**** BEGIN LOGGING AT Thu Mar 11 02:59:57 2021 Mar 11 04:57:06 nbd: hope the massive netfilter patch series goes in smoothly! https://patchwork.kernel.org/project/netdevbpf/list/?series=445827 Mar 11 08:34:15 stintel: What do you mean, doesn't work on DSA interfaces? I never had any issues… :/ Mar 11 09:00:33 rsalvaterra: tcpdump on e.g. lan3 shows only mcast/bcast frames Mar 11 09:01:06 stintel: Hm… weird. What router is that? Mar 11 09:01:27 rsalvaterra: Netgear GS108T v3 Mar 11 09:02:01 Ah, Realtek switch. Mar 11 09:03:23 I was going to say I never had any problems with the Marvell 88E6176 on the Omnia (or the MT7530 on the RM2100) Mar 11 09:05:31 Probably some issue with the switch driver? Mar 11 09:07:49 Pet-peeve #345345: FFS, why do they put the Ethernet and DC connectors on *opposite* sides of desktop (not rack-mount) switches?! Mar 11 09:07:57 stintel: having issues with your gs108t v3? Mar 11 09:10:53 Borromini: dunno if tcpdump on lanX is supposed to show all packets on that port? Mar 11 09:11:29 no idea :( Mar 11 09:12:43 maybe ask in #rtl83xx ? Mar 11 09:12:49 yeah probably \ Mar 11 09:12:52 later Mar 11 09:12:53 work :( Mar 11 09:42:52 getting a bunch of No related RPC reply in luci lately... Mar 11 09:45:25 known issue by any chance? Mar 11 09:50:25 nyt: using ssl? Mar 11 09:58:32 yes Mar 11 09:58:53 it automatically redirects to https Mar 11 09:58:58 i just have the defaut luci selected Mar 11 09:59:30 RPCError: RPC call to uci/get failed with ubus code 4: Resource not found Mar 11 09:59:30 at handleCallReply (https://rooter/luci-static/resources/rpc.js?v=git-21.060.51374-cd06e70:15:3) Mar 11 10:00:41 and the menu doesnt load right after login Mar 11 10:00:43 have to refresh to get it to Mar 11 10:01:15 just went to it by ip instead of name for first time in a while and saw this also... Mar 11 10:01:16 HTTP error 400 while loading class file "/luci-static/resources/view/dashboard/index.js?v=git-21.060.51374-cd06e70" Mar 11 10:01:16 at compileClass (https://192.168.1.1/luci-static/resources/luci.js?v=git-21.060.51374-cd06e70:170:16) Mar 11 10:02:38 I think this is caused by ustream-ssl-wolfssl Mar 11 10:02:47 ever since it was made default, luci was not functioning properly Mar 11 10:03:01 there's been no related changes to luci or uhttpd in months Mar 11 10:03:17 weird, since that doesnt seem to be an ssl related error... Mar 11 10:03:23 what should I switch it out to? Mar 11 10:03:43 try changing to ustream-ssl-openssl if you have the sapce, see if it improves it Mar 11 10:03:58 x86, so have all the space in the world Mar 11 10:04:30 afair, the errors above are caused by the ustream-ssl-wolfssl layer mixing up data, or someow returning data in such a way that uhttpd's state machine gets confused Mar 11 10:04:55 that leads to all sorts of issues... timeouts, random 400 errors, even on static ressources etc. Mar 11 10:05:09 BTW, any improvement over using nginx as webserver? Or is it more of "no one is caring so no coding is done" I'm also fine with any answer, craving for info to determine future actions =) Mar 11 10:05:14 luci-ssl-nginx or luci-ssl-openssl ? Mar 11 10:05:22 nyt: luci-ssl-openssl Mar 11 10:05:51 [nyt@cc:~/archive/openwrt/x86]: make -j30 Mar 11 10:05:53 (heh, timing of our questions 😀 ) Mar 11 10:06:27 olmari: I personally don't care for nginx and I keep receiving bug reports about it handling things differently in subtle ways Mar 11 10:06:45 so it likely has hidden issues with LuCI as well Mar 11 10:07:11 mm Mar 11 10:07:20 make -r world: build failed. Mar 11 10:07:21 ew Mar 11 10:07:42 that failed after package/install Mar 11 10:07:55 nyt: might be that you need to manually deselect conflicitng stuff Mar 11 10:07:56 jow: howdy, wondering if you had time to look at the fw3 dscp/mark fixes I posted? Mar 11 10:08:05 guidosarducci: briefly, looks okay Mar 11 10:08:23 jow: Would have