**** BEGIN LOGGING AT Tue Feb 02 02:59:57 2021 Feb 02 03:39:24 any openwrt buildroot experts online? Feb 02 03:40:55 any idea why prepare target doesn't work for packages under feeds/packages/kernel/*? Feb 02 03:42:07 but works for another kernel package, such as package/network/services/wireguard/prepare Feb 02 03:42:52 is it because the ones under feeds/packages/kernel don't define virtual or meta packages? Feb 02 06:38:06 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_ath79.html has been updated. (99.1% images and 98.0% packages reproducible in our current test framework.) Feb 02 09:28:39 blocktrron: I'm not sure. I upgraded (remotely) from 18.06 to 19.07, that broke the config (due to changes in usbmuxd) so I got the router to me, fixed the config then dnsmasq started acting up, not delivering leases on br-lan. I "fixed" that by ticking "force even if other dhcp detected", Feb 02 09:29:57 but I didn't perform a final cold boot test after that, which turns out to be a mistake because now that the router is back at its location, it's not connecting. The phone (iPhone) shows a connection being shared but the router doesn't get link it seems and the small watchdog script (taken from wiki) keeps resetting the iface (seen on the phone cycling state) Feb 02 09:30:13 iphone running iOS 13 I should mention. Feb 02 09:31:22 wasn't there an issue with iOS 13 tethering a while ago? one that got fixed (at least in master) Feb 02 09:31:32 ios 14. Which is why I stayed on 13 Feb 02 09:31:49 (to my knowledge) Feb 02 09:33:05 zorun: Hauke: lechner: I'm just building some images for runtime testing https://gitlab.com/ynezz/openwrt/-/pipelines/250122010/builds of that wolfSSL update to 4.6.0 in 19.07 Feb 02 09:35:57 f00b4r0: oh, sorry Feb 02 09:36:01 np Feb 02 09:36:37 still can't explain why it seemed to work when I had my hands on it and won't now that it's 200km away Feb 02 09:36:45 maybe it misses me ;^P Feb 02 09:37:24 zorun: Hauke: lechner: working in progress https://openwrt.org/advisory/2021-02-02-1 Feb 02 09:37:58 Kernel guys (and aparcar[m]), one question: we're on a zstd-all-the-things (or as many as possible) rampage, aren't we? Or have I misinterpreted? Feb 02 09:41:10 f00b4r0: sensing hardware ;) Feb 02 09:41:16 :) Feb 02 09:49:32 rsalvaterra: what is zstd again? Feb 02 09:50:04 aparcar[m]: The compression algorithm? :P Feb 02 09:51:41 aparcar[m]: facebook made it ;) Feb 02 09:51:46 Sorry just kidding. What's your plan? Feb 02 09:52:11 Well, sometimes Facebook does one or two nice things… :D Feb 02 09:52:42 I backported my upstream patch to select the default compression algorithm for zram, breaking it's current strict dependency on lzo. This means it can default to zstd now. Feb 02 09:52:53 We're mostly on a "let's have a release soon" rampage so if you introduce sone ground breaking zstd changes please wait a bit Feb 02 09:52:58 And lzo can be completely deselected. Feb 02 09:53:24 Oh. I wager kernel patches aren't on the menu, then… :P Feb 02 09:54:13 :P Feb 02 09:55:16 And this probably warrants some LuCI zram section changes, but I don't do LuCI… Feb 02 10:00:52 rsalvaterra: some people do. I'm happy to review patches and also we got a experimental flag which can be used to disable such things for a release. However if you can please test and fix DSA 🙄 Feb 02 10:01:55 DSA? It's working fine here. Or do you mean in LuCI? Feb 02 10:03:30 rsalvaterra: good question. I need to run now but from what I know mostly DSA is missing Feb 02 10:06:09 zorun: Hauke: lechner: I've made them aware https://github.com/wolfSSL/wolfssl/issues/3709 Feb 02 10:24:01 zorun: Hauke: 4.6.0 builds on ath79/ipq40xx/x86 https://foo.true.cz/minio/openwrt/staging-builds/testing-openwrt-19-07/e72ab876/ and upgrade from 4.5.0 to 4.6.0 on 19.07 didn't breaked ABI of uclient-fetch/hostapd so seems fine and I'm going to push it Feb 02 10:46:07 nbd: git bisect pointed to the latest commit included in the 2021-01-27 bump, but with or without that, i am seeing no issues (pushed tens of gigabytes over the radio). I am seeing very bad performance though with latest master (which has two mac80211 fixes). Could the mac80211 patches play a role here? radio hasn't died yet but bitrate is all over the place (between 300-ish to 0) Feb 02 10:46:21 latest openwrt master, that is. Feb 02 11:20:28 Borromini: what was the openwrt version that you were seeing no issues with (with the 2021-01-27 bump)? Feb 02 11:21:59 nbd: cbedb5de75 Feb 02 11:23:52 Borromini: can you test if reverting openwrt commit 84fa59b5a80f8522239755d75d7ae51855c47fd2 makes a difference? Feb 02 11:24:45 nbd: will do. Feb 02 11:35:02 nbd: sorry for the noise about the bitrates. wife was sitting in the room where the AP is and that seemed to cause a lot of interference (a floor up) Feb 02 11:35:18 i'll keep an eye on the message 73 timeout but so far it hasn't popped up yet. Feb 02 11:37:28 well nevermind, i was on the wrong AP it seems. will revert 84fa59b5a80f8522239755d75d7ae51855c47fd2 and report back Feb 02 11:52:03 Borromini, when you have time, can you please check if you also have this problem? https://github.com/openwrt/mt76/issues/494 Feb 02 11:53:07 My hardware is MT7615, but the MT7613 uses the same driver, right? Feb 02 11:56:08 yeah same driver Feb 02 11:56:11 different firmware Feb 02 11:59:36 That's nice. At this point, it could be a firmware issue, for all I know. Feb 02 12:00:00 :P Feb 02 12:23:16 rsalvaterra: can you try latest openwrt master to see if it makes a difference for your issue? Feb 02 12:28:20 nbd: already running it, will test later, thanks! :) Feb 02 13:08:57 guys is tmomas on irc? Feb 02 13:09:09 and what's his handle Feb 02 13:09:21 * Borromini always gets confused about different handles for the same person Feb 02 13:36:28 * stintel couldn't resist and ordered the HiFive Unmatched RISC-V board =) Feb 02 13:42:06 stintel: you know about the beaglev? Feb 02 13:43:58 https://beaglev.seeed.cc/ Feb 02 13:45:47 too bad about the gigabit ethernet :P Feb 02 13:45:53 16 GB DDR4 is nice though. Feb 02 13:48:32 russell--: different category, that's more of an RPi clone Feb 02 13:49:42 the unmatched is kind of a desktop board, and I happen to have an empty mini ITX case lying around, and I'd like to play with RISC-V and not be frustrated by it being too slow for really playing around Feb 02 13:50:12 and since restaurants and malls etc are closed, travel is restricted, ... Feb 02 13:50:15 one can only game so much :P: Feb 02 13:50:42 although I'm trying to reach Battle Pass lvl 100 this season on CoD Black Ops Cold War Feb 02 13:53:54 but I might order a beaglev or alternative too at some point Feb 02 13:54:17 if you come across a Pi Zero W clone that is RISC-V based, please let me know :) Feb 02 13:56:47 stintel: obviously the next Raspberry Pi will the the Pi V ;-) Feb 02 13:56:58 :P Feb 02 13:58:51 Borromini: "too bad about gigabit ether" means what? you wanted 10G or what? Feb 02 13:58:57 I did :P Feb 02 13:59:11 but doesn't make sense on this kind of board Feb 02 13:59:11 :P Feb 02 13:59:18 karlp: 2,5Gb would be nice Feb 02 13:59:34 stintel: why not? (just curious, not familiar with it) Feb 02 13:59:43 si 2.5G really that common to be worthwhile? is it really going to be prevalent? it seems like such a halfway state? Feb 02 14:00:06 seems like 2,5 is the next affordable thing in the years to come Feb 02 14:00:41 yep, and compatible with multigigabit