**** BEGIN LOGGING AT Fri Jan 08 02:59:57 2021 Jan 08 07:26:14 Hello, comrades! Please help me! My script not work in cron job. Who help me? My script for cron job is: https://pastebin.com/wFbjE3rP Jan 08 07:33:00 aliceussr: what happens with you run the script by hand? Jan 08 07:34:06 normal script debugging technique would be to sprinkle "echo " commands to get an idea where it's going wrong Jan 08 07:37:05 russell--: It works manually! Jan 08 07:39:06 russell--: I know it. But there is no cron debugging option in OpenWRT! Jan 08 07:39:06 russell--: 33 10  * * *   /root/develop.sh >> /sd/backup/site/backup.log 2>&1 Jan 08 07:39:06 russell--: This debug not work for me. Jan 08 07:42:30 something that works manually but not from cron usually implies an environment difference Jan 08 07:50:36 russell--: PATH=/usr/sbin:/usr/bin:/sbin:/bin:/root Jan 08 07:50:36 russell--: I use right environment from OpenWRT system in crontab and my script. Jan 08 08:14:46 Hello, comrades! Please help me! My script not work in cron job. Who help me? My script for cron job is: https://pastebin.com/wFbjE3rP Jan 08 09:03:04 Hello, comrades! Please help me! My script not work in cron job. Who help me? My script for cron job is: https://pastebin.cowFbjE3rP Jan 08 09:03:04 Hello, comrades! Please help me! My script not work in cron job. Who help me? My script for cron job is: https://pastebin.com/wFbjE3rP Jan 08 09:05:33 aliceussr: stop spamming the same repeated line. Jan 08 09:06:50 Currently, fw3 seems to translate a WAN port forwarding rule as something like this: -i pppoe-wan --dport 123 -j DNAT --to-destination . This has the undesired effect of also doing DNAT for a roadwarrior's internet connections, if a VPN server happens to run on the router. I tried specifying an external ip for the WAN zone in the port forward rule, but found out the ip was recorded statically. It stops working once WAN ip changes. Jan 08 09:06:57 aliceussr: "does not work" isn't a fault description. Jan 08 09:07:04 Any way to ask fw3 to resolve the WAN ip? It already does so for reflection rules. Jan 08 09:10:08 i have a question about the board_name variable. octeon target still uses 'codenames' like 'er' for EdgeRouter, 'erlite' for EdgeRouter Lite. However, EdgeRouter 4 (a more recent addition) is a readable 'edgerouter-4'. Is there a specific reason to use the er/erlite ones? can I replace them with edgerouter/edgerouter-lite? Jan 08 09:11:33 Borromini: probably Jan 08 09:11:50 there aren't many octeon devices so not much development Jan 08 09:12:21 mangix: yeah, i read that Jan 08 09:12:58 at first sight though it looks more like loose ends (and less conventions at the time when the boards got added) Jan 08 09:13:04 SwedeMike: I cannot with debug cron task in OpenWRT. Jan 08 09:13:55 mangix: e.g. board_name is prefixed with vendor almost everywhere now (e.g. ubnt,edgerouter-4) but octeon has very few of those. Just the new Itus stuff and the EdgeRouter 4, which got added recently Jan 08 09:14:02 aliceussr: 08:34 < russell--> normal script debugging technique would be to sprinkle "echo " commands to get an idea where it's going wrong Jan 08 09:14:20 Borromini: yeah. the other stuff wasn't added recently :) Jan 08 09:14:27 propose a PR if it bothers yuou Jan 08 09:14:30 *you Jan 08 09:14:59 mangix: working on it, was testing the waters :) Jan 08 09:15:05 SwedeMike: What information do you need? My environment variables are set correctly, all files are present. Jan 08 09:15:26 testing on my edgerouter lite :) Jan 08 09:16:08 aliceussr: I am trying to help you help yourself. I put in "echo "something" >> /tmp/mydebugfile.txt and try to figure out where it's going wrong. Jan 08 09:21:28 SwedeMike: Where to insert it? Jan 08 09:22:34 aliceussr: at any point where you are performing an action Jan 08 09:22:53 'doing $this now' Jan 08 09:22:56 then see how far it gets Jan 08 09:27:30 Borromini: My script in the console manually works fine. It doesn't work only in cron! My script work in cron Linux, not work only in OpenWRT! Jan 08 09:27:58 openwrt is ash, make sure it's POSIX compliant Jan 08 09:31:12 aliceussr: start by adding a ">> /tmp/cronlog.txt 2>&1" at the end of the crontab line, and see what the output is Jan 08 09:33:19 I mean, you try and run bash, do you have bash installed at least? Jan 08 09:35:34 SwedeMike: 29 12  * * *   /root/develop.sh >> /sd/backup/site/backup.log I have this line in crontab - it`s not work! Jan 08 09:36:03 does it work from the command line _on openwrt_ ? Jan 08 09:36:28 ie, do you have bash actually installed in your openwrt system? Jan 08 09:36:46 aliceussr: "doesn't work" is not a fault report. You have to provide more information than saying "it doesn't work" Jan 08 09:37:34 karlp: I've tried both bash and sh and ash. The result is the same: manually script works in cron does not work! Jan 08 09:37:44 ok, I'm done :) Jan 08 09:39:35 I, here, think what kind of idiots do you develop firmware for some devices and then they upload their junk to your official repositories! Jan 08 09:39:35 Yes it works from the command line in OpenWRT, but doesn't work in cron OpenWRT 07.19.5. Jan 08 09:39:35 SwedeMike: Go to Mommy, Troll! Jan 08 09:39:50 me too, he just wrote some weird shit to me in #openwrt after I called him out for spamming the same line there whilst we were helping him here. Jan 08 09:41:36 Go to your fagots! Jan 08 09:42:05 … Jan 08 09:43:37 *HOSPITAL* Jan 08 09:43:37 If you could help, you would ask technical questions about my working environment in OpenWRT and find out other technical details. And so, you are trolls and fagots, not apostates! Jan 08 09:44:56 i tend to treat this as a sign that the community is now pretty big. Jan 08 09:45:37 The price of popularity? :P Jan 08 09:45:38 Hello, comrades! Please help me! My script not work in cron job. Who help me? My script for cron job is: https://pastebin.com/wFbjE3rP Jan 08 09:46:31 yeah, also true for all kinds of issue trackers :) Jan 08 09:55:45 lol Jan 08 09:59:15 who has ops in this channel... Jan 08 09:59:58 .!.. Jan 08 10:07:16 aliceussr: your script cannot work because the interpreter line is wrong Jan 08 10:07:33 "#!/bin/bash /etc/rc.common" will execute /etc/rc.common and nothing else Jan 08 10:07:54 since your script declares none of the prcoedures (start(), boot() etc.) expected by rc.common, it'll end up doing nothing Jan 08 10:08:38 the reason why it work when running manually is because you likely invoked it as "sh /root/develop.sh" and not simply "/root/develop.sh" Jan 08 10:09:01 apart from that, there's no /bin/bash on openwrt by default Jan 08 10:09:09 so you might want to change it to "#!/bin/sh" Jan 08 10:12:03 jow: I have bash in my OpenWRT. And script work right by hands. Not work only in cron OpenWRT. Jan 08 10:12:33 how do you invoke it exactly when you test it by hand? Jan 08 10:13:02 as "bash /root/develop.sh" or as "/root/develop.sh" ? Jan 08 10:14:04 jow: I try change to "#!