made one of mine bigger router nice lil RP and webserver for things, but we'll manage other ways :) Mar 11 10:08:46 I can fully see your point in openwrt context in general tho Mar 11 10:09:24 https://pastebin.com/raw/ac727yXt Mar 11 10:09:25 yeah some conflicts Mar 11 10:09:29 jow: is that something you could squash/merge from my branch, or prefer sending to mailing list? Mar 11 10:09:30 luci-ssl-openssl and lubistream-openssl Mar 11 10:09:45 err libustream-mbedtls20201210 Mar 11 10:10:17 guidosarducci: yes I can take it from your branch Mar 11 10:10:29 cshark select libustream-mbedtls :/ Mar 11 10:11:18 so seems PACKAGE_libustream-mbedtls and PACKAGE_libustream-openssl conflict Mar 11 10:11:30 jow: great, appreciate that, and thanks for your help getting it going. Mar 11 10:16:37 whats the dif between libustream-mbedtls and libustream-openssl? Mar 11 10:16:55 oh one is mbedtls and one is openssl Mar 11 10:17:11 - DEPENDS:=+libjson-c +libpcap +libuci +libubox +libuclient +libustream-mbedtls Mar 11 10:17:11 + DEPENDS:=+libjson-c +libpcap +libuci +libubox +libuclient +libustream-openssl Mar 11 10:17:13 seeinf if this helps any Mar 11 10:20:19 jow that seems to have fixed it Mar 11 10:21:23 broke cshark though lol =[ Mar 11 10:22:51 deselecting that shit though, since it's pay only now Mar 11 10:36:34 so that's interesting ... Mar 11 10:36:38 /usr/lib/lua/luci/dispatcher.lua:725: No valid theme found Mar 11 10:36:52 deselected the themes since i saw luci-theme-bootstrap was selected Mar 11 10:36:58 and i twas the default Mar 11 10:37:03 apparently, it has no joy Mar 11 10:38:52 jow, which theme will be the default moving forward? Mar 11 10:38:59 2020 or regular? Mar 11 10:45:09 hm looks like mediaurlbase was clobbered by openwrt2020 when i had that in there Mar 11 10:45:42 getting those same errors again Mar 11 10:57:50 jow: i didn't follow discussion on DSA that was mostly between you and nbd_ i think Mar 11 10:57:57 jow: but still let me ask 1 question Mar 11 10:58:06 jow: i see this syntax: option ports 'lan2=pt lan3' Mar 11 10:59:13 shouldn't that be list maybe? Mar 11 10:59:21 list ports lan2=pt Mar 11 10:59:24 list ports lan3 Mar 11 11:04:41 NetworkError: HTTP error 400 while loading class file "/luci-static/resources/fs.js?v=git-21.060.51374-cd06e70" Mar 11 11:05:04 works fine loading directly in browser... Mar 11 11:06:14 https://i.imgur.com/q5s2Ool.png loading the site for the first time Mar 11 11:09:11 and i have uhttp-mod-ubus installed Mar 11 11:44:21 Heh… just did a time make -j 3 after a dirclean, for the lulz… real 153m0,236s. Mar 11 12:29:46 jow: same errors without ssl btw, and with ustream-openssl Mar 11 12:43:07 nyt: okay, interesting Mar 11 12:53:21 ubus stuff is commented in uhttpd Mar 11 12:53:26 looks like uncommenting makes it a bit happier Mar 11 12:53:47 ubus_prefix and ubus_socket Mar 11 12:55:32 that seems to have been the issue Mar 11 13:03:46 jow: speaking about luci, what about mine HTTPS improvements https://github.com/openwrt/luci/pull/4659 and https://github.com/openwrt/luci/pull/4660 ? Mar 11 13:04:17 would be nice to get it into 21.02 Mar 11 13:12:26 have you figured out how to get me a cert that is respected yet? Mar 11 13:12:35 otherwise I'm certainlyagainst this going into 21.02, Mar 11 13:12:50 I thought the pint was thebranch so you go and do this sort of hacking in master, instead of backdooring it back into stable Mar 11 13:22:31 uh? Mar 11 13:24:28 It's about SSL opt-in, making it more visible to the users Mar 11 13:25:18 yeah, but we're still at this, "ok, you get ssl, just accept this self signed cert warning, hope you aren't being mitm'd" right? Mar 11 13:25:38 ssl is disabled by default, after first login you'll get a notice, that "HTTPS is available, make router management more secure with HTTPS. Please read https://openwrt.org/docs/guide-user/luci/luci.secure first" Mar 