**** BEGIN LOGGING AT Sat Jan 05 02:59:56 2019 **** BEGIN LOGGING AT Sat Jan 05 06:02:09 2019 Jan 05 06:55:45 mangix: some Jan 05 06:56:36 here's a good place to start: https://www.wireguard.com/quickstart/ Jan 05 07:00:18 i ask since i couldn't figure out how to set it up Jan 05 07:00:31 i'm guessing the interface needs to be in br-lan Jan 05 08:12:33 mangix: https://openwrt.org/docs/guide-user/services/vpn/wireguard Jan 05 08:12:54 i would guess it has its own interface Jan 05 08:15:53 https://forum.openwrt.org/search?q=wireguard Jan 05 22:08:14 build #1174 of ar71xx/nand is complete: Failure [failed targetupload] Build details are at http://phase1.builds.lede-project.org/builders/ar71xx%2Fnand/builds/1174 blamelist: Anton Arapov Jan 06 00:11:19 greearb: hi Ben, perhaps not related directly to CT, but I was wondering if your NDA also includes access to proprietary wifi driver (ie one used in qsdk given to various vendors) Jan 06 00:12:38 as I was researching the most suitable wifi module to buy for an x86 build, I saw some users reporting very mediocre performance on Qca9994 (from compex), ie only pairing on highrates if in line of sight, and speed not being very high Jan 06 00:13:33 this same module is used in 'enterprise' APs costing $1000+ , like aruba and friends, and people report very good rates/performance Jan 06 00:14:10 is this because of some fancy things aruba does on their modules/amps or is the proprietary driver able to reach those speeds compared to ath10k? Jan 06 00:16:07 I didn't know before that the Philips Hue gateway 2.0 uses OpenWrt 15.05 with ath9k and 3 tls libraries Jan 06 00:16:52 did someone already installed a recent OpenWrt on it? Jan 06 00:21:22 greearb: https://www.compex.com.sg/forum/t/5#Post354 Jan 06 00:21:29 abenz: I can compare QCA9984 with vendor firmware and OpenWrt/ ath10k against each other, in terms of range there is no difference. there is a difference when it comes to maximum throughput (you can easily push your system load to maximum), which probably comes down to the nss offloading Jan 06 00:22:20 pkgadd: 9984 and 9994 seem to be identical, 9994 is only rated for higher operating temps and such Jan 06 00:22:45 abenz: in general QCA9984 works very well for me (using OpenWrt/ master) Jan 06 00:23:01 pkgadd: so you're on CT firmware/ath10k-ct then? Jan 06 00:24:47 abenz: planned, but not yet - due to the wds 4addr regression in mac80211, I first wanted to keep the number of moving parts to a minimum (by doing my regression tests with ath10k; but it's a stack issue, so both ath10k and ath10k-ct are affected alike). Jan 06 00:26:02 pkgadd: I was under this impression, that our boxes (eg IPQxxxx) are unable to realize the full potential of the wifi modules because the main ARM cores are not fast enough, ie the whole SoC is built with NSS offloading in mind.. but if one were to purchase eg QCA9994 and place it in a more powerful system/x86, then its full potential will be realized Jan 06 00:26:29 that should be the case, yes Jan 06 00:26:32 but seeing that thread on compex forum makes me rethink my assumption, it could be the driver. not a cpu bottleneck Jan 06 00:26:52 https://www.compex.com.sg/forum/t/5#Post354 Jan 06 00:27:12 I've seen ~45 MByte/s thoughput (wget/ httpd) on ipq8065/ QCA9984, with both cores loaded at 100% Jan 06 00:27:27 that's with OpenWrt, so no NSS Jan 06 00:28:23 pkgadd: another thing is 9984 seems to receive firmware updates from time to time, most recent less than a month ago.. while 9980 (one in my archer c2600) last got an update in 2015 Jan 06 00:28:52 I think that was the initial release, so it was never updated.. have to rely on CT firmware Jan 06 00:29:35 pkgadd: are you currently using the patch from the mailing list for the 4addr regression? Jan 06 00:30:05 abenz: yep, I'm currently running firmware ver 10.4-3.9.0.1-00008 api 5 features no-p2p,mfp,peer-flow-ctrl,btcoex-param,allows-mesh-bcast,no-ps crc32 7eed6f20 Jan 06 00:30:24 nice feature set, mine only says: Jan 06 00:30:58 Hauke: I'm currently using the proposed patch from linux-wireless, the one keithwky suggested in https://github.com/openwrt/openwrt/pull/1640 --> https://gist.github.com/keithwky/23fdff907c23df57d35ee052fbd44cf6 Jan 06 00:31:30 but both the revert and the proposed fix (which is supposed to be pending a v2 submission) are working equally well Jan 06 00:31:43 ok thanks Jan 06 00:31:43 firmware ver 10.4.1.00030-1 api 5 features no-p2p crc32 d2901e01 Jan 06 00:32:44 pkgadd: the pull reqeust still contains the revert Jan 06 00:32:49 abenz: I've been tempted by an ipq8064 device with QCA9980, pretty cheap, but only a single firmware update for QCA9980 ever... so I really don't know yet if I'm going to buy it Jan 06 00:33:55 Hauke: yes, I didn't want to change it, to keep both options on the table - either the revert or the fix as provided by keithwky Jan 06 00:34:02 I can update it, if you like Jan 06 00:35:44 pkgadd: yes seems 99x0 is EOL Jan 06 00:38:38 well, the price makes it attractive Jan 06 00:39:54 just looking for something with ipq806x I can 'break' without affecting my network ;) Jan 06 00:41:44 meaning perfect wlan support isn't the top checkbox to cross out (actually ipq8064+quantenna would also do the job, if the price would be accordingly) Jan 06 00:45:23 pkgadd: I suppose its a good buy if the price is right Jan 06 00:46:16 someone on forum was able to compile NSS support (for lede 17.01 though), with NSS cores enabled it can do wire speed with main cores free to do other tasks Jan 06 00:49:17 it's interesting to follow, but fortunately not that critical for me (WAN is limited to 100/40 MBit/s anyways and my wlan throughput is good enough for my needs as well, the system that profit most are wired anyways) Jan 06 00:51:28 Hmm, no ath79 port for the UAP-Pro yet? Jan 06 00:53:27 apparently QCA is still focussed on NSS cores for ipq807x as well (ipq40xx seems not to have them) Jan 06 00:54:44 pkgadd: i don't have an ipq40xx device but from the commit message adding support for ipq4019, it seems they can do 550mbps on 5ghz (when irq is distrubted on all 4 cores) Jan 06 00:54:57 which is pretty good given the typical price of one of those Jan 06 00:56:03 I unfortunately didn't test it with iperf, the figures should be higher with that than my quick and simple wget/ http test **** ENDING LOGGING AT Sun Jan 06 02:59:57 2019