**** BEGIN LOGGING AT Thu Apr 11 02:59:58 2019 Apr 11 06:53:32 hi all, i'm preparing ag71xx for mainline. just want to make sure, no body else is working on this Apr 11 07:44:10 olerem: ar71xx? Apr 11 07:48:50 ath79 Apr 11 07:49:01 devicetree variant Apr 11 07:55:16 olerem: nope. nobody. Apr 11 07:55:38 ok, then I do Apr 11 07:56:47 well Apr 11 07:57:00 it will require a lot of cleanup for mainline Apr 11 07:57:06 yes Apr 11 07:57:42 actually Apr 11 07:57:48 while on the topic Apr 11 07:57:51 https://github.com/neheb/source/commit/51f07d483ec6ef64bb0981c7c3e16b2dab07b399 Apr 11 07:58:44 I also carry this one. ~20mbps improvement Apr 11 07:58:47 https://github.com/neheb/source/commit/4324e14f08f0cd3d7f81c9b074ecbf711d26896a Apr 11 08:07:23 olerem: ^^ Apr 11 08:11:28 mangix: done Apr 11 08:23:37 ynezz: lgtm Apr 11 08:41:46 list winds Apr 11 08:41:58 sorry rong window Apr 11 08:49:28 oh joy, wallystech spam. Apr 11 08:49:49 Suggestion: Ought AR71xx be planned to deprecate, make new default choice buildroot menuconfig? (: Apr 11 08:54:22 olmari: it's a good point! I'd suggest X86 as default. Apr 11 08:56:21 Came in mind as just now looking it :D Apr 11 08:57:55 well... x86, ath79... could "none" be option? (I see maybe technical difficulties for "nothing" in slection that needs exactly one) Apr 11 08:58:08 Also, overthingking it maybe, but at least today I'm thinking :P Apr 11 09:08:27 ldir: the pragmatic choice would be mt7621 :) Apr 11 09:09:52 I sort of don't care - anything but ar71xx - as much as I hate to select Intel it's at least a choice that isn't likely to go away :-) Apr 11 09:10:31 "no choice" would also ofrce user to choose something :) Apr 11 09:11:09 ar71xx is probably default due to popularity Apr 11 09:11:15 yes, if a choice *has* to be preselected, else no selection would be better Apr 11 09:11:41 at the moment I don't even know where that pre-selection is made! Apr 11 09:12:39 mangix: well it was AFAIK choosed exatly as it was "the new wrt54gl" at times when tplink wndr1043nd was most popular stuff Apr 11 09:14:03 ..and 4/16 devices was decided to "can't be sanely supported by default" anymore, thus bcm platform was obsoleted which was then default Apr 11 09:15:03 i still remember running 2/8 devices Apr 11 09:15:12 what a nightmake Apr 11 09:15:37 i had to disable httpd and dhcpd Apr 11 09:15:51 else it would swap like crazy Apr 11 09:16:09 Anyway, I too would be plenty happy as "non-ar71xx" as default, "no selection" would be a bonus, but nothing major Apr 11 09:38:42 I consider pushing out 18.06.3 and 17.06.7 later this week, mainly to address the EAP weaknesses found in hostapd Apr 11 09:39:21 if there's other urgent things you like to see fixed, consider submitting/pushing them to the appropriate branches Apr 11 09:41:48 jow is the hostapd stuff just for WPA3? Apr 11 09:42:36 mostly WPA3 (SAE) and some medium weakness/vulnerability in EAP (wpa enterprise) Apr 11 09:43:04 it is not a doom scenario like CRACK (which in turn was a little bit of a storm in a water glass as well) Apr 11 09:43:06 jow OK how do you use WPA3 in openwrt? Apr 11 09:43:40 is it in luci or do you have to put it in the config file for the radio Apr 11 09:43:55 Tapper: iirc it needs to be configured manually in /e/c/wireless Apr 11 09:44:06 KK thanks Apr 11 09:46:38 Not that it's specifically an OpenWrt thing, but I'd sure like an system where multiple versions of WPA could be supported, as there becomes to be many of them and I don't see sudden hop happening to WPA3 at client sides :) Apr 11 09:47:10 jow: where on earth is the default arch selected? Apr 11 09:51:11 ldir: scripts/target-metadata.pl - default TARGET_ar71xx Apr 11 09:52:07 ldir: last change was 2012 with 84b43f112a569ece93cf62894828db8c0446cdb6 Apr 11 09:52:21 (the script was called differently back then) Apr 11 09:53:35 so I see :-) my git grep / pickaxe isn't as good as yours - helps if you know what you're looking for too Apr 11 09:55:42 wouldn't mind a change to mt7621 Apr 11 09:55:52 https://downloads.openwrt.org/stats/#extra1 - ar71xx stil lseems to be the most popular though Apr 11 09:56:04 so it should likely get switched to ath79 Apr 11 09:56:56 ath79 is the functional replacement yes. Apr 11 09:57:13 wow, no downloads for brcm2708 at all?! Apr 11 09:58:00 iirc not so long ago someone in here said it was one of the more popular targets Apr 11 09:58:44 hmm removes default TARGET_ar71xx and it still selects that as the default. Apr 11 09:58:45 stintel: is this the Pi devices? Maybe people all use weird debian derivates now to power their nodejs projects ;) Apr 11 09:58:56 ldir: rm -r tmp/ Apr 11 09:59:02 ha! Apr 11 09:59:09 jow: haha ;) Apr 11 09:59:53 actually here I see rpi3 images: https://downloads.openwrt.org/stats/awstats.downloads2.lede-project.org.downloads.html Apr 11 09:59:56 and others Apr 11 10:00:14 * ldir tries very hard not to type rm -r /tmp Apr 11 10:00:50 hehe Apr 11 10:01:27 wouldn't be a disaster though, just recreate it and reboot Apr 11 10:02:15 hmmm, removing the default makes it select Allwinner A1x/A20/A3x - which makes absolute logical sense - but not ideal - votes for ath79 as new default Apr 11 10:04:06 * jow tends back to this k8s cluster setup Apr 11 10:04:22 jow: you becoming a hipster? :D Apr 11 10:05:35 * jow grumbles Apr 11 10:06:28 k9 would bark more (aaand other older than old shit? :) ) Apr 11 10:06:35 stintel: I'll bore you one day with my accidental trashing of a /usr partition on a live system involving mirrored discs (aka RAID 1) Apr 11 10:07:39 goes out for the day with missus ldir and will pick up on the default target change later if not beaten to it. Apr 11 10:07:57 ldir: backups, backups, backups :) Apr 11 10:08:12 ldir: and enjoy the day Apr 11 10:09:14 jow: no offense meant; we're going to have to use it too Apr 11 10:10:06 I actually have minikube running here Apr 11 10:16:02 before I really go.... Cuba? !!! Cuba really? Apr 11 10:25:07 yeah I had the same thought Apr 11 10:33:51 backup was that step when you git push right ? Apr 11 10:58:24 How will "Force 40MHz mode" apply to 11ac networks? is it more of "force 80MHz" or is it not read at all in this case? Apr 11 11:19:22 stintel: I went ahead with kube-spray which does a decent job at deploying a proper ha cluster Apr 11 11:19:54 also takes care of the CNI, kube-router in my case because I wanted to extend the BGP peerings directly to the bare metal LBs Apr 11 11:20:45 that default overlay network mesh routing business is not ideal for high bandwidth applications Apr 11 11:21:51 kubespray could be interesting, passed the link to my colleagues Apr 11 11:22:13 you think we should supporting running openwrt as a kubernetes node ? :) Apr 11 11:23:32 not sure what exactly that involves apart from making it Docker-friendly Apr 11 11:24:04 s/docker/cr-i/ I guess Apr 11 11:24:06 compete with alpine? Apr 11 11:24:26 or cri-o Apr 11 11:40:26 karlp: no, I think it makes no sense to compete with alpine. Its just to have openwrt available for certain networking tasks Apr 11 11:41:00 at least I wouldn't want to fight with OpenWrt, trying to get all this Go and NOde and Ruby etc. stuff running Apr 11 11:42:53 no I wouldn't really either. Apr 11 12:41:28 did anyone try to build OpenWrt on IBM Power