11 13:27:11 ynezz: So, uh… this interest in getting LuCI working over HTTPS is because suddenly we stopped trusting the LAN? :P Mar 11 13:27:28 openwrt2020 looks kind of not so great Mar 11 13:27:29 :( Mar 11 13:27:47 but we trust it enought to accepet self signed certs from it.... Mar 11 13:27:52 magic security! Mar 11 13:28:14 pffft, http or gtfo :P Mar 11 13:28:31 Sounds a bit pointless to insist in HTTPS, to be honest… Mar 11 13:28:35 rsalvaterra: yeah, you can use telnet if you want Mar 11 13:28:38 ... if youre being mitm'd on your own network after you just setup and connected to your router, maybe there's an issue Mar 11 13:28:55 ynezz: Only from the WAN, over WireGuard, thank you. XD Mar 11 13:29:04 also i just disabled ssl for testing Mar 11 13:29:18 karlp: this PR is about promotion of optional HTTPS, no intention to fix your self-signed issue Mar 11 13:29:18 was getting weird login issues and 403 without it Mar 11 13:29:39 nyt: that's issue with cookie Mar 11 13:29:41 ah Mar 11 13:29:42 except that by turning it, you're _giveing_ self signed, and thereby encouraging the behaviour of "just accept the warning" Mar 11 13:29:44 TBH, Acme client in owrt is easy enough, and even more so uhttpd/luci-openssl has awesome defaults too :) Mar 11 13:29:52 and I strongly feel that's NOT OK. Mar 11 13:30:14 karlp: you can create/improve that wiki page and make it OK Mar 11 13:30:16 but hey, this is the same coversation as last time. Mar 11 13:30:23 eh, how are you going to actually get a signed cert for your router with no fqdn Mar 11 13:30:25 not gonna happen Mar 11 13:30:37 on organizational stuff you'd obviously have your own CA to sign certs with and company computer policies would include that in cert trust store ;) Mar 11 13:31:04 nyt: you don't need one, you can have your own PKI and trusted CA, so just upload your own cert Mar 11 13:31:30 nobody forces you to use self-signed certificate generated on a router with crappy entropy Mar 11 13:31:35 ynezz: so, seeing as it's the same discussion as las titme, and it's a lost cause,can we at least leave it out of 20-branch? it certainly seems like new functionality, and was highly controversial at the itme, Mar 11 13:31:44 slamming it in now sounds like backdooring it Mar 11 13:32:09 ynezz, and if youre going to go through that trouble, you can just specify the certs? Mar 11 13:32:10 it says Mar 11 13:32:15 i dont see the issue Mar 11 13:32:16 karlp: "ynezz wants to merge 6 commits into openwrt:master" Mar 11 13:32:17 you _can_ upload your own cert, but is that the "click here to turn on https! magic!" is doing? Mar 11 13:32:21 where do you see 21.02? Mar 11 13:32:25 the main point is to have trusted certificate, either by public CA, or by yourself, or as last resort, accept it in browser =) Mar 11 13:32:30 13:04:17 ynezz | would be nice to get it into 21.02 Mar 11 13:32:37 that's what I'm against. Mar 11 13:32:39 would be, yes Mar 11 13:32:55 let it play out in master, fine, I'm never going ot be able to stop progress anyway :) Mar 11 13:35:03 Hey, in alternative we could move LuCI to a 802.1AE secured management VLAN… Mar 11 13:35:06 * rsalvaterra runs Mar 11 13:36:07 well... nothing in https relates to how you would run stuff on local network, nor VLANning methods =) Mar 11 13:38:55 … or, you know… just keeping untrusted devices in an isolated VLAN (or not keeping them at all). Mar 11 13:41:17 mhm, "management VLAN" alone is an good practice, while nothing stops to use non-ssl stuff at local network perioid... all within decisions of network owner Mar 11 13:43:31 no my field but i would argue that ssl by default would make some sense if luci would have a default password set so that it doesnt go into the plain, but luci has no default password set either.. so that argument doesnt fly Mar 11 13:43:37 s/no/not/ Mar 11 13:44:23 olmari: Yeah, but I'm not seeing