**** BEGIN LOGGING AT Thu Sep 19 03:01:42 2019 Sep 19 05:03:53 rmilecki: ping Sep 19 05:04:00 jow: pong Sep 19 05:04:20 rmilecki: I just stumbled over 2c3dd70741 procd: add procd_running() helper for checking running state Sep 19 05:04:47 rmilecki: this needs some changes to work with init service names that contain things like dashes Sep 19 05:05:05 jow: could be, i probably never tried that Sep 19 05:05:27 jsonfilter -e "@.$service.instances.${instance}.running" => jsonfilter -e "@.$service.instances['${instance}'].running" Sep 19 05:05:50 makes sense Sep 19 05:06:00 actually jsonfilter -e "@['$service'].instances['${instance}'].running" Sep 19 05:06:05 goodp oint Sep 19 06:04:19 hm, has anybody tried wpa3 with wpa_supplicant in master yet? Sep 19 06:05:16 trying to get an ath9k client joining an mt76 wpa3 master Sep 19 06:12:03 does any mainstream OS/distribution ship with WPA3 support yet? Sep 19 06:12:16 (STA side) Sep 19 06:14:03 allegedly any somewhat modern linux Sep 19 06:19:08 reason for me asking, is that I have an WRT1200AC here which I installed snapshot on earlier this week. How can I test WPA3? Sep 19 06:19:53 ubuntu 19.04 as client? Sep 19 06:20:27 SwedeMike: any linux distro shipping with wpa_supplicant >= 2.7 (better 2.8) should (to some extent, the networking daemons on top of wpa_supplicant might not, yet); Debian 10/ buster does Sep 19 06:23:24 whohas suggest only 2:2.6-21ubuntu3.3 for "Disco Dingo"/ 19.04, that would be a bit old Sep 19 06:24:27 hm, ok, so debian10 I can do. If the network manager doesn't support this then it becomes harder to test. Sep 19 06:24:49 I wanted to see if I could get it working out of the box as a "normal" user, but I guess that's too early then Sep 19 06:24:55 my primary problem is that I can't manage to extract any debug output out of wpa_supplicant on openwrt Sep 19 06:25:03 both ifupdown and systemd-networkd can (as they leave it up to wpa_supplicant) Sep 19 06:25:42 I already rebuilt it using full debugging, yet "wpa_supplicant -d -d -Dnl80211 -iwlan1 -c /tmp/wpa_supplicant-wlan1.conf" just prints "Successfully initialized wpa_supplicant" and then stalls Sep 19 06:26:00 when stracing it I see that its receving scan results via netlink, yet it does not print any single debug line Sep 19 06:26:34 I usually start with running wpa_cli interactively and watch what it has to say and then poke it a little Sep 19 06:28:36 tried that too, it just keeps saying no network found Sep 19 06:28:43 while it clearly shows the bss when dumping the results Sep 19 06:32:37 during my first attempts with wpa3, I stumbled over trying to use a hex psk, instead of a passphrase. but I only have ath5k, ath9k and ath10k here to (really) test. not sure about rt2x00's (rt2800pci/ rt2800usb and earlier) or b43's (bcm4306/3) state in regards to 802.11w and SAE (but not very hopeful) Sep 19 06:33:33 when the mt76 ap side is sae-only, my ath9k wpa_supplicant ignores the network entirely Sep 19 06:34:03 jow: maybe it's problem with wpa_supplicant logging? Sep 19 06:34:15 jow: are you running it manually? without syslog options in the .conf? Sep 19 06:34:33 when the ap side is switched to mixed wpa2-psk/wpa3-sae, it fails with wlan1: CTRL-EVENT-ASSOC-REJECT bssid=50:64:2b:b1:a8:d2 status_code=31 Sep 19 06:35:18 rmilecki: yes, I do run it manually without syslog options, I've also recompiled it with WPA_MSG_MIN_PRIORITY=0 Sep 19 06:35:42 I have also disabled hw encryption on the