**** BEGIN LOGGING AT Mon Dec 09 04:42:13 2019 Dec 09 05:20:00 DNS over HTTPS Proxy is not working in snapshots Dec 09 06:23:08 Tapper: compare the configs from /rom/etc/config/foo /etc/config/foo Dec 09 06:30:35 Tapper: in particular the resolver Dec 09 08:33:38 pkgadd About https over dns proxy it works fine in build r11582 Dec 09 08:33:56 I tried flashing to a newer build eg 11638 Dec 09 08:35:42 And redoing configs. I just can not get it to work. If i reuse my old configs i do not get any internet no matter what I do and when I redo the configs I get internet but it uses my ISPs dnsservers, even after unticking the tick box to not use ISPs DNS servers. Dec 09 08:39:39 pkgadd can I send you a dm? Dec 09 08:46:08 f00b4r0: pong Dec 09 09:38:21 blocktrron: Thanks for testing. I notice in your tree you indeed encountered the stuff I was still working on. Dec 09 09:39:02 blocktrron: One of the issue I was checking last was the slow compilation of the kernel. I notice Linus also talks about it in a mail stating that make was running single-threaded at some point Dec 09 09:39:37 The malformed patches was from manually editing them to run some quick-testing. Thanks for fixing those properly. Dec 09 09:40:12 blocktrron: Maybe you should update my patch with your changes and add your SoB? Dec 09 09:41:30 mangix: I've read some technical stuff regarding cake and I was wondering how it performs in terms of cpu-usage compared to fq-codel Dec 09 09:41:55 keeping in mind to reduce cpu load as much as possible for the slowest single-core devices Dec 09 09:46:30 blocktrron: In the meantime. I'll add your fixes to my initial 5.4 WIP patch to advance further. Please let me know if I can add your SoB Dec 09 10:07:21 xback: sure thing, you can include it Dec 09 10:07:50 I did not had the feeling that compilation takes longer than usual Dec 09 10:08:12 I currently only tested x86 Dec 09 10:08:18 and x86_64 Dec 09 10:08:49 From my first testing, on mpc85xx everything works as expected, did not yet test performance though Dec 09 10:09:30 On ath79, a lot of kernel symbols needs refreshing, while this is not the case on mpc85xx - did you refresh the generic symbols? Dec 09 10:10:39 yes, but I do notice a lot of changes Dec 09 10:11:12 also some stuff moved Dec 09 10:11:39 some crypt lib called Coric or so Dec 09 10:11:48 which seems to be widely used Dec 09 10:12:12 blocktrron: merged your changes and added your SoB to the WIP patch. pushed to staging Dec 09 10:13:23 I'll continue from this point. If you also encounter further stuff, please let me know so we get this in a proper state soon :-) Dec 09 11:20:50 xback: thanks, notice that the mpc85xx change still needs it's symboled refreshed, yurrently it's c&p from 4.19 Dec 09 11:21:10 But i think this is obvious ;) Dec 09 11:21:45 blocktrron: yeah. I just pulled it in to have the "latest state" for all targets Dec 09 11:22:03 currently adding cns3xxx and imx6 so I can do more extensive testing on more platforms Dec 09 11:23:25 (currently building x86 and cns3xxx here now) Dec 09 11:44:52 xback: nice, just ping me once you've something ready for imx6 Dec 09 12:05:04 * andy2244 maybe someone can give me a hint, i have this problem with iconv: I try to build a static binary with glib2 and a automake project, yet i cant get iconv to link statically without patching the actual makefile.am? Dec 09 12:05:13 I get this error: /lib/libglib-2.0.a(libglib_2_0_la-gconvert.o): in function `g_convert_with_fallback': Dec 09 12:05:13 gconvert.c:(.text+0xb08): undefined reference to `libiconv' Dec 09 12:06:05 from my knowledge this should work but its not? TARGET_LDFLAGS += liconv Dec 09 12:06:11 from my knowledge this should work but its not? TARGET_LDFLAGS += -liconv Dec 09 12:07:21 if i add -liconv to the makefile.am via 'LIBS = -liconv' than it build correctly? Yet why does this work and not the LDFLAGS? Dec 09 13:48:44 Hello all, is anyone able to lead me to the correct documentation ? Im searching how to display a system text log in LuCi for a package im developping. Dec 09 14:45:05 hello Dec 09 14:46:10 how to know which version of openwrt the patch on patchworks is based? Dec 09 14:47:05 ynezz: quickly added basic support for imx6, but some work is still needed :-) https://pastebin.com/raw/ZQhM14j8 Dec 09 15:12:43 