/bin/sh", is result: work hands, not work in cron! Jan 08 10:15:08 you added it to /var/spool/crontabs/root ? Jan 08 10:15:56 correction, /etc/crontabs/root Jan 08 10:16:05 jow: /root/develop.sh Jan 08 10:16:38 did you restart crond after adding the cron entry to /etc/crontabs/root or at least reboot once? Jan 08 10:16:55 by defualt, cron is not started if no cron entries exist, to save resources Jan 08 10:17:44 also is cron even attempting to execute your script? check "logread | grep crond" Jan 08 10:18:05 should print lines such as "Fri Jan 8 09:55:00 2021 cron.err crond[2813]: USER root pid 26817 cmd /root/develop.sh" Jan 08 10:18:07 jow: /etc/crontabs/root - cron run this job, not run script! Jan 08 10:20:08 I'm not an idiot, I reloaded everything. Jan 08 10:20:08 It`s not work in OpenWRT 19.07.5 Jan 08 10:20:44 Q1) /etc/crontabs/root contains "29 12  * * *   /root/develop.sh" ? Jan 08 10:21:03 Q2) `pidof crond` reports a pid? Jan 08 10:21:34 Q3) `logread | grep crond` shows a line like "Fri Jan 8 09:55:00 2021 cron.err crond[2813]: USER root pid 26817 cmd /root/develop.sh" ? Jan 08 10:22:09 Fri Jan  8 09:59:57 2021 cron.info crond[7693]: crond (busybox 1.30.1) started, log level 5 Jan 08 10:22:09 Fri Jan  8 10:06:09 2021 cron.info crond[7727]: crond (busybox 1.30.1) started, log level 5 Jan 08 10:22:09 Fri Jan  8 10:06:35 2021 cron.info crond[7760]: crond (busybox 1.30.1) started, log level 5 Jan 08 10:22:09 Sorry! Jan 08 10:22:09 Cron not run my script! Jan 08 10:22:10 07 13  * * *  /root/develop.sh > >/tmp/debug.log 2>&1 ---- It`s not work! Strange... Jan 08 10:22:55 your lock shows "10:06:35" as local time Jan 08 10:23:32 the cron entry is supposed to run daily at 13:07 Jan 08 10:23:43 how did you test that it does not work? Jan 08 10:23:57 so you mean it didn't run yesterday? Jan 08 10:24:10 pidof crond 7870 Jan 08 10:26:11 jow: In `logread | grep crond` not shows a line like /root/develop.sh Jan 08 10:26:11 jow: My time is 13.23 Moscow (UTC +3) not 10:06:35! Jan 08 10:27:42 fwiw, my cronjobs run fine Jan 08 10:28:12 jow: Ask the OpenWRT developers this question: why their web gui (luci) time does not match the time when using the data (System) command? Jan 08 10:30:11 uci show system.@system[0].timezone Jan 08 10:30:12 I will try changing the time to UTC 0. Jan 08 10:30:13 russell--: .!.. Jan 08 10:32:13 jow: Big sorry! In Web Luci i have time UTC 0, i change it`s to Moscow time. Jan 08 10:34:17 for testing you might want to use something like this so it fires more often: */5 * * * * Jan 08 10:39:30 jow: Please wait, i testing. Jan 08 10:42:04 aliceussr: sorry, I had a phone call Jan 08 10:42:07 what russell-- said Jan 08 10:42:28 also in the line you quoted "07 13  * * *  /root/develop.sh > >/tmp/debug.log 2>&1" Jan 08 10:42:34 there's a small syntax error Jan 08 10:42:41 it should be ">>" not "> >" Jan 08 10:43:31 jow: Fri Jan  8 13:37:00 2021 cron.info crond[8299]: USER root pid 8307 cmd /root/develop.sh > /tmp/debug.log 2>&1 Jan 08 10:43:49 that looks good, so at least it attempts to run your script Jan 08 10:44:02 anything in /tmp/debug.log ? Jan 08 10:45:32 jow: Problem solved! Probles is complex. Big thanks for you! It is a pity that I can not send you tinctures of Russian moonshine on oak bark - better than brandy! Jan 08 10:45:32 jow: Yes! I known! Jan 08 10:47:33 My script is already working, I can see it: the archives are downloading! Hurrah! Jan 08 10:51:10 aliceussr: i credit the insults for getting results Jan 08 10:55:36 jow: The problem was complex: i change crontab file (add SHELL, PATH, MAIL), i change to "#! /Bin/sh" in script and add PATH to my script. Jan 08 11:42:36 jow: You here? I have new small problem with run script in cron: can you please tell me how to increase its execution time? The script was interrupted and the files