**** BEGIN LOGGING AT Fri Apr 26 02:59:56 2019 Apr 26 05:24:55 . Apr 26 05:25:20 good yawning Apr 26 05:26:33 anyone else have the problem that you start the day with 2 terminal windows and when you get to the screen next morning there are ~20 left over from the day before ? Apr 26 05:36:46 o/ Apr 26 05:43:19 blogic: ;) Apr 26 05:54:48 jow: you only replied to contact@ and i dont have the prior mail in my inbox Apr 26 05:57:32 blogic: it was one of the usual "we're a chinese brand agency and our client wants to register openwrt as brand and domain name, did you authorize that" Apr 26 05:57:45 ok Apr 26 05:57:50 well you replied to us only Apr 26 05:57:52 I usually ignored these but figured it couldn't hurt to reply. Not that I expect it to have any impact Apr 26 05:58:28 no I replied with contact@ in cc Apr 26 05:58:47 to: contact@ ?! Apr 26 05:59:15 To: Arno Liang / Cc: General OpenWrt Contact Apr 26 05:59:31 showed up different here... anyhow Apr 26 06:00:52 was probably rewritten Apr 26 07:11:31 blogic: to me it happens for browser tabs Apr 26 07:12:02 blogic: i clean my tabs, going down to 20, next day I;m baack to 50 Apr 26 07:13:36 rmilecki: that aswell Apr 26 07:14:10 yeah usually browser tabs here, terminal tabs not so much as I tend to start closing once I reach > 10 Apr 26 07:14:52 i start cleaning browser tabs when I cannot see favicons anymore Apr 26 07:15:30 I am happy to see i am not the only one .... Apr 26 07:16:16 multitasking ... still screwing up several things at once in 2019 ;) Apr 26 08:01:12 jow: any opinion on this? http://downloads.openwrt.org/snapshots/faillogs/aarch64_cortex-a53/packages/libxerces-c/compile.txt Apr 26 08:02:02 $(FPIC) defaults to fPIC for PPC only. not sure if also applicable here Apr 26 09:39:00 * aparcar[m] sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/BHZgTqvQpIIkSqaPPPYkWOhO > Apr 26 10:14:38 https://iot.mozilla.org/ OpenWrt Inside (TM) Apr 26 10:18:55 npm install openwrt Apr 26 10:19:38 :-) Apr 26 10:20:01 opkg install npm; npm install openwrt; Apr 26 10:41:52 that retrojeff guy. sigh Apr 26 10:42:39 there, I fixed it Apr 26 12:36:54 could sombody please mirror ucert.git and procd.git to GitHub? Apr 26 12:39:34 blogic you maybe? Apr 26 12:41:41 why? Apr 26 12:42:53 so I can create PRs Apr 26 12:43:18 that implies that the maintainers actually want to deal with PRs Apr 26 12:43:52 and since its just a mirror, the effort to merge these PRs will be even higher Apr 26 12:45:10 Github brings the peanut gallery Apr 26 12:47:05 Monkeh strong statement Apr 26 12:47:24 jow I'll send mails send Apr 26 12:47:33 *then Apr 26 12:49:50 is here someonen from the ToH crew? Apr 26 12:52:37 aparcar[m]: best is to get touch with alberto or tmomas in the forum Apr 26 12:52:59 aparcar[m]: afair they're not available via irc Apr 26 12:53:11 jow: great, thank you! Apr 26 13:35:11 BTW, what does it mean "peanut gallery", I've never read it before ? :) Apr 26 13:36:44 people who are perhaps not directly stakeholders, but commenting anyway. Apr 26 13:37:03 sitting in the seats around a show, but not actually particpating in it. Apr 26 13:52:53 karlp: ah, thanks Apr 26 14:05:02 Also the cheap seats, usually meaning low quality occupants ;) Apr 26 14:06:03 is anyone else facing problems building the current git master? Apr 26 14:22:23 aparcar[m]: I just built 9424b6f998917e2926c0b7afb8d7a968590da335 x86/64 fine Apr 26 14:23:22 ah but there's 1 new commit. anyway, pastebin your error somewhere Apr 26 14:24:08 https://paste2.org/aBnEdM1F