ynezz: the NAND symbols got renamed. running a test with the new ones enabled now Dec 09 15:24:14 xback: almost there :] Dec 09 15:24:23 nice, even u-boot bump? Dec 09 15:32:01 It's a personal u-boot bump. (corporate internal one) Dec 09 15:32:27 basically we just select nand as primary boot devicem shaving a few seconds off. and also add a password to enter uboot :-) Dec 09 15:34:52 and same password on all devices :) Dec 09 15:36:34 anyway, I don't follow u-boot development, but IIRC there was some deprecation warning regarding board removal if not migrated to DT after some time, so I'm wondering what has changed since then (April?) Dec 09 15:37:03 probably going to bump u-boot to some latest while playing with 5.4 on Apalis Dec 09 15:41:02 Yeah, same password. but we dont do this for security Dec 09 15:41:12 the uart port is actually used to talk to a Iridium modem Dec 09 15:41:31 so if the device boots, and the modem talks .. we enter u-boot which we dont want it to do Dec 09 15:41:41 so we just use some long password to avoid this ;-) Dec 09 15:42:01 and eventually make that modem crash because of unexpected u-boot messages :p Dec 09 15:42:18 that's were AT protocol comes in ;-) Dec 09 15:42:42 parsers are hard Dec 09 15:44:53 yeah .. nand attaches properly now Dec 09 16:17:20 I'm noob in this, how to handle the situation, where I've probably fixed some issues in the main system component, which could be possibly (ab)used for getting CVEs if I post it for review on the mailing list Dec 09 16:36:02 ynezz: maybe post the patch to contact@ ? Dec 09 16:47:56 well, this is actually a blind spot I would say Dec 09 16:48:56 I mean, you get the proof-of-concepts over plain text email sometimes weeks before the fix is even deployed Dec 09 16:54:46 "We haven't set up a PGP key to send encrypted mail for this mail address yet. " :-) One doesn't even realize this until you would like to send something confidentially Dec 09 16:55:56 for me it's easy, I can just contact Hauke for example Dec 09 16:56:34 seems like I got first topic for Barcelona meeting :] Dec 09 16:58:32 aparcar[m], ynezz ping hi are there any other repositories except docker.io? Dec 09 17:06:39 aparcar[m], i see https://github.com/openwrt/docker/blob/master/Dockerfile#L19 sudo and build as member of sudo group, but in sdk i can't run sudo Dec 09 17:24:52 greearb_: I've got the CT firmware bump in my staging. tested it today and it seems fine (Wave 1) Dec 09 17:24:58 'lo Dec 09 17:26:25 greearb_: I think if the multicast issue could be solved (regarding ARP) I could start testing on dozens of devices to iron out the last few bits Dec 09 17:26:27 xback, thanks for testing Dec 09 17:27:01 I need a way to reproduce that mcast problem locally to make progress on that sort of thing Dec 09 17:27:13 unless someone can bisect, and some of these things are nearly impossible to bisect Dec 09 17:28:43 greearb_: using the latest state, it's pretty easy to replicate Dec 09 17:28:53 use 2x 9882 wave 1 Dec 09 17:29:03 put both in IBSS using WPA2-PSK Dec 09 17:29:05 no fancy routing Dec 09 17:29:36 the devices will link up properly. but no arp data is set in the arp table Dec 09 17:30:37 greearb_: I'll push the FW updates tomorrow to master and stables Dec 09 17:32:16 greearb_: I'm currently bumping kernel to 5.4 so I can test mac80211 v5.4 CT asap Dec 09 17:32:59 ok, there are some changes to ath10k-ct driver, modest, but a couple of upstream patches in 5.4 Dec 09 17:33:59 I'm curious to find out wheter mcast packets are not being sent, or they get dropped on reception Dec 09 17:34:12 can you reproduce the ibss problem w/out encryption? Dec 09 17:34:19 no Dec 09 17:34:28 without it seems to work fine Dec 09 17:35:08 But i'll double check the latest state tomorrow disabling encryption Dec 09 17:45:16 in the past, it appeared that supplicant had some bugs regarding ibss encryption. Can you reproduce this problem on 9984 ct firmware and/or ath9k driver by chance? Dec 09 17:47:54 and NIcolas Cavallari just posted some ibss patches to linux wireless that might be worth trying Dec 09 18:25:04 Hope this is OK to post in -devel. Found (in my view, buggy, unhelpful default settings), using OpenWRT 19.07rc2 on Lantiq XWAY VDSL-modem in VDSL, PPPoE mode.... the 'dsl' driver uses MTU 1500, which then