OpenWrt configured out of the box with a management VLAN. That would scare the s**t out of Joe User. Mar 11 13:45:32 well, you should first configure your device before plugging it into your network Mar 11 13:45:48 so just your device and your laptop Mar 11 13:46:02 does that not imply setting up any ssl then? Mar 11 13:46:27 ynezz: Then why insist on HTTPS out of the box? :) Mar 11 13:46:27 Having PW and being able to sniff data is separate things Mar 11 13:46:45 cool, at least someone understands Mar 11 13:49:20 yes i dont mind eitherway. i guess theres some worthy "data to sniff" Mar 11 13:49:29 root password? Mar 11 13:49:37 but its not set? Mar 11 13:49:45 with ssh password login by default Mar 11 13:49:57 yes but then you manually set up ssl? Mar 11 13:50:06 using ssh Mar 11 13:50:37 anyhow ill drop it, i am fine with it Mar 11 13:50:56 https "breaks" things even with a valid certificate Mar 11 13:51:06 so ce'st la vie Mar 11 14:09:15 i generally admire a pro-active approach, but there could be back lash. try allowing sites with self-signed certificates in some chrome browsers on android phone. you need a PhD for it Mar 11 14:09:43 in my selinux policy i dont try to do things like this (i use to) Mar 11 14:10:28 i basicly assume that device owners know what they are doing Mar 11 14:10:47 have to trust someone ... Mar 11 14:15:05 it's opt-in, so you can just ignore it, but those who get it have it one click away Mar 11 14:15:25 and you know i was going to say but you cannot compare selinux with ssl in openwrt but you actually cal Mar 11 14:15:34 because selinux is opt-in as well Mar 11 14:16:00 so you can argue that if someone has to knowledge to enable selinux then you can expect that someone to know what that might mean Mar 11 14:16:05 well, you need kernel for that, so you can't opkg install selinux Mar 11 14:16:12 and i guess the same applies for ssl since thats not a deffault either Mar 11 14:16:26 thats true Mar 11 14:16:31 but SSL libs are already there, in every image Mar 11 14:16:32 but there are some similarities Mar 11 14:16:36 yes Mar 11 14:17:19 and I can't imagine selinux on most of the targets Mar 11 14:17:35 thats not the point though Mar 11 14:17:36 like for example currently cheap/popular mt7621/ath79 Mar 11 14:17:56 the point is should we be treating our audience like little children? Mar 11 14:18:06 ynezz: There's a lot of pending stuff I can't imagine for ath79 and friends… :P Mar 11 14:21:13 grift: BTW it was me who has made that SSL optional, as it was always on by default in commit 0cf3c5dd7257dff1c87b61c5e53e5b1787ab7015 Mar 11 14:21:17 How would you get a valid certificate for it? Or would you create a self-signed certificate? Mar 11 14:21:28 so actually I don't know why the people are complaining :p Mar 11 14:21:51 Either you trust your network not to do an MITM attack, or you use ssh. Mar 11 14:21:58 i am not complaining for the record (i just find it an interesting topic and i am bored) Mar 11 14:22:02 you mean ssh Mar 11 14:22:05 with a self generated key Mar 11 14:22:06 .... Mar 11 14:22:13 that you have to accept the first time Mar 11 14:22:16 lol Mar 11 14:22:57 yes, that's exactly the same issue as with HTTPS, but people don't care, because SSH clients just make it easy to accept the MITM key Mar 11 14:23:16 who checks the fingerprints, right Mar 11 14:23:36 the thing is that accepting a self signed certificate is not easy with modern web browsers Mar 11 14:23:57 i tried it a while ago on my phone Mar 11 14:25:11 for sane ciphers etc it is mostly easy.. for broken ciphers etc its almost impossible Mar 11 14:27:49 Shamelessly hijacking the topic, what we *really* need is zstd compression support for jffs2. Mar 11 14:28:49 tofu is not the same as how brwosers do it. Mar 11 14:29:19 you don't get the "HOST KEY HAS CHANGED!" warning rom a browser, you just get a "new" "trust