Apr 26 14:24:14 might as well flash my image now that I built one :) Apr 26 14:24:17 19d uptime Apr 26 14:25:06 aparcar[m]: can you post /home/a/src/openwrt/build_dir/toolchain-x86_64_gcc-7.4.0_musl/gcc-7.4.0-initial/config.log ? Apr 26 14:29:25 stintel I just removed the build_dir folder and let it rerun, due to the musl update I guess some recompiling is required anyway Apr 26 14:52:57 Does anyone know what/where enables init.d scripts in the build system? rc.d ends up with symbolic-link to init.d already. Apr 26 14:58:03 stintel please see the paste here http://sprunge.us/htNydK Apr 26 14:58:08 still doesn't work Apr 26 15:03:01 Figured it out it is in package/Makefile target $(curdir)/install Apr 26 15:09:00 stintel any idea? I have no idea at all Apr 26 15:09:31 aparcar[m]: I was just looking at the config.log but unfortunately I have no idea either Apr 26 15:10:00 there is "g++: error: unrecognized command line option '-V'" and "g++: error: unrecognized command line option '-qversion'; did you mean '--version'?" but it's not clear to me if this is the actual problem Apr 26 16:06:49 stintel git clone openwrt.git did the trick :P Apr 26 16:41:01 i have two mini pic-e wlan cards QCA9984 4x4:4, an+ac and another QCA9984 4x4:4, bgn. is there a board i can use both cards in with openwrt? Apr 26 16:42:10 PCEngines APU2 Apr 26 16:44:17 Is there an easy way to build the MTD tests like `flash_speed`, `nandpagetest`, `nandbiterrs`, ... I didn't see them in `build_dir/target...` Apr 26 16:44:39 ref: Apr 26 16:45:24 Slimey: not the APU2 afaik Apr 26 16:45:45 only 1 is suitable for pcie wlan cards Apr 26 16:45:55 the other two are usb for modems etc Apr 26 16:46:54 no, they have one m-SATA and two mini-pcie: https://pcengines.ch/apu2.htm Apr 26 16:47:27 but with two 4x4 cards, you would have to drill holes in the case to mount all the antennas :) Apr 26 16:47:33 (or find a suitable case) Apr 26 16:48:25 abenz: it's the APU3 that has limitations on the mini-PCIE ports: https://pcengines.ch/apu3c4.htm Apr 26 16:48:40 Apr 26 16:48:48 3 LAN, 6 SMA, USB, red anodized or painted (two SMA on the back side, two each on the left and right side of the cover). Apr 26 16:50:19 oh nice, didn't know this one Apr 26 16:51:37 Been in the back of my head for a while -- but a 1/4" or equivalent drill works too -- never been able to justify the cost of an SMA-specific chassis punch Apr 26 16:57:32 I see. thanks for the info Apr 26 17:38:37 not long ago compex (manufacturer of the pcie wireless nics people with APUx usually buy) setup a forum on their website Apr 26 17:39:05 in it I read a few users who complained about poor range of compex cards Apr 26 17:39:36 could be isolated.. but I was wondering if any here can share their experience Apr 26 17:39:57 eg an ac class card, compared to, say, R7800 or archer c2600 Apr 26 17:39:59 range wise Apr 26 17:40:58 A lot may have to do with antenna and feedline choice. I don't have a lot of confidence in "eBay" antennas, and none in eBay feedlines, even the 5 or 10 cm ones Apr 26 17:41:25 Easy to lose 3 dB or more over an OEM setup through poor choices there Apr 26 17:42:32 aparcar[m]: why would we want to ? Apr 26 17:42:57 ah you jow asked already Apr 26 17:43:31 jeffsf: what would you use? Apr 26 17:43:39 do you have a compex card? Apr 26 17:43:54 No, though I do use Sierra Wireless LTE cards Apr 26 17:46:06 I order my pigtails through Mouser here in the US. I've ordered 900 MHz antennas from an eBay seller that focuses on