**** BEGIN LOGGING AT Tue Mar 09 02:59:57 2021 **** BEGIN LOGGING AT Tue Mar 09 03:54:27 2021 Mar 09 05:19:49 stintel: yes, downgrade to the version prior to your update Mar 09 05:19:55 stintel: that is what I was seeing Mar 09 06:03:18 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_mediatek.html has been updated. (58.3% images and 97.9% packages reproducible in our current test framework.) Mar 09 07:34:33 blogic Mar 09 07:34:37 whoops Mar 09 08:18:02 lipnitsk: It's possible. I hadn't dircleaned in months. But I could also have screwed up my kernel configuration, I'm not sure yet. Mar 09 08:59:08 blogic: but how can that fix anything if I'm not seeing frames in tcpdump?! Mar 09 09:42:18 mangix: Uhh… have you tested musl 1.2.2 on ramips (mt7621)? Mar 09 10:23:55 mangix: what issue are you having? Mar 09 10:24:53 btw libselinux was fixed with version 3.2 Mar 09 10:25:02 sorry rsalvaterra Mar 09 10:27:54 mangix: It's hard to say, since I don't have a serial UART on this device. I compiled/installed an image from master, only with the musl 1.2.2 patch set applied, and for some reason Dropbear isn't responding, as if it had crashed (or didn't even start). Mar 09 10:29:21 mangix: Yeah, I dropped the libselinux fix in my tree and rebased. Mar 09 10:32:15 dropbear not responding is almost always keygen related Mar 09 10:32:59 mangix: Not after a sysupgrade -n, right? Mar 09 10:33:35 -n would do it Mar 09 10:33:47 -n removes dropbear's key Mar 09 10:33:57 That's what I thought. Mar 09 10:34:26 do you have urngd installed? Mar 09 10:34:27 For testing, I'm always doing sysupgrade -n, to be sure. Mar 09 10:34:51 mangix: Yes, I do. Mar 09 10:36:33 hmm yeah no easy solution Mar 09 10:36:49 i wonder if dropbear uses /dev/urandom or /dev/random Mar 09 10:36:54 should be using the former Mar 09 10:38:38 mangix: I have no reason to suspect of a low entropy issue… the router is wired to both WAN and my laptop, the network interface is also a source of entropy, if I'm not mistaken. Mar 09 10:40:29 you could try building with openssh and then installing dropbear to see what's happening Mar 09 10:41:04 Yeah, or (ugh…) telnet. Mar 09 10:41:29 (Is it even possible to use telnet in OpenWrt, nowadays?) Mar 09 10:41:52 yeah Mar 09 10:42:00 jow has a package in the packages feed Mar 09 10:44:15 Actually, in the age of the VPN, telnet might not be such a stupid idea… we just delegate security to another layer. Mar 09 10:44:37 :) Mar 09 10:44:52 And BusyBox has a telnetd applet… Mar 09 10:44:55 I stopped exposing ssh to the WAN after wireguard Mar 09 10:45:11 mangix: That makes two of us. ;) Mar 09 10:47:50 the busybox applet was retired for a specific reason. don't remember the details Mar 09 10:47:52 If it weren't for a couple of forwarded ports, I'd be 100 % stealth. Mar 09 10:48:19 i think it can still be enabled Mar 09 10:48:35 having said that, my 1.33 update removed a maybe needed telnet patch Mar 09 10:49:20 https://github.com/openwrt/openwrt/blob/0275ee5dde7c36c925396779dd23d4f470ab40e1/package/utils/busybox/patches/240-telnetd_intr.patch Mar 09 10:50:58 mangix: do you mean mac-telnet? thats not a classical telnet Mar 09 10:51:10 ordinary, rather Mar 09 10:51:44 jow: What's your opinion on reviving telnetd (for smart people with VPNs :P)? Mar 09 10:54:08 rsalvaterra: it's still there under customize busybox Mar 09 10:54:26 This is 3 years old… https://openwrt.org/inbox/howto/telnet_enable Mar 09 11:03:52 mangix: Not a first class citizen, though. You can enable the applet, but there's no init script for it out of the (Busy :P)box. Mar 09 11:06:45 And speaking of first class citizens, it's high time we'd document (at least with a placeholder) /etc/udhcpc.user and /etc/odhcp6c.user. These scripts can be incredibly useful. Mar 09 11:09:50 For example, in my case, with udhcpc.user, updating