devices Feb 02 14:01:09 so you can have multigbe switch with "expensive" 10Gbe fileserver, cheaper 2.5GbE clients Feb 02 14:03:07 lots of high and mid-end x86 boards already come with 2,5 Gbit NICs it seems Feb 02 14:03:32 stintel: you gonna turn yours into a file server? Feb 02 14:04:00 Hi! I'd like to ask if I should do something with this: https://github.com/openwrt/openwrt/pull/3135 Feb 02 14:04:05 Borromini: no my fileserver is xeon with dual 10GbE Feb 02 14:04:09 ;) Feb 02 14:04:37 but looking at https://www.asrockrack.com/general/productdetail.asp?Model=EPYC3451D4U-2L2T2O8R maybe to replace the board + PCIe NIC Feb 02 14:04:43 if only I can find it somewhere Feb 02 14:08:23 :) Feb 02 14:49:55 there is probably no nice way to say this but instead of trying to reach level 100 in CoD you could also offer you help to others who could use help, maybe then one wouldnt have to critizise others for reluctantly prioritizing due to lack of time and resources Feb 02 14:53:02 grift: I spend enough time on my day job, all time and effort spent on open source is voluntary, doesn't pay any bills, etc Feb 02 15:05:22 grift: correct, there's no nice way to have tried saying that :) Feb 02 15:09:56 grift: For a lot of us, OpenWrt is a (useful) hobby which we do for fun. Speaking for myself, I already have $dayjob. If this starts to smell like work to me, I'll fuck right off. :P Feb 02 15:10:43 yes and how would you like it if i would critize your work becaus of that? Feb 02 15:10:54 because thats exactly my point Feb 02 15:11:31 what i do is also my hobby but i get blamed that i dont do the work for others as well Feb 02 15:12:24 I don't mind being blamed if things break for changes *I* made voluntarily, and I have the responsibility to fix them. Feb 02 15:13:12 That's just common sense, I believe. Feb 02 15:16:47 i guess so. i guess next time ill shrug it off with: "ok thats my bad, i will address it when i get to it" Feb 02 15:17:41 some things arent fixed that easily Feb 02 15:17:45 Well, if I break something, I try to fix it ASAP, of course. Feb 02 15:18:04 well thats what i said "when i get to it" Feb 02 15:19:33 I can't speak for everybody else, but that works just fine for me. :) Feb 02 15:25:39 so i guess that means that when you add that zram lzo functionality you will also address the luci integration Feb 02 15:28:00 because we might not have the luxury to say "i dont do luci" Feb 02 15:30:13 if you're dealing with a sense of entitlement from user x, tell user x you accept donations for accelerated development. or patches :) Feb 02 15:31:16 grift: BTW what's the context? I don't follow here... Feb 02 15:31:53 well obivously my selinux policy changes are very intrusive and fragile so i have to prioritize my work Feb 02 15:32:09 and so one afternoon i mentioned the i focus first on the stuff i use Feb 02 15:32:34 but people interpretted as saying that i dont care for users (or something) Feb 02 15:33:05 it just needs time, you'll learn to ignore it :) Feb 02 15:33:08 but i was just saying "well because i cant do it all in reasonable time alone anyway i will focus on my own use case first Feb 02 15:33:28 i almost did ignore it Feb 02 15:33:38 slip of the tongue i suppose Feb 02 15:33:39 AFAIK it's optional, right? Feb 02 15:33:47 sure Feb 02 15:40:05 you were calling very common stuff "exotic", insinuating it's not worth your time, and this is my main problem. if something is in openwrt.git, and selectable via menuconfig, it's not exotic, it's supported, and should not be ignored Feb 02 15:40:09 ynezz, You also commented the PR months ago. Could you please have a short look? https://github.com/openwrt/openwrt/pull/3135 Feb 02 15:42:08 stintel: patches accepted (i have to prioritize, its a huge task, i do it when i get to it (if i dont get any help before that) Feb 02 15:44:35 grift: sure, just keep in mind that openwrt is very customizable, it's actually one of the reasons I'm using it over something like dd-wrt, and I'm sure many people are using it for the same reason. so if someone has a problem with e.g. libmbedtls, which is in openwrt.git but not enabled by default, while testing your selinux policy stuff, then at least don't call it exotic Feb 02 15:45:11 stintel, the policy is also customizable ... Feb 02 15:46:27 ok i wont call it exotic anymore promise Feb 02 15:46:56 how about "optional" sound better? Feb 02 15:47:36 yes :) Feb 02 15:47:50 i was just saying the first priority is the stuff that is a default and is not really considered optional Feb 02 15:48:26 eventually obivously i would love to address all, but thats not going to happen any time soon especially if no one else cares Feb 02 15:49:00 I might have misread that, so maybe just a miscommunication Feb 02 15:49:20 probably also in part my fault Feb 02 15:49:50 i might have been trying to incentivise other to get involved too much Feb 02 15:49:54 communication is hard sometimes, we have similar issues on company slack :) Feb 02 15:50:34 if I ever manage to finish all my other WIP I might spin an image with SELinux enabled for one of my test devices Feb 02 15:50:50 that would be appreciated, thanks Feb 02 15:51:57 but my stuff is so customized that it's going to break almost everything, and right now motivation for doing anything is pretty low Feb 02 15:52:27 if you dont use luci then things might just surprise you Feb 02 15:52:42 luci integration is the hardest part Feb 02 15:53:14 did I suggest to have something like $packagename-selinux for per-package policy? might make more sense than using a single policy package Feb 02 15:53:37 I think I did but I don't remember Feb 02 15:53:44 yes but that is not possible because the policy is loaded from squashfs Feb 02 15:53:55 ie the lower layer Feb 02 15:54:29 so it wouldn't work for anything installed via opkg install .. Feb 02 15:54:35 so if you would do that and you would reboot then your $packagenaame-selinux would be ignored Feb 02 15:55:04 in theory everyhing works with opkg, you just cannot alter the policy at runtime Feb 02 15:56:05 its probably more understandable if you give it a whirl when you find a dead moment Feb 02 15:56:25 I honestly considered it, on my spare ERL Feb 02 15:56:47 but I probably got distracted by something random on my todo list Feb 02 15:57:12 or random breakage of something in my infra Feb 02 15:58:20 anyway, back to powershell for a bit :P Feb 02 15:59:23 before anyone asks: no, it's *not* for fun ;) Feb 02 16:01:04 for the record i dont mind having fun, "all work and no play makes jack dull" Feb 02 16:01:15 i was just a bit touched by your critisism Feb 02 16:02:37 well let's consider it sorted, no hard feelings Feb 02 16:02:49 alright Feb 02 16:05:22 Simple question: Should I just close and reopen pull requests so they will get attention after months? Feb 02 16:06:50 FooDeas: ping @maintainer ? Feb 02 16:15:04 are we back to why luci renders all the markup with js again? :) Feb 02 16:23:29 jow: give us some static html to play with! Feb 02 16:59:41 stintel: I don't know who exactly may be maintaining this module in OpenWrt. Feb 02 17:00:09 FooDeas: is it openwrt.git or one of the feeds ? Feb 02 17:00:43 stintel: openwrt.git Feb 02 17:00:49 I automatically expected a feed because github PR, forgot there are also github PRs for openwrt.git Feb 02 17:01:02 FooDeas: url ? Feb 02 17:01:06 (of the pR) Feb 02 17:01:17 URL: https://github.com/openwrt/openwrt/pull/3135 Feb 