**** BEGIN LOGGING AT Mon Dec 14 03:00:00 2020 Dec 14 03:18:38 mangix: okay, will do. this is the change i was referring to: https://github.com/openwrt/packages/commit/9b4a8f12a3e84de3edfd2605d8f572c63af3746e it just moved around the CONFIGURE_ARGS Dec 14 05:20:05 rsalvaterra: your kernel lzo patch should wait until 5.11 so zstd is actually in use Dec 14 05:20:12 if not we'd just waste 6kb on every device Dec 14 05:22:17 the master branch is fine, I bricked my own build, it took me 8 hours to debug Dec 14 05:33:31 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_tegra.html has been updated. (0% images and 97.1% packages reproducible in our current test framework.) Dec 14 08:22:57 hm, why was the ABI bump needed for `libubox: utils: introduce mkdir_p` ? Dec 14 08:25:02 ynezz: I don't know the criteria for ABI bumps, but maybe because a new function was introduced? Dec 14 08:28:38 new function shouldn't case any issue Dec 14 08:28:45 s/case/cause/ Dec 14 08:30:21 Yeah, removing functions is usually where ABIs are bumped. Dec 14 09:28:25 nice, `opkg find luci-theme*` -> Out of memory: Killed process 4921 (opkg) total-vm:52416kB, anon-rss:51468kB, file-rss:0kB, shmem-rss:0kB, UID:0 pgtables:140kB oom_score_adj:0 Dec 14 09:35:45 ynezz: LOL. What machine is that? Dec 14 09:36:19 x86/64 with 128M of RAM Dec 14 09:36:38 Hm. No swap? Zram? Dec 14 09:36:45 why? Dec 14 09:37:10 128M should be plenty for OpenWrt Dec 14 09:37:34 Sure, but you weren't just "running OpenWrt". ;) Dec 14 09:38:02 opkg find is probably a memory hog. Dec 14 09:43:21 jow: BTW why do you use the GalanoGrotesqueW00-Regular.woff2 font for the theme? I think, that this font is only mandatory for the logo, not for the branding Dec 14 09:46:06 jow: it would be nice to use this openwrt-2020 theme as default in 20.12 release Dec 14 09:53:15 * karlp decides he'd better at least try it out :) Dec 14 09:53:51 jow: BTW I'm using Firefox Nighlty 85.0a1 (2020-12-02) (64-bit) and am unable to enter the LuCI admin interface http://ynezz.true.cz/openwrt/luci/luci.har Dec 14 09:54:01 jow: It works fine in Chromium :] Dec 14 09:54:35 20.12? Is that official? ;) Dec 14 09:54:35 jow: I mean, default firstboot install, without password, clicking on Login button does nothing Dec 14 09:55:09 (Maybe we should start making yearly releases on Christmas… :P) Dec 14 09:57:33 there used to be "Go to password config" button, it's missing as well Dec 14 09:57:55 ah, the code is still there `` Dec 14 09:58:27 probably `<% if disp.lookup("admin/system/admin") then %>` doesn't work anymore Dec 14 10:24:09 ynezz: that not working login happens if firefox still has a sysauth cookie cached Dec 14 10:24:15 try deleting it Dec 14 10:41:05 jow: thanks, that fixed it Dec 14 10:41:18 its some Browser heuristics Dec 14 10:41:33 it remembers that the old sysauth cookie came from https Dec 14 10:41:44 so the http-only login on the new isntall is not allowed to overwrite it Dec 14 10:42:02 I was tyoing with the idea to use differently named cookies for http and https, thne accept both Dec 14 10:42:20 but not fully thought through if it'd weaken security Dec 14 10:44:16 browsers are weird operating systems Dec 14 13:51:22 adrianschmutzler: I think we should continue updating the minor kernel version in master to release with a recent minor version Dec 14 13:52:23 aparcar[m]: I'm not sure about make 4.1 .. breaking CentOS 7 is something we might want to avoid Dec 14 13:53:18 Hauke: thanks. I wasn't sure what was discussed with respect to that in the meeting. And I didn't really see a reason for not doing minor kernel bumps ... Dec 14 13:53:53 we do minor