**** BEGIN LOGGING AT Thu Nov 26 02:59:57 2020 Nov 26 04:21:19 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_mediatek.html has been updated. (100.0% images and 98.4% packages reproducible in our current test framework.) Nov 26 08:06:47 I've sent out a patch to fix the broken builds for mt7621 (sorry for that) Nov 26 08:13:09 svanheule[m]: thanks, pushed Nov 26 08:21:07 mangix: I think, that "cmake.mk: set C/CXX compiler for host builds as well" is proper fix for those gcc/clang macOS woes, so rejecting the other clang related fixes Nov 26 08:50:58 ok Nov 26 09:17:29 ynezz: I'm of the opinion that both patches should be merged but w/e. Nov 26 09:18:53 the openssl is wrong Nov 26 09:19:10 or the description doesn't make sense Nov 26 09:19:33 that libjson one seemed wrong as well Nov 26 09:20:11 adding just compiler error into the commit description is not enough, you need to provide context Nov 26 09:30:43 openssl's config system wrongly picks up host CC instead of target. that's really it. Nov 26 09:31:18 libjson-c has a strange math patch that fixes...i have no idea what/ Nov 26 09:36:25 then fix the openssl config system or report it upstream and backport fix, removing the compiler flag is not proper fix Nov 26 09:37:03 if you want to remove that math patch for libjson, then just propose the patch to do that, don't hide it behind some strange macOS foo Nov 26 09:38:06 maybe it was just needed back in the days and now it's obsolete, who knows Nov 26 09:39:41 :+1: Nov 26 10:04:21 If it helps, I am able to build on macos catalina, xcode 12, on master plus the nettle patch that I submitted. Nov 26 10:05:19 nbd's recent macos fixes allowed building using xcode 12, which AFAICT made clang fussier, ie. more warnings enabled by default. Nov 26 10:10:42 'morning! Nov 26 10:11:46 I'm a bit confused… what are the canonical OpenWrt repositories? The ones on GitHub, or the ones on git.openwrt.org? Nov 26 10:12:31 On git.openwrt.org, it says the packages repository is a *mirror* of the packages feed. Nov 26 10:12:43 yep Nov 26 10:12:49 so packages is github. Nov 26 10:12:52 the others are git.openwrt.org Nov 26 10:12:56 like it says... Nov 26 10:13:01 for the main repo and projects like netifd, git.openwrt.org is primary Nov 26 10:13:11 for the feeds, github is primary Nov 26 10:13:39 What a mess. :P Nov 26 10:13:49 no? Nov 26 10:14:13 no, packages feed is a community project, with many more committers than openwrt core and everything related to it Nov 26 10:14:22 Oh, well… I guess I'll reconfigure my git remotes, then. Nov 26 10:15:06 so what is this libroxml package in base? Nov 26 10:29:37 Thanks, stintel! I had pages of these. :) https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=26c26e11a2131e3e07a9322c9bea6dcc107cddad Nov 26 10:29:52 can we maybe merge my commit soon? https://github.com/openwrt/openwrt/pull/3596 Nov 26 10:30:42 commits to core are always second class citizens if they come from github, you'll get more traction on it posting to the list. Nov 26 10:31:26 I never understood the aversion to git send-email… Nov 26 10:32:07 rsalvaterra: generation Z :P Nov 26 10:32:35 rsalvaterra: and welcome. are they all gone now for you? Nov 26 10:33:25 stintel: Not yet, haven't built today, but the kernel was bumped too, I have two wonderful reasons for it. :) Nov 26 10:33:47 on the mailing list it was already discussed that the ipq40xx soc has now issues Nov 26 10:33:48 some other person wanted to revert the