DDNS entries *only when needed* is trivial: https://paste.debian.net/1188502/ Mar 09 11:42:17 rsalvaterra: my opinion? waste of space Mar 09 11:44:58 jow: On telnetd? Yeah, possibly… Mar 09 11:48:46 rsalvaterra: or did you mean revive i nthe sense of packaging a standalone telnetd? Mar 09 11:49:07 that'd be fine for me, unless it involves me doing work of any kind Mar 09 11:50:33 jow: I was thinking more of something for the advanced user… A standalone package which would select the required BusyBox features and install the necessary scripts. Of course, it would not installable with opkg (or even available in the official images). Mar 09 11:50:48 which makes it a non-starter Mar 09 11:51:14 packages triggering conditional compilation elsewhere are highly discouraged Mar 09 11:51:25 in the offical feeds at least Mar 09 11:51:44 jow: Fair enough. :) Mar 09 11:52:20 Are you opposed to documenting udhcpc.user and odhcp6c.user? If not, I can take care of that. Mar 09 13:27:39 why is quilt used? what is its added value? Mar 09 13:27:48 i can't help but to hate it Mar 09 13:30:44 ability to manage stacks of patches Mar 09 13:30:53 refresh, interactively add/rebase etc. Mar 09 13:32:03 always complaining about dirty patches ... Mar 09 13:32:21 doesn't match my experience Mar 09 13:32:24 its just a diff Mar 09 13:32:30 quilt is probably some middle ground between package without patches and full external git tree package Mar 09 13:32:38 exactly Mar 09 13:33:08 grift: It's extremely useful. Granted, it has its learning curve, but allows you to use the upstream versions of the code, applying only the patches you need for your specific use case. Mar 09 13:33:40 you don't need to use it if you have better tooling available to manage the patch stacks, like importing the vnailla sources into git and add modifications as commits which are thne exported with git-format-patch or similar Mar 09 13:33:46 The alternative would be having repositories for all the apps whose code needs to be patched. Mar 09 13:33:52 so its telling me this: Mar 09 13:33:54 2021-03-09T13:11:27.3040826Z Dirty patches detected, please refresh and review the diff Mar 09 13:34:06 never seen that message Mar 09 13:34:12 so what am i supposed to do? patch Refresh the patches…? Mar 09 13:34:41 I've also never seen quilt printing messages with timestamps Mar 09 13:34:44 how? please dont tell me to install quilt? Mar 09 13:35:08 How are you supposed to do OpenWrt development without having quilt installed…? Mar 09 13:35:12 I thought quilt is the one complaining? Mar 09 13:35:37 lol i get the message ... Mar 09 13:36:27 eh "Dirty patches detected, please refresh and review the diff" Mar 09 13:36:32 is not related to quilt or openwrt Mar 09 13:36:47 that seems to be some random ci check the folks added for the packages feed Mar 09 13:37:30 Yeah, never seen quilt saying that… Mar 09 13:37:44 o ok my bad Mar 09 13:37:56 anyway the patches apply fine locally Mar 09 13:38:11 maybe it detects fuzz Mar 09 13:40:26 but yes imagine this, your updating to upstream. everything is fine in the build system. so now you create a pull request, and the ci rejects it. no reason just say's "review the diff" Mar 09 13:40:30 what diff? Mar 09 13:40:53 and why does it build fine in openwrt build system but not that ci? Mar 09 13:40:56 no idea, I'd love to help but I fail at even finding the code for the ci Mar 09 13:41:11 what is the fuzz with a bit of fuzz Mar 09 13:41:11 seems to be hidden in random containers (?) Mar 09 13:46:11 jow: Who knows the MT7621 platform better? Mar 09 13:46:18 blogic Mar 09 13:46:28 I guess Mar 09 13:46:55 I'm seeing strange things in the device tree… I mean, both the GDMA and the HSDMA nodes are disabled, but we have drivers for them… Mar 09 13:47:55 And I see both GDMA and HSDMA enabled in the vendor firmware… Mar 