kernel bumps in stable releases, so it makes no sense imo to stop them for the freeze Dec 14 13:53:59 are the meeting minutes available somewhere guys? Dec 14 13:54:07 Borromini: not yet Dec 14 13:54:20 Hauke: ok. just curious =) Dec 14 13:54:35 stintel: that's exactly what I thought, too. But I wasn't what other kernel bump would have been the subject of that short comment then ... Dec 14 13:55:05 stintel: do we need that make dependency adjustment _now_? Dec 14 13:57:10 adrianschmutzler: the patch is on the ML / patchwork Dec 14 13:57:48 and so far I have seen no objections Dec 14 13:57:59 I would definitely not merge it before branching Dec 14 13:58:12 and even then, it might be too soon to break building on EL7 Dec 14 13:58:14 I've seen the patch, I'm not so sure what the actual reason is Dec 14 13:58:37 and whether that reason to introduce it has more weight than disabling certain systems as you mentioned Dec 14 13:59:27 I think we have a build problem with older make version Dec 14 14:00:03 So, it doesn't work anyway on EL7? And the question is whether to fix that problem or abandon the platform? Dec 14 14:19:10 hostapd seems to have a bad implementation of Proactive / Opportunistic PMKID Caching (same thing) Dec 14 14:20:14 It does have an ability to check if the STA claims capability Dec 14 14:20:47 Will look to patch this and submit upstream Dec 14 14:22:09 *It does NOT have an ability to check if the STA claims capability Dec 14 14:46:41 Hurrah, the 5.10 kernel has landed Dec 14 14:46:58 As in has been released upstream :-) Dec 14 14:47:19 is there anything particularly important about this one? Dec 14 14:47:25 it's the new LTS Dec 14 14:56:51 so the next OpenWrt kernel as well probably :) Dec 14 14:58:25 karlp: I don't get when people ask those kinds of questions. It's a new kernel. It's the most important part of the operating system. :/ Dec 14 14:58:46 https://kernelnewbies.org/Linux_5.10 Dec 14 14:58:50 Jeez… Dec 14 14:58:59 I would disagree. Dec 14 15:00:09 karlp: I would too, but Linux isn't a microkernel (thankfully). Dec 14 15:24:02 is 2020.12 official? 2020 is a really bad year, maybe just 2021.xx something with 5.10 LTS kernel and skip 2020 all together Dec 14 15:25:02 by then the new luci is probably ready to ship, and openwrt-2020 theme can be renamed to something not related to 2020 :) Dec 14 15:25:38 rr123: next release will use 5.4. bumping kernel would cause additional delay Dec 14 15:26:24 5.4 is ok, new luci matters a lot, not sure when it will be the default Dec 14 15:28:35 new theme is ~irrelevant. Dec 14 15:28:49 luci in master has major behind the scenes changfes that are interesting/new/important Dec 14 15:29:06 even on the "omg so boring" current theme. Dec 14 15:30:09 working on remedying the "omg so boring" theme issue, if that's how you categorize it, karlp Dec 14 15:30:24 not the theme, not sure when client-side-luci will replace the lua-server-side luci in the release Dec 14 15:31:01 * rr123 has been using openwrt2020 for a few months and can't go back Dec 14 15:35:49 well, just get used to "no big changes until the release" Dec 14 15:36:46 rr123: feel free to have a look at the new theme I'm working on, based on bootstrap: https://github.com/openwrt/luci/pull/4617 Dec 14 15:37:27 rr123: lots of luci in master is already client side, it's not some single big bang, Dec 14 15:37:34 it's not some switch to "make it the default" Dec 14 15:38:00 When's master likely to fork to a stable branch? Dec 14 15:38:35 (deployment is straightforward, just rename cascade.css in /www/luci-static/bootstrap and upload the latest revision of the cascade.css