**** BEGIN LOGGING AT Wed Feb 12 03:00:34 2020 Feb 12 05:23:24 why does my hostapd.conf have an ap_isolate=1 in it? Feb 12 06:20:25 # grep isolate /tmp/run/hostapd-phy0.conf Feb 12 06:20:25 ap_isolate=1 Feb 12 06:20:48 no isolate option in my /etc/config/wireless Feb 12 06:39:57 it's getting set via the json_get_vars call in /lib/netifd/hostapd.sh:hostapd_set_bss_options() Feb 12 06:43:33 russell--: it's auto-enabled when multicast_to_unicast is enabled Feb 12 06:50:57 * russell-- didn't set multicast_to_unicast either Feb 12 06:52:54 russell--: I have a feeling it's enabled by default for wireless interfaces. I didn't set it too but /sys/class/net/wlan0/brport/multicast_to_unicast gives 1 Feb 12 06:54:37 it never was before, it's recent behavior afaik Feb 12 06:54:55 ap_isolate that is. Feb 12 07:12:18 maybe it's older than i thought Feb 12 07:24:49 the scripts seem to suggest it defaults off, e.g.: set_default isolate 0 in /lib/netifd/hostapd.sh Feb 12 07:29:59 thats a diferent isolate Feb 12 07:30:15 or rather the "option isolate" does not directly control ap_isolate=1 Feb 12 07:30:52 if [ "$isolate" -gt 0 ]; then Feb 12 07:30:52 append bss_conf "ap_isolate=$isolate" "$N" Feb 12 07:30:52 ap_isolate is a precondition for multicast to unicast iirc, thats why it is enabled even if client isolation is disabled Feb 12 07:31:38 [ ${multicast_to_unicast:-1} -gt 0 ] && json_add_boolean isolate 1 Feb 12 07:32:06 what is ap_isolate then, because i'm confused ;-) Feb 12 07:32:40 ap_isolate means direct client to client communication is inhibited and forced via the routers bridge interface instead Feb 12 07:33:10 which is needed for multicast to unicast conversion to work properly Feb 12 07:33:17 ah, gotcha Feb 12 07:33:56 and that multicast_to_unicast conversion happens to be default on Feb 12 07:38:55 jow: you see a chance backporting the infamous JSON info file patch to the next 19.07 point release? Feb 12 07:40:28 As in me backporting and you accepting Feb 12 07:51:58 rmilecki: did you see my previous email on that ipv6/mld regression? Feb 12 07:52:13 nbd: i just replied Feb 12 07:52:42 ah, ok Feb 12 08:04:26 jow: opinion on this? https://github.com/openwrt/packages/pull/11294 Feb 12 08:04:31 the , at the end feels wrong Feb 12 08:33:31 mangix: looks fishy to me. Can't tell offhand whether this works and is legal Feb 12 08:41:51 build #262 of layerscape/armv8_64b is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/layerscape%2Farmv8_64b/builds/262 Feb 12 10:14:34 jow: I don't need this, but I'm just reading the jsapi docs, and read that fs.read() is "unsuitable for obtaining binary data" is there another method or should that be done just via making them available via http or via lua controllers? Feb 12 10:51:11 karlp: fs.exec_direct("/bin/executable", ["arg1", "arg2", ...], "blob").then(function(blob) { /* do whatever with the JS Blob */ }) Feb 12 10:52:32 I am using that to capture vnstati and rrdtool graph output to directly render the png output in the browser Feb 12 10:52:43 hat avoid the need for temporary files or extra http endpoints Feb 12 11:01:04 updated the docs now Feb 12 11:03:42 is there a way to change the default interface selected in Luci's wake on LAN page from eth0 to something else? Feb 12 11:07:05 ldir: you could edit /www/luci-static/resources/view/wol.js Feb 12 11:07:49 ldir: and add an `o.default = "whatever";` after `o = s.option(widgets.DeviceSelect, ...)