09 13:50:21 grift: ok, so the CI check performs a package/xxx/refresh Mar 09 13:50:46 grift: if the end result produces output in "git diff", it means the patches contained in the commit are not in the canonical format Mar 09 13:51:14 yes, they might apply cleanly, yes they might work but patch headers could be different (e.g. -p a b style, truncated context lines) or there could be fuzz Mar 09 13:51:25 that seems to be what this check is about Mar 09 13:51:33 https://github.com/openwrt/gh-action-sdk/blob/master/entrypoint.sh#L68 Mar 09 13:52:07 thanks for digging that up Mar 09 14:18:45 i have another question about this: https://github.com/openwrt/openwrt/blob/master/package/utils/policycoreutils/Makefile Mar 09 14:19:08 my package depends on policycoreutils-setfiles on the host Mar 09 14:19:39 currently i have a PKG_BUILD_DEPENDS:=policycoreutils/host Mar 09 14:20:26 however i have a feeling that this will not ensure that policycoreutils-setfiles is installed on the ho... o wait Mar 09 14:20:30 it wouldnt work anyway Mar 09 14:20:41 sigh Mar 09 14:41:31 blogic: ping. Any specific reason for the GDMA and HSDMA nodes being disabled in the MT7621 device tree, when we have drivers for both of them? Mar 09 14:44:04 grift: make package/x/refresh Mar 09 14:48:16 mangix: ? Mar 09 14:49:51 if thats about the dirty patch challenge i faced, then nevermind Mar 09 14:50:30 i was doing duplicate work anyway someone else already addressed the package updates Mar 09 19:53:27 Ooo.. The ER10x is here.. Now to poke and prod at it Mar 09 19:54:03 Grommish: You should try the "world's most powerful wireless router". :P Mar 09 19:54:48 rsalvaterra: Someone sent this to me to try and getr Owrt on it hehe Mar 09 19:55:06 https://www.mcsweeneys.net/articles/congratulations-on-your-purchase-of-the-worlds-most-powerful-wireless-router Mar 09 19:55:10 Old, but gold. ;) Mar 09 19:55:13 and ramips looks scary Mar 09 19:55:39 and the whole console has to be set to 57600 was a nice curve Mar 09 20:00:37 That ER10x is ramips/mt7621, right? Mar 09 20:00:42 rsalvaterra: My hazmat suit into rated level-b.. wil I be alright? ;p Mar 09 20:00:59 Grommish: It's optional… XD Mar 09 20:01:01 rsalvaterra: yes, with the realtek switch on top Mar 09 20:02:40 rsalvaterra: That link is amazing.. Thank you ;p Mar 09 20:02:51 Ok, it would be nice if you could give mangix's musl 1.2.2 pull a spin, eventually. After dropping the pull, I managed to get my RM2100 on Linux 5.10.20 just fine… :/ Mar 09 20:03:16 rsalvaterra: This ia NiB, so its still running EdgeOS Mar 09 20:03:29 rsalvaterra: I want to see about pulling a working dts from it before I go playing Mar 09 20:03:48 Oh, of course! :) Mar 09 20:04:06 I found drivers for the upper ports, but the 5.12.x kernel removes the openwrt "second, but never upstreamed" switch defines Mar 09 20:04:15 :/ Mar 09 20:06:57 https://github.com/torvalds/linux/commit/c88c76c7286ef857216434d245c6032b1813376b Mar 09 20:08:09 but that should be for teh boards switch, not the realtek chipset switch Mar 09 20:19:17 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_x86.html has been updated. (100.0% images and 97.8% packages reproducible in our current test framework.) Mar 09 20:19:33 Grommish: You need to figure out how the Realtek switch connects to the SoC… but the device tree will be helpful, for sure. Aren't the sources available? Mar 09 20:22:51 Speaking of device trees… Mar 09 20:24:03 noltari: ping. I have decompiled the device tree from an undocumented (closed source, GPL violating) BCM96838 device… is this interesting? Mar 09 20:26:19 noltari: It's this thing: https://www.alticelabs.com/content/datasheets/FL_GR241AG_FiberGateway-4x4_ALB_EN.pdf (But seems to be missing a lot of stuff.) Mar 09 20:34:07 rsalvaterra: `If its closed source an obtained questionably, I don't