were not fully downloaded. Jan 08 11:46:38 Who here? I have new small problem with run script in cron: can you please tell me how to increase its execution time? The script was interrupted and the files were not fully downloaded. Jan 08 12:03:38 aliceussr: the script runs until it's done. the interruption must come from somewhere else. but be careful if you're script is running long (e.g. 5minutes) and you execute it every minute. so it will stack up. Jan 08 12:10:50 lynxis: Thanks for the information. I think I found a solution to my problem using the timeout program from coreutils. Jan 08 12:12:51 lynxis: I have a very slow download speed from the Internet, so my script takes a very long time. Jan 08 12:15:42 aliceussr: you could try `lock` to protect the script from running multiple times and also add a timeout to ensure it doesn't run too long. hope that helps Jan 08 12:18:02 Calling people Faggats should get you band! Jan 08 12:42:27 hey guys, I wanted to start working on a new target Jan 08 12:42:48 it won't be completely working yet Jan 08 12:43:12 so I'll like to add it a and just comment out "BuildTarget" call Jan 08 12:43:28 i hope it's OK? Jan 08 12:53:01 rmilecki: can't you mark it as broken? Jan 08 12:53:19 lynxis: can i? how? Jan 08 12:53:29 lynxis: i thought only devices can have BROKEN := y Jan 08 12:53:53 or source-only Jan 08 12:54:12 in FEATURES Jan 08 12:54:13 tmn505: ah, there was sth like that Jan 08 12:54:16 tmn505: thanks! Jan 08 13:12:32 adrianschmutzler: hi Jan 08 13:40:12 lynxis: Ok, i tryy "lock". "Timeout" i try current use and test it`s. Jan 08 13:51:45 damex: ping Jan 08 13:58:11 there it is :) [PATCH] bcm4908: initial work on the Broadcom BCM4908 target https://patchwork.ozlabs.org/project/openwrt/patch/20210108135232.31360-1-zajec5@gmail.com/ Jan 08 14:46:04 rmilecki: BTW if you add new device to the tree it's recommended (or rather required) to add information about this device in the commit message (hardware details, flashing details) so it can then be added by the documentation folks to openwrt.org/toh automagically Jan 08 15:04:26 dangole, ping Jan 08 15:06:23 ynezz: ah, more work, duplicating info from the .dts file ;) Jan 08 15:07:46 dangole, output before the patches: https://paste.debian.net/hidden/fcc2496b/ Jan 08 15:10:13 shibboleth: please do `ubus call iwinfo info '{"device":"wlan0"}'` as well Jan 08 15:10:54 after: https://paste.debian.net/hidden/c0cd12a7/ Jan 08 15:10:59 will do Jan 08 15:15:08 sure that command is right? Jan 08 15:15:27 shibboleth: yes, but you need rpcd-mod-iwinfo installed for it to work Jan 08 15:15:41 one moment Jan 08 15:17:23 is there a package somewhere on downloads.openwrt.org? Jan 08 15:19:58 shibboleth: yes, it's part of base packages Jan 08 15:20:40 https://downloads.openwrt.org/snapshots/targets/ipq806x/generic/packages/ ? Jan 08 15:21:29 shibboleth: https://downloads.openwrt.org/snapshots/packages/arm_cortex-a7_neon-vfpv4/base/rpcd-mod-iwinfo_2021-01-06-ea7f4717-1_arm_cortex-a7_neon-vfpv4.ipk Jan 08 15:31:54 dangole, https://paste.debian.net/hidden/87bbbe17/ Jan 08 15:32:51 shibboleth: looks good, hwmode now says 'ad' which makes more sense than 'a/g' on that hardware Jan 08 15:33:19 shibboleth: if you like, try encryption 'psk2', that should now also work out-of-the-box and correctly use GCMP, and show nicely in iwinfo Jan 08 15:34:36 https://paste.debian.net/hidden/3471e8ae/ Jan 08 15:34:40 i'll try psk2 Jan 08 15:36:40 https://paste.debian.net/hidden/a949e2d9/ Jan 08 15:37:22 https://paste.debian.net/hidden/200b03e8/ Jan 08 15:37:52 seems psk2 works better than gcmp(-foo) with re to iwinfo Jan 08 15:39:12 shibboleth: 'gcmp' has never been a valid option for 'encryption'. 