**** BEGIN LOGGING AT Fri Aug 03 03:00:00 2018 Aug 03 04:27:24 moin Aug 03 06:12:56 Hello Alll! Aug 03 06:13:02 Any dev. here can help me? Aug 03 06:17:16 xcom169: help you do what? Aug 03 06:18:28 Help me to understand DS-Lite settings, and possible differences in the new 18.06 version Aug 03 06:18:46 I upgraded yesterday to 1806 but my ds-lite broke Aug 03 06:26:17 russell--: What is my AFTR number ? How can I detect it? Aug 03 06:26:26 russell--: Is it mandatory? Aug 03 06:34:31 noob git question, if i want to apply certain commits without cherry-pick, is there a better/more proper way than e.g.: git show 53ec1c83a28cd504808c42307bf7243727345494|patch -p1 ? Aug 03 06:34:45 i don't want the commit message Aug 03 06:36:07 oh, git cherry-pick -n i guess Aug 03 06:39:55 xcom169: you are likely hitting https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=327c711da495483099a30e3f80e0fd8a85e8192d - try setting network.wan6.encaplimit_dslite to 'ignore' Aug 03 08:05:51 KanjiMonster: Thanks, I'll try this. Can you answer some questions regarding ipv6? Aug 03 08:06:23 xcom169: don't know if I can, but I can try ;) Aug 03 08:07:15 KanjiMonster: What is my AFTR number ? How can I get it? How do I know how big range I get from my ISP ? Should I assign /64 to my lan ? Should I use Dhcpv6 or slaac or both ? Aug 03 08:09:12 xcom169: Never heard of an AFTR number. AFTR is usually the upstream router/tunnel endpoint for the IPv4 connectivity Aug 03 08:09:56 https://www.isc.org/downloads/aftr/ Aug 03 08:10:06 Address Family Transition Router Aug 03 08:10:20 Also never heard of it. Aug 03 08:10:32 KanjiMonster: https://wiki.openwrt.org/doc/uci/network#protocol_dslite_dual-stack_lite Aug 03 08:10:41 peeraddr is the other name Aug 03 08:10:49 I must set it, or it's automatic? Aug 03 08:11:03 ifstatus wan6 should tell you the assigned prefix(es) from your ISPs including their size, the info is likely also somewhere in LuCI as well. If you did get prefixes from your provider, OpenWrt will automatically assign them to lan (or rather a subset of it) Aug 03 08:11:44 usually - at least in an ISP setting - the AFTR details are send as a DHCP option along with IPv4 DHCP responses Aug 03 08:11:54 this ^ Aug 03 08:12:08 usually no DHCPv4 though Aug 03 08:12:12 openwrt will pick up that option and spawn an appropriate dynamic ds-lite interface Aug 03 08:12:59 jow: Because in Luci it's a mandatory field, and not filled in automatically Aug 03 08:13:18 jow: How can I know this DHCP option? Aug 03 08:13:24 usually you should not need to declare a ds-lite interface Aug 03 08:13:35 jow: quick question: Is it correct that the Gemini target has "samba36-server" as config default? Aug 03 08:13:39 it all happens automatically Aug 03 08:13:48 andy2244: good question, guess not Aug 03 08:14:09 might be that gemini targets only NAS devices, where it might make sense Aug 03 08:15:44 ah oki i just do "CONFIG_TARGET_gemini=y" and than doe a "make defconfig" and get this configuration Aug 03 08:16:24 trying to create a script to build all my samba4 targets for the stupid cross-answer files.... Aug 03 08:18:17 i also assume i don't have to build the "broken" ppc40x,ppc44x targets or will those get fixed any time soon? Aug 03 08:25:28 andy2244: you can ignore broken targets Aug 03 08:26:04 oki Aug 03 08:38:25 andy2244: samba36 is part of the default packages for gemini nas devices (but not all devices): https://github.com/openwrt/openwrt/blob/master/target/linux/gemini/image/Makefile#L95 Aug 03 08:40:02 yeah i noticed, i just unset it now for my build script Aug 03 08:40:28 KanjiMonster: And what does it do actually? try setting network.wan6.encaplimit_dslite to 'ignore' ? Is it a setting in /etc/config/network ? Aug 03 08:40:31 btw is there a other name or shorthand for the "ARC" platform? I cant find a compatible qemu target Aug 03 08:41:26 xcom169: yes, add a option encaplimit_dslite to the wan6 interface with the value ignore Aug 03 08:42:23 KanjiMonster: This option is not yet documented as I see on the website :) Aug 03 08:43:24 xcom169: it's quite new Aug 03 08:43:56 andy2244: ARC is ARC. Looks like support was submitted, but never merged -> https://lists.gnu.org/archive/html/qemu-devel/2016-09/msg01906.html Aug 03 08:48:42 mhh than i have to "guess" the answer.... Is there any platform that might be similar? Aug 03 08:50:49 not really, arc is quite unique afaik Aug 03 08:52:05 andy2244: there are some autoconf default values, maybe they answer some of the questions -> https://github.com/openwrt/openwrt/blob/master/include/site/arc Aug 03 08:52:19 https://www.irccloud.com/pastebin/HCHe85I8/ Aug 03 08:53:11 stuff concerning libc functions should be the same on all targets Aug 03 08:53:29 only things like data type sizes, signedness etc. should differ Aug 03 08:54:28 yeah, but i also get slightly different results for some targets, trying to see what differs and what not atm by building like 20 targets Aug 03 09:02:40 jow: where are