` Feb 12 11:09:19 ah, great! Thank you! Feb 12 11:10:38 * ldir regrets enabling 'minify js' - errgh! Feb 12 11:15:01 /msg Feb 12 11:16:21 jow: looks like a hack to me Feb 12 11:23:18 jow: cool, wasns't something I needed, just things in the back of my head while I'm working, tryign to use more of your stuff, less of my hacks :) Feb 12 11:31:39 jow: like so? https://paste.ubuntu.com/p/SRxyt6V9yV/ Feb 12 11:32:09 ldir: yes Feb 12 11:32:28 where so != work :-/ Feb 12 11:32:53 intersting. let me check here Feb 12 11:34:28 works for me, you need to clear the browser cache one though (or try holding shift while reloading the page) Feb 12 11:34:34 *once Feb 12 11:36:39 Yeah, I've just opened the javascript console thing and noticed it has an older page. Feb 12 11:37:15 ha! - solved. problem exists between chair and keyboard! Feb 12 11:37:19 sorry :-( Feb 12 11:40:24 blogic: did you have wired network working with 5.4 on bcm53xx? Feb 12 11:40:43 safari needed a 'shift reload' not just a 'cmd shift reload' Feb 12 11:40:58 blogic: i don't see any errors in dmesg, swconfig looks OK, it displays ports etc. but I don't get IP over DHCP and for static I can't ping my router Feb 12 11:41:20 Hauke: ^^ Feb 12 11:52:22 after switching from 5.4 back to 4.19 I get: Feb 12 11:52:26 MODPOST 1 modules Feb 12 11:52:28 FATAL: parse error in symbol dump file Feb 12 11:52:35 make[5]: *** [scripts/Makefile.modpost:92: __modpost] Error 1 Feb 12 11:52:50 I already tried rm -fR build_dir/target-arm_cortex-a9_musl_eabi/linux-bcm53xx_generic/ tmp/ Feb 12 11:53:28 any hint? Feb 12 12:12:44 fixed by rm -fR staging_dir/target-arm_cortex-a9_musl_eabi/ Feb 12 12:15:56 rmilecki: nope Feb 12 12:16:03 i just did basic boot up testing so far Feb 12 12:16:11 I plan to do more today Feb 12 12:16:32 rmilecki: i also have a 80% complete rewrite of the FA offload engine done Feb 12 12:16:38 it all compiles but does not work yet Feb 12 12:16:43 need to figure out why :-) Feb 12 12:16:48 nice!!! Feb 12 12:19:49 i see you tracked down the memleak Feb 12 12:19:52 very cool Feb 12 12:21:36 rmilecki: fyi, https://forum.openwrt.org/t/ipq806x-help-needed-for-kernel-5-4-porting/53875/283?u=xback Feb 12 12:21:50 blogic: still not fix though Feb 12 12:32:45 xback: thanks for that link Feb 12 13:58:10 hi all, is there any example of feed 'included-in-source'? i mean i have some package/library/whatever and for cd/ci i want to have feed for building from included directly in upstream source Feb 12 14:01:20 I've not see a _feed_ inside a package, normally the other way around. though I've seen people keep the openwrt package makefiles in a directory of their project and symlink it directly into the packages dir. Feb 12 14:01:25 not a _feed_ like that per se Feb 12 14:10:12 indy: yes Feb 12 14:10:31 indy: https://github.com/OLSR/olsrd/tree/master/openwrt Feb 12 14:11:29 neat, a single package feed. Feb 12 14:11:53 isn't olsr also in the routing feed? Feb 12 14:12:36 jow, +1 thanks a lot, that is what i was looking for :) Feb 12 14:12:38 yes it is. But years ago when olsrd v1 was still under development, the guys there wanted a simple way to always build latest as feed without having to deal with external repos or altering makefiles Feb 12 15:47:33 in javascript if I want to get a value from uci is it ' uci.get('thing', 'section, 'var'); ' Feb 12 15:48:37 config, section, option Feb 12 15:50:49 yes Feb 12 15:58:17 hmmm Feb 12 15:58:34 I bet you can't guess where this is going jow ;-) Feb 12 15:59:38 * ldir is copy & pasting code again Feb 12 16:00:27 uci get etherwake.setup.interface Feb 12 16:00:27 eth1 Feb 12 16:02:40 so naive logically - var iface = uci.get('etherwake', 'setup', 'interface') || 'eth2'; should work. Feb 12 16:02:54 yes, but only if the etherwake uci has been loaded Feb 12 16:03:10 I guess you're extending wol.js ? Feb 12 16:03:24 am I *that* transparent ? :-) Feb 12 16:04:13 * ldir goes to hunt down the missing 'load a uci section' function Feb 12 16:04:13 you need to extend the load function. After this.callHostHints(), add uci.load('etherwake') Feb 12 16:05:17 btw I was using the || 'eth2' thing as part of my debugging. It's