24 21:34:14 might get someone to soler it and try the serial cable properly, but not much hope frankly. Dec 24 21:34:50 Garbage might mean lack of gnd connection too. Dec 24 21:36:35 at least a completely unpopulated header (without existing solder in the holes) would be trivial to solder (getting solder out of the holes needs a little more equipment, solder wick and solder sucker, and some basic experience) Dec 24 21:37:31 PaulFertser: i had all those wires angled (gnd as well), and inserted into the serial cable. Dec 24 21:38:00 pkgadd: i'm not in a hurry, i'll leave that to someone who knows how, but thanks for the tips Dec 24 21:39:37 the garbage sounds like you had it right Dec 24 21:39:50 rx anyway ;-) Dec 24 21:39:57 yep (just bad GND) Dec 24 21:40:16 oh well... Dec 24 21:40:35 i've learned to just not move for a few minutes Dec 24 21:41:53 how so? :P Dec 24 21:42:14 yep, I nanddump/ backed up and flashed my bthub5 by pressing small cut/ insulated tinfoil leads to the solder points, using latex finger cots and keeping my fingers steady for around an hour Dec 24 21:42:52 (given that the vias to connect to are so fragile) Dec 24 21:43:11 ok you won. i'll think about it :P Dec 24 21:43:22 :) Dec 24 21:43:39 i'm trying to find what's breaking the n radio on my DIR-860L. it's on and off. very weird. Dec 24 21:43:53 (hijacked the ramips 5.10 pr) Dec 24 21:44:48 that's getting precedence atm :P Dec 24 21:47:10 Borromini: 19.07 works? Dec 24 21:47:42 mangix: it does, albeit not reliably i think. i remember seeing the 2,4 GHz radio request resets all the time Dec 24 21:48:33 i had to revert to 19.07 through the bootloader, then it worked, sysupgraded to master again, there sometimes it works, sometimes it doesn't. Dec 24 21:48:44 Borromini: i don't immediately see why that commit would break anything, do you have a link to an explanation of the bug? Dec 24 21:48:45 paraka already posted suggestions, trying those. Dec 24 21:49:08 russell--: sorry what commit are you referring to? Dec 24 21:49:15 your bricking Dec 24 21:49:26 ok just finished bisecting, and yes it's 5cf28ef4795 Dec 24 21:50:35 russell--: swegener shared a fix, he says the sysupgrade tarball code is leaking into other sysupgrade binaries Dec 24 21:50:49 https://github.com/openwrt/openwrt/pull/3714 Dec 24 21:51:45 swegener: is the empty definition already there in ipq806x? because someone in #openwrt reported having an ipq806x device break (tp-link c2600) Dec 24 21:52:13 this leakage killed my holiday :( Dec 24 21:56:09 Borromini: the issue 5fc28e introduced is specific to ath79. ipq806x has some devices that uses sysupgrade tarballs, but it should be correct Dec 24 21:56:11 it's sad how many vendors still don't care about a robust and reliable recovery mechanism Dec 24 21:57:33 swegener: ok, thanks Dec 24 21:57:58 pkgadd: i'm not buying tp-link anymore without checking available recovery methods first. learned my lesson. Dec 24 21:58:27 Borromini: or join the dark side with breed :) Dec 24 21:58:28 Borromini: tftp recovery worked well for me, i have 4 tplinks at home Dec 24 21:58:55 not 100% robust but 95%, plus I can reach to uart if needed Dec 24 21:59:07 but yes breed/never-die-uboot will be nicer Dec 24 21:59:07 mangix: i might yes, next time. Dec 24 21:59:43 * rr123 tplink firmware is actually openwrt/luci undernearth as far as I can tell Dec 24 21:59:45 rr123: well this is the second 1043 v2 that bricks on me this way. hadn't tried serial with the first one though, gotta be honest about that Dec 24 22:00:11 rr123: a lot of vendor firmware is. just outdated SDKs though. Dec 24 22:00:37 Borromini: that's actually the main reason why I didn't buy the archer c2600 (ipq8064), because they intentionally 'broke' the serial header (you need to bridge two solder points under the SOC tin and cooler, add some voltage level shifter, etc. pp. --> https://openwrt.org/toh/tp-link/tp-link_archer_c2600_v1#serial) - no, thanks Dec 24 22:01:05 no uart, no deal, absolutely Dec 24 22:01:31 pkgadd: yeah, i've seen a few pictures on the wiki, even the 1043 v5 etc seems to need bridging like that nowadays Dec 24 22:01:34 nb6817 --> easily accessible (unpopulated) header; r7800 --> populated header, a bit hidden under the cooler, but accessible with some bending of the pins Dec 24 22:02:32 that (and the newer ipq8065 SOC) was worth the ~40 EUR markup for me Dec 24 22:05:29 swegener: the author and committer of the problematic commit should probably be tagged Dec 24 22:06:09 * enyc meows at BT-HHv5a's working nicely on master =) will eep testing builds -- curious what sort of invasive changes expected into 20.x Dec 24 22:19:23 enyc: DSA might be possible, afaik the mainline DSA drivers are close, not quite 100% ready, but close Dec 24 22:21:42 are there any targets that might still move to DSA before 20.x/21.x gets branched? Dec 24 22:21:42 (for lantiq specifically) Dec 24 22:22:24 Borromini: at this point, I hope not - quite quickly after probably ipq40xx (ipq806x shouldn't be too far away either), that should solve the