architecture ? Apr 11 12:43:56 I believe a few years back we had one contributor building on ppc macs Apr 11 12:44:13 but I might've misremembered Apr 11 12:44:20 ok. I'll figure it out soonish Apr 11 12:44:28 stintel: I did, a few months ago (to benchmark a new machine from https://cfarm.tetaneutral.net/machines/list/ ) Apr 11 12:44:36 zorun_: cool. did it work? Apr 11 12:44:38 I don't remember running into any issue Apr 11 12:44:44 cool! Apr 11 12:44:45 but I did not build all packages Apr 11 12:45:20 stintel: if you want to test yourself, just apply for a gccfarm account, you'll get a shell on it Apr 11 12:45:27 oh that's nice Apr 11 12:46:40 I have a 10-core Power8 system on the way, I'm suspecting it can be slightly faster than my i7 3930k :) Apr 11 12:48:18 possibly :) Apr 11 12:49:23 and this could help me decide whether or not I want a Thalos II as next workstation, or stick with x86 Apr 11 12:49:41 these systems have a funny hyperthreading model (the power8 on the farm has 20 cores / 160 threads) Apr 11 12:49:57 yeah, SMT* Apr 11 12:50:00 SMT8* Apr 11 13:03:20 Hi will WPA3 be added to luci in 19.xx? Apr 11 13:06:25 if you add it, why not? :) Apr 11 13:09:58 ynezz I wish I could code I can just about type this response! 😀 Apr 11 13:10:24 My job is best bigging up OpenWrt on websits and twitter Apr 11 13:10:58 Like I am doing here: https://arstechnica.com/information-technology/2019/04/serious-flaws-leave-wpa3-vulnerable-to-hacks-that-steal-wi-fi-passwords/?comments=1&post=37177047 Apr 11 13:11:22 in the COMMENTS Apr 11 13:16:46 I previously had WPA3 enabled on my OpenWRT device, but when I installed Android Q Beta 2 on my phone, it stopped connecting to my WAP at all and I had to turn WPA3 back off. :( Apr 11 13:19:01 same here Apr 11 13:36:48 ldir: I've created almost final version https://github.com/ynezz/linux/commit/6ca37d5f3570927af7ca22907b7dceceae55f0ed can you please test it? Can I include your tested-by tag as well? Apr 11 14:52:32 ynezz: I think your reply about GL.iNet GL-X1200 is a bit harsh Apr 11 14:54:06 ynezz: Fix for ar8216 here: https://github.com/openwrt/openwrt/pull/1984 Apr 11 14:57:04 mangix: I guess this may fix the iperf regression you mentioned a while ago :) Apr 11 15:04:41 stintel: I wonder what panned out to be harsh exactly, which part, I will try to improve that part next time, I didn't meant to be harsh, just tried to explain my view why it's not a good idea to submit ar71xx anymore Apr 11 15:07:38 (and I've deleted my response to his "urgent customer need" before actually sending the reply) Apr 11 15:08:59 it's just the "wasting our/your time" that comes across as harsh, at least to me Apr 11 15:10:29 ok, noted Apr 11 15:10:44 might just be me though ;) Apr 11 15:13:19 gch981213: thanks a lot! I'll look at it in the following days Apr 11 15:16:39 stintel: and I think, that it's better to express my view on this patch in some way, let them understand, that it's probably not going to be merged and instead focus on ath79 then to let it rot in the patchwork until the 19.03 is branched and patch rejected with "not applicable" status Apr 11 15:17:04 no objection there Apr 11 15:17:31 but maybe we should officially announce that we will no longer accept new ar71xx devices Apr 11 15:18:03 and write it on the wiki somewhere, then we can simply reference that announcement as of then Apr 11 15:18:03 yeah, let branch 19.03 :) Apr 11 15:42:46 Hello guys!! and thank you for your great work. I am facing a situation where I need to create and destroy network interfaces upon some events, for a mobile connection. My program connects the modem and then invokes a