**** BEGIN LOGGING AT Mon Apr 22 03:00:08 2019 Apr 22 06:03:22 olerem: ping Apr 22 06:09:08 mangix: pong Apr 22 06:09:41 i sent an email regarding ag71xx. anyway, when did you branch your work? Apr 22 06:11:01 mangix: branched in january, compared with ../ath79/.. state for two weeks Apr 22 06:12:21 ok that makes sense. my comments are basically these two commits: https://github.com/openwrt/openwrt/commit/f73b2d64ed56c1ba41c1ef906ef292d07f839def#diff-6f733a71ff3054a7dc0229c9320ad3e6 https://github.com/openwrt/openwrt/commit/96e0fa94c700fa95bb12bee1c0b75dfeea091b29#diff-6f733a71ff3054a7dc0229c9320ad3e6 Apr 22 06:14:08 mangix: ok. will do it Apr 22 06:21:37 olerem: out of curiosity, does the driver support hooking up to a switch using qca8k or something? Apr 22 06:23:38 mangix: currently I still use the old ag71xx_ar7240.c moved to drivers/phy/ Apr 22 06:26:16 mangix: currently i don't have enough resource to manage switch driver in this upstreaming round Apr 22 06:34:23 ok Apr 22 06:41:33 mangix: done. v3 is out Apr 22 06:52:14 hmm interesting. I thought free_netdev was unnecessary with devm_alloc_etherdev Apr 22 06:54:42 A while back I looked at QCA's patchset to see if anything useful could be taken. A few were. Probably more can: https://portland.source.codeaurora.org/patches/external/qca/qsdk-base-1.0.112.patch Apr 22 06:59:38 it is a huge diff Apr 22 06:59:50 yeah Apr 22 06:59:58 the interesting stuff happens at patch ~900 Apr 22 07:01:31 the most recent version of QCA's driver has the ring size set to 224. No idea why. Apr 22 07:01:47 last I checked I couldn't change using ethtool Apr 22 07:07:43 found their latest driver: https://source.codeaurora.org/quic/qsdk/oss/kernel/linux-msm/tree/drivers/net/ethernet/atheros?h=NHSS.ILQ.2.15 Apr 22 07:23:24 mangix: patches like this https://source.codeaurora.org/quic/qsdk/oss/kernel/linux-msm/commit/drivers/net/ethernet/atheros/ag71xx?h=NHSS.ILQ.2.15&id=7b3504ee3cea29500c86db080e02f7520f25501a Apr 22 07:24:20 make me feel, there is bigger problem then this Apr 22 08:18:53 Hello, wondering if the build system has issues with recent pre-req programs. I'm on Arch linux (new install) I cannot get the toolchain built. I try on a virtual machine ubuntu 19.04 and it builds Apr 22 08:19:03 Tried adjusting cores etc Apr 22 08:19:27 ubuntu 8 cores works fine, arch -j1 -> -j21 does nothing to help Apr 22 08:21:30 thagabe: you'd need to be a little more specific than 'cannot get it to build' Apr 22 08:22:15 Borromini I realize that, sorry Apr 22 08:22:29 It seems to be stopping around package m4 Apr 22 08:23:09 logs thagabe Apr 22 08:23:17 when i update the feeds i get errors for missing packages (libpam tho its installed on the host) Apr 22 08:23:31 ok give me a min Apr 22 08:43:50 thagabe: ? Apr 22 16:13:20 * ldir has at last switched to using his mir3g as main router...for the moment...until I consider the apu2 again. Apr 22 16:23:01 ldir: what went wrong? Apr 22 16:24:49 In a mach-based kernel (ar71xx), where is the SPI CS line generally defined? I'm not seeing it in `target/linux/ar71xx/files/arch/mips/ath79/mach-gl-ar750s.c` Apr 22 16:24:59 (from vendor code) Apr 22 16:26:00 nothing - I've had the mir3g sat there doing nothing for a long time. The archer c7 has done excellent service but is now a bit slow. in the process of considering an apu2 I pondered splitting out the 'wifi' bits from the 'routery' bits ready for APU2 and then thought in preparation for that I could use my mir3g as the router and convert the c7 to AP only