always returning 'eth2' at the moment Feb 12 16:10:31 thanks jow - it works! expect a PR shortly Feb 12 16:28:30 ldir: I didn't know that etherwake has a uci config Feb 12 16:31:38 you can't be expected to know *everything* ;-) Feb 12 16:50:09 Hi there, I need some help with OpenMPTCP router on Raspberry Pi configs. There are 4 USBs (2 blue, 2 black) on Raspberry pi 4. MPTCP installed on it, but I do not understand which port is which? for example WAN1 is the upper blue USB port ? I tried to set the static IP but it does not seem right... how can I be sure WAN1 or WAN2 or... are which Feb 12 16:50:10 one on the kit for example look at this picture. which one of these ports are WAN1, WAN2, OMRVPN, OMR6IN4, OVPWAN1, OVPNWAN2, ...? Feb 12 16:50:39 for example I connect 192.168.9.1/24 and 192.168.10.1/24 ( which go to different WANs) to the blue USBs on Raspberry Pi. How can I set up the MPTCP router? should I set the static IP on WAN1 and WAN2 or what? because it seems it randomly changes... is there any other way to do that? Feb 12 17:00:03 jow: what's the trick again for getting the present value of another cbi/js form field? I'm trying to have a button that can test some parameters, and looking at all the users of form.Button isn't giving me the insights I was hoping for. Feb 12 17:05:08 karlp: https://github.com/openwrt/luci/search?q=lookupOption&unscoped_q=lookupOption Feb 12 17:05:13 that should give some real world examples Feb 12 17:07:13 in essence its var opt = this.map.lookupOption('that_other_option', section_id); var other_val = opt[0].formvalue(section_id); Feb 12 17:09:46 alternatively you could simple store those other options in a variable to access them Feb 12 17:19:20 rmilecki: please ping me when you pushed that swconfig patch Feb 12 17:19:28 i'll add it to my tree with 5.4 then Feb 12 17:25:24 jow: thanks, that worked! console.log("looking up: ", opt); Feb 12 17:25:27 https://zerobin.net/?178432efb3b679f2#WrXmyhLWJqOGvy24bHAvbHZRDeVv/9IjpiMINIIRT44= Feb 12 17:27:23 karlp: you could also simply use `key` instead of `opt[0]` and lookpOption Feb 12 17:27:45 lookupOption() is onyl usefule if you cannot hold references to the other opts for whatever reasons Feb 12 17:28:44 hrm, I think I had tried doign formvalue stuff before and it wasn't workign on things. but perhaps that was inside write callbacks or something. Feb 12 17:28:54 the problem is obtaining the current section_id Feb 12 17:29:13 I think I'll extend form.js to pass that as first argument to the onclick handler Feb 12 17:29:27 eyah, getting the section id is tedious. Feb 12 17:29:30 its an oversight in form.js Feb 12 17:29:44 I have an s in s = m.section(), Feb 12 17:29:51 if I could just get it from there somehow Feb 12 17:30:26 or in the callback, that's fine too :) Feb 12 17:37:43 can you try this? http://sprunge.us/CIZzZR - that will pass the sid as 2nd arg Feb 12 17:37:54 can't make it first since that owuld break existing code Feb 12 17:39:48 eh, I've only got minified on this device at the moment, can try that in the morning. Feb 12 17:40:05 working on 19.07 anyway, so it's just improvements for the future. Feb 12 17:46:49 damn CORS problems now anyway. Feb 12 18:41:32 rmilecki: I also had the problem in scripts/Makefile.modpost:92 Feb 12 18:41:45 I did the same workaround like you with rm Feb 12 19:08:53 Hi, I am looking to implement flow offload for the AR8227N. The openwrt PHY driver defines switch_dev_ops but no net_device_ops, would it be possible to plug into the ndo offload callbacks at this level ? Is it only possible for ethernet drivers ? Feb 12 21:15:39 xback: ping, done Feb 13 00:06:37 i wonder if theres any 802.11ax device supported now or nah? Feb 13 00:32:37 nickname_ not yet mate **** ENDING LOGGING AT Thu Feb 13 02:59:57 2020