VLAN issues Dec 24 22:26:42 that would be neat Dec 24 22:27:54 mvebu and mt7621 imho switched rather late already Dec 24 22:32:28 :) Dec 24 22:32:44 it's a pity realtek won't be in the new stable release, but understandable given how fresh it is Dec 24 22:33:30 it's looking quite cool, just keeping my fingers crossed for gs1900-24e support ;) Dec 24 22:34:47 :D Dec 24 22:41:49 i'm a bit in limbo on whether to stick with bkobl's tree or with openwrt master. the latter works. but i'll have to wait till someone ports fixes from it to openwrt master, if there are any Dec 24 22:43:36 the target renaming made the situation a bit more complex to compare Dec 24 22:47:32 pkgadd: humm i see. All I can say is so-far my lantiq VLANs are working fine! Dec 24 22:47:59 enyc: different story, the VLAN situation is specific to ipq40xx with swconfig Dec 24 22:48:05 pkgadd: that as well. Dec 24 22:49:26 age-old query when to fork openwrt 20.12, is the idea to do so before new year anyhow!?!? judgement call to make with ''new stuff' constantly on horizon, anyhow? Dec 24 22:50:06 I hope it'll become 21.01 (especially considering the date), more reasonable to explain Dec 24 22:51:55 :P Dec 24 22:53:12 * enyc reads https://openwrt.org/docs/guide-developer/releases/goals/20.xx Dec 24 22:54:03 imho it would be better to branch off as 21.01 in early january, than to squeeze it in within the coming week, just to hit the magic numbers (2020 date) Dec 24 22:54:42 especially as you can expect 6+ weeks for the -rc phase Dec 24 22:54:57 excuse me while i whip this out Dec 24 22:55:33 lol, naming something 2020 is going to weaken adoption ;-) Dec 24 22:57:46 pkgadd: i think we should christen it 'the covid edition' :P Dec 24 22:57:48 nbd Sorry to bother you in the IRC channel. But after a couple days I finally was able to make time to respond in the forum. While the GF is out getting groceries I have a small window of a time to do any kind of further testing that might be useful. Let me know if I can do anything useful Dec 24 22:58:18 the page lists hostapd 5.10 backports, ?!!? can't see 510 anywhere. .. and the DSA already mentioned. Dec 24 22:58:32 Borromini: just find a cocktail receipe with corona beer to go full circle Dec 24 22:58:47 enyc: nbd has a tree with 5.10 bringup and there's a PR for ramips e.g. Dec 24 22:58:50 blocktrron: hehehe Dec 24 22:59:41 enyc: oh 5.10 mac80211. sorry. hauke has those in his tree as well i think (the wireless stack, not the full kernel bump) Dec 24 23:00:05 Borromini: uerrr aaaah, to backport extra device support? Dec 24 23:00:43 enyc: no to backport wireless drivers. e.g. 19.07 uses 4.19 wireless stack (whereas most targets use a 4.14 kernel) Dec 24 23:00:59 allows openwrt to stick closer to upstream for drivers where it matters a lot (wireless) Dec 24 23:01:16 while not needing the most recent (and relatively more unstable) kernel Dec 24 23:02:10 Borromini: I see. In x64 / desktop land I'm seeing all sorts of consternation about 5.4 vs 5.8 5.9 5.10 and new Ryzen support ;p Dec 24 23:02:58 desktop is different :) Dec 24 23:03:12 a desktop is something you shut down or reboot. you don't want your wifi to be down Dec 24 23:03:17 sure, but still lots of 'what is stable' grumble grumble Dec 24 23:03:28 arguments-arguments going on ;p Dec 24 23:06:12 Borromini: hrrm you make me wonder if ath10k client mode works in updated release I shall try....... Dec 24 23:06:46 enyc: check what patches are in the mac80211 bump, usually gives an idea Dec 24 23:08:47 Borromini: happy to look if you have link to reference, i'm not yet that level of git voodoo =) Dec 24 23:11:19 enyc: you can check hauke's tree on git.openwrt.org Dec 24 23:11:33 for some reason i am unable to paste my buffer into irc atm >_> Dec 24 23:11:53 his mac80211 updates for 5.10 are working fine for me Dec 24 23:11:58 :) Dec 24 23:13:02 enyc: https://bit.ly/37KAvNm Dec 24 23:13:18 sorry no idea what's happening with the shitty touchpad here. Dec 24 23:13:54 enyc: commit 2 and 3 from the top is what you'll surely need Dec 24 23:21:47 Borromini: i'm trackpoint'ing on a thinkpad ;p Dec 24 23:23:41 as long as it works ;) Dec 24 23:27:13 i'm out, goodnight guys. Dec 24 23:43:03 * enyc OoOOOooooOo at the patches to support tg582n Dec 24 23:43:20 I have DANT-T about I believe, not just the DANT-1 and DANT-V mentioned.... Dec 25 00:35:28 w00t, updated OpenWRT master seems to *work* in client mode now Dec 25 00:35:49 on the ath9k 5ghz, QCA9880 802.11acn Dec 25 00:43:15 what the... Dec 25 00:43:23 libubox supports jsoncpp? Dec 25 00:51:59 I wonder why these snapshot images don't have luci (web interface) by default, in contrast to 19.07.5 etc Dec 25 01:01:54 Hauke: just sent a workaround for the pkgconf issue on the mailing list. Dec 25 01:34:38 mangix: sorry, was with the kids. do you use IPsec? Dec 25 01:35:18 need someone to test both the X.509 changes to UCI classic as well as the new swanctl-based generator… Dec 25 01:59:43 philipp64: I do not Dec 25 02:00:13 okay. **** ENDING LOGGING AT Fri Dec 25 03:00:10 2020