operation. Apr 22 16:26:40 a dual core dual thread mips cpu is better than a single core single thread cpu :-) Apr 22 16:27:48 (I also haven't found anything obvious in the maze of symlinks around `/sys/devices/platform/ath79-spi/`) Apr 22 17:04:04 ldir: ok, but the wifi bits of the c7 are not that much better than the wifi bits of the mir3g Apr 22 17:05:45 the mir3g is a temporary install until the apu gets ordered/configured etc - I'm not using the wifi on the mir3g at all. Apr 22 17:38:34 ldir: overclock the c7 Apr 22 17:40:32 max comfortable clock is 1ghz Apr 22 17:41:06 5 days uptime so far (I sysupgrade too much) Apr 22 17:41:50 oh jeffsf left again. Apr 22 17:42:01 (mach-rbspi.c around line 650 has wha tthey're looking for...) Apr 22 17:42:16 ldir: i meant that there wasn't much point to deploying the c7 Apr 22 17:45:41 Does openwrt have exfat-support (usb)? Apr 22 17:52:10 hsp: there is a package for extfat, but I do not know if we ship it as a binary because you normaly need a patent agreement with microsoft to do so Apr 22 17:53:36 ok, thanks Apr 22 18:39:18 someone is reporting a splat related to cfg80211_calculate_bitrate, bottom of this: https://github.com/greearb/ath10k-ct/issues/61 Apr 22 18:39:51 looks to me like this might be user space and/or cfg80211/mac80211 issue, not ath10k-ct. Anyone debugged this yet? Apr 22 20:40:08 Hello, I wonder what I miss from enabling WPA3 (sae-mixed) support in openwrt with tplink wdr4300, using self-comipled from yesterday... I have "full" version of wpad but doens't help so far Apr 22 20:40:26 log shows "Mon Apr 22 23:35:12 2019 daemon.err hostapd: Line 69: invalid key_mgmt 'SAE'" so at least somehing doesn't work Apr 22 20:42:37 on another note, working psk2 config I see similar for both radios "Mon Apr 22 23:36:03 2019 daemon.notice netifd: radio0 (19893): sh: out of range" Apr 22 20:48:54 also... as looking now at menuconfig, diffirent wpad options are kind of confusingly arranged... first there is wpad (full), then "under" it as tree is wpad-mini, under that is wpad-openssl (full), but if it is full, why is it under mini, and not wpad alone? :) Apr 22 20:49:15 yeah, that's appeared broken for a while now. Apr 22 20:49:37 ie is the wpad-openssl full version using openssl, or some mini variant, and how does it differ from wpad (lone) Apr 22 20:50:15 ..and/or do I need specifically for example openssl version for wpa3 or something? (well would e 'easy' as I do have openssl anyways included for other reasons Apr 22 20:53:06 so... for now I think I have 2 separate questions: what wpad variant(s) I need for sae-mixed (and is there somethign else needed?), and then second, is wpad-openssl really full with using openssl, or some mini variant? :) Apr 22 21:08:40 olmari: https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=package/network/services/hostapd/Makefile;h=235f178bc9776956109a2597e7176b7437509fc2;hb=HEAD#l228 is the answers, I don't know how to fix the tree view in menuconfig though... Apr 22 21:11:32 Me neither, sadly :) Apr 22 21:18:31 olmari: I'll take a jab at it. I've been messing with the package metadata generator, and I might as well take a look at that. Hopefully I can at least answer why it does it that way. Apr 22 21:21:11 Yeah... Funnily enough I don't see really how menuconfig "builds" the tree how it does for wpad and such from that file linked :) Also I can't tell how it should really look, but I guess any "variant" as it's own top selection and if add-in options, then tree branches.. or something... =) Apr 