changes Nov 26 10:33:55 I introduce a kernel flag to switch between port isolation, and nested vlans Nov 26 10:34:43 ah nice, I'll go for a bunch of image builds then today Nov 26 10:34:44 rsalvaterra: because it requires functional MTA, which is decidedly non trivial. Nov 26 10:35:31 karlp: You've got to be kidding me. Nov 26 10:35:53 yeah and the gmail webinterface is user friendly Nov 26 10:35:56 ¯\_(ツ)_/¯ Nov 26 10:36:30 it's a frickin' abomination. not to mention I have too much browser tabs open already, one more for mail is just asking to not look at mail Nov 26 10:36:36 Three Debian packages: git-email libauthen-sasl-perl libmailtools-perl. Nov 26 10:36:41 That's all. Nov 26 10:36:51 (we use google mail for work) Nov 26 10:37:01 hey but gmail just revamped its interface. again. it got even better! Nov 26 10:37:03 /s Nov 26 10:37:16 Of course, the configuration is email-service specific, but Gmail is trivial to configure with STARTTLS. Nov 26 10:37:19 half of the time I miss meetings because the tab stopped refreshing and don't get notifications Nov 26 10:37:32 stintel: I'd consider that a feature Nov 26 10:37:47 oh but we're moving to office365 without IMAP/SMTP/... Nov 26 10:38:03 Ah, meetings… where minutes are kept and hours are lost. Nov 26 10:38:04 I guess I'll be installing evololution Nov 26 10:38:06 stintel: RIP Nov 26 10:38:24 stintel: i kind of like geany. Nov 26 10:38:30 although i think it uses the evolution backend Nov 26 10:38:34 svanheule[m]: acquisitions ftl Nov 26 10:39:05 I might soon receive an ubuntu laptop with 4 cores and 16GB RAM, with ubuntu and no sudo access Nov 26 10:39:11 stintel: my university used to have SMTP/IMAP, but then they went full MS a few years ago :-( Nov 26 10:39:12 as a frickin' contractor Nov 26 10:39:35 because I will no longer be allowed to use my own hardware Nov 26 10:40:02 stintel wtf? lol What are you going to do with that? Nov 26 10:40:10 That just sounds mad. Nov 26 10:40:20 Tapper: probably take it to a shooting range for target practice Nov 26 10:40:35 A sound plan! Nov 26 10:40:41 not sure if I can do long range with 50 cal here though Nov 26 10:40:48 hahah Nov 26 10:40:50 because I'd love to :D Nov 26 10:40:59 Would make a nice vid for youtube. Nov 26 10:41:15 to hand in with my notice :P Nov 26 10:41:20 haha Nov 26 10:41:32 Would be like russian FPS. Nov 26 10:41:56 Rush B? :P Nov 26 10:42:24 https://www.youtube.com/watch?v=rC9LFLpHmeE Nov 26 10:43:38 Oh, my, the stereotypes…! XD Nov 26 10:47:13 * nitroshift invites stintel over for a shooting session Nov 26 10:47:39 though i don't have 50 cal, only 30x06 magnum Nov 26 10:48:19 nitroshift: do I need a PCR test though :P Nov 26 10:48:32 rsalvaterra Yeah I know, I bet that dude is not even from Russia Nov 26 10:48:40 umm... i ain't got a clue about that Nov 26 10:48:55 rsalvaterra, 0/ Nov 26 10:49:04 It doesn't sound a lot like a Russian accent, for sure… Nov 26 10:49:10 nitroshift: o/ Nov 26 10:49:13 nitroshift never mind that laptop needs to die. lol Nov 26 10:49:28 speaking about shooting: https://www.newsbreak.com/news/2107406337575/trans-species-man-who-self-identifies-as-a-deer-accidentally-shot-by-hunters?s=ws_tw Nov 26 10:49:40 Tapper, be mean, don't feed it ^_^ Nov 26 10:50:09 stintel holey crap! Nov 26 10:50:17 stintel, rofl Nov 26 10:50:25 WTF?! Ok, we can close the internet for today. Nov 26 10:50:27 serves him right! Nov 26 10:50:55 * nitroshift waits to be kicked in his back Nov 26 10:51:13 "“When