script passing needed information via environment variables. What's the best way to hook up into openwrt and configure the system dynamically? I may have multiple connections at any Apr 11 15:42:53 time, and I may need to remove stale ones Apr 11 15:43:12 I mean - the script is invoked both when building and when tearing down interface; but I may need to clean up interfaces because sometimes I may not be invoked on teardown Apr 11 15:44:08 I was looking around the add_dynamic method of the network object, I guess by netifd Apr 11 15:46:49 Until now I managed the situation creating and destroying interfaces via UCI / ubus calls to reload network config / ifup / ifdown; I avoided committing changes to avoid useless writing to flash memory, but uci changes get accumulated over time, and this definitely doesn't scale Apr 11 15:47:01 is https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commit;h=2b9d3bd3655d0a3a1bf82a632d78af11e3484dfb a fair summary of the ar71xx/ath79 switch ? Apr 11 15:47:11 any recommendation would be greatly apreciated. And sorry for the long series of messages Apr 11 15:50:54 I will of course translate the commit message into something approximating English as well ;-) Apr 11 15:51:34 ynezz: building with your latest perf irq tweak... will let you know Apr 11 15:52:45 ldir: that ar71xx switch, lgtm Apr 11 15:54:59 ldir: ack Apr 11 15:56:56 I also tought about a protocol handler, but I don't know how well it may interact with my current case Apr 11 16:20:40 ynezz: ok so I've had a go with your latest perf irq patch - it works - I notice a change in behaviour in that the perf int is no longer listed in /proc/interrupts - I have no idea if this is actually a problem or not. Apr 11 16:56:05 gch981213: aaaahhhh that makes sense. will test Apr 11 16:57:03 in other news, github seems to be working poorly Apr 11 16:57:23 that's no news :P Apr 11 17:03:07 Hi. I'm looking into changing the default OPENSSL_ENGINE option. People are getting in trouble over it because it is not on by default, but libp11 forces it. Apr 11 17:03:27 So when you build openssl with default options, engine support is off. Buildbot-build packages, otoh, are built with engine support on. Apr 11 17:03:38 When a user tries to run a package installed from the repo, they get errors looking for engine symbols in the library. Apr 11 17:03:43 It was never an issue in 1.0.2 because the option to turn it off was mispelled, and engine support was always built. Apr 11 17:03:51 Engine support is mostly useful for systems with hw-crypto, so I would make the default y for those archs, and off for all others, Apr 11 17:03:56 and remove the + in +OPENSSL_ENGINE in libp11, so the default option will be respected. Package size increases from 1048 to 1064 for wrt3200acm. Apr 11 17:05:08 I was looking to use CONFIG_CRYPTO_HW to signal availability of hardware crypto, but I can't use it directly. Apr 11 17:05:13 grep -hr '^ARCH *:\?=' $(for f in $(egrep -rl '^CONFIG_CRYPTO_HW'); do dirname $f; done | sort -u) | sort -u | sed 's/ARCH *:\?=//g' Apr 11 17:05:16 returns aarch64 arm powerpc. Does this sound right? Is the size increase small enough to leave it on for all? Apr 11 17:09:35 cotequeiroz: yeah probably. openssl is not meant to be mbedtls. Apr 11 17:09:49 btw, mips also has support for hardware crypto Apr 11 17:10:17 it's not in the kernel config but can be selected by the specific kmod Apr 11 17:13:19 probably makes things slower though.... :) Apr 11 17:16:42 The problem with leaving it off is that the packages linked to just libcrypto (and not libssl, where default init enables all engines) may not be