unknown?" Mar 11 14:30:17 Right now, zstd is the perfect jack-of-all-trades of compression. We can use it for zram, overlay compression and whatever other compression needs which might arise. And having a single compressor for everything saves space. Mar 11 14:32:24 I have patches in my trees (core and fstools) enabling zstd overlay compression in ubifs, but unfortunately jffs2 doesn't support zstd. Mar 11 14:34:59 dunno what to say rsalvaterra any suggestions? Mar 11 14:36:52 Well, obviously we need zstd compression in jffs2. I started reading the zstd API, in order to do it myself, but I still don't understand it completely (and jffs2 is stuck in 2005, or so, as it doesn't use the crypto API). And there are possible locking issues to solve too… Mar 11 14:38:06 are there no possible replacements for jffs2? (excuse my ignorance, just showing interest) Mar 11 14:38:19 btrfs Mar 11 14:38:21 * stintel hides Mar 11 14:38:23 2005 is a long time Mar 11 14:38:37 grift: Not for small NOR flash (up to 16 MiB, or so). Mar 11 14:39:21 For larger filesystems, ubifs scales better, but has higher (space) overhead. Mar 11 14:39:32 i guess that say's something .... Mar 11 14:39:51 about nor flash upto 16 MiB i mean Mar 11 14:40:28 grift: That's probably the vast majority of the devices supported by OpenWrt. ;) Mar 11 14:41:18 * grift avoids a trap Mar 11 14:41:31 I have three 8 MiB devices (ath79), one 128 MiB NAND device (RM2100) and The Beast which is the Omnia. Mar 11 14:42:11 i guess the status quo it is Mar 11 14:42:33 And I still wonder why the Omnia's eMMC isn't socketed… missed opportunity, I believe. Mar 11 14:44:26 stintel: I won't touch btrfs with a two meter pole. :P Mar 11 14:44:45 until bcachefs goes upstream there's no alternative ;) Mar 11 14:45:05 stintel: I thought I was the only one looking forward to it! :) Mar 11 14:45:19 rsalvaterra: testing bcachefs in production for 3 years Mar 11 14:45:39 stintel: You wrote "testing" and "production" in the same sentence. Mar 11 14:45:47 best testing happens in production Mar 11 14:45:49 prove me wrong :) Mar 11 14:46:10 stintel: I can't. I'm a programmer. :P Mar 11 14:46:31 But it still feels wrong… ;) Mar 11 14:47:52 well it's home/home office production, but it's still production Mar 11 14:48:32 my home automation runs on it, it holds VM images exposed via iSCSI that I'm using for my day job, etc Mar 11 14:49:35 it's been a bumpy road, but still, I like it a lot Mar 11 14:49:48 had to wipe and restore twice because fubar Mar 11 14:50:14 but using even the most robust fs is no substitute for backups, so I have those anyway Mar 11 14:50:14 Heh… I fail to recount the number of times btrfs ate my data. Mar 11 15:20:17 stintel: hi, do you have news of your HiFive Unmatched board? I just heard from a colleague that they cannot ship to Europe for now (something about missing CE compliance) Mar 11 15:24:27 isn't CE just China Export? :P Mar 11 15:24:28 * Borromini ducks Mar 11 15:24:54 zorun: delayed Mar 11 15:25:23 08|19:30:25< stintel> estimated shipping date of the hifive unmatched has changed from 11/03 to somewhere june :( Mar 11 15:29:34 aw, ok :( Mar 11 15:48:00 zorun: reminds me of my order of a macchiatobin double shot being canceled because missing US re-export authorization Mar 11 15:48:34 nitroshift suggested to try via Romania but I never tried anymore Mar 11 16:16:31 zorun: mouser website now also displays shipping alert: Due to government regulations, Mouser is unable to sell this product in your country. Mar 11 17:36:49 stintel: ah, right Mar 11 17:39:08 zorun: I requested more info from them Mar 11 17:40:57 dwmw2: ping. Quick question, since you're one of the authors… is jffs2 compression single-threaded? :) Mar 11 17:41:08 It looks like it, but I'm not sure yet. Mar 11 19:10:02 Jesus… compiled at first try without errors. Mar 11 19:13:09 169 lines of C, that's