linked on that pull request) Dec 14 15:39:20 Nick_Lowe: there's been talk of 20.12 here so i reckon branching might happen this month. and hauke talked about not pushing the 5.10 mac80211 bump until branching happens Dec 14 15:41:20 dorf: i saw your diff, how to directly download the new cascade.css Dec 14 15:41:59 ok i will download from your github Dec 14 15:42:19 you'd hit the commit and then follow the links, or -> https://github.com/openwrt/luci/raw/b496350bbcd9892502df6d5e51ef04d1265d5e98/themes/luci-theme-bootstrap/htdocs/luci-static/bootstrap/cascade.css Dec 14 15:42:32 oh, wait.. not that. Dec 14 15:43:47 > https://github.com/vituperative/luci/blob/master/themes/luci-theme-bootstrap/htdocs/luci-static/bootstrap/cascade.css Dec 14 15:57:11 dorf: https://pasteboard.co/JEUbi8C.png tested on pc chrome looks fine, except all fonts appears to be bold, this img showed Edit and Delete button are 'out of space' on android phone Dec 14 15:58:27 https://pasteboard.co/JEUc0x4.png this one has ? info not aligned up on the phone Dec 14 16:00:36 https://pasteboard.co/JEUczKI.png this one, the letter in the select box(e.g. "accept") was not fully visible, the bottom portion disappeared Dec 14 16:02:57 dorf: yeah, for the fonts to look good, you need to install a multi-weight font on the system viewing LuCI. Dec 14 16:03:15 open sans recommended if your OS doesn't provide one by default. Dec 14 16:04:29 phone-wise, I probably need to add some media queries to improve that. Dec 14 16:05:23 and clipped text, hmm.. not seen that, thanks for pointing it out. probably that issue goes away with open sans. Dec 14 16:06:07 or maybe that's another issue to be addressed with a phone-specific media query. Dec 14 16:06:36 Did someone mention me yesterday? I saw a highlight, but my scrollback is not long enough Dec 14 16:07:03 neoraider: ping Dec 14 16:07:07 Anyone know why tcpdump-full (x86-64) breaks on master? Dec 14 16:07:09 checking for pcap_loop... no Dec 14 16:07:11 configure: error: This is a bug, please follow the guidelines in CONTRIBUTING and [...] Dec 14 16:07:21 dorf: tried midori,firefox,brave,chrome, all worked well enough, nice job. Dec 14 16:07:52 thanks, rr123 :) work in progress.. at some point I'd like to get just a little more color in there. Dec 14 16:08:17 two feedbacks are: 1. the font all bold a little tiring 2. the font bottom cut off inside select-box. I can use chrome devtool and select screen size to see that without using a phone Dec 14 16:09:44 yup, thanks, got that. try with Open Sans if you can :) Dec 14 16:10:06 i was playing with browser in the last two days trying to find a lightweight one on memory usage so I can run a browser inside a browser in the cloud, turns out all of them, including the little midori, will pretty much ask for 1GB with one or two tabs Dec 14 16:10:36 on a debian based distro, you should be able to snag Open Sans with the package: fonts-open-sans Dec 14 16:10:58 midori is yesterday! falkon is the new midori. Dec 14 16:11:31 as long as it uses chromium or webit or whatever backend, it's memory hungry Dec 14 16:11:37 yeah, sadly. Dec 14 16:11:49 the only lightweight is w3m/lynx Dec 14 16:11:57 I'd hesitate to recommend dillo. Dec 14 16:12:06 hunting a regression between 18->19, any hints where a wifi device "ifname" is set? is that in netifd or somewhere in a script? Dec 14 16:12:08 but it's a bit more capable than lynx, at least. Dec 14 16:12:29 true, have not used that for a while, thanks for bring it up, i essentially forgot that one Dec 14 16:13:01 you can guarantee it'll make a dog's dinner of anything css3 :) Dec 14 16:13:23 falkon is qt's