the build logs for 18.06 release? Aug 03 09:04:54 Monkeh: this test was on Unifi AP AC Pro, my other AP is D-Link DAP-2695 Aug 03 09:05:46 tmn505: there are none. What are you looking for specifically? Aug 03 09:06:40 jow: 18.06 build is getting changed with point releases? So tomorrow's build might be different from today's ? Aug 03 09:06:54 xcom169: no Aug 03 09:07:04 only packages are updated Aug 03 09:09:41 jow: I see few packages missing from base for aarch_cortexa53, for example wireguard-tools. Snapshot faillogs show: https://downloads.openwrt.org/snapshots/faillogs/aarch64_cortex-a53/base/wireguard/compile.txt Aug 03 09:10:06 *aarch64_cortexa53 Aug 03 09:10:20 18.06 likely had the same error Aug 03 09:11:44 tmn505: ah I do upload faillogs Aug 03 09:11:50 I thought I had disabled it Aug 03 09:11:53 http://downloads.lede-project.org/releases/faillogs/ Aug 03 09:12:12 jow: thanks Aug 03 09:12:14 they're populated by both 17.01 and 18.06 builders atm, need to straighten that out eventually Aug 03 09:14:20 seems it fails in the same moment, I'll try to build them localy, and investigate. Aug 03 09:15:05 tmn505: hey i just pushed this https://github.com/tofurky/openwrt/commit/cbe89a60a8380bc3e63112dbecfaa1248005a07a Aug 03 09:15:06 it probably fails sicne it is built by the SDK Aug 03 09:15:16 i put you in signed-off-by, if it looks good i can just turn it into PR Aug 03 09:16:13 nothing changed except commit message Aug 03 09:17:37 maybe i should remove targets from the title, it seems to wrap ;( Aug 03 09:18:01 tmn505: the problem is that the wireguard package was not marked nonshared, so the userspace parts are built by the SDK to put in the shared package repository Aug 03 09:18:17 tmn505: this then fails because it attempts to compile a kernel module within the SDK Aug 03 09:18:23 o, yeah, that was nano. i removed it. Aug 03 09:20:34 m4t: looks good, thanks. Aug 03 09:20:48 cool Aug 03 09:21:04 * m4t does one last build before doing pr Aug 03 09:21:32 jow: thanks. Aug 03 09:22:09 tmn505: you did test ath5k though right? i didn't make that up? Aug 03 09:23:58 yes, but that was short test, only create ap and connect to it. I'll do one more with Your commit. Aug 03 09:29:03 tmn505: seems to only affect some targets. WIll take a look Aug 03 09:29:20 I suppose it affects ald those having specific assembly optimizations Aug 03 09:29:24 *all Aug 03 09:30:59 It' built with SDK for rpi which is on 4.9 kernel, the rest of cortex a53 is on 4.14 Aug 03 09:32:19 maybe location of opcodes.h changed beetwen kernel versions. Aug 03 09:34:24 tmn505: the ar71xx/generic SDK includes no opcodes.h, and wg builds fine there Aug 03 09:34:54 tmn505: the brcm2708/bcm2710 does ship an opcodes.h and wg fails Aug 03 09:37:35 tmn505: ah okay, its set another ARM special souce... sigh Aug 03 09:38:05 so when packing the SDK, we copy arch/$(LINUX_KARCH)/ Aug 03 09:38:13 which for the affected targets is arm64 Aug 03 09:38:35 however the includes in arch/arm64/ reference includes in arch/arm/ Aug 03 09:38:38 which is not bundled Aug 03 09:39:08 Thanks for diging this. Aug 03 10:29:19 tmn505: okay i gotta go, here is the pr: https://github.com/openwrt/openwrt/pull/1237 Aug 03 10:29:28 KanjiMonster: maybe give it a spin if you feel like it ^ Aug 03 10:32:50 jow: out of curiosity, I suppose it's _desired_ that the phase1 buildbots run a full make world, correct? i.e. there would be no interest in speeding up the process by e.g. not rebuilding toolchain every time? Aug 03 10:36:00 yes Aug 03 10:36:21 we had the no-toolchain rebuild in the past and it caused various issues Aug 03 10:36:31 openwrt buildroot is not mature/stable enough for such things Aug 03 10:36:42 ok, that's what I figured Aug 03 10:36:59 good to know still ;) Aug 03 11:18:27 *yawn* Aug 03 11:18:34 morning ;) Aug 03 11:18:39 pfff Aug 03 11:18:47 crazy weather Aug 03 11:19:01 come to Bulgaria, here it's cold :P Aug 03 11:19:25 only 25C today, and forecast for the next week maxes at 28 Aug 03 11:19:46 abnormally cold and wet summer here this year :( Aug 03 11:20:12 sounds like they stole ours :P Aug 03 11:20:38 stintel: that is way better than here in NL. +30C :( Aug 03 11:26:56 would quite like 25 Aug 03 11:27:24 27 in prague but its really humid :( Aug 03 11:27:46 I wouldn't mind 30+ Aug 03 11:27:52 hiding in mcdonalds Aug 03 11:27:59 AC on 23C ;) Aug 03 11:28:03 coz aircon Aug 03 11:28:05 heh Aug 03 11:28:18 yeah here AC is pretty common due to us normally having such hot summers Aug 03 11:28:42 I had my office in a room without direct AC before, and omg it went over 40C in there :P Aug 03 11:28:50 im british, not used to the heat :D Aug 03 11:29:29 jow: btw, seems like there is open bug for the mentioned issue: https://bugs.openwrt.org/index.php?do=details&task_id=1725 Aug 03 11:29:44 does anyone know if linux-firmware.git is shipped as tar files of latest correlating to kernel versions ? Aug 03 11:30:06 and/or can a owrt meta makefile tell the buidsystem to do