electronics, actually had specs and a long, positive seller history Apr 26 17:46:41 The antennas don't suck at 900 MHz, but I can't say anything about their 2.4 and 5 GHz offerings Apr 26 17:48:55 I think it was "rfsuppliers" on eBay -- again, can't comment on their 802.11 antennas as I haven't bought any of them. Apr 26 17:49:36 I will have to look further into it Apr 26 17:49:38 L-Com makes some nice gain antennas. Their patch antennas did a great job for me back in the WRT54G days on 2.4 GHz Apr 26 17:49:45 I was thinking perhaps its due to cheaper construction Apr 26 17:49:47 They aren't cheap though! Apr 26 17:49:56 board/component wise that is Apr 26 17:50:07 (L-Com, no idea on the Compex boards) Apr 26 17:50:39 The FCC cert info on the Compex cards will at least tell you power output. Receive chain is another question Apr 26 17:51:01 eg, see this compex qca9994 (highest end in ac wave2): https://ae01.alicdn.com/kf/HTB1PErJf3nH8KJjSspcq6z3QFXap/JINYUSHI-For-Brand-new-Compex-WLE1216V5-20-Qualcomm-QCA9994-chip-1733Mbps-4x4-Wave-2-802-11ac.jpg_640x640.jpg Apr 26 17:51:25 and this is another card based on the same chipset: Apr 26 17:51:32 https://aruba.i.lithium.com/t5/image/serverpage/image-id/6671iCDCDFD7805F93A57/image-size/original?v=mpbl-1&px=-1 Apr 26 17:51:42 Apr 26 17:51:47 granted the second is from an enterprise AP, but the difference is vast Apr 26 17:52:20 both 4 chains Apr 26 17:52:50 Remember that the high-power cards may not be able to get what they need from a standard slot. Some require 5 V, all require significant current Apr 26 17:53:53 If you're putting out 4x1W, the card needs to get 4W + more, so you're talking 1 A or more on 5 V, perhaps 2 A on 3.3 V Apr 26 17:54:02 abenz:the aruba one is broadcom Apr 26 17:54:05 this would result in the card not operating wouldn't it? rather than it operating with lower performance Apr 26 17:54:31 so you shoujldn't even try and compare them Apr 26 17:54:37 karlp: ops, wrong pic. hold on Apr 26 17:54:39 it's also a pcb design from 2013... Apr 26 17:54:52 Three options; (a) you get lucky, (b) the card sort of works, (c) you fry the regulator on your motherboard Apr 26 17:54:53 no the 9994 based designs are similar Apr 26 17:55:53 Apr 26 17:56:26 3.3 V, 8.5 W ~ 2.6 A Apr 26 17:56:54 (That's their high-end, Wave 2 card) Apr 26 17:57:15 karlp: https://fccid.io/Q9DAPIN0324325/Internal-Photos/Internal-Photos-2682059 Apr 26 17:57:42 Which may not fit in/on any specific slot, due to the heatsink Apr 26 17:58:18 karlp: the ap 325 radios are qca, they look even more complex than the broadcom Apr 26 17:58:35 Hence why I chose to use my aging Archer C7v2 units, or my new EA8300s as APs Apr 26 17:58:48 jeffsf: I'm essentially trying to build a compact system that can break 50MBps Apr 26 17:59:32 My EA8300s can do that out of the box ;) Apr 26 17:59:33 the R7800 and friends are bottlenecked by their cpu Apr 26 18:00:33 Use the x86 for processing, the bridging of wireless to Ethernet is not a big challenge for the current flock of ARM-based all-in-ones. Apr 26 18:01:00 wikidevi seems to be down Apr 26 18:01:07 jeffsf: is that based on ipq40xx ? Apr 26 18:01:21 The single-core MIPS units tend to bind up around 400-500 mpbs over Ethernet Apr 26 18:01:46 EA8300, yes, IPQ4019 + third 5 GHz radio on QCA9888 Apr 26 18:01:53 I have a B1300, which is based on ipq40xx, I get max 500mbps (not consistent, those are peaks) Apr 26 18:03:01 I think the current ethernet driver for this platform is taxing as well, hopefully with dsa it