02 17:04:21 FooDeas: well you pinged one of the team members already (adschm), maybe you can ping other team members that previously commented, we don't have target maintainers anymore so there's no maintainer to ping, one other option is to git send-email --to openwrt-devel@lists.openwrt.org, it gets more love than github afaik Feb 02 17:05:24 stintel: Thank you very much for your assessment! Have a nice evening! Feb 02 17:13:29 build #735 of apm821xx/nand is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/apm821xx%2Fnand/builds/735 Feb 02 18:31:38 nbd: ping Feb 02 18:53:42 rsalvaterra: pong Feb 02 18:54:59 nbd: cautiously optimistic news, I'm able to access my dad's laptop on the 5 GHz band (MT7615). Feb 02 18:56:13 It's still not obvious to me which commit might have fixed the broadcast issue. Feb 02 18:57:29 interesting Feb 02 18:57:59 one candidate is 84fa59b5a80f8522239755d75d7ae51855c47fd2 Feb 02 18:58:26 it fixes a bug where mac80211 was doing rate table updates on stations that had not been uploaded to the driver yet Feb 02 18:58:43 I thought you might say that, and I was already looking at it. ;) Feb 02 18:59:06 this bug specifically affects 7615 and not the other chips Feb 02 18:59:33 earlier chips upload the station before assoc and newer chips don't use software rate control Feb 02 19:00:05 Software rate control. You mean minstrel_ht, right? Feb 02 19:00:25 yes Feb 02 19:00:49 i found it while developing some code to allow user space to control minstrel Feb 02 19:02:19 I see. So, starting with the 79xx series, rate control is done in hardware, right? Feb 02 19:05:22 Anyway, I need to test this more thoroughly, since I have the feeling that sometimes I was be able to connect, only to lose connection after several hours (due to ARP ageing, possibly?)… Feb 02 19:07:56 nbd: i have hammered the mt7613be radio, it seems the message timeout issue only pops up when there's no clients connected). so i'm redoing the bisect. Feb 02 21:07:36 ynezz: thanks for writing the advisory Feb 02 21:31:24 seconded Feb 02 21:46:34 nbd: I am having problems with my iwl8265 to connect to my MT7603E/MT76x2E since January, the iwl8265 needs pretty long to connect and sometimes it looses the conenction and then it needs abour 5 minutes to connect again. I didn't had this problem with OpenWrt from 5. Dezemebr, but have it at least in 96017a60138bdf97d2f1f7e7b519b4ad6ff13b88 from mid of January. Is this a known problem or should I try Feb 02 21:46:36 to bisect it? Feb 02 21:47:20 MT7603E/MT76x2E is inside a Xiaomi Mi Router 3G and iwl8265 inside a T480 Feb 02 21:48:09 this is OpenWrt master + debian stable (kernel 4.19) Feb 02 21:48:23 scanning on iwl8265 also does not see my wifi Feb 02 22:01:18 Hauke: I've had similar issues (8265 / T470). Connecting to MFP networks would often time out, Some 2.4 GHz networks not visible in search. Feb 02 22:02:10 However, this also occured with APs running vendor firmware (AVM / Aruba). Feb 02 22:03:13 intel ax200 is also far away from perfection (current stable kernels) Feb 02 22:04:03 These issues were not there with a 7265 / AX200, so i went along there. Not to say that it might not be mt76 related, just sharing my experiences there Feb 02 22:04:38 pkgadd: try an AX210 where bluetooth disappears from the USB bus as soon as iwlwifi comes up :P Feb 02 22:06:14 all my clinets are now on the 2.4GHz Wifi which is strange Feb 02 22:06:16 * Borromini has an 8260AC and eternal issues with BT on Linux but not so much with WiFi Feb 02 22:08:01 blocktrron: hehe, actually I've been looking for ax210 - because of 6 GHz support, but my biggest worry in that regard would be overzealous regdom settings rendering that feature useless Feb 02 