a bare git clone ? Aug 03 11:30:19 i would like to point ath10k-firmwae at latest stable Aug 03 11:30:39 but that git tree takes eons to clone, throtteling our build infra Aug 03 11:31:12 we might have a problem with sysupgrade here, https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=package/base-files/files/sbin/sysupgrade;h=a378b029500ac5981d504889b0b3e6af3cf92a0d;hb=HEAD#l167 - if there is any error with gzip or tar, the error isnt caught and the device starts without any configuration after the upgrade Aug 03 11:31:26 the error is just directed to dev null Aug 03 11:31:45 rotanid: send a patch please Aug 03 11:31:51 or open a bug report Aug 03 11:32:13 i wanted to ask at first, maybe i'm wrong Aug 03 11:32:46 looks like what you say Aug 03 11:32:47 I was having sysupgrade -v -n not reset the disks just earlier too, had to manually do firstboot; reboot. Aug 03 11:33:14 but that would be the opposite. Aug 03 11:34:28 blogic: could ~all of the git cloning infra be set to just do shallow clones by default? are there any repos we pull from that we need the history? Aug 03 11:34:46 it's conveninet for devs sometimes to step in and see the whole tree maybe, but there's a heap of cloning Aug 03 11:34:49 karlp: i was wondering the same Aug 03 11:44:42 blogic: set a MIRROR_HASH, then we'll not re-clone Aug 03 11:49:03 tmn505: fixed Aug 03 11:49:26 should trickle down the build chain so that wg userspace pacakges appear in the repos during the next one or two days Aug 03 11:50:00 verified that building wg now works on a self compiled brcm2708/bcm2710 sdk Aug 03 11:51:49 ty, much appreciated. Aug 03 12:02:46 KanjiMonster: option ip6assign 60 , means the size of the subnet ? Aug 03 12:04:26 jow: ula_prefix is mandatory, or just optional ? What does it do actualy? Aug 03 12:08:18 ²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²²² Aug 03 12:08:51 :P Aug 03 12:09:04 what's gebeurt Aug 03 12:14:32 stintel:pressng by accident a key while moving my laptop :) Aug 03 12:16:09 xcom169:ip6assign is the prefix size of the delegated prefix Aug 03 12:17:32 dedeckeh: If I got /57 from my ISP , then the 2a02:ab88:6385:xxxx::/57 is the prefix ? Aug 03 12:18:29 xcom69:57 is the prefix size you received from your ISP Aug 03 12:20:42 dedeckeh: infant or cat would have been more random, hence my asking ;) Aug 03 12:21:56 stintel:lol Aug 03 12:22:14 hmm. 35C outside, 24C inside. Gotta love old houses Aug 03 12:32:48 stintel: https://git.openwrt.org/?p=openwrt/staging/blogic.git;a=shortlog;h=refs/heads/mac80211 Aug 03 12:39:37 blogic: <3 Aug 03 12:41:13 dedeckeh: What should be my delegated prefix? Aug 03 12:41:43 xcom169:what do you mean ? Aug 03 12:43:30 xcom169:in a dual stack scenario you receive a prefix from your ISP via DHCPv6; via ip6assign you configure which refix size needs to be delegated to the downstream interface Aug 03 12:43:57 downstream interface being the interface on which you configure the ip6assign prefix size Aug 03 12:53:55 m4t: ath5k works without problems, but no benchmarks yet used. Aug 03 13:08:24 anyone had any experience with things like this, that claim "simultaneus" 2.4/5gig? http://www.eenewsembedded.com/news/mouser-now-stocking-pan9026-dual-mode-wi-fi-and-bluetooth-5-module?news_id=108591 Aug 03 13:10:02 rapid frequency switching, sounds like duplex operation Aug 03 13:10:12 I assume this thing is meant for client mode? Aug 03 13:15:51 I think that's a bit of false marketing going on Aug 03 13:16:08 It needs to be able to switch the radio between 2.4GHz and 5GHz to support bluetooth. Aug 03 13:17:56 "dynamic rapid channel switching" sounds like hopping back and forth Aug 03 13:18:05 Exactly. Aug 03 13:18:12 And it does that just for bluetooth operation. Aug 03 13:18:44 that makes sense, I think bluetooth protocol is better suited for passive oepration / being absent for a while now and then Aug 03 13:18:58 It's not going to be simultaneous dual band for wifi Aug 03 13:19:14 not sure how well it would work to keep simultaneous wifi connection with frequency hopping Aug 03 13:19:27 I guess there'd be a pretty big chance to miss crucial things Aug 03 13:19:35 afaik most solutions use two radios for this - one for the bluetooth and one for the wifi, and an RF switch Aug 03 13:19:55 karlp: you might also want to test my mac80211 update Aug 03 13:20:00 or anyone else for that matter Aug 03 13:20:24 oh, I've got that R6100 yesterday and tested 18.06 on it Aug 03 13:20:31 was basically flawless, no error encountered Aug 03 13:20:38 The two radio solution can receive on both radios simultaneously, this.. Aug 03 13:20:42 so really no idea what the issues in the bugtracker are about Aug 03 13:20:45 Either they've got two RX blocks or some serious magic. Aug 03 13:20:52 jow: great Aug 03 13:20:55 but, finally a NAND device for playing with ubi :) Aug 03 13:20:56 50 euro well invested Aug 03 13:21:06 and an ath10k one Aug 03 13:21:07 Hey, more hardware is better Aug 03 13:21:53 certainly more