'psk2+gcmp' should have worked with robimarko's patch before i made gcmp the default cipher on 11ad Jan 08 15:39:32 nope, then it fails completely Jan 08 15:39:39 at least two days ago, lemme retry Jan 08 15:40:40 ok, now it works, but it failed completely before the reverted patches Jan 08 15:41:50 shibboleth: what was changed now after the revert and committed again was unrelated to 11ad. Ansuel fixed some stuff in the channel analysis stuff in iwinfo which wasn't safe otherwise. Jan 08 15:43:00 right, i've failed to bring up the ap with psk2+gcmp as late as wednesday, lemme look into what changed Jan 08 15:43:04 shibboleth: but even more before, i also didn't understand fully how it was supposed to work with gcmp, hence i made it the default for 11ad and remote-tried with lynxis' pair of mikrotik devices Jan 08 15:47:30 he has the wap60g? Jan 08 15:50:37 txpowerlist would have been ridiculous had it been n/ac :) Jan 08 15:52:02 40 dBm (10000 mW) <--- yeah, let's put that in the living-room :P Jan 08 15:53:43 shibboleth: 2 wap60 v2, 3 wap60 v1. a friend also has wap60v3. i think wap60v2 has the newer chip (wil64xx?). unsure what v3 has Jan 08 15:54:02 iirc that's the x3 and v2 Jan 08 15:55:13 been following the forum thread, the wap60g (especially the x3) would have made a *sweet* radar if the fw could be unlocked Jan 08 15:56:36 still, quite useful as a high-bw link. same goes for the lhgg60ad over distance Jan 08 15:59:36 shibboleth: a friend of mine has the 6 hap60g in production. he would like to replace routeros with openwrt. Jan 08 16:00:49 would've been something wrong with him otherwise :) Jan 08 16:01:53 How many minutes does one hour have? Jan 08 16:02:19 I'm just asking because my Linksys shows this interface uptime: Jan 08 16:02:21 Laufzeit: 0h 60m 26s Jan 08 16:02:35 lynxis, i thought work on the x3 had stalled: https://forum.openwrt.org/t/support-for-mikrotik-wap-60gx3/59177 Jan 08 16:03:07 :) Jan 08 16:03:45 shibboleth: would be great if someone can take a look onto the borad files and decrpytion. i looked into the forum thread but would like to take a newer board file and decode it myself with the attached script. validating the seed woudl be great Jan 08 16:05:15 also the wil62xx drivers some work. ibss and 4addr need more investigation if both are possible. it looks promising. also high latency on "idle" links needs to be fixed. mikrotik did a patch which fixed that in recent build. Jan 08 16:05:48 also, is 11ad SOL with re to wpa3? Jan 08 16:07:48 >KGB-0< https://tests.reproducible-builds.org/openwrt/openwrt_omap.html has been updated. (0% images and 98.2% packages reproducible in our current test framework.) Jan 08 16:46:02 shibboleth: what do you mean? Jan 08 16:46:46 i'm looking for the ieeee 802.11ad as pdf if someone have it around. Jan 08 16:52:33 hm, something is quote broken with recent master Jan 08 16:52:39 mac80211-hwsim used to work just fine Jan 08 16:52:49 now I can't even get it to start hostapd and/or wpa_supplicant Jan 08 16:53:26 Fri Jan 8 16:48:38 2021 daemon.notice netifd: radio0 (7086): Command failed: Request timed out Jan 08 16:53:29 Fri Jan 8 16:48:38 2021 daemon.notice netifd: radio0 (7086): Command failed: Not found Jan 08 16:53:32 Fri Jan 8 16:48:38 2021 daemon.notice netifd: radio0 (7086): WARNING (wireless_add_process): executable path /usr/sbin/wpad does not match proc) Jan 08 16:53:35 Fri Jan 8 16:48:38 2021 daemon.notice netifd: radio0 (7086): Command failed: Invalid argument Jan 08 16:53:45 jow: /etc/init.d/wpad restart Jan 08 16:54:05 this used to happen for me sporadically until master as of last week or so Jan 08 16:54:58 nope, not helping Jan 08 16:55:03 now i get some "variable not set"-something and i have to manually ifconfig wlanx down Jan 08 16:55:21 ./mac80211.sh mac80211 setup radio0 {"config":{"channel":"1",...