**** BEGIN LOGGING AT Fri Sep 20 02:59:57 2019 Sep 20 05:53:06 jow: this fixes scanning using iwinfo for me, https://github.com/openwrt/openwrt/pull/2429 Sep 20 06:05:27 pkgadd: nice, just a small nitpick, you're missing the package: prefix in the commit message Sep 20 06:06:23 ynezz: yep, just noticed that myself (going from git, to quilt, to git) Sep 20 06:07:33 ...and fixed Sep 20 06:55:02 I added per-interface uci options and want to parse them on interface events. I currently do this via /etc/hotplug.d/ ... Sep 20 06:55:14 and I am just curious if it can be done with netifd scripts. They're not protocol handlers, just handlers for some per-interface options. Sep 20 07:45:42 where is wifi-device.path generated/detected? Sep 20 07:46:06 it doesnt appear to be hardcoded in a board specific config or anything like that Sep 20 07:47:34 aep: /lib/wifi/mac80211.sh ? Sep 20 07:48:50 aah that easy, thanks! Sep 20 07:49:21 so i can reliably put the sysfs path in there? generating config from somethign else and i'm worried it might break Sep 20 07:50:53 aep: probably Sep 20 07:51:25 aye thanks Sep 20 08:00:16 rmilecki: ok, I will generate them without numbers, but keep in mind that the has will change anyway Sep 20 08:00:46 Noltari: ah, i forgot that Sep 20 08:00:59 hash* Sep 20 08:01:05 yeah, if they rebase, it will **** BEGIN LOGGING AT Fri Sep 20 09:13:30 2019 Sep 20 09:16:07 ah nvm, forgot about the hostapd.sh brokeness Sep 20 09:16:32 it turns mode sta / encryption owe into key_mgmt=NONE instead of key_mgmt=OWE atm Sep 20 09:37:17 Hauke: I'm going to push a change to hostapd that aligns OWE ieee80211 default value handling for sta mode with the one for ap mode Sep 20 09:38:09 Hauke: right now we default ieee80211w to 2 (forced) for mode ap / encryption owe but not for mode sta / encryption owe Sep 20 09:38:31 which means that naively setting owe on an openwrt client will not work against an openwrt ap with the same settings Sep 20 09:39:18 same for sae and sea-mixed, it results in ieee80211w 2 and 1 respectively in ap mode, but not for sta mode Sep 20 09:39:45 no idea if similar logic is required for mesh Sep 20 10:12:28 does it look reasonable https://git.openwrt.org/?p=openwrt/staging/ynezz.git;a=commitdiff;h=bde4792723e8597c44498d90eb8f6d3f4925a1be;hp=3a95804ef32e2180414c5091c96ffa19327d7af5 ? Sep 20 10:13:47 I've just checked it, and $(shell which pkg-config) works fine on macOS Sep 20 11:14:41 ynezz: from a quick lance this looks rather brittle and dangerous Sep 20 11:17:16 ynezz: I'd prefer a fix similar to https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=fe43969336201f2cc7d103b68fd6e65989bee184 Sep 20 12:09:47 regarding WPA3 discussion here yesterday. It seems Apple has said that iOS13 supports WPA3, so I thought I'd test that since I installed iOS13 yesterday on one device here. I saw https://forum.openwrt.org/t/wpa3-support-in-openwrt/10554/58 which recommended to install wpad-wolfssl so I did that, but I still do not see any WPA3 related options in LuCI. What more is needed? Sep 20 12:10:22 this is on snapshoy I downloaded 15 minutes ago Sep 20 12:10:34 snapshot... and on WRT1200AC. Sep 20 12:11:17 SwedeMike: I don't know ohw latest master is, but when I made WPA3 to work it didn't yet have LUCI options Sep 20 12:11:24 tough that few months ago so :) Sep 20 12:11:49 olmari: right, so what did you have to set to get WPA3 only, without luci? Sep 20 12:12:42 just set wireless.default_radio0.encryption to WPA3-SAE ? Sep 20 12:12:56 or OWE ? Sep 20 12:13:22 at least wpad-openssl (but wolfssl should work all the same), hmm... I'm not sure about was there some other very related package needed to be installed, but anyway then I just looked wiki/documents what to put into /etc/config/wifi Sep 20 12:13:53 .wireless ofcourse Sep 20 12:14:11 I seem to have 'sae-mixed' as encryption Sep 20 12:16:22 https://openwrt.org/docs/guide-user/network/wifi/basic#wpa_modes Sep 20 12:16:39 ah, thanks. Sep 20 12:16:46 ..So WPA2 will still work also :) Sep 20 12:17:34 since