able to use hw crypto. Apr 11 17:18:19 just turn it off and never let it be turned on :) Apr 11 17:18:29 are there any platforms where it actually makes things faster? Apr 11 17:18:56 karlp: on ramips, AES goes from 8MB/s to ~40MB/s Apr 11 17:19:22 still slow but :) Apr 11 17:20:42 is that on a useful application or just openssl speed? Apr 11 17:20:53 cryptsetup benchmark Apr 11 17:40:24 If you want it off, all you have to do is not install the engine package. They are independent now. Apr 11 17:48:51 ldir: I would be suprised if that patch would be accepted in the current state, so lets try it and see, thanks for testing Apr 11 18:14:33 hello, is there a way to see the required deps for a local .ipk package? Apr 11 18:18:34 hmm - this is a bit odd /dev/sdb1 7302304 60472 6847848 1% /mnt/sdb1 Apr 11 18:19:43 *yawn* Apr 11 18:19:55 i'll be dreaming 11ax HE capab crap tonight Apr 11 18:20:05 i am thi f'in close to gettint a proper HE assoc done Apr 11 18:20:17 i am this f'in close to getting a proper HE assoc done Apr 11 18:20:48 * blogic notes that ofdma is actually awesome and wifi will actually work in 2-3 years Apr 11 18:21:17 nice, so no need to waste money and time on anything 11ax yet :) Apr 11 18:21:24 no Apr 11 18:21:28 give it 2-3 years Apr 11 18:21:35 tries again https://pastebin.com/h6DnWhMe Apr 11 18:22:09 i am drained, i am off to bed and its only 8:30 :-D Apr 11 18:22:17 We'll probably also need to wait until ath11k-ct comes out or the FW will always crash. Apr 11 18:22:27 not been this hyped on code in a while Apr 11 18:22:41 mamarley: its actually pretty stable Apr 11 18:22:55 Good! Apr 11 18:36:48 ldir: http://paste.debian.net/1077169/ might be worth adding as well Apr 11 18:55:27 pkgadd: don't know if I'm *that* brave Apr 11 18:55:55 ;) Apr 11 19:03:21 aww come on Apr 11 19:03:24 the hard part's already done ldir Apr 11 19:03:50 has been a while since we've seen a brit with balls on the world stage ;) Apr 11 19:06:11 in the include/kernel-version.mk there are multiple kernel versions defined, how is decided which one is used in the build? This is an example for the file: https://pastebin.com/raw/k08deHgB Apr 11 19:06:59 starlays_: that's decided by the individual targets, see e.g. target/linux/ath79/Makefile --> KERNEL_PATCHVER:=4.14 Apr 11 19:08:31 don't touch that starlays_ if you're still on your kernel/module mismatch problem Apr 11 19:08:42 Borromini: lol - I'll put it out to an advisory referendum ;-) Apr 11 19:08:49 :D Apr 11 19:09:26 Borromini, I have solved that problem, now trying to understand things a little bit more Apr 11 19:09:32 starlays_: KERNEL_PATCHVER is meant to switch between actual kernel releases (e.g. 4.4 -> 4.14 -> 4.14), not patchlevels (4.14.109 -> 4.14.111) Apr 11 19:09:36 starlays_: good :) Apr 11 19:09:56 did you get opkg to install a local package then? Apr 11 19:09:57 ok, and the patchlevels how are build? Apr 11 19:10:16 sorry? Apr 11 19:10:39 the patchlevels are defined in include/kernel-version.mk Apr 11 19:11:38 pkgadd: s/DEPRECATED/FIXES ONLY/ Apr 11 19:12:02 or that, yes Apr 11 19:12:23 starlays_: so if you were to bump a particular kernel, you'd do it there. however, due to the amount of patches openwrt applies, it's not recommended to just bump releases, you'd need to refresh the patches too Apr 11 19:12:49 regarding the opkg, yes I have deleted the mainstream cache for opkg from /tmp/ and after that retried manualy install and I have observed that the kernel mod had dependency comming from mainstream and where defined for 4.14.111, used the 4.14.109 version and all things have worked well, the router have survived after