useful for regression testing than a wr1043nd with broken antenna and a wrt54gl without power supply Aug 03 13:22:07 What, you don't have a tub of 12V warts? Aug 03 13:22:29 not anymore Aug 03 13:22:40 Used them all or downsized? Aug 03 13:22:51 got rid of some stuff Aug 03 13:23:08 And it always ends up being the useful stuff. :P Aug 03 13:28:45 m4t: don't know how I missed that, should be UBNT_TYPE := RSx for routerstation Aug 03 14:02:43 jow: i am looking at that nbg6616 patch just now Aug 03 14:16:45 when adding support for more ar71xx devices in master at this point, should they just be ath79 instead? Aug 03 14:19:16 I would keep accepting for ar71xx until ath79 is no longer WIP Aug 03 14:19:34 and we should probably announce it on the MLs once we stop accepting new devices for ar71xx Aug 03 14:19:36 that would be great, stintel Aug 03 14:20:01 one device was already rejected, maybe i will get beaten for bringing it up again :D Aug 03 14:21:18 DonkeyHotei: yes Aug 03 14:21:26 DonkeyHotei: so we will use 4.14 for 19.01 Aug 03 14:21:31 and we wont use v4.9 Aug 03 14:21:34 ath79 wont be ready Aug 03 14:21:45 so its a-ok for a ar71xx v4.14 bumpage Aug 03 14:22:08 i will do so next week hopefully BUT i wont accept new boards into the target after august/september Aug 03 14:22:18 all new boards will need to be added to ath79 Aug 03 14:22:28 and i plan to drop those board supported by ath79 from ar71xx Aug 03 14:24:42 [1381103.993565] CPU3: Package temperature above threshold, cpu clock throttled (total events = 1090710) Aug 03 14:24:45 yay Aug 03 14:24:53 just wondered why my build is taking ages Aug 03 14:25:01 Intel FTL Aug 03 14:25:15 turbo but don't use for longer than 30 seconds! Aug 03 14:49:13 gninrom Aug 03 14:53:33 ultito: and there goes that password Aug 03 14:54:36 he's been typing that in this channel every day at around the same time Aug 03 14:55:51 rotanid: the Zsun? Aug 03 14:56:07 read it backwards :-) Aug 03 14:57:32 ah ok Aug 03 14:57:58 :-) Aug 03 14:58:24 first router in my grid runs 18.06 very well :-) Aug 03 14:58:54 ultito: ¡ooʇ noʎ oʇ ƃuıuɹoɯ poo⅁ Aug 03 14:59:42 looooool Aug 03 15:01:29 anyone knows if the usbip api is stable and platform agnostic? Aug 03 15:02:47 no idea, I don't use usbip Aug 03 15:03:07 wonder if I can make the usbip userspace shared Aug 03 15:06:23 rotanid: ah, you meant a board that is already ath79-only Aug 03 15:07:50 DonkeyHotei: depends. its another name for a board already supported, another formfactor with the same hardware inside Aug 03 15:08:12 ok, not 100% the same, the name is differentiable Aug 03 15:10:19 i was asking in reference to the board that Slimey has been hoping to get supported for ages Aug 03 15:10:42 dont know about that one Aug 03 15:11:10 used to post about it on the old forum Aug 03 15:13:28 didnt use the old forum (and not much the new either) Aug 03 15:48:18 blogic: can't fetch backports-v4.18-rc7.tar.xz Aug 03 16:01:33 well. mt76 really doesn't work on 2.4GHz Aug 03 16:05:07 indeed Aug 03 16:05:37 it's a bit frustrating that this is not clearly stated anywhere, I'd have spared a fair bit of money had I known beforehand :P Aug 03 16:09:01 got bit twice Aug 03 16:09:22 the thing is, it reportedly *used to* work Aug 03 16:09:49 quit Aug 03 16:09:52 because "wifi disable" does nothing, it's "wifi down". And the default action for "wifi" if the keyword is not recognized is "enable", which is a bit odd Aug 03 16:10:13 it's a bit frustrating that this state of "support" is not clearly stated anywhere, I'd have spared a fair bit of money had I known beforehand :P Aug 03 16:10:33 it's obviously a driver bug Aug 03 16:10:50 indeed Aug 03 16:11:45 during the openwrt-dd/lede divergence, reportedly mt76 in dd worked better than in lede Aug 03 16:11:57 interesting Aug 03 16:12:26 at least according to a post on the old forum Aug 03 16:13:40 * f00b4r0 is trying to figure where /sbin/wifi lives, to provide a patch Aug 03 16:14:06 found Aug 03 16:22:51 DonkeyHotei: looks like this issue also affected DD https://dev.archive.openwrt.org/ticket/17556.html#comment:11 Aug 03 16:23:30 it appears to have been there forever Aug 03 16:23:36 pretty much. Aug 03 16:24:26 tmn505: thanks, i made the change and push --force, PR is updated Aug 03 16:26:21 i remember explicitly changing it using faulty logic that RSx was a standin that signified both RS and RSPRO as a placeholder or something dumb lol Aug 03 16:27:04 happened while i had pasted both sections and was removing the redundancies Aug 03 16:28:06 jow: iirc it is dependent on the same kernel version or something like that Aug 03 16:28:36 i remember trying to connect one kernel to another kernel ver and having to override some #define Aug 03 16:31:15 f00b4r0: that's not m76, that's rt2x00 Aug 03 16:31:31 DonkeyHotei: don't ask me why it's what my mt7620 device displays Aug 03 16:31:54 as soon as I hit some amount of traffic Aug 03 16:31:59 rt2x00 is known-broken Aug 03 16:32:54 [ 0.000000] SoC Type: MediaTek MT7620A ver:2 eco:6 Aug 03 16:33:12 that's the SoC, not the wifi Aug 03 16:33:52 i thought 2.4 was embedded Aug 03 16:33:54 * m4t checking some old backup to see what he had to do Aug 03 16:34:36 f00b4r0: mt76 = mt7603e, mt76x2e, mt76x8 Aug 03 16:35:02 f00b4r0: MT7620 is the old rt2x00 driver, and sucky galore Aug 03 16:35:16 *sigh* Aug 03 16:35:40 i guess that's why these devices are cheap ;P Aug 03 16:35:48 Yes. Aug 03 16:36:12 That said, 7603e and 76x8 aren't great. Aug 03 16:36:22 well Aug 03 16:36:28 might be quicker to list what is ;) Aug 03 16:36:38 I have a longstanding issue with those devices as clients Aug 03 16:37:04 Which I hope nbd will find some time to look at more Aug 03 16:37:09 ath10k seems to suck, mt76 has issues... rt2x00 is awful. Have to wonder what one has to buy to go trouble-free Aug 03 16:37:24 ath9k is pretty solid it seems. 2.4 only tho Aug 03 16:37:29 I have no big issues with ath10k on my hardware Aug 03 16:37:39 I'll give a shot at https://github.com/sgruszka/wireless-drivers-next/commits/rt2800-draft Aug 03 16:37:50 ath9k is available for 5GHz, but it's .11n only Aug 03 16:37:51 listed from kernel bugzilla Aug 03 16:38:22 I also haven't had any issues with mt76x2 Aug 03 16:38:26 I'm mostly interested in 2.4 anyway tbh Aug 03 16:38:31 Performs quite well IME Aug 03 16:38:40 I get no coverage to speak of with 5GHz Aug 03 16:44:32 ath10k-ct is decent on 2.4 Aug 03 16:56:53 jow nevermind, actually i think it was actually because of a large difference in kernel versions, i.e. 3.18 vs 4.x. i.e. "staging: usbip: userspace: increase version to 2.0" in configure.ac Aug 03 16:57:06 tools/usb/usbip/configure.ac:AC_DEFINE([USBIP_VERSION], [0x00000111], [binary-coded decimal version number]) Aug 03 16:57:57 m4t: okay, so there's nothing speaking against making it kernel-agnostic Aug 03 16:59:58 i think the api is relatively stable at this point, i want to say it should be fine Aug 03 17:00:23 i've seen a lot of talk about atk10k-ct since the 1806 release, does this mean that wifi on the archer c7 v4 (qca9880 iirc) isn't all that? Aug 03 17:01:09 last change for tools/usb/usb/usbip/configure.ac was 2014 Aug 03 17:01:21 and even then it was just a rename Aug 03 17:02:55 jow: would like to see buildbot results for https://github.com/openwrt/packages/pull/6667 after travis is happy, so maybe this can be merged over the weekend plz? Aug 03 17:03:43 andy2244: sure: https://downloads.lede-project.org/snapshots/faillogs/x86_64/packages/samba4/compile.txt Aug 03 17:03:50 probably not what you expected :D Aug 03 17:04:49 yeah thats for the current version, the PR switches waf and hopefully all targets compile now (fingers crossed) Aug 03 17:05:28 is there a way of doing 'iptables -t mangle -A FORWARD -p udp -s 192.168.219.12 --sport 6981 -j DSCP --set-dscp-class CS1' in fw3 ? Aug 03 17:08:38 o/ ldir Aug 03 17:09:13 Borromini: hello Aug 03 17:09:19 are you melting as well. Aug 03 17:10:19 ha, yeah well, Mrs ldir decides that today is the day (along with mother-in-law) to visit Kew gardens in Richmond. It's not quite the hottest day of the year but..... Aug 03 17:11:02 * f00b4r0 tries to remember if one can get away with editing files in the linux kernel tree in build_dir Aug 03 17:11:58 today is also day 2 or absolute rage ranting at useless Sky ISP 'cos my line has apparently gone from 80/20 to 74/3.6. and that apparently *isn't* a fault. Aug 03 17:11:59 ldir: https://forum.openwrt.org/t/advanced-policy-routing-where-does-it-go/12402/12 Aug 03 17:12:26 *of* absolute rage ranting Aug 03 17:12:42 ldir: It's still working, not a fault Aug 03 17:12:46 :) Aug 03 17:13:06 ldir: i feel for you :) Aug 03 17:13:07 yes. Aug 03 17:13:45 ldir: 20:1 ratio? Sounds like piece of CAKE ;D Aug 03 17:13:45 well they've till Monday to fix it, and then they lose my custom for TV, phone & internet Aug 03 17:14:04 They don't want to deal with Openreach Aug 03 17:14:11 They have to convince them it's a fault Aug 03 17:14:12 f00b4r0: CAKE is making the resulting challenge a PIECE OF PISS ;-) Aug 03 17:14:19 hehe Aug 03 17:14:24 Openwretch Aug 03 17:14:43 actually Openretch usually. Aug 03 17:14:58 I was about to say. Aug 03 17:15:10 I have special cause to hate on them, the muppet in law works for them Aug 03 17:15:19 only 1 thing worse than openretch - mj quinn Aug 03 17:15:40 correction on 2 things worse than openretch - 1 mj quinn Aug 03 17:15:43 * Borromini was thinking openwretch was a foul allusion to openwrt Aug 03 17:15:48 can you guess the second? Aug 03 17:16:02 ldir: The government? Aug 03 17:16:04 :) Aug 03 17:16:14 ooh i know Aug 03 17:16:16 BREXIT Aug 03 17:16:20 3 things worse than openretch Aug 03 17:16:37 4 things worse than openretch - bloody hell it's all gone a bit Python in here! Aug 03 17:16:48 haha Aug 03 17:16:51 anyway... Kelly!!! Aug 03 17:17:33 If a Kelly van turns up, I'm taking a sledge to the vdsl cabinet.... so they *have* to send BT out Aug 03 17:19:09 premeditated vandalism. omg. Aug 03 17:20:17 ldir: Just accidentally drive into it Aug 03 17:20:45 Borromini: I am at that level, sadly. This is what customer disservice in this country is like. Aug 03 17:21:01 isn't it like that with any monopolist? :( Aug 03 17:21:07 i remember trying to get my vdsl installed Aug 03 17:21:17 provider would set an appointment, i'd take a day off Aug 03 17:21:33 and if you're a vaguely technical customer.... well best off shooting yerself now. Or driving into the cabinet :-) Aug 03 17:21:37 call them at 16:00... 