he was eight he believed he was a porcupine, then at 12 a squirrel, now he’s a deer. It could be worse, he could be a homosexual,” his mother argued when reached by phone." Nov 26 10:51:23 That is so sad. Nov 26 11:09:17 Tapper: fparussia is a professional russian, not a real one Nov 26 11:10:43 mangix yeah I thought so! Nov 26 11:12:08 he also happens to be in jail Nov 26 11:15:07 rsalvaterra: hi! Any news about ubifs/zstd ? Nov 26 11:16:05 mrkiko: It's working fine in my tree, but the whole thing is brittle as hell. Nov 26 11:17:28 mrkiko: https://github.com/rsalvaterra/openwrt/commit/2c6030dd84fcfb7d65824be3ecd0b625165832d9 Nov 26 11:18:02 And also: https://github.com/rsalvaterra/fstools/commit/e90abe7ff4aab60f57f6bba4e8d5ad9635dba9ac Nov 26 11:18:32 Of course, you have to compile the kernel with zstd support for ubifs. Nov 26 11:20:10 The main problem is that we treat jffs2 and ubifs (and, well, basically any filesystem on the overlay) as if it supported the same mount options. Nov 26 11:28:24 videos feature Kyle Lamar Myers, an American from Franklin County, Georgia, playing the role of Dimitri, a heavily accented Russian from Moscow. Nov 26 11:28:47 mangix you were write Nov 26 11:46:23 see on a recent build of openwrt master branch on an APU4 and a pcengines 4GB pSLC SD card: http://paste.debian.net/1174423/. I am not seeing the error on the same card with r5244-f0c37f6ceb or another SD card from the same batch with r10863-e1dcfe02b2 (a little over a year ago) on APU2. Nov 26 11:48:30 a SanDisk Ultra SDSQUNS-016G-GN3MN 16GB seems to work with the fresh build, seems to be an interaction between the kernel and the card Nov 26 15:15:01 stintel: New build deployed, no more sh complaints. :) Nov 26 15:26:43 rsalvaterra: \o/ Nov 26 15:29:47 Has anyone played with the new airtime policy settings? Nov 26 15:30:33 I'm seeing (what I interpret as) some odd behaviour with a per-BSS-limited configuration. Nov 26 15:30:52 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_kirkwood.html has been updated. (100.0% images and 98.4% packages reproducible in our current test framework.) Nov 26 16:49:05 rsalvaterra: I was setting it to mode 3 , weight 4 on lan, weight 1 on dmz and iot Nov 26 16:49:13 but didn't do it on both my APs yet Nov 26 16:51:58 Hmm… I see all stations with the same weight (256) on the AP, though. Nov 26 16:54:31 stintel: https://paste.debian.net/1174469/ Nov 26 16:55:02 apparently I'm seeing the same. but then I just remembered the AP I'm looking at doesn't even support ATF Nov 26 16:55:43 or at least not on the 5GHz radio Nov 26 16:56:05 Well, these are two mt76 radios, so they do claim to support airtime fairness. Nov 26 16:57:55 but I did not configure per client weight, only per bssid Nov 26 16:58:27 svanheule[m]: did you figure out why the 5GHz radio on the eap245v3 doesn't support ATF? Nov 26 16:58:32 Yeah, that's what I did too. Nov 26 16:58:39 Did you see the paste? Nov 26 16:58:49 stintel: no :-/ Nov 26 16:59:04 rsalvaterra: so maybe try configuring per-client weight and see if that changes anything ? Nov 26 16:59:28 there were some people on the forum looking at throughput for the QCA9982, but ATF wasn't mentioned I think Nov 26 16:59:40 rsalvaterra: I ran through the paper with half an eye while in bed yesterday so really don't know much about it Nov 26 16:59:56 svanheule[m]: it feels kind of slower than my wave 1 unifi Nov 26 17:00:04 stintel: I'll try it, just for teh lulz, but I don't really