probably a personal record. Mar 11 19:14:16 rsalvaterra: In case you haven't see it yet.. https://forum.openwrt.org/t/major-european-funding-program-is-taking-submissions/90749 Mar 11 19:14:58 Oooh! Nice! Mar 11 19:18:04 I think I'm going to test this on the AirGrid. Easier to recover when things go south (and they will, I'm messing with the guts of a filesystem). Mar 11 19:20:00 rsalvaterra: Make sure you play appropriate jaunty/adventerous music while you do it.. The Indiana Jones theme is nice Mar 11 19:35:58 Grommish_: Actually… https://www.youtube.com/watch?v=-calvsGLRfs :P Mar 11 19:47:06 Let's see if I can kill the abomination that is 530-jffs2_make_lzma_available.patch… Mar 11 20:00:36 Borromini: about nlbwmon vs IPv6, did you try to "/etc/init.d/nlbwmon restart" and check if it manages to count properly then? I have a feeling that nlbwmon might have a race condition with wan6 coming up Mar 11 20:03:03 pkgadd: yeah it does Mar 11 20:03:37 https://forum.openwrt.org/t/solved-nlbwmon-vs-ipv6/42478/4 < found this in the meantime Mar 11 20:08:12 did you just add the ULA prefix or the global one? Mar 11 20:09:00 i didn't add any Mar 11 20:09:16 when nlbwmon restarts it looks like it adds both though. Mar 11 20:43:33 so now there's 2 PRs for 5.10... lol. Mar 11 20:49:47 Borromini: Too many people found out that update_kernel.sh exists… :P Mar 11 20:49:54 :P Mar 11 20:50:01 Actually, why aren't those scripts part of the core…? Mar 11 20:50:27 no idea Mar 11 21:41:44 rsalvaterra: Oh, That is a very catchy tune Mar 11 21:42:21 I want that playing in my funeral. B) Mar 11 21:42:27 hahah Mar 11 21:42:59 Denis Leary's I'm an Asshole was the top of my lists.. but, this is strong contender Mar 11 21:43:44 for funeral music Mar 11 22:23:57 sigh Mar 11 22:24:51 I dunno who maintains camembert bot but seriously fix the damn excess flood Mar 11 22:25:13 * stintel grabs some brie from the fridge Mar 11 22:26:18 camembert bot? Mar 11 22:26:29 11|23:24:25 -!- KGB-1 [~kgb@camembert.tina.pm] has quit [Excess Flood] Mar 11 22:27:01 ah Mar 11 22:33:27 Cheesy bot… Mar 11 22:35:08 =) Mar 11 22:35:21 I'd say cheeky but hey Mar 11 22:35:27 ain't KGB always cheesy? ;P Mar 11 22:41:12 Building an image to test jffs2 zstd compression… I'm still worried about the locking, though… It doesn't *seem* to be needed, but I may be completely wrong. And I'll be testing on a single CPU device… Mar 11 22:51:24 OMFG Mar 11 22:52:11 /dev/mtdblock5 on /overlay type jffs2 (rw,noatime,compr=zstd) Mar 11 22:52:27 rsalvaterra: either the world became rainbow coloured unicorno or everything just died ;D Mar 11 22:52:29 is work? o_O Mar 11 22:52:56 stintel: It seems so. No stack traces. Mar 11 22:53:35 amazeballs Mar 11 22:53:48 But I had to manually mount it. I probably need to teach a couple of things to fstools. Mar 11 22:55:34 sounds plausible Mar 11 22:56:15 Let me just push this somewhere, so I can show you guys… :) Mar 11 22:56:47 (I actually developed in the Linux kernel tree, then generated a patch to apply in OpenWrt.) Mar 11 22:57:26 that' Mar 11 22:57:28 grmbl Mar 11 22:57:35 that's how it's supposed to be done ;) Mar 11 22:58:52 Seemed logical enough. :) Mar 11 23:00:19 https://github.com/rsalvaterra/openwrt/commit/e850af61e96dff633a65df0697d09f57cebaf498 Mar 11 23:01:48 rsalvaterra: planning to send upstream ? Mar 11 23:02:08 zstd is awesome in zfs relating stuff, and I can very well see the awesomeness in jffs2 and owrt standpoint too Mar 11 23:02:51 stintel: After fixing the mounting in fstools and retest, for sure! Mar 11 23:04:47 The fact this compiled and ran without issues *the first time* scares me. Nothing works the first time. Mar 11 23:05:16 maybe it's silently eating your data ;) Mar 11 23:05:47 like btrfs Mar 11 23:05:49 SCNR Mar 11 23:07:19 stintel: Edited a file with vi, adding some rubbish. Saved, exited, synced. All fine. Mar 11 