web engine, i will assume it is as memory hungry as others Dec 14 16:13:55 actually, not so much. Dec 14 16:14:14 6 open tabs, under 400M usage here. Dec 14 16:14:31 ymmv Dec 14 16:16:13 dorf: no way, I have one tab here it's 1GB, you need check both falon and all its QtWeb* processes, the latter is the memory eater, the backend engine Dec 14 16:16:26 s/falon/falkon/ Dec 14 16:17:03 ah, you're right. my bad. still, not as bad as firefox/chrome. Dec 14 16:20:16 rr123: no open sans in your software repo? https://fonts.google.com/specimen/Open+Sans .. you can download the entire family. Dec 14 16:25:34 dorf: not a frontend guy, how to use this with bootstrap them then? copy them all to $HOME/.fonts? Dec 14 16:25:56 no Dec 14 16:26:10 by the way just tested dillo, can't open many sites, netsurf will crash for any http->https auto-switch site, e.g. http://yahoo.com will crash it Dec 14 16:26:10 install them on the device your viewing openwrt from. Dec 14 16:26:30 then ctrl+shift+r in a sane browser (not dillo!) Dec 14 16:26:44 dorf: which is, by default, my PC Dec 14 16:27:12 sorry, yeah, copy them to /usr/share/fonts/ if you want them globally accessible. Dec 14 16:27:22 or whatever location is for user-specific fonts. Dec 14 16:28:15 then sudo fc-cache -f to refresh the font cache. Dec 14 16:29:05 yeah just did that Dec 14 16:30:10 refresh luci, nothing changed, how to switch fonts, or is Open sans in the preferred font list in the css already Dec 14 16:30:15 it's not picked up somehow Dec 14 16:30:26 Open Sans is the default font. Dec 14 16:31:00 you might need to restart your browser? Dec 14 16:31:21 (available fonts may be scanned at initialization) Dec 14 16:37:24 dorf: used chrome whatfont extension to check, I'm still using ubuntu fonts after restarted chrome Dec 14 16:37:52 a quick comparison between opewnrt vs openwrt.2020 I would say the latter does better on screen size and colors so far Dec 14 16:39:18 rr123: weird. if you're on ubuntu, sudo apt install fonts-open-sans is the best method to get those fonts installed, followed by a sudo fc-cache -f Dec 14 16:40:05 in theory the font names should be the same from google fonts and repo install, I'll take a look at that. Dec 14 16:40:36 also, if you're viewing with the Ubuntu font, that should also remedy the bold font issue.. ubuntu sans is multi-weight, which is why it's supported in the css. Dec 14 16:51:05 have a good evening guys!!!!! :) Dec 14 17:58:30 Just noticed... https://openwrt.org/docs/guide-user/network/wifi/basic does not seem to be indexed by Google Dec 14 17:58:59 That's not very helpful for people searching for wifi config options to understand how to use them Dec 14 18:00:29 Who 'owns' this area? Dec 14 18:02:09 https://developers.google.com/search/docs/advanced/crawling/ask-google-to-recrawl Dec 14 18:02:55 It could be worth somebody following this, which requires an ability to demonstrate domain control/ownership Dec 14 18:25:26 If anyone is bored and feels like giving me a hand... https://forum.openwrt.org/t/collecting-build-triple-information-community-help-requested/82062 Dec 14 18:42:46 jow: why luci has a "Network" on the left column at login page? openwrt theme does not have that Dec 14 20:17:10 Hauke: Will it be possible to applied this patch series https://lore.kernel.org/linux-wireless/20201214172118.18100-1-jouni@codeaurora.org/T/#r9d45c97d87353beef1de1e22d630de6a8beb2e3e to your's mac80211 backports? Dec 14 20:27:26 Pepe: you can sedn a patch to add this to openwrt Dec 14 20:27:44 Sure, will do it. Dec 14 21:28:54 rr123: if it's no tin the normal theme, ask the them author? Dec 14 21:49:32 