want to micromanage that stuff. :) Nov 26 17:00:24 stintel: https://forum.openwrt.org/t/adding-openwrt-support-for-tp-link-eap245/57583/164?u=svanheule Nov 26 17:01:03 rsalvaterra: I suspect the per-bssid weight is handled differently and not visible in `iw wlanX station dump` output - but just a wild guess Nov 26 17:01:07 stintel: I didn't really push it, or make an in-deep comparison to my Wave-1 AP Nov 26 17:01:47 stintel: I thought of that possibility too, but I don't know who could enlighten us about it… there's almost no information available. Nov 26 17:02:45 rsalvaterra: ask Toke Nov 26 17:02:58 Does he lurk around here? Nov 26 17:03:01 can't recall his nick Nov 26 17:03:45 tohojo ? Nov 26 17:04:16 he's in #openwrt :) Nov 26 17:04:36 Hah! Nov 26 17:05:03 but I decided not to stay there, I tend to get annoyed easily, so that's my way of following our "be nice" rule Nov 26 17:05:36 I'm on both channels, but I rarely interact in #openwrt… Nov 26 17:06:40 Unfortunately I also don't have a lot of patience to answer questions which can be answered with a simple search. Nov 26 17:07:05 (And this doesn't get any better with age. :P) Nov 26 17:11:16 and I guess 2020 also doesn't help Nov 26 17:11:58 shitshow at work is finally settling down a bit. glad I managed to get a few commits in again Nov 26 17:12:49 wow, those are my first commits since August 1, wtf how time flies Nov 26 17:17:18 yay, just managed to load an fw4 nftables reference ruleset Nov 26 17:17:39 almost reached feature parity, some obscure corner cases require Kernel 5.9+ though Nov 26 17:18:04 jow: Well, bump it, then! :D Nov 26 17:18:09 * rsalvaterra runs Nov 26 17:19:01 :P Nov 26 17:19:13 jow: cool. I didn't even know fw4 was a thing :( Nov 26 17:19:21 it's not public yet Nov 26 17:19:45 stintel: It's fw3, but for nftables instead of iptables. Nov 26 17:19:45 but once finished, it is supposed as drop-in replacement for fw3, using the same uci config but translating to nftables isntead of iptables Nov 26 17:19:55 I guess it'll be something for > 20.x Nov 26 17:20:04 likely, yeah Nov 26 17:20:25 I hope so, since this will force me to learn nftables. :P Nov 26 17:20:26 if we even manage to get a 20.x Nov 26 17:21:57 Until the 31st December we can have a 20.12… :) Nov 26 17:26:20 https://pastebin.com/sQnXzV3s Nov 26 17:26:37 translates ruleset starts at line 298 Nov 26 17:27:48 this covers both ipv4 and ipv6 simultaneously, restricting to inet or inet6 if needed Nov 26 17:34:28 stintel: https://bugs.openwrt.org/index.php?do=details&task_id=3405 Nov 26 17:43:06 and back to yak shaving :/ Nov 26 17:43:48 https://bpa.st/3JGQ Nov 26 17:47:16 is Florian Eckert on IRC ? Nov 26 17:49:06 ynezz: i'm now building malta/mips32be to also try what happens in initramfs with ujail Nov 26 18:02:28 stintel: airtime fairness should work with ath10k radios. i can only speak for the non-ct flavor, but the airtime weight displayed on "iw station dump" depends on the airtime policy mode Nov 26 18:03:25 i have this running @work for the guest wifi. But you are right, i should document this on the wiki Nov 26 18:03:41 blocktrron: driver supports it but apparently not on all devices Nov 26 18:03:45 Oh, and thanks for the set_default bugfix, i idn't catch that. Nov 26 18:03:49 blocktrron: np Nov 26 18:04:07 Hmm, we have UniFi AC lites there Nov 26 18:04:22 hadn't done any updates for almost 3 months, then saw the out of range spam and debugged until fixed :) Nov 26 18:04:48 i was