} is timing out Jan 08 16:55:37 the wireless handler is unable to reconfigure netifd Jan 08 16:55:54 well, disable all the radios in config/wireless, run wifi Jan 08 16:56:02 then ifconfig, are any of the wlan ifaces still up? Jan 08 16:56:19 if so, ifconfig down Jan 08 16:56:22 then restart wpad Jan 08 16:56:41 and reenable the radios/ifaces, wifi Jan 08 16:57:28 you could try wpad stop, start, ofc Jan 08 16:57:33 is wpa_supplicant running? Jan 08 16:57:48 Fri Jan 8 16:56:37 2021 daemon.notice netifd: radio1 (8938): WARNING: Variable 'data' does not exist or is not an array/object Jan 08 16:57:54 yup Jan 08 16:58:09 now stuff is not running because the ubus calls that instruct netifd to configure the radios are timing out Jan 08 16:58:23 logread | grep crash Jan 08 16:58:30 something about wpad? Jan 08 16:58:36 no Jan 08 16:58:40 wpad is never started Jan 08 16:59:21 ok, and is /usr/sbin/wpa_supplicant running? Jan 08 16:59:58 jow: for me the wpad restart used to work Jan 08 17:00:14 until i figured it out i had to powercycle/reset to get back up Jan 08 17:00:24 could be unrelated Jan 08 17:01:22 i was getting some could not connect to something, something, wpa, crashes 79 times Jan 08 17:01:49 hostapd and supplicant both start, then eventually terminate Jan 08 17:03:02 lynxis, with how the 11ad drivers require gcmp Jan 08 17:06:11 definitely some netifd bug apparently Jan 08 17:06:24 starting hostapd manually with the generated config makes things working instantly Jan 08 17:09:58 starting wpa_supplicant... boom kernel oops Jan 08 17:13:40 who's the maintainer for adding new devices to ramips? or should i just send it to the openwrt-devel mailing list. tried looking in some makefiles and couldnt find anything Jan 08 17:14:33 there are no target maintainers anymore Jan 08 17:14:38 just send to openwrt-devel Jan 08 17:15:15 dangole: found the reason for the broken LuCI signal reporting :) Jan 08 17:15:56 figgyc: git.openwrt.org/9ba09653 Jan 08 17:16:01 dangole: https://github.com/openwrt/luci/blob/master/libs/rpcd-mod-luci/src/luci.c#L809 - pretty obvious in hindsight Jan 08 17:16:23 thanks Jan 08 17:18:18 and mac80211-hwsim started working just fine ootb after qemu-system-x86-64 rebooted due to the kernel oops Jan 08 17:18:45 to not delve into another yak shaving session I'll simply pretend that I didn't see those issues Jan 08 17:32:39 jow: fun =) Jan 08 17:43:51 jow: nice you found it. i was getting desperate. i had no idea that rpcd-mod-luci is dlopen'ing iwinfo.so... Jan 08 17:44:58 jow: to reproduce the issue you stated above, all i need to do is build for x86/64 with hwsim enabled and boot that with qemu? Jan 08 17:46:06 yep Jan 08 17:46:35 once I learned that mac80211-hwsim can do actual associations I started to really like it Jan 08 17:47:07 make phy0 an ap and phy1 a sta Jan 08 17:47:11 or do ap-sta on the first phy Jan 08 17:49:03 jow: that's what i usually do for testing hostapd/wpa_supplicant before moving to real hardware :) Jan 08 17:50:22 regarding dlopen: yeah, its kind of a hack since I was too lazy building the C code to chain async calls to iwinfo via ubus within the luci rpcd plugin Jan 08 17:50:27 but that would be the cleanest way Jan 08 17:51:41 basically there's a