know how it could be used.. Any derivitive work would be tainted by it Mar 09 20:35:01 Grommish: Uhh… not in the EU…? Reverse-engineering is allowed. Mar 09 20:35:18 Yes, but what about Intellectual Property rights? Mar 09 20:35:28 is that covered as well? Mar 09 20:35:39 I'm not versed in EU standards, so I'm asking :) Mar 09 20:36:18 You can Reverse Engineer in the US, as long as it's a clean-room dev environment.. which means you can't have seen or interacted with the closed source Mar 09 20:38:44 Well, IANAL, but this is the law… https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32009L0024&from=EN Mar 09 20:44:16 rsalvaterra: Article 6, 2 (c) might bite you Mar 09 20:44:42 ) Mar 09 20:44:42 Mar 09 20:44:42 to be used for the development, production or marketing of a computer program substantially similar in its expression, or for any other act which infringes copyright Mar 09 20:44:47 Oof. Sorry Mar 09 20:45:07 Grommish: I interpret it in the opposite way. The keyword is interoperability. :) Mar 09 20:45:30 Production or Marketing... Mar 09 20:45:45 You are trying to recreate their IP for us outside of their control Mar 09 20:46:02 Dunno.. I would find true legal advice though Mar 09 20:46:09 because I'm not it (obviously) Mar 09 20:46:57 and unless you can prove you were "fixing bugs in their binary"..... Mar 09 20:47:09 rather than "trying to use their binary to fix their OS" Mar 09 20:47:23 which would be a different pig altogether Mar 09 21:19:09 isn't reverse engineering also allowed to build a compatible product then the interface description is not available? Mar 09 21:20:09 e.g. When you want to implement a network protocol which is not documented to talk to some device Mar 09 21:20:10 Hauke: Dunno, but I bet BCM would argue using their IP for OpenWrt would cut into their profits somehow Mar 09 21:20:34 Hauke: You know as well as anyone how squirrley Copyright IP can be Mar 09 21:20:38 clean room reverse engineering should be ok Mar 09 21:20:51 Hauke: Right, but he has decompiled source Mar 09 21:21:02 Hauke: which is why I questioned it Mar 09 21:21:20 if it is linked to the Linux kernel it should be GPL Mar 09 21:21:22 ;-) Mar 09 21:22:12 hehe Well.. Should and is.. Broadcom SHOULD make their drivers source public.. but... heh Mar 09 21:24:26 rsalvaterra: Of course, nothing says you can't take that knowledge and use it locally for yourself :) Mar 09 21:24:35 Grommish: there is some code for it and it can also be seen in their dts file from the GPL sources of their stock firmware Mar 09 21:25:43 decke[m]: *nod* I'm wanting to pull the full tree from the running config on the device and compare it to the existing mt7621.dtsi and see what is different so I can find the switch info Mar 09 21:25:49 Grommish: so the realtek is connected via the rgmii link to the SoC Mar 09 21:26:10 I think I'm going to have to re-arrange my setup to be near a network drop for this Mar 09 21:27:18 But, I can always YOLO it.. I need to read up on debricking Mar 09 21:28:24 decke[m]: https://help.ui.com/hc/en-us/articles/360018189493-EdgeRouter-Manual-TFTP-Recovery :) Mar 09 21:31:11 decke[m]: Easy enough it looks like Mar 09 21:32:49 Grommish: yeah looks well documented Mar 09 21:36:34 Grommish: i found this post to be very informative about the ER-X Mar 09 21:36:38 https://kazoo.ga/re-visit-the-switch-in-edgerouter-x/ Mar 09 21:41:29 i've noticed that master doesn't have an option build "generic" ath79/mikrotik images (16/64mb, nor/nand ac/non-ac) like those available for 19.07 Mar 09 21:42:14 is this an omission or will each specific device have to be supported "manually"? Mar 09 21:42:23 support for each even Mar 09 21:55:01 shibboleth: i think that's the consequence of generic profiles being removed Mar 09 21:56:16 commit 9b5e9b65cf Mar 09 21:56:21 for ath79 e.g. Mar 09 