reboot and usb is mounted and accessible Apr 11 19:13:31 that's good Apr 11 19:13:45 if you want to bump a kernel yourself, use this script: https://github.com/KanjiMonster/maintainer-tools/blob/master/update_kernel.sh Apr 11 19:14:12 however, patches might break and need manual adjustment, so it really depends on how well you can read C code to fix stuff up. Apr 11 19:17:23 ok, basically the *.109, *.111 are the applied patched levels Apr 11 19:18:04 yeah Apr 11 19:18:07 I understand a little bit of C but I didn't did it until now Apr 11 19:18:26 at some point it becomes easier to just build your own image Apr 11 19:18:39 if you spend lots of time tinkering, installing stuff and adding scripts Apr 11 19:20:35 I'm still looking for a way to see the deps for local packages. It will be useful if I build it wrong to understand some things Apr 11 19:21:31 I hope that the opkg force option --noaction will help a little bit Apr 11 19:21:53 it would be nicer to see it like it is for mainstream, opkg info Apr 11 19:30:47 Can anyone point me to where the buildsystem translates the 'Depends:' information from tmp/.packageinfo to the 'Depends:' line in the package's 'control' file? Apr 11 19:36:28 cotequeiroz: it's probably all in include/package-ipkg.mk Apr 11 19:36:36 163 $$(call addfield,Depends,$$(Package/$(1)/DEPENDS) Apr 11 19:36:49 155 Package/$(1)/DEPENDS := $$(call mergelist,$$(foreach dep,$$(filter-out @%,$$(IDEPEND_$(1))),$$(dep)$$(call GetABISuffix,$$(dep)))) Apr 11 19:39:45 cccccchdgggnllbttdtjntfeieniljvtrrdbjr2tdujc Apr 11 19:39:46 Thanks. I'll take a look. Apr 11 19:41:51 mamarley: stintel: does you android phone support WPA3 or why do you think it does not work? Apr 11 19:42:28 I tested WPA3 only with wpa_supplicant of more or less the same version Apr 11 19:45:54 Hauke: Android Q apparently is supposed to officially introduce WPA3 support, and the phone sees the WAP as supporting WPA3, but the connection fails without any log output on either side. Apr 11 19:48:05 the tricky part which totally got me with WPA3 was using hex keys (which I normally do, remembering the early days of wlan when vendors interpreted passphrases differently - don't, just don't...), --> 128 hex values instead of 64 (and wpa_supplicant/2.7+git20190128+0c1e29f doesn't seem to accept a 128 hex value key either). I finally got it to work with ath5k on my notebook 2 days ago Apr 11 19:50:58 i read a statement (by an openwrt user on a dutch forum) that there would be an 18.06.3 release because of the wpa3 vulnerabilities - but 18.06 does not support WPA3 does it? Apr 11 19:51:07 only the hostapd in master does i thought? Apr 11 19:52:01 11:38:41 < jow> I consider pushing out 18.06.3 and 17.06.7 later this week, mainly to address the EAP weaknesses found in hostapd Apr 11 19:52:41 ah WPA2 is affected too then Apr 11 19:52:41 Borromini: the defects also apply to WPA2-EAP Apr 11 19:52:43 thanks ynezz Apr 11 19:52:45 and pkgadd Apr 11 19:53:14 so feel free to suggest patches which should be included Apr 11 19:54:14 ynezz: good point. but the dropbear bump won't be accepted i suppose :) Apr 11 19:54:36 the area these patches are changing has been under active development recently, so getting them to work on older hostapd sources gets progressively harder Apr 11 19:55:10 Borromini: well, there were some CVEs as well if I remember Apr 11 19:57:31 ynezz: quite possible, i have no idea though if the list of dropbear patches in 18.06 already remedies those. i've been running .77 for a while (and recently .78) on my local 18.06 branch Apr 11 20:01:20 I'm unsure about this one https://github.com/openwrt/openwrt/pull/1349#issuecomment-480397828 