luci-rpc/getWirelessDevices call which is a proxy for network.wireless/status Jan 08 17:52:02 it returns the original network.wireless/status data but adds iwinfo details to each radio and network section Jan 08 17:52:23 jow: regarding hwsim wpad crashing: you selected wpad-basic-wolfssl, i assume? Jan 08 17:53:03 I booted a snapshot off the website and opkg-installed mac80211-hwsim and wpad Jan 08 17:53:36 dangole: ok, lemme do exactly that then Jan 08 17:53:36 that didn't really work well until the system crashed due to a reboot Jan 08 17:53:43 erm rebooted due to a crash Jan 08 17:53:57 afterwards everything just worked Jan 08 17:54:25 wI did restart netifd (/etc/init.d/network restart) after the opkg install stuff, theoretically that should be enough Jan 08 17:54:41 but apparently stuff was still in a wonky state Jan 08 17:55:07 mac80211.sh was stuck in "ubus wait_for hostapd" / "ubus wait_for wpa_supplicant" Jan 08 17:55:13 jow: i see. i usually build with hwsim included from the start, maybe adding wifi phy dynamically didn't go well Jan 08 17:55:21 as if netifd failed to start the processes Jan 08 17:55:42 oh yes, I also did "wifi config" after opkg install to populate /etc/config/wireless Jan 08 17:55:47 jow: maybe some ieee80211 hotplug handler making a mess, i'll see it in a minute Jan 08 18:01:14 jow: which changes did you make to /etc/config/wireless after running `wifi config`? because i did all that now and it worked. i obviously also had to put the station interface into another network, so i added that to /etc/config/network as well (with proto 'none'). Jan 08 18:02:01 flipped disabled to 0, ecnryption to psk2, key to 98765432 Jan 08 18:02:32 but maybe it was just a fluke. I mena if its working properly after a reboot then that is good enough for me Jan 08 18:03:03 opkg install sequence was opkg install mac80211 (acidentally) - thatp ulled iw-full hsotapd-common, cfg80211 etc. Jan 08 18:03:12 then opkg install mac80211-hwsim Jan 08 18:03:17 little bit of playing around Jan 08 18:03:22 eventually opkg install wpad Jan 08 18:10:09 chrome remembers all my username/password on other sites, but not luci(master branch), is this because, luci's login is totally client-side-js-generated? Jan 08 18:17:20 rr123: good question. In my case browsers insist on putting the login password into the wifi psk config field Jan 08 18:18:14 I think at some point we introduced hidden dummy password fields which are prefilled by the browser to not populate the actual config fields with garbage Jan 08 18:18:35 that probably has a negative effect on the login form though Jan 08 18:25:44 i actually like the way to not let browser to remember certain site's password by design(so I don't need tell chrome), in this case i don't want my kid to get access to the router when i leave my computer without locking it Jan 08 18:26:19 so he can never mess around with openwrt at mid-night for gaming, for example. Jan 08 18:27:47 there is no way to hide passwowrd/wifi-psk-key/802.1x key as either the OS or the browser happily remember them so a walk-by can easily reveal them, nobody can remember lock up their screen all the time these days Jan 08 18:29:13 so i want sites providing those no-autofill, no-autocomplete, always-input-password js-driven login form Jan 08 18:29:50 rr123: some people use ubikey and such, and when they pull or remove it, that locks the session. Jan 08 18:32:35 dangole: these days all are in the house, and i have more than 10 devices running, the only way to secure login is to make sure the browser won't remember my password is to make sure