under the impression variables should just be empty in that case Nov 26 18:04:51 my DAP-2695-A1 and Unifi AP AC support it, the TP-Link EAP245v3 apparently not Nov 26 18:04:56 Ah Nov 26 18:05:06 blocktrron: yeah but empty and -eq -gt -ge -> out of range Nov 26 18:05:17 EAP245 is QCA9886, using a different firmware Nov 26 18:05:41 yeah, maybe an el-cheapo radio that doesn't support it Nov 26 18:05:41 i see the problem Nov 26 18:05:47 will take better care next time Nov 26 18:06:02 Have you tried both firmwares? Nov 26 18:06:16 I didn't Nov 26 18:06:38 svanheule[m]: did you try ath10k stock firmware also? Nov 26 18:06:49 dangole: hi, thanks! Nov 26 18:15:01 mangix: https://gist.github.com/stintel/d64298b03192411212395aa40f070bc9 Nov 26 18:15:24 reverting your bump fixes it Nov 26 18:15:56 yness: @champtar is right, we will need hauke's patch. i just read up on it, switch_root involves free'ing the initial ramfs. Nov 26 18:16:11 ynezz: ^^ Nov 26 18:17:06 ah, apparently there already is a fix for it Nov 26 18:17:08 https://github.com/openwrt/packages/pull/14030 Nov 26 18:20:37 stintel: can you please respond to https://github.com/openwrt/packages/pull/14028/files#r530699883 ? Nov 26 18:22:08 How does the linux kernel override U-Boot "bad" bootargs? I see ATAG mangle stuff but that's not what U-Boot passes to the kernel, right? Nov 26 18:22:19 philipp64: https://github.com/openwrt/packages/pull/14028#issuecomment-734201851 Nov 26 18:25:58 stintel: i will look into it. Nov 26 18:26:10 mangix: nvm, I merged the PR already, it's an upstream fix Nov 26 18:29:37 cool Nov 26 18:30:09 dedeckeh: you around? Nov 26 18:33:15 dangole: is there a better fix for the kmod-nft-reject dependency issue? Nov 26 18:39:33 build #530 of ramips/mt7621 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/ramips%2Fmt7621/builds/530 Nov 26 18:42:00 blocktrron: EAP245v1 is QCA9880, EAP245v3 is QCA9982 Nov 26 18:42:33 EAP225v3/EAP225-Outdoor/EAP225-InWall are QCA9886 IIRC Nov 26 18:44:12 stintel: I think I did, when we were discussing this with greearb Nov 26 18:46:07 hmm, maybe that was for another issue https://gist.github.com/svanheule/33f43e2495581db7910e43682649f0b8#file-iw-phy0-info-patch Nov 26 18:49:52 no, `iw phy0 info` should show airtime fairness, but it didn't. Not on ath10k, not on ath10k-ct Nov 26 18:50:45 it still doesn't, although there appear to be other new features Nov 26 19:38:42 dangole: which patch? Nov 26 19:41:39 Hauke: https://patchwork.ozlabs.org/project/openwrt/patch/9231D502B07C5E4A8B32D5115C9F19991F9A9D55@IRSMSX108.ger.corp.intel.com/ Nov 26 20:05:20 Is arc_arc700 currently building on master? https://github.com/openwrt/packages/pull/14028/checks?check_run_id=1460427485 Nov 26 20:18:18 ynezz, Hauke: just tried a local malta/mips32be build and that went fine with ujail with your patch from a year ago. you reckon we should include that by default? Nov 26 20:31:06 blocktrron: I chatted with tohojo on the other channel, since I also have issues with per-BSS-limited airtime fairness. Case in point: https://paste.debian.net/1174469/ Nov 26 20:31:58 Basically, all stations display an airtime weight of 256, regardless of the BSS. Nov 26 20:32:30 In my case, I have mt76 hardware, which should be well supported, I think. Nov 26 20:58:41 philipp64: uClibc-ng issue. I have a fix for it but nobody is merging. Nov 26 20:59:44 link? Nov 26 21:00:11 https://patchwork.ozlabs.org/project/openwrt/patch/20201029005257.101772-1-rosenp@gmail.com/ Nov 26 21:13:56 rsalvaterra: hehe hmhm Nov 26 21:15:33 mangix: Yes? :P Nov 26 21:16:54 mt76 kind of sort of depends on having a good EEPROM. Nov 26 21:17:51 I've had two mt76 pcie cards that didn't work Nov 26 21:18:05 And… how do I know I have a good EEPROM? :P Nov 26 21:18:12 one because of bad eeprom and the other because of power issues Nov 26 21:18:41 besides manual inspection I have no idea Nov 26 21:18:57 I think mt76 exposes some info via debugfs Nov 26 21:19:03 Oh, I have no power issues, that's for sure… ;) Nov 26 21:19:29 26 dBm at 5 GHz, 27 dBm at 2.4 GHz. Nov 26 21:19:31 by power issues I mean it needed another voltage line. Nov 26 21:20:10 Ah, power *delivery* issues, you mean. That's fine too, fortunately. Nov 26 21:21:05 yeah. something about pcie not having a 5V line or something Nov 26 21:21:21 or only specific ones do Nov 26 21:21:27 There are some insane MT7615 cards on AliExpress, which are both oversized and require a totally non-standard extra +5 V in a reserved PCIe pin(!!). Nov 26 21:21:37 Exactly!! Nov 26 21:21:42 yeah I think that was it Nov 26 21:21:54 in my case, it was a full sized mt7612e Nov 26 21:22:08 I *almost* got one, until I saw the schematic and cried. Nov 26 21:22:22 what's wrong with the schematic? Nov 26 21:23:11 I only noticed it was non-standard when I looked at the schematic the vendor provided. Nov 26 21:23:57 But these aren't discrete cards, it's a Redmi AC2100 router. ;) Nov 26 21:24:30 hmm since it's a name brand it should have a proper EEPROM Nov 26 21:25:04 I'd be happy if it had a proper NAND. I got four units, one of them has a dead NAND. Nov 26 21:25:27 NOR master race Nov 26 21:25:40 (Coincidence or not, the only unit with a ESMT NAND… the others are Toshiba.) Nov 26 21:26:15 Yeah, NOR for the win! NAND is just cheap. :P Nov 26 21:26:33 I remember the NAND driver being a disaster. Then conveniently, a mediatek developer wrote a new driver for it. Nov 26 21:27:15 which immediately got rejected by upstream Nov 26 21:29:05 mak Nov 26 21:29:50 (keyboard splat, pay no mind) Nov 26 21:47:43 mangix: LGTM Nov 26 21:47:50 not sure what the delay is. Nov 26 22:12:05 mangix: can libiconv move to packages.git? Nov 26 22:14:13 mangix: nevermind Nov 26 22:14:23 lol absolutely not Nov 26 22:15:38 philipp64: the delay is no one caring for it. uClibc-ng is only kept here because of ARC. Once glibc 2.32 gets in the tree, nbd plans to remove uClibc-ng completely. Nov 26 22:16:49 dangole: is it possible to use tmpfs directly in intramfs as default fs? Nov 26 22:18:48 mangix: just didn't see it in the packages folder, so it's not compiled as a installable package but only as a dependency? Nov 26 22:20:52 hmm, there seems to be a very recent issue with lantiq/ xrx200, factory/ sysupgrade images aren't being built (neither locally, nor the official master snapshots) Nov 26 22:22:16 git bisecting points at the kernel bump to 5.4.79 (https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=9100566267755ea642f08c203d03a9b8f13478c9), but I don't see how that one could be responsible - but I bisected it three times now (even more strange, I did patch r15021-b0ecae504b with that kernel bump manually before, without any issues) Nov 26 22:23:03 mangix: meanwhile… Nov 26 22:23:43 mangix: glibc 2.32 is there Nov 26 22:23:53 gdb should not alos support arc Nov 26 22:24:04 aparcar[m]: libiconv-full is installed as a package. libiconv is only an InstallDev library AFAIK Nov 26 22:24:24 mangix: do