22:08:12 I think the 5GHz is borken, I will have a look in the next days Feb 02 22:08:54 pkgadd: Is 6Ghz already allowed in the EU? Feb 02 22:09:29 I think the FCC approved it some time ago for the US and I am sure it will also be approved by the EU soon Feb 02 22:10:45 Germany will get 6GHz Wifi in Q2 2021: https://www.bundesnetzagentur.de/SharedDocs/Pressemitteilungen/DE/2020/20201125_6GHz.html?nn=265778 Feb 02 22:11:00 Hauke: i would stay away from now, as it is not even capable of HE in it's current state Feb 02 22:11:47 I thought 6GHz wifi should be HE only Feb 02 22:11:56 I have the second one I've ordered from china still around, if you are interested i can send it to sou for the amount I've got it from aliexpress Feb 02 22:12:13 It only does VHT on 5 GHz Feb 02 22:12:23 what second one? Feb 02 22:13:03 Ordered two (for both notebooks) but only installed one into my spare one, discovered it's current state and stayed away for my main machine Feb 02 22:13:33 blocktrron: no I do not really need a ax210 Feb 02 22:13:46 I do not have any other 6GHz devices Feb 02 22:13:57 That's what I was thinking to myself after using one for 5 minutes :P Feb 02 22:15:22 Hauke: iirc, it should come available in summer (I'm not in a hurry to buy an ax210 (actually with desktop-PCIe adapter), but given the sources of these devices, I worry about region code clashes nevertheless) Feb 02 22:17:06 often there are engineering samples sold on ebay and aliexpress, they will work as long as Intel and their customers also use engineering samples, but then they remove support for them from the drivers later Feb 02 22:17:22 some people complained about this for older generations Feb 02 22:17:42 yep, that as well Feb 02 22:26:35 ynezz: thanks! Feb 02 22:28:29 stintel: where did you buy the Unmatched? Feb 02 22:29:25 zorun: mouser (pre-order) Feb 02 22:30:06 stintel: do you already have it? Feb 02 22:30:21 Hauke: (pre-order) ;-) Feb 02 22:30:27 ETA early march Feb 02 22:30:39 oh ok, I thought it was more widely available already Feb 02 22:32:51 this looks intresting: https://www.cnx-software.com/2021/01/13/beaglev-powerful-open-hardware-risc-v-linux-sbc-targets-ai-applications/ Feb 02 22:33:32 zorun: no, it should have been but they decided to upgrade 8GiB to 16GiB RAM Feb 02 22:33:49 or that's how I understood it Feb 02 22:35:13 I see Feb 02 22:36:28 but that's a good thing, I don't think I would have ordered it if it was 8GiB RAM Feb 02 22:41:43 the mouser page still mentions 8 GB RAM Feb 02 22:42:01 https://eu.mouser.com/new/sifive/sifive-hifive/ Feb 02 22:42:38 yes but read the crowdfunding page Feb 02 22:42:43 forgot which one Feb 02 22:42:58 crowdsupply Feb 02 22:43:06 ok Feb 03 01:17:10 problem is that most of the desktop pci-e 11ax cards i've seen are just ax200's mounted on adapters Feb 03 02:11:19 Namidairo: anything wrong with that? Feb 03 02:12:24 Namidairo: that's not a problem at all, I have one of those - iwlwifi (and that of its firmware) reliability however is another topic Feb 03 02:12:40 the one in this box works fine and dandy Feb 03 02:13:49 just a little choice in the matter would be nice if you don't like Intel always setting them to NO-IR etc Feb 03 02:14:07 yeah, aside from the weird latencies (dropped packets, losing the connectiong, etc. - AP's are QCA9984/ ath10k, QCA9880+BR4A/ ath10k, QCN5054/ ath_pci) Feb 03 02:15:21 and very firmware version dependent bugs Feb 03 02:18:10 why am I still looking for ax210 on a PCIe riser? because of availability, pricing (at least from ebay, aliexpress, amazon, etc.), wifi6e and BT (actually the system I'm having this in mind for needs BT and occassional wifi/ client usage) **** ENDING LOGGING AT Wed Feb 03 02:59:57 2021