certain sites's form can never be remembered by design, like what luci does now(please don't fix that :) ) Jan 08 18:33:44 after certain time, the session times out, and you must relogin, and chrome/firefox can never remember your password, is what I want, i actually want to design the same login scheme to my site but i'm still learning js these days Jan 08 18:34:04 rr123: in firefox it still offers to auto-generate and remember all password fields (login, password-change form, wifi key, ...) just checked Jan 08 18:34:54 my firefox does not auto remember luci password, just the username part Jan 08 18:35:35 rr123: and on the android device (also firefix) i'm quite thankful i don't need to type a long password to check whats happening on the router (eg. restart uqmi interface while driving the car...) Jan 08 18:36:39 rr123: obviously, on Android other ways of permanently authenticating the browser/webapp are thinkable. i'm not into that at all, but i imagine it can be useful for some people. Jan 08 18:36:46 dangole: agree for cellphones, then you can add a 'remember me' button to turn that on/off Jan 08 18:38:02 rr123: for luci we use an additional before the actual password Jan 08 18:38:09 and hide that with CSS Jan 08 18:38:54 browser are smart enough to ignore fields with "display:none" but not yet smart enought to detect off-screen rendered fields (position absolute; left: -10000px) Jan 08 18:39:14 it's an uphill battle though Jan 08 18:39:14 that's a nice trick, let's study that code Jan 08 18:39:37 chrome will remember the form whenever it detect type password Jan 08 18:40:05 i was thinking about use a type-text input and to let JS to hidden the alphanumerics(****) but that's too much Jan 08 18:40:40 basically the js converts all inputs for this fake password input field to ***** Jan 08 18:40:59 the browser has no interest in remembering as it detects no password input type Jan 08 18:43:35 jow: where is the left -10000px? could not find it in the css Jan 08 18:44:02 Borromini: are the changes introduced by the realtek macros safe? Jan 08 18:44:14 e.g. removing sfp property Jan 08 18:44:39 jow: it's in cascade.css, will study more Jan 08 19:47:18 jow: now that we are already breaking things, what about "iwinfo: add BSS load element to scan result"? would also be nice to have... Jan 08 19:56:28 rr123: jow : is there any reason you could not just leave it to users to decide whether they want to remember passwords or not? Jan 08 19:58:02 I mena, we _want_ to let that happen right? we're not trying to actively prevent it? Jan 08 19:58:28 that's what 'remember me' button under password normally did Jan 08 19:59:01 but these days the browser will store them anyways unless you exclusively disallow it, which many people won't bother Jan 08 19:59:29 I still realllllllly want that to be left ot users, not have luci decide "we'r egoing to fight you to tyr and prevent you from ever saving the password" Jan 08 20:00:27 i agree, as long as it can be selected, in my case, i don't want to certain sites including luci ever be stored by browser Jan 08 22:29:47 karlp: I too, but not if the consequence is bug reports about the browser resetting the wifi psk on every save operation because it sees a password field and incorrectly autocompletes nonensense then "leafing it to the users" is no choice Jan 08 22:29:59 *leaving Jan 08 22:30:48 once browsers start adhering to their own specs again (autocomplete=off, autocomplete=new-password and all the other funny non-features) then the workarounds could be dropped **** ENDING LOGGING AT Sat Jan 09 02:59:57 2021