23:07:37 that's what btrfs promised 😂 Mar 11 23:08:59 stintel: I don't think jffs2 is as complex as btrfs, thankfully… :P Mar 11 23:10:02 yeah don't mind me really, I'm mostly talking bs again Mar 11 23:10:22 I was curious about btrfs back in.. like... well 15 years ago or something alike that... and it still is kinda... well.. not there =D Mar 11 23:11:01 all the licensing quirks aside, then came zfs and never looked back =) in computer side of things Mar 11 23:11:46 for most router stuff I tend to install openwrt then, even when I'm thrown gazillion beefy xeon "do us router" =) Mar 11 23:12:48 I have this conspiracy theory that oracle paid some developers to make btrfs so fundamentally flawed in order to lure people to slowaris/zfs Mar 11 23:13:40 fortunately zfs was made good enough by Sun time, and then when oracle came into play, now oracle has their zfs and rest of the work uses openzfs =) Mar 11 23:14:03 world* Mar 11 23:14:16 though work isvery suitable word there too, by acident =D Mar 11 23:16:25 is Eneas U de Queiroz on IRC ? Mar 11 23:16:41 Why does Oracle kill everything it touches? ¬_¬ Mar 11 23:17:30 CEO is evil. I remember him openly saying he wants to kill open source Mar 11 23:18:16 unfortunately that's what money does with most people Mar 11 23:18:36 Pfft… I hope he lives to see PostgreSQL kill Oracle. :P Mar 11 23:18:41 amen Mar 11 23:19:53 Hmm… can't see what's wrong in fstools, should be mounting with compr=zstd… Mar 11 23:20:35 I've been carrying this for a while (for ubifs, until now)… https://github.com/rsalvaterra/fstools/commit/6fb942b99fd97f7b513e960923856e0428b9888d Mar 11 23:21:20 … and the buildroot counterpart… https://github.com/rsalvaterra/openwrt/commit/14013a3f46d486fe668ff953a17ef9d4fc89e015 Mar 11 23:21:45 This works fine for ubifs. Mar 11 23:22:49 Oh, well… I probably just need to get some sleep- Mar 11 23:26:35 Amen Mar 11 23:30:09 rsalvaterra: FYI I didn't want to wait and upstreamed a limits.h patch, with you as reporter. If two people had related problems, probably a safe bet. Mar 11 23:35:34 guidosarducci: Great! Thanks! :) Mar 11 23:36:38 upstream is imo always the best place to get feedback, so even if the patch is wrong/incomplete/imperfect/..., it never hurts sending it upstream Mar 11 23:38:47 stintel: I've found upstream quite helpful and open to thoughtful fixes and discussion, unlike ... Mar 11 23:39:26 Wait! It mounted correctly!! I probably hadn't noticed at first boot. XD Mar 11 23:39:37 Sweeeeeeeet! Mar 11 23:40:35 I love problems which fix themselves without a single line of code. :P Mar 11 23:42:54 stintel: also often easier to get something backported from upstream than accepted first downstream Mar 11 23:43:13 guidosarducci: well upstream has much more manpower usually Mar 11 23:44:45 I simply completely ignore stuff that is over my head / outside my comfortzone, which effectively means I barely accept stuff at all, as I'm a system engineer for living, barely touch any real code at work Mar 11 23:44:46 In this case, I'm testing on OpenWrt because this is jffs2 and it's easier to test here. But I'm *not* sending this to OpenWrt before being accepted upstream. Mar 11 23:49:06 stintel: yup, very true. The "maintainer" role is also clearer there. And FWIW, the context that comes from a systems mindset is key but often undervalued. Mar 11 23:50:25 guidosarducci: it's probably why I'm always complaining in here ;) Mar 12 01:38:28 stintel: i don't think btrfs or bcachefs are flash friendly Mar 12 01:39:36 stintel: he is not active. cotequeiroz is his nick when he is Mar 12 02:11:48 Rene__: ping Mar 12 02:39:32 build #680 of mediatek/mt7629 is complete: Failure [failed pkgbuild] Build details are at http://buildbot.openwrt.org/master/images/builders/mediatek%2Fmt7629/builds/680 blamelist: Bj?rn Mork , Daniel Golle **** ENDING LOGGING AT Fri Mar 12 03:00:01 2021