you have any ARC hardware? Nov 26 22:24:28 I do not Nov 26 22:24:43 does anyone other than those snps guys who asked for it to be accepted and then left? Nov 26 22:25:04 I do not think so. All of the OpenWrt ARC boarc are dev boards. Nov 26 22:25:20 I think quantenna uses Linux on ARC Nov 26 22:25:22 *ARC boards Nov 26 22:25:35 given how unhelpfull the snps guys were post merge, I'd be in favour of simply deleting the target. Nov 26 22:25:42 see if they decide to be more helpful Nov 26 22:25:54 yes the stuff which is supported are probably mostly FPGAs or something emulated in HAPS Nov 26 22:26:14 you can probably also buy this when you bug a ARC core for your SoC ;-) Nov 26 22:26:23 s/bug/buy/ Nov 26 22:29:45 mangix: is there a simple way to reproduce this? https://github.com/aparcar/openwrt/commit/d21d42e882fe62aace381bc460a62ae952d743bb Nov 26 22:30:12 talking about quantenna, quite sad that there's now mainline driver support for their topaz line (QT3840BC), but no firmware anywhere - leaving it just as unusable as before Nov 26 22:31:00 is it not possible to extract the FW from some vendor binary? Nov 26 22:31:29 aparcar[m]: make package/glib2/host/compile && make package/libiconv-full/host/compile Nov 26 22:31:30 from all I know, the firmwares used by the previous vendor driver and qtnfmac are incompatible Nov 26 22:31:50 too bad Nov 26 22:31:55 aparcar[m]: erm, add make package/libiconv-full/host/clean in between Nov 26 22:34:01 Hauke: no from what I understood. The documentation says to do what you did, ie. let PID 1 handle ramfs (which is just uncommitted fscache from what I understood) and use switch_root when you're done Nov 26 22:34:49 dangole: will the data for the initramfs be freed up? Nov 26 22:35:51 Hauke: yes, after switch_root it's gone (that's what the manpage says at least) Nov 26 22:38:37 adrianschmutzler: if you ever feel like let's continue with the metadata stuff, I had a nice talk with Thomas and he's interested as well Nov 26 22:40:26 aparcar[m]: btw for that patch you will need to run ldd on the resulting binaries to see what I mean. Nov 26 22:40:49 mangix: It currently doesn't compile at all Nov 26 22:40:53 I'll see whats happening Nov 26 22:44:11 dangole: ok then it should be ok Nov 26 22:44:45 aparcar[m]: just realized that patch is for a different package... Nov 26 22:45:26 to test, make package/glib2/host/compile && make package/gettext-full/clean && make package/gettext-full/compile Nov 26 22:45:40 then run ldd on the resulting binaries to see failure Nov 26 22:46:34 it's going to take forever to compile since gettext-full is a pile of **** Nov 26 22:46:40 rsalvaterra: merged your sstrip patch Nov 26 22:48:06 mangix: thanks looking at it right now Nov 26 22:49:22 ynezz: do you have any feelings about 18.06.x in the firmware selector? As JSON isn't backported it means some awkward hack around. cc mwarning Nov 26 22:51:28 aparcar[m]: wasn't 18.06 supposed to be archived and no longer maintained after 18.06.9 anyways? Nov 26 22:51:54 pkgadd: yes that's kinda my main point against it, but there seem to be some devices only available in 18? Nov 26 22:52:39 aparcar[m]: most of those 4/32 devices don't really work with 18.06.x anyways (not without stripping down the default package set) Nov 26 22:53:39 a user who can build from source (or use the imagebuilder) shouldn't have any issues locating the firmware files Nov 26 22:53:57 amen Nov 26 22:54:55 there are also devices where 17.01 is the