22 21:33:44 Allright at least wpad-openssl seems to enable wpa3 too, (so that answers that questiond definately :) ) Apr 22 21:36:29 olmari: wpad-openssl or wpad-wolfssl are required for wpa3, make sure to use passprhases instead of hex keys (that drove me to insanity) Apr 22 21:36:57 hostapd needs one of those two crypto libraries for sae Apr 22 21:38:31 pkgadd: yeah... I gathered that after asking here (well the wpad-openssl one), hexkeys, aren't those really ever used in WEP? :P (well talking about WPA PSK style and not enterprise, with any WPA version ) Apr 22 21:40:24 I've always used hex keys for wpa2 so far, remembering interoperability problems between the way different interpreted the passphrase in the past --> wpa_passphrase foo barbazfoo Apr 22 21:41:24 but WPA2 and WPA3 need different lenghts (and pure wpa3 networks aren't possible for me yet, aside from quick testing) - and hostapd seems to reject the hex form for sae alltogether Apr 22 21:48:10 wpa2 isn't going away anytime soon, but I do want to support wpa3 when possible, and fortunately there is sae-mixed :) Apr 22 21:54:12 pkgadd: If I remember correctly, for the way that SAE works, hex keys won't be supported. That might be the reason hostapt is rejecting it. Apr 22 21:56:01 luaraneda: at least doing functional testing it looks that way, but that issue really drove me mad - especially as I couldn't be confident that the client side was WPA3 compatible in the first place ;) Apr 22 21:56:24 Hey! I am an aspiring student for GSoC 2019. And a part of my project is to create a LuCI module. I created basic modules using vim on my router. But moving forward, I'd like to do things faster. How can I achieve that? Any suggestions for the dev setup?? Apr 22 21:57:03 only testing a proprietary firmware/ webinterface made me realize that being a potential issue - and it was Apr 22 21:57:42 ..still need to check my various old WLAN cards for compatibility (ipw2200 doesn't seem to cope) Apr 22 22:05:46 pkgadd: fun fact: iphone refuses to connect if WPA3 is enabled. I'm talking about mixed WPA2/3 Apr 22 22:06:14 might be fixed with 12.2 but i refuse to give up my jailbreak Apr 22 22:08:53 fortunately there are no iphones around, just non-WPA3 compatible androids, so they will need mixed mode till their untimely demise (kernel too old, so little chance that future lineageos updates will retrofit WPA3 support) Apr 22 22:09:55 another fun fact, iOS 6 and older fail to connect when running 802.11w optional Apr 22 22:10:01 actually no Apr 22 22:10:16 they fail if the authentication is set to SHA1/SHA256 mixed Apr 22 22:10:31 plain SHA1 works Apr 22 22:10:51 gotta love broadcom for their broken drivers Apr 22 22:11:40 the android broadcom drivers don't seem to be that bad Apr 22 22:29:31 pkgadd: right but they have to use wpa_supplicant. Apple has some weird integration Apr 22 22:41:48 mangix: are macOS and iOS using wpa_supplicant? Apr 22 22:42:07 i do not believe so Apr 22 22:42:26 at least it didn't last i used it (Tiger) Apr 22 22:54:37 is there a generic way of fetching dhcp leases within openwrt? if so, does this work with dnsmasq as well as odhcpd? Apr 22 23:02:11 Hauke: no Apr 22 23:29:00 mirko: i do know that luci does exactly that Apr 22 23:49:51 DonkeyHotei: i was hoping for an ubus interface or alike, but luci itself also only just parses /tmp/dhcp.leases Apr 22 23:55:38 luci parses the odhcpd file too Apr 23 00:09:43 DonkeyHotei: saw it, so no generic API **** ENDING LOGGING AT Tue Apr 23 02:59:57 2019