'where's your guy?' 'erm sir we have no appointment on schedule' Aug 03 17:22:44 sounds lovely Aug 03 17:23:23 I wonder who's been taking lessons from whom? BT from Belgium telecom, or Belgium telecom from BT. Aug 03 17:23:46 I think ours was first Aug 03 17:23:47 ;P Aug 03 17:23:54 yes the britons did it first Aug 03 17:24:02 except for driving on the right side of course :P Aug 03 17:24:17 I had the opposite problem; a while back my cableco kept insisting I had an appointment to have TV and phone service installed. I had to call them several times to completely cancel it. Aug 03 17:24:19 I mean, we did invent it Aug 03 17:24:23 And we do drive on the correct side Aug 03 17:24:38 mamarley: they wanted to forcibly switch you to a newer modem or whatever? Aug 03 17:25:03 Borromini: I'm not even sure. I never requested that they do anything, and they didn't even supply the modem I do have. Aug 03 17:27:00 macOS appears very optimistic about what is good wifi signal. It shows me 5 bars when I'm down to MCS 1, 7 Mbps Aug 03 17:27:32 that's why Mac is better than windows Aug 03 17:27:34 and 72% packet loss Aug 03 17:27:39 i'd love linux to tell me everything was hunky dory. Aug 03 17:27:50 f00b4r0: have you thought about running an ISP ? Aug 03 17:27:55 Borromini: lol Aug 03 17:28:04 ;) Aug 03 17:28:48 ldir: I understand it's a regulation nightmare ;P Aug 03 17:28:48 although some boutique ISPs here are apparently thriving Aug 03 17:29:06 like the one I'm using right now. They're super cool. And very competent :) Aug 03 17:29:35 hmm did the forum dns just get fixed recently or am i misremembering? Aug 03 17:30:29 and at least they're not doing CG-NAT Aug 03 17:30:37 despite probably running out of IPV4s Aug 03 17:30:46 Do they have IPv6? Aug 03 17:31:08 they do when the infra operator allows it. Mine doesn't (yet, presumably) Aug 03 17:31:16 :( Aug 03 17:31:20 (this is FTTH in rural areas) Aug 03 17:31:50 i think it's very clear you need to be small to be competent, as a company Aug 03 17:32:05 once it's only size that matters, you just buy regulators instead of offering quality, it seems Aug 03 17:32:19 Borromini: well, it's more like you can't _afford_ to be incompetent as a small company, otherwise you're quickly out of business Aug 03 17:32:27 sadly that rule doesn't apply to the big players Aug 03 17:32:28 Why compete when you can anticompete? Aug 03 17:32:43 true enough Aug 03 17:32:45 f00b4r0: yeah, just don't answer the phone.... you'll be fine! Aug 03 17:32:52 hehe Aug 03 17:34:05 Is there a way to retrieve a single text file from travis-ci for a PR test run? Aug 03 17:34:06 IPV6 comes with strings attached for some ISPs here. Free is typically switching their clients to shared IPV4 when they "upgrade" you to fiber (they do CG-NAT) Aug 03 17:34:32 or say that the fault is in your internal house wiring... which we already know in your case is true... you're an expert! Aug 03 17:34:41 lol Aug 03 17:35:09 it's not even my house ;) Aug 03 17:35:56 sorta hard to blame ISPs for CGNAT though. we're out of v4, and it's just one of the consequences of that Aug 03 17:36:10 it's the other way around Aug 03 17:36:24 (on the other hand, it's not very hard to blame them for not having done v6...) Aug 03 17:36:32 instead of working harder to deploy they're just patching to death Aug 03 17:36:38 Sure you can blame ISPs for that. If they had adopted IPv6 more quickly, there would be no need for CGNAT. Aug 03 17:36:42 ^ Aug 03 17:37:00 no, it would still be needed for v4 Aug 03 17:37:11 they just found another way to delay the inevitable, all the while further breaking the internet Aug 03 17:37:36 you still need some way of reaching legacy v4 sites Aug 03 17:37:49 DNS64/NAT64 Aug 03 17:37:55 you need that as long as they exist. The point is to get rid of v4, eventually Aug 03 17:38:36 NAT64 at the ISP level is usually paired with CGNAT Aug 03 17:39:25 what ISPs didn't anticipate is IoT. That really threw their projections out the window Aug 03 17:39:34 (it's possible to avoid that by assigning each v6 client of the NAT64 its own v4 address, but obviously we don't have that many v4 addresses) Aug 03 17:40:45 Does IoT even affect ISP's IPv4 allocations? Aren't most IoT devices behind customer-premises NAT devices? Aug 03 17:41:18 hehe Aug 03 17:41:25 most are. Not all, AIUI Aug 03 17:41:27 not the iot devices that are part of mirai Aug 03 17:45:46 jow: I'm trying to do lan to wan dscp tweakage - that example shows wan to lan Aug 03 17:48:34 anyone want to try compile https://github.com/openwrt/packages/pull/6667 i have no clue why travis complains about it, it compiles fine locally, so i need to figure out if this is distro/host specific triggered? If it fails i need the cross-answer from the build_dir, thanks. Aug 03 17:51:34 hmmm so lede's forum is openwrt's now Aug 03 17:51:48 ldir: src lan, dest * Aug 03 17:51:56 yes, breaking all existing links to the openwrt forum Aug 03 17:51:57 there are no certainties in life anymore :'( Aug 03 17:52:11 DonkeyHotei: that is not true Aug 03 17:52:15 DonkeyHotei: and the link to the old forum someone scavenged? Aug 03 17:52:32 nice. forum.archive.openwrt.org Aug 03 17:52:43 this is the forum that died with apparently no hope of recovery Aug 03 17:52:49 yeah :) Aug 03 17:52:52 related, for some reason google seems to prefer wiki.openwrt.org to openwrt.org links too Aug 03 17:53:02 f00b4r0: yeah that's a habit that dies hard:( Aug 03 17:53:17 and inconveniently, the link in the "read only banner" isn't clickable Aug 03 17:53:20 bites me every time as well, google sth, get pointed to wiki.openwrt.org Aug 03 17:53:26 links in the form https://forum.openwrt.org/viewtopic.php?id=67411 still work and redirect to the forum archive Aug 03 17:53:42 https://forum.openwrt.org/viewtopic.php?id=67215 Aug 03 17:54:04 "Oops! That page doesn’t exist or is private." Aug 03 17:54:42 same with your link Aug 03 17:54:45 jow: that happens on both, and i'm logged in Aug 03 17:54:55 (not that being logged in should make any difference) Aug 03 17:55:15 i suppose adding a noindex header to the wiki.openwrt.org domain isn't an option Aug 03 17:56:20 it should be moved to oldwiki.openwrt.org just like it was the last time the wiki was replaced Aug 03 17:57:18 DonkeyHotei: point is, if you want search engines to notice, adding X-Robots-Tag http header helps... Aug 03 17:57:41 it's fairly trivial to do, usually Aug 03 17:57:49 Borromini: thanks, fixed. one of the two server had outdated nginx config Aug 03 17:58:33 thanks :) Aug 03 17:58:58 they both work now Aug 03 18:00:03 jow: https://pastebin.com/mWwNV1r6 - how am I being stupid? Aug 03 18:00:37 ldir: might be a false alarm, check "fw3 print | grep DSCP" Aug 03 18:01:07 you could add "option target ACCEPT" to make it shut up Aug 03 18:01:14 the -j in extra should override it Aug 03 18:04:45 ha, yes the rules (in filter) do actually exist. Aug 03 18:23:17 jow: does your nand ar71xx board has bad blocks by any chance? Aug 03 18:24:29 mkresin: nope, nand seems to fine yet Aug 03 18:25:15 jow: meh. I suspect the ar71xx out of tree nand driver doesn't return an error when reading a bad block Aug 03 18:26:02 anyone around who is using flow offloading with mt7621? Aug 03 18:26:13 *hw flow offloading Aug 03 18:27:57 ahh, I knew there was an issue somewhere "x_tables: ip_tables: DSCP target: only valid in mangle table, not filter". sorry jow Aug 03 18:31:46 stintel: ah of course Aug 03 18:31:50 stintel: let me mail it Aug 03 18:36:50 ldir: mind if i pick your brain on the 18.06 kernel bumps if i need to? Aug 03 18:38:46 Borromini: I'm pretty much out of the loop on them to be honest, since stintel and koen did a whole load, and dealt with some ext4 issue. Aug 03 18:41:34 ok. koen is saying thanks but no thanks on the bump, kinda Aug 03 18:51:55 blogic: too large to mail? :P Aug 03 18:55:36 stintel: got distracted Aug 03 18:55:39 on its way Aug 03 18:55:43 blogic: me too :) Aug 03 18:57:16 blogic: tha Aug 03 19:00:55 omg. ar71xx is such as mess a we managed to maek it worse with the tiny split Aug 03 19:01:14 mkresin: people cumplaining? Aug 03 19:01:23 stintel: I do Aug 03 19:02:05 we support boards of different flash sizes with the same mach file. hence the MACH symbol is now in the tiny and generic config Aug 03 19:02:07 andy2244: https://pastebin.com/NbcJ48D5 (just tried to build it with the ar71xx SDK) Aug 03 19:02:33 and it is a pain to get if it is intentional or a mistake... Aug 03 19:02:55 cool thanks, no clue why my local version dont need "POSIX capabilities are available" .. Aug 03 19:02:55 mkresin: and ... nbg6616 is generic (SPI-NOR) and 6716 is NAND both in the same mach file hidden behind a 6716 symbol only built by nand yet imagerized by generic Aug 03 19:03:14 mkresin: however i am currently building a fixup patch for it Aug 03 19:03:25 a big pile of shittery-do Aug 03 19:03:50 the qca brainfarty is viral Aug 03 19:04:22 My gli ar750s arrives today. Any special requests? Going to dissect it and stuff. Aug 03 19:04:28 blogic: flashing my DAP-2965 with your mac80211 branch now Aug 03 19:04:34 stintel: good Aug 03 19:04:41 blogic: okay, I drop my nbg6616 patch then. the mess isn'T worse the work. ar71xx need to die Aug 03 19:04:56 stintel: i've used that ath10k for almost 48 hours now and had amazon prime streams running for several hours Aug 03 19:05:11 mkresin: it will, rest assured Aug 03 19:05:31 blogic: fingers