last supported version Nov 26 23:05:30 mangix: I compiled iconv but it still shows 1.11 and doesn't work as expcted Nov 26 23:05:31 iconv (GNU libiconv 1.11) Nov 26 23:05:40 the package is however called 1.16 Nov 26 23:06:31 fcb42af695 (HEAD -> staging) libiconv-full: Makefile polishing Nov 26 23:06:31 8e6213b2b0 libiconv-full: update to version 1.16 Nov 26 23:06:32 these 2 are added Nov 26 23:08:10 nevermind it's just the lib Nov 26 23:09:56 ok merged Nov 26 23:43:08 aparcar[m]: \o/ Nov 26 23:54:20 philipp64: btw, "local" is not posix compliant Nov 26 23:55:08 okay… so… should I be adding them or taking them away? was trying to make everything consistent. Nov 26 23:57:00 we've got local though, so ignore shell check not being tuned for our environment Nov 26 23:57:26 that's a shell check failure, 100%, we don't need, nor, IMO, should we strive for strict 100% posix and only posix. Nov 26 23:57:32 (what year posix anyway?! ;) Nov 26 23:59:50 I tend to write posix Nov 27 00:17:44 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_x86.html has been updated. (100.0% images and 98.0% packages reproducible in our current test framework.) Nov 27 00:45:46 hi all… the x.509 changes to strongswan are coming along and I have a connection up, according to swanctl. but… I’m not sure I’ve configured the default zone or the firewall rules appropriately to test this correct. can someone look at this with me? thanks Nov 27 00:55:44 * pkg_hash_fetch_best_installation_candidate: Packages for libftdi found, but incompatible with the architectures configured Nov 27 00:55:45 * opkg_install_cmd: Cannot install package libftdi. Nov 27 00:55:47 wtf Nov 27 00:56:06 * stintel sighs and goes to be Nov 27 01:18:42 build #102 of rtl838x/generic is complete: Failure [failed checkarch] Build details are at http://buildbot.openwrt.org/master/images/builders/rtl838x%2Fgeneric/builds/102 blamelist: Rosen Penev , John Crispin , Rui Salvaterra , John Audia , Kevin Darbyshire-Bryant @darbyshire-bryant.me.uk>, Sander Vanheule , Josef Schlehofer Nov 27 01:29:48 great. apu2 image doesn't boot anymore and apparently I don't have the required cable to use its serial console Nov 27 01:29:51 just fucking great Nov 27 01:44:37 i start to regret that this fast-forward switch to wolfssl happened: transmission-daemon segfaults weirdly with it's cyassl crypto (from 2005), while mbedtls and openssl still work great. and thanks to libcurl there is no easy way to switch ssl library. Nov 27 01:45:17 imho libcurl should have goddamn variants for each crypto library. leave alone gnutls, for that we got gnurl alredy Nov 27 01:46:25 https://github.com/transmission/transmission/issues/1537 Nov 27 01:59:38 wonderful, console just hangs, hitting enter inserts line breaks on the console but not getting any prompt Nov 27 02:00:07 seriously folks, we all need to start testing more Nov 27 02:00:17 master != let's break the fucking shit out of it every fucking day Nov 27 02:23:47 especially since bisecting is near impossible Nov 27 02:24:00 Collected errors: Nov 27 02:24:00 * satisfy_dependencies_for: Cannot satisfy the following dependencies for lldpd: Nov 27 02:24:00 * libnl-tiny1 Nov 27 02:24:00 * opkg_install_cmd: Cannot install package lldpd. Nov 27 02:24:00 * satisfy_dependencies_for: Cannot satisfy the following dependencies for luci: Nov 27 02:24:00 * libnl-tiny1 Nov 27 02:24:01 * opkg_install_cmd: Cannot install package luci. **** ENDING LOGGING AT Fri Nov 27 02:59:56 2020