ath9k radios Sep 19 06:35:53 as this allegedly prevents wpa3 from working Sep 19 06:36:04 jow: if you tried recompiling it after it was already compiled once, it may not work, hostapd has broken Makefile (or whatever it uses) Sep 19 06:36:18 rmilecki: I used make package/hostapd/{clean,compile} Sep 19 06:36:24 jow: you may try /clean, /prepare, then edit WPA_MSG_MIN_PRIORITY and then /compile Sep 19 06:36:30 ah Sep 19 06:36:38 i run out of ideas Sep 19 06:37:12 and "strings /usr/sbin/wpa_supplicant" confirms that all debug messages are included Sep 19 06:37:21 at least for the very old/ draft-n ar5008/ ar9103 chipsets (e.g. tl-wr1043ndv1) disabling hwcrypt is necessary, yes Sep 19 06:42:50 ah, status code 31 is "Robust management frame policy violations" Sep 19 06:43:04 looks like some 11w related issues Sep 19 06:44:42 bingo, at least mixed association works now Sep 19 06:45:10 on Debian this configuration works for me http://paste.debian.net/1101541/ (ath5k), I haven't quite checked what OpenWrt's wpa_supplicant support generates Sep 19 06:46:19 yeah, after adding ieee80211w=1 to my client test config, it started to work against an mixed ap Sep 19 06:46:34 seems openwrt allows to generate invalid wpa3 client configs yet Sep 19 06:46:54 ieee80211w >= 1 seems mandatory to authenticate to sae-enabled aps Sep 19 06:48:04 yes (well, strictly speaking, no - you can explicitly disable 802.11w on a wpa3/sae AP; but it's enabled by default) Sep 19 06:48:13 on the ap side, openwrt already enforces ieee80211w=2 apparently Sep 19 06:48:29 which explains why a client without ieee80211w > 0 can't connect Sep 19 06:49:52 correction, s/enforces/defaults to/ Sep 19 06:50:58 yes, wpa3 as a standard mandates using 802.11w (but using sae without wpa3 blessing itself doesn't strictly require it) Sep 19 06:54:01 the wpa_supplicant config generator does not handle uci option ieee80211w at all yet Sep 19 06:54:52 okay, so the sum of all these issues explains why it does not work at all - the ap gui side misrepresents the implicit ieee80211w=1 default for sae-mixed and ieee80211w=2 default for sae-only Sep 19 06:55:16 and the client side uci does not handle the configured ieee80211w at all Sep 19 06:56:23 so there is no gui-only way right now to setup wpa2/wpa3 mixed or wpa3 only ap-sta connections Sep 19 09:32:30 jow: is the same true on 18.06? Sep 19 10:03:24 DonkeyHotei: I think 18.06 does not have any sae support in the first place Sep 19 10:03:37 but assuming there is, then yes, the gui would be broken there too Sep 19 10:11:10 [Wed 2019-09-18 11:48:28 PM PDT] which explains why a client without ieee80211w > 0 can't connect <------ i was asking about this in particular Sep 19 10:14:50 DonkeyHotei: no, I don't think this applies Sep 19 10:15:33 because i noticed that winxp cannot connect if ieee80211w > 0 Sep 19 10:15:59 not even at "optional" Sep 19 10:31:57 that sounds like a different issue to me Sep 19 14:24:45 rmilecki: ping Sep 19 14:33:39 https://answers.microsoft.com/en-us/windows/forum/all/network-adapter-failure-after-installing-kb4515384/d9b0e1d8-0d6a-4b31-bdd4-a746d92369ce?auth=1 Sep 19 14:35:57 https://docs.microsoft.com/en-us/windows/release-information/status-windows-10-1903#706msgdesc Sep 19 14:38:38 fyi this is not just NEC systems Sep 19 14:43:45 And they probably only screwed over a few tens of thousands of users with their forced updates before it got caught.. Sep 19 14:45:11 i work at a uni so