21:57:13 Reading that edgerouter X article, it says there is 2 gigabit/s between the cpu and the switch. I think I also saw that the nat hardware acceleration can do 2 gigabit/s. But if I test that, I can do 930 mbit/s in 1 direction, or do about 450 in both directions. Mar 09 21:58:20 I guess I should read the rest of the artcile Mar 09 22:05:29 Hauke: It's actually linked to the first version of the Linux kernel which contains code I wrote. :P Mar 09 22:05:35 (4.4) Mar 09 22:07:24 decke: So do you think OpenWrt should be able to support 1 gbit/s symmetric? Mar 09 22:18:53 Aw, wtf, sent the same patch twice. *facepalm* Mar 09 22:19:18 rsalvaterra: at least you can still send them :P Mar 09 22:19:27 * Borromini gives his vps the side eye Mar 09 22:19:52 Borromini: VPS? You don't develop locally? :P Mar 09 22:22:35 oh no i just have my mail server on the vps. it's just a bit weird that both dovecot and postfix aren't receiving/sending anything from/to the ML anymore Mar 09 22:23:52 Borromini: Ah, right, I forgot you have your own domain. Mar 09 22:25:02 i mostly stick to it because protonmail doesn't do plain smtp etc. ... but i found out about hydroxide and will be looking into that Mar 09 22:27:56 Ok, now I have a 100 % working base to test musl 1.2.2 on the RM2100. Mar 09 22:28:27 I'm almost sure it's broken on ramips/m7621, but I can't confirm before testing. Mar 09 22:39:35 cheesus anyone here can help me figuring out opt passing between perl functions? Mar 09 22:39:41 ./scripts/feeds is no fun Mar 09 23:11:08 rsalvaterra: This setup is completely janky but I think You'd appreciate it. One laptop connected via WiFi, sharing the connection to the Ethernet port, which is connected to the WAN of the ER10x in the WAN port.. Eth1 is connected to another laptop so I can use it internal to the device (also a wifi connection from laptop2 to the network) >_< Mar 09 23:12:11 Grommish: That's exactly my setup, right now! XD Mar 09 23:13:35 In my case, I have the RM2100 connected to my laptop (Ethernet). The laptop is connected via Wi-Fi to my home network, and I'm accessing it from another machine. Mar 09 23:13:51 It Works™. Mar 09 23:14:11 Yes :) It also means i don't have to make a 75m cat6a cable hehe Mar 09 23:14:20 So I'm super happy it just works Mar 09 23:16:14 It works because Wi-Fi has a lower metric than Ethernet… ;) Mar 09 23:16:49 Wait, no, the other way around. Mar 09 23:17:01 Ethernet has a lower metric than Wi-Fi. :) Mar 09 23:17:39 Yes, but in my case, the metric is not important because I'm bridging the ethernet port and the wifi adapter so that the wifi is the upstream Mar 09 23:18:04 I'm using the alienware m17xr4 as a wireless bridge hahaha Mar 09 23:18:23 (and a platform for resting the other laptop on actually). its buff Mar 09 23:18:36 4kg laptops are beefy Mar 09 23:18:51 as long as it's not IPoAC/ RFC 1149 ;) Mar 09 23:19:20 No, I don't have a power modulator so I can do that.. and I"m running thru a power block hehe Mar 09 23:19:53 and no PoE out here Mar 09 23:19:55 AC ~= Avian carriers, not mains voltage in this case ;) Mar 09 23:20:07 AH Mar 09 23:20:10 https://en.wikipedia.org/wiki/IP_over_Avian_Carriers Mar 09 23:20:33 The coops aren't allowed in teh garage ;( They smell Mar 09 23:20:50 Only _I_ can smell bad in here damnit Mar 09 23:20:59 house rules Mar 09 23:22:26 ah ha! and now I can do apt-get on device Mar 09 23:31:01 https://gist.github.com/Grommish/7886f7be0d07b59978226844f2f10d9e Mar 09 23:59:26 pkgadd: And it's not restricted to IPv4… ;) https://tools.ietf.org/html/rfc6214 Mar 10 00:03:35 [OT] Holy hell, rpc.idmapd still requires dnotify… in Debian Sid…?! This was fixed in 2018…! o_O http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=commitdiff;h=cec1278cdb385188cdf6b693fda1b00f9b93df55 **** ENDING LOGGING AT Wed Mar 10 02:59:56 2021