Apr 11 20:02:02 rather this comment https://github.com/openwrt/openwrt/pull/1349#issuecomment-475069391 Apr 11 20:04:33 the fact noone tested (or bothered to confirm?) Apr 11 20:04:56 from what i read it seems to make a lot of sense to remove ath10k from the c7 v1 image. Apr 11 20:05:24 i suppose (and hope) recovery has gotten better with recent tp-links but once your device starts bootlooping you have a problem on your hands :( Apr 11 20:06:14 ynezz: that patch with s/-ct// is obviously correct for the archer c7-v1, there is no way to make the ath10k/ 5 GHz radio work anyways, so in an ideal way (with ath10k getting fixed) it would just no longer crash, but never work (unless $user replaces the card themselves, in which case kmod-ath10k is just an opkg install away) Apr 11 20:07:12 wireless is disabled by default anyway so the fact ath10k is not there doesn't make it unusable Apr 11 20:07:16 there is no driver or firmware support for QCA9880-AR1A (as in the c7 v1) at all Apr 11 20:09:52 sorry, I don't have (any revisions of) the c7, so I can't test it - but the patch removing kmod-ath10k and ath10k-firmware-qca988x is obviously correct and can't break anything Apr 11 20:10:50 well, that's obvious, I'm just not 100% sure if it really fixes the bootloop Apr 11 20:11:14 don't have it neither, sorry Apr 11 20:11:34 maybe open a request on the forum? Apr 11 20:12:59 it's imho worth a shot, judging from the discussions it has a good chance of fixing it (and zero regression potential) Apr 11 20:13:58 yeah Apr 11 20:14:42 ok, so I'm adding your reviewed-by, so we can share the potential anger :) Apr 11 20:15:25 fine with me, Reviewed-by: Stefan Lippers-Hollmann Apr 11 20:15:31 thanks Apr 11 20:17:20 mamarley: I do not have an andoid Q capable device Apr 11 20:17:33 I will try to look into this in the next few days Apr 11 20:17:59 I hope someone else will fix this hostapd ;-) Apr 11 20:18:08 *in hostapd Apr 11 20:18:36 Hauke: For what it is worth, I would see the same behavior with Android 9 if I had both WPA3 and 802.11r turned on at the same time. Apr 11 20:19:00 I still need a new phone ;-) Apr 11 20:19:03 what's the cheapest phone with Android Q? Apr 11 20:19:17 but I also saw some prblem with 802.11r Apr 11 20:19:33 ynezz: Used OG Pixel, probably. I think only Pixels get the beta so far. Apr 11 20:20:08 I read windows will also support WPA3 soon Apr 11 20:20:50 mamarley: ah, didn't knew it's still in beta, thanks Apr 11 20:22:37 ynezz: a second hand xperia xz1 compact would work too Apr 11 20:22:53 my gf's on stock and got bumped to 9 quite a while ago. Apr 11 20:23:04 (i'm still on LOS 15.1 :-P ) Apr 11 20:24:04 Borromini: is the kernel new enough? that's the big question, I have LOS 16, but only kernel 3.4, so no dice with WPA Apr 11 20:24:07 3 Apr 11 20:24:36 Borromini: No, we are talking Android Q Beta, which would be Android 10. Apr 11 20:24:37 pkgadd: yeah i have 4.4 Apr 11 20:24:50 o.k., that should do it Apr 11 20:25:10 mamarley: sorry. was confused by the 9 and Q being used at the same time. 9 is P indeed. Apr 11 20:26:04 pkgadd: my nexus 7 tablet has an old kernel too, but it can kinda run 15.1. of course those kernels don't get bumps. Apr 11 20:26:45 the sad story of android development, yes (klte here) Apr 11 20:27:19 yeah. Apr 11 20:33:27 thanks guys! have a good night! Apr 11 20:33:31 sorry, disconnecting Apr 11 20:35:48 bed time for me too. Apr 11 20:48:00 mamarley: stintel: the first thing I would try is to use a git snapshot of hostapd and one without our patches Apr 11 22:06:45 nigth @all **** ENDING LOGGING AT Fri Apr 12 02:59:57 2019