yeah, freakin blah Sep 19 14:45:33 our updates are controlled but the kids are on there own Sep 19 14:52:53 Hauke: ping Sep 19 15:44:42 ldir: pong Sep 19 15:47:06 rmilecki: I'm sort of around for the mount problem Sep 19 15:48:55 and I've recreated the very strange behaviour Sep 19 15:55:10 ldir: i'm afraid I need you to debug that Sep 19 15:55:14 i couldn't reproduce it Sep 19 15:56:25 ah, ok. My misunderstanding. Sep 19 16:47:03 jow: pong **** BEGIN LOGGING AT Thu Sep 19 17:58:17 2019 Sep 19 18:53:45 rmilecki: hi, I'm not sure if you get emails for @mentions on GH, so bothering you here as well, could you please review this https://github.com/openwrt/openwrt/pull/2378/commits/b6da613e90ae36098c25017a1671abba8b007cd3 target/linux/generic/files/drivers/mtd/mtdsplit/mtdsplit_fit.c fix? Sep 19 19:18:24 Is there an image for archer c7 without luci? Sep 19 19:24:45 snapshots have images without luci Sep 19 19:24:56 http://downloads.openwrt.org/snapshots/targets/ Sep 19 19:25:17 or you can build your own with image builder Sep 19 19:26:45 ynezz: k, thx Sep 19 20:14:15 ynezz: sure, thanks for pinging me Sep 19 20:35:50 Noltari: hey, i've a little suggestion for brcm2708 patches, can you generate them using --no-numbered ? Sep 19 20:36:17 Noltari: that will result in a simple [PATCH] (without a number) so updating all patches will resutlt in much smaller diff Sep 19 20:43:37 The Easybox 904 LTE is now dead after flashing the u-boot file from https://forum.archive.openwrt.org/viewtopic.php?id=44676&p=11 Sep 19 20:43:52 Just random crap on serial console Sep 19 20:47:09 it's lantiq, meaning if you find the bootsel pin, you can recover it Sep 19 20:55:13 pkgadd: It sounds as if shortcutting something during powerup would bring it into UART mode and then you can upload another uboot .asc file or so Sep 19 20:55:49 yes (pulling the bootsel pin down to GND) Sep 19 20:57:05 e.g. the BT Home Hub 5 Type has a rather elaborate explanation, the eb904 has the same functionality - but where the bootsel pin is hiding would be another question Sep 19 20:59:00 some say that the 904 XDSL has a R148 pin for that while the 904 LTE does not have such a thing .. Sep 19 20:59:08 how much effort you want to invest into this particular device would be another question though Sep 19 20:59:26 all lantiq VRX2xx SOCs have that functionality Sep 19 20:59:34 so far it's at 10EUR .. Sep 19 21:00:14 where (or if) the vendor exposes that pin on the PCB is another question, I don't think xDSL and LTE differ in that specific regard though Sep 19 21:00:25 I just need a good openwrt supported LTE gateway. No wifi no USB Sep 19 21:00:42 I wouldn't have picked that device then ;) Sep 19 21:01:08 Let me get to my soldering lamp with magnifying glass Sep 19 21:01:16 and look out for that R148 Sep 19 21:01:40 pkgadd: I'm open for your opinion - which one would you have chosen? Sep 19 21:02:59 I don't know, something with a mini-pcie slot (with USB lanes attached). I'm not really on top of LTE equipped routers Sep 19 21:04:45 I'd nick the LTE card out of it and put it into a different router Sep 19 21:08:49 perhaps ZBT WE1326 (don't take that as an endorsement, just as an example, I have zero personal expierence with mt7621/ mt76 or ZBT) Sep 19 21:16:39 Found the R148 resistor it's close to a realtively large chip labeled as MRL952AKW33-E-IR MAIN U13 V01 . 03t8 D74A Sep 19 21:17:32 Connecting Pin 9 to Gnd **** ENDING LOGGING AT Fri Sep 20 02:59:57 2019