limits PPPoE data to 1492 quite needlessly. Dec 09 18:25:46 doing "ip link set dsl0 mtu 1508" then "ip link set dsl0.101 mtu 1508" then reconnecting, FIXES this, temporarially! Dec 09 18:26:06 Should I report a bug somewhere? Dec 09 18:30:10 enyc: you can add "option mtu '1508'" to the relevant interface section in /etc/config/network Dec 09 18:30:58 ldir: I have WAN interface pppoe-wan , but that seems to be separate, not the same thing Dec 09 18:31:13 at least can't se how to change this in web frontend Dec 09 18:32:10 interfaces -> advanced - right at the bottom (usually) Dec 09 18:33:43 ldir: there is an advanced -> Override MTU but that doesn't work , that seems to be setting pppoE-wan mtu, rather than dsl0 mtu! Dec 09 18:34:22 oh I have no idea then - don't use pppo* or even an on-board modem Dec 09 18:34:23 ldir: there is a "DSL" tab (alongside 'interfaces'), but that is Annex/Tone/Encap/DSL-mode/SNR/Firmware file ....... no MTU setting Dec 09 18:38:23 There's some discussion https://bugs.openwrt.org/index.php?do=details&task_id=94 that might be helpful Dec 09 18:44:15 ldir: fwiw https://ec2.sheer.us/~enyc/noidx/etc-network-interfaces Dec 09 18:45:11 ldir: thankyou, seems outdated.. and old-suggestions hrrm Dec 09 18:47:00 hrrm, question is really is this actually an issue with underpinning dsl driver, or netifd not picking up right things, or what Dec 09 18:47:04 indy: whats the problem exactly? Dec 09 18:47:09 indy: how to reproduce it? Dec 09 18:50:03 indy: docker run --rm --tty --interactive openwrtorg/sdk:ath79-generic-master bash -c "id && sudo id" Dec 09 18:50:07 works here as expected Dec 09 18:51:18 enyc: no idea, being a sky customer I don't have to deal with PPPoE - it just runs dhcp over the E bit that comes out of the modem :-) Dec 09 18:59:17 ldir: well yes but you may find the sky-modem under the hood is doing PPPoE on the VDSL bit Dec 09 18:59:17 ldir: just not, ontop physical ethernet Dec 09 19:01:31 The sky modem is the HG612 - There's no PPP involved - It's ethernet over PTM, instead of PPP over ethernet over PTM Dec 09 19:04:51 In the early days SKy required DHCP option 61 for user authentication but that's gone the way of the dodo. Dec 09 19:16:41 ldir: ooo i see, that figures,, ad only works like that on Sky lines on Sky-in-exchange unbundling, i guess Dec 09 19:50:23 Just a quick check, is what transmissions luci package does ok to replicate? Aka not selecting the actual service package, since it has multiple variants and does not know which one the user want? Dec 09 19:51:14 So being able to offer more variants, but loose the auto-install feature via just installing/selecting the luci package? Dec 09 19:55:37 ldir: ping? Dec 09 20:02:13 can procd configure to not hold onto /dev/watchdog? apparantly atheros can only support one at a time Dec 09 20:05:27 ynezz, i use podman instead of docker (more secure and for building good enough) - https://paste.centos.org/view/3f6d338b Dec 09 20:06:14 indy: openwrtorg/sdk:x86-64-18.06.4 is missing sudo command? Dec 09 20:06:23 yep Dec 09 20:06:27 ano :) Dec 09 20:10:26 so it should be fixed with 18.06.5 Dec 09 20:11:03 as the sudo was added some time after 18.06.4 Dec 09 20:22:42 temporary workaround http://paste.ubuntu.com/p/nzgbbXs6Cs/ Dec 09 20:30:34 f00b4r0: pong Dec 09 21:58:01 Hrrm, any $clue on why i can't see tl-wr740n images in 19.07.0-rc* ar71xx/tiny ? can't see in changelog disabled .... i'd expect it to be there as a 4/32 device on a still-upported arch. Dec 09 22:02:52 afaik 32/4 devices aren't supported anymore because openwrt won't fit. also I think ar71xx has been deprecated in favor of ath79 Dec 09 22:08:28 ynezz, yes i did something similar Dec 09 22:08:43 ynezz, any timeframe for 18.06.5 release? Dec 09 22:09:51 indy: you mean as in (exactly) one month ago? http://downloads.openwrt.org/releases/18.06.5/targets/ Dec 09 22:22:15 indy: I've just triggered build of 18.06.5 images https://gitlab.com/openwrt/docker/pipelines/101832084 Dec 09 22:24:15 it has to be done manually, only master/19.07 is build periodically https://gitlab.com/openwrt/docker/pipeline_schedules Dec 09 22:26:16 aand its borken Dec 09 22:49:26 hell__: OooOOooo thankyou for the pointer to ath79, despeite my ogoogling hadn't found that **** ENDING LOGGING AT Tue Dec 10 02:59:58 2019