I want to test this, I want to just support WPA3. Sep 20 12:18:23 if I set it to SAE, then my iphone seems unable to connect to it. I just get "incorrect password" Sep 20 12:19:43 same with OWE Sep 20 12:20:55 I don't remember what hurdles I had exactly, but I do remember looking openwrt log at the wifi up and so on helped Sep 20 12:21:48 been a while since I tried wpa3... seems that even with psk3-mixed, my OnePlus One running LineageOS does not connect Sep 20 12:23:28 Well first I'd check owrt logs to see is there any errors (on wifi up even), AFAIK when all was set, WPA2 has been working just the same like without wpa3.. but when I didn't have everything needed wifi up failed anyways :) Sep 20 12:24:29 ..not to say there couldn't be bug somewhere, but who knows... and again, my latest wpa3 setup experience from some months ago :) Sep 20 12:24:32 it doesn't even seem to try to connect Sep 20 12:25:39 with sae or sae-mixed, I see the SSID in the phone, but get password error when trying to connect. No errors in "logread" Sep 20 12:27:22 Irritating that I can't remember what I needed to do exactly... I _think_ it was some additional hostapd/wpad/something package that was sitll needed, but hrhm... I do remembe having basically "same" errors you see now Sep 20 12:28:37 my raspberry pi's running openwrt still connecting fine, so are my ESP32's and my ESP8266, my receiver, huawei p20 pro, sony tv, samsung tablet running lineageos (old version), and chromecasts Sep 20 12:28:51 so this seems to be a bug in lineageos or the oneplusone Sep 20 12:29:44 khadas vim3 running coreelec also fine Sep 20 12:32:25 oneplus 2 running old version of lineageos connects as well Sep 20 12:32:29 but nexus 5x doesn't Sep 20 12:36:04 with psk3-mixed and ieee80211w=0 both OnePlus One and Nexus 5x can connect. interesting Sep 20 12:36:39 did you have taht 1 or 2 before? Sep 20 12:36:54 1=optional, 2=required Sep 20 12:37:09 and/or I don't know what's teh default especially with wpa3 Sep 20 12:38:34 olmari: I had it set to 1 before Sep 20 12:39:20 I know I have it as 1 also (even before wpa3) Sep 20 12:39:51 same here Sep 20 12:40:02 stintel: psk3-mixed isn't in https://openwrt.org/docs/guide-user/network/wifi/basic#wpa_modes , do you mean sae-mixed ? Sep 20 12:40:20 SwedeMike: psk3-mixed*|sae-mixed*) Sep 20 12:40:29 see scripts/netifd-wireless.sh in netifd source Sep 20 12:40:34 I btw replaced wpad-wolfssl with wpad-openssl with no change in behavior Sep 20 12:41:01 stintel: ok. I tried psk3 as well, same result as sae , which I guess is not strange then Sep 20 12:41:10 indeed, it's exactly the same :) Sep 20 12:41:31 psk3 is more meaningful for me Sep 20 12:41:36 the weird thing is that it doesn't work with sae-mixed wither. psk2 works of course. Sep 20 12:41:37 as I've been used to using psk2 Sep 20 12:47:24 jow: this wouldn't solve the problem here a) kernel_menuconfig is defined in include/toplevel.mk so it lacks STAGING* vars b) we're using ncurses libs from host anyway to build mconf Sep 20 12:59:44 I mean, we don't build host ncurses Sep 20 13:08:14 when setting psk3 I get the following in hostapd-phy0.conf: ieee80211w=2 wpa_key_mgmt=SAE I this expected? Anything else I should look out for? Sep 20 13:14:29 SwedeMike: looks correct Sep 20 13:15:08 SwedeMike: well I do know WPA3 mandates 11w also, but AFAIK that shouln't make "mixed" mode work any less (I don't really know how 11w is handled there(?) ) Sep 20 13:15:31 is there something hw/driver specific that might make this silently not work? WRT1200AC uses Marvell wifi... Sep 20 13:15:44 yeah, it requires firmware support Sep 20 13:16:21 jow: what about this https://git.openwrt.org/?p=openwrt/staging/ynezz.git;a=commitdiff;h=8d5c70d84f43f5ab37c836abace84d470de0a5a9;hp=3a95804ef32e2180414c5091c96ffa19327d7af5 ? Sep 20 13:17:59 ynezz: this looks slightly better Sep 20 13:18:25 ynezz: did you check whether simply providing a filtered PATH would work too? Sep 20 13:19:45 something like kernel_menuconfig: export PATH=$(subst $(space),:,$(filter-out $(TOPDIR)/%,$(subst :,$(space),$(PATH)))) Sep 20 13:20:37 oh, nice! Sep 20 13:21:11 jow: shouldn't there be an error if this is not supported? is this 802.11w ? Sep 20 13:21:56 SwedeMike: well the error and status reporting infrastructure in mac80211 / wpa_suipplicant / hostapd is not really good to begin with Sep 20 13:22:10 SwedeMike: the fact that openwrt patches out many debug messages does not make it simpler Sep 20 13:22:48 on top of that there's various quirkcs and bugs in both luci and the uci translation scripts Sep 20 13:23:20 so no, there mustn't necessarily be an error Sep 20 13:23:33 right. Sep 20 13:25:07 check the generated /var/run/hostapd-phy*.conf or /var/run/wpa_supplicant-wlan*.conf Sep 20 13:25:14 depending on whether you're ap or sta Sep 20 13:27:32 jow: https://pastebin.com/4cs9aVFZ I'm AP. I don't see anything strange in there? Sep 20 13:28:05 SwedeMike: yep, matches my working setup Sep 20 13:28:17 what client are you trying? Sep 20 13:28:52 jow: iPhone8 with iOS13. Supposedly it should have WPA3 support, but I am not sure. Sep 20 13:28:56 jow: no luck, seems like the build still needs access to some of the stuff in host/bin 'mkhash: command not found' Sep 20 13:29:25 meh, yeah - probably due to the stampfile generation magic Sep 20 13:34:40 SwedeMike: I would try with a linux/wap-supplicant first Sep 20 13:35:00 *wpa-supplicant, this way you can at least debug and assert that the ap is wokring to begin with Sep 20 13:37:28 jow: ok, so debian10 is my best bet for a recent enough distribution etc? Sep 20 13:38:22 Is there something new about 19.07? Sep 20 13:40:57 rmilecki: nice @ crash reporting Sep 20 13:41:11 jow: thanks Sep 20 13:41:11 wonder what upstream thinks about it Sep 20 13:41:14 me too! Sep 20 13:41:32 for those interested check openwrt-devel@ for [PATCH RFC] cfg80211: add new command for reporting wiphy crashes Sep 20 13:41:56 or just linux-wireless@ directly Sep 20 13:42:06 Pepe: there's some outstanding bugs in ustream-ssl we have to take care of, then we should be good to go for rc0 Sep 20 13:42:15 or rc1, rather Sep 20 13:42:30 SwedeMike: yeah, I think so - with a well supported mac80211 driver Sep 20 13:42:42 like ath9k, ath10k, iwlwifi etc. Sep 20 13:48:43 A moot point as I do'nt have nor have immediate plans to get, but is there any idea about "wifi 6" aka 11ax? (damn I hate the new naming stuff) Sep 20 15:40:51 Hello Sep 20 15:41:06 Need some help with ath10k ACK timing Sep 20 15:45:20 ? Sep 20 16:09:09 no one? Sep 20 16:10:46 Need some help with ath10k ACK timing Sep 20 16:18:49 maxnet: Would help to even describe what kind of help and so on... not that I can guarantee anyone can help :) Sep 20 16:19:09 My problem is with ACK timing Sep 20 16:19:26 I have a 3km link and i can't get datarates although the signal is -60 Sep 20 16:20:34 I am using APU2 boards with QCA9994 card Sep 20 16:28:35 jow: ok thanks Sep 20 16:56:05 jow: I've expanded your filter idea https://git.openwrt.org/?p=openwrt/staging/ynezz.git;a=commitdiff;h=4faf5f30e4479e4f033963c70b312035fa1774ab;hp=3a95804ef32e2180414c5091c96ffa19327d7af5 which seems to work Sep 20 19:41:18 any plan to support nftables in openwrt, default in debian now and looks neat Sep 20 19:41:59 i understand it may require invasive changes but there is an iptables-ntf tool that may mitigate the pain a bit Sep 20 19:42:49 https://openwrt.org/meetings/hamburg2019/start#switch_to_nftables Sep 20 20:14:47 I think I found an issue with the qmi.sh proto(col) script. Sep 20 20:14:47 It tries to set the mode to raw-ip but the device is up so the kernel rejects the change. Sep 20 20:14:47 Trying to figure out the best way to fix it. Trivially I can down/up the iface Sep 20 23:03:14 ynezz: thanks, a good read, openwrt.org should have a donate button, so those items can be turned into action items with a deadline slightly easier. Sep 20 23:04:12 https://buildroot.org/downloads/manual/manual.html openwrt needs a manual like buildroot badly **** ENDING LOGGING AT Sat Sep 21 03:00:00 2019