crossed Aug 03 19:05:37 stintel: lets see Aug 03 19:05:39 I know ath79 is a WIP, but it works great on my UAP-LR. :) Aug 03 19:05:43 blogic: are you using 802.11w with ath10k by any chance? Aug 03 19:05:52 dont know Aug 03 19:06:00 building for my uapacpro now, flashing that and then go get drunk Aug 03 19:06:40 stintel: How does that device work for you? I have considered obtaining one. Aug 03 19:06:52 mamarley: ath10k is le suck for me :( Aug 03 19:06:59 Sorry :( Aug 03 19:07:22 mamarley: flyspray #333 if you care Aug 03 19:07:30 it keeps biting me again and again Aug 03 19:07:35 same here Aug 03 19:09:43 I guess you have tried the -ct firmware too? Aug 03 19:10:51 mamarley: yeah talked about that yesterday, they were so slow for me that I couldn't even stream a 720p video in my bedroom Aug 03 19:12:15 ath10k pretty stable here with 18.06 Aug 03 19:12:32 f00b4r0: enable 802.11w and wait a few days Aug 03 19:13:10 mkresin: ah. Was noticeably worse in 17.01 so there seems to be an improvement still Aug 03 19:19:50 ldir: try option target MARK, then Aug 03 19:20:11 that should select the mangle table Aug 03 19:20:24 ok, will try Aug 03 19:26:09 now it wants a set-mark option! Aug 03 19:27:12 such a b.tch Aug 03 19:27:42 ldir: what's wrong with adding your rule to custom rules? Aug 03 19:28:14 f00b4r0: you mean firewall.user ? Aug 03 19:28:27 i mean whatever script is executed as "Custom rules" in Luci :) Aug 03 19:29:28 dunno, not looked - firewall.user works...except my rules keep on being added to every time the firewall bumps. Aug 03 19:29:47 ah. Sounds like a bug then ;) Aug 03 19:30:12 it would be nice (tm) if fw3 could be persuaded to do this sort of thing somehow natively. Aug 03 19:31:21 so I'm as far as 'DSCP tcp -- anywhere anywhere tcp spt:6891 dpt:smtp /* !fw3: Apply DSCP mark */ DSCP set 0x08' but note I didn't specify a dest port...so that's a mystery! Aug 03 19:32:12 dpt:smtp Aug 03 19:32:28 yes, I know, I don't want that Aug 03 19:32:59 https://pastebin.com/Znzi6s2v Aug 03 19:33:16 iptables -t mangle -A PREROUTING -p tcp -m tcp --sport 6891 -m comment --comment "!fw3: Apply DSCP mark" -j DSCP --set-dscp 0x01 Aug 03 19:33:19 iptables -t mangle -A PREROUTING -p udp -m udp --sport 6891 -m comment --comment "!fw3: Apply DSCP mark" -j DSCP --set-dscp 0x01 Aug 03 19:35:16 oh hell, what a turd - PEBKAC Aug 03 19:35:21 there's only --set-dscp? should be trivial Aug 03 19:35:45 fixed - a rogue # ;-) Aug 03 19:37:31 https://pastebin.com/xZuKLHF1 - same difference really :-) Aug 03 19:39:49 * ldir goes back to his wiring loom spreadsheet Aug 03 20:20:02 cp: cannot create directory '/home/varenet/openwrt/staging_dir/hostpkg/lib/lua/5.1/': No such file or directory Aug 03 20:20:17 what did I break again Aug 03 20:30:50 ldir: https://git.openwrt.org/?p=project/firewall3.git;a=commitdiff;h=12a7cf9db1f902f359038d9f52851da83c153407;hp=aa8846bb101054392b81f09bade120c021695892 Aug 03 20:31:36 f00b4r0: https://github.com/openwrt/luci/commit/bb336671fff613b812b308c7eeeeda38608bda9d Aug 03 20:32:10 jow: thanks. I used mkdir -p that solved it :) Aug 03 20:34:19 stintel: is -ct slow with the firmware or driver? you can mix and match Aug 03 20:51:02 if I want to test changes to drivers/net/wireless/ralink/, I should tweak files in build_dir/target.../linux.../backports/, correct? Aug 03 20:59:23 f00b4r0, hey nice to see you here :) Aug 03 20:59:40 guerby: likewise ;) Aug 03 21:17:25 [ 0.678730] spi-nand: Giga SPI NAND was found. Aug 03 21:17:27 [ 0.683329] spi-nand: 128 MiB, block size: 128 KiB, page size: 2048, OOB size: 128 Aug 03 21:17:29 hrm interesting. Aug 03 23:41:35 jow: would it be possible to add the uclient update to lede-17.01, too? Aug 03 23:44:13 hey Aug 03 23:44:19 i noticed i cant enter lede-dev anymore Aug 03 23:44:22 because it's +i Aug 03 23:44:35 i just annoyed nbd in a query about it, but then it occurred to me... maybe lede-dev channel is gone? Aug 03 23:44:38 because of the merge? Aug 03 23:45:04 zx2c4: correct, there's an automated redirection to this channel in place Aug 03 23:45:39 and of course that does nothing if you're already here Aug 03 23:46:24 DonkeyHotei: oh thats what must have happened Aug 03 23:46:29 let me try this for real Aug 03 23:46:44 oh hey Aug 03 23:46:57 it worked Aug 04 00:39:27 mangix: all combinations were slow for me Aug 04 00:43:33 out of curiosity, what is the maximum speed you achieved with ath10k so far? my test with ipq8065/ qca9984 achieved 45 MByte/s (lan-to-wlan, http on a LAN server, wget on a client) yesterday, both CPU cores loaded to the maximum Aug 04 00:44:15 I know it was well over 400Mbps Aug 04 00:46:31 anyway, I just flashed an image built from blogic's staging tree with mac80211 and friends backported from 4.18-rc7 on both my APs Aug 04 00:46:40 fingers crossed it will work better Aug 04 00:47:50 I hope to test that this weekend/ early next week - to much other stuff on the schedule first :( **** ENDING LOGGING AT Sat Aug 04 02:59:59 2018