Risk64: ping Dec 14 21:50:10 mangix: ping back Dec 14 21:50:41 you have issues with tcpdump? Dec 14 21:51:03 yes, i've done a distclean now... Dec 14 21:51:08 ultima ratio Dec 14 21:51:47 it has something to do with musl/pcap whatever... no one seems to know the exact cause. Dec 14 21:52:35 I have patches for libpcap and tcpdump. please test: https://github.com/neheb/openwrt/commits/mangix Dec 14 21:53:58 I can't test as I don't encounter this issue. Dec 14 21:54:15 Previous person was using arch linux. Dec 14 22:00:44 can anyone tell me what the "list addresses" entry is for in the wireguard [interface] config? docs say it is not mandatory but my vpn will not start without it. is it an analog for the "address" option in wireguard wg-quick config? interestingly a /32 mask also broke my vpn, whereas /16 and /24 did not. how is this used? Dec 14 22:01:29 Risk64: you check them out? Dec 14 22:01:56 mangix: If i encounter the bug again, yes. Dec 14 22:02:30 When are these commits go upstream? Dec 14 22:02:52 There are some more interesting ones... like musl 1.2.1. util-linux 2.36 Dec 14 22:04:24 Risk64: musl is being held until after the release. it works fine for me. util linux I need to send I think Dec 14 22:04:40 Is there a review system for merge requests for OpenWrt? Dec 14 22:04:48 patchwork Dec 14 22:05:35 http://patchwork.ozlabs.org/project/openwrt/list/ Dec 14 22:19:43 mangix: distclean is still broken. I will inject the 2 commits for libpcap and tcpdump. Dec 14 22:46:50 mangix: The patches work.. compiles properly now. Thx! Dec 14 22:53:15 Risk64: what distro are you compiling on? Dec 14 22:53:21 arch Dec 14 22:53:29 thought so Dec 14 22:54:01 Risk64: can you test gnutls to see if it compiles? Dec 14 22:54:27 compilation just passed it Dec 14 22:54:50 This one: make[3] -C feeds/packages/libs/gnutls compile Dec 14 22:55:02 ok. someone else was complaining about gnutls Dec 14 22:55:22 jow: ping Dec 14 23:00:08 mangix: pong. libnetfilter-conntrack ? Dec 14 23:08:07 jow: ) Dec 14 23:08:10 :) Dec 14 23:19:50 mangix: pushed, thanks for pinging Dec 15 00:07:37 How do I debug procd when it crashes at bootup? Dec 15 00:11:32 Hauke: btw. your mac80211 updates are working fine for me on ipq806x (nbg6817), ipq40x (map-ac2200) and lantiq (bthub5), and build-tested for ath79 (haven't flashed those yet, as those devices aren't in regular use); so ath9k and ath10k appear to be fine Dec 15 00:13:12 pkgadd: thanks for testing, I am using them currently with mt76 Dec 15 00:17:20 Hauke: at this stage is it crashing? no output at all after /sbin/init is executed? Dec 15 00:22:04 I am using the address sanitizer in my system and it does not work correctly for the init process Dec 15 00:22:10 just found the log Dec 15 00:22:21 with some other qemu settings I get it Dec 15 00:36:01 Hauke: interesting... might be interesting to just run procd in a container and have valgrind crunching around it Dec 15 00:37:36 dangole: I will just try to deactivate it for the init process Dec 15 00:37:57 it wants to access files in /proc and fails Dec 15 00:38:26 Hauke: and /proc isn't yet mounted? Dec 15 00:39:20 dangole: no procd mounts it Dec 15 00:39:41 or /sbin/init Dec 15 00:39:45 this is currently failing Dec 15 00:46:12 No VFP support on a SoC means. Softfloat? Dec 15 00:46:37 specifically armv7 if that makes a difference Dec 15 00:55:40 Grommish: https://developer.arm.com/architectures/instruction-sets/floating-point Dec 15 01:01:48 pkgadd: Appreciated :) So, yeah softfloat it seems. The tech specs on devices don't specify, so I've got to go digging I guess **** ENDING LOGGING AT Tue Dec 15 02:59:57 2020