will be able to squeeze a bit more Apr 26 18:04:04 see an interesting observation I noticed when I was trying to reach the b1300 limits, was that streaming a 4k file to a tv connected via LAN cable, would result in stuttering. After I connect the TV to 5G on the same router, I can stream the same file without any stuttering Apr 26 18:04:14 minidlna is running on the router itself Apr 26 18:04:16 There are some "unique" features to the way that the switch and interfaces interact on the IPQ40xx -- OpenWrt typically "ties" one of the two interfaces to the WAN port, and the other one to the LAN ports Apr 26 18:05:07 As long as you're aware of that, you might be able to bond the two if you can really get more than 500 mbps from any real-world client in a real-world situation Apr 26 18:05:34 jeffsf: bond the two? can you elaborate Apr 26 18:05:38 I don't have any IPQ8xxx devices, so I can't comment on their architecture Apr 26 18:06:41 I do, while generally good, ipq40xx actually performs better in many scenarios (with openwrt of course). The ipq80xx are impotent due to the NSS cores being disabled.. on stock FW those two cores do all the heavy lifting Apr 26 18:06:56 Bond: Use Linux to pass one logical stream of Ethernet packets over two physical interfaces and cables. Requires bonded interfaces on the clients (switches, other boxes, ...) Apr 26 18:07:22 Bonding itself is "standard" so a Linux bond should be able to interoperate with a Cisco bond or a FreeBSD bond Apr 26 18:07:48 (though each end may choose how to send "the next frame" differently) Apr 26 18:08:07 I thought you were talking about something else, namely cpu port or so to be able to provide more bandwidth to the wifi chip Apr 26 18:08:52 No magic there, and I don't think a bottleneck, as the chips are typically on a high-speed internal bus or on PCIe these days Apr 26 18:09:15 what kind of speeds do you get on your ea8300 ? Apr 26 18:09:24 running openwrt or stock? Apr 26 18:09:38 I'm asking about wifi 5g specifically Apr 26 18:10:08 500 mbps tested in both directions, not simultaneous. I'm more interested in robustness and latency with low or moderate signal levels Apr 26 18:10:30 Using OpenWrt from the related PR Apr 26 18:10:38 I see. similar numbers here Apr 26 18:10:38 so this should prob work http://www.mini-box.com/ALIX-APU-2D2-AMD-G-Series-GX-412TC?sc=8&category=1361 Apr 26 18:11:05 although the b1300 hovers around ~430, just peaks at 500mbps now and then Apr 26 18:11:11 could be my client too Apr 26 18:11:32 Was doing testing to be reasonably confident that I had the right pre-cal/board data before I submitted the PR Apr 26 18:11:50 Client for my tests was a 2018 MacBook Pro Apr 26 18:12:21 I haven't tested EA8300 <=> EA8300 yet -- I wanted a three-radio device for wireless backhaul Apr 26 18:13:18 jeffsf: you are the person who sent emails to ML, re dsa on ipq40xx? Apr 26 18:13:42 Yes, I poked into that for a long while and decided it wasn't quite ready for prime time for my applicaiton Apr 26 18:14:40 I'll come back to it once the patches around DSA stop flying at the linux-netdev list! Apr 26 18:14:41 I see. I was following that thread with great anticipation :) Apr 26 18:16:46 I traded that for spi-nand on ath79 -- at least that's upstream on Linux 5.x ;) Apr 26 18:17:43 :) Apr 26 18:31:05 heh speaking of ath79, anyone want to convert a ar71xx device to it? :P Apr 26 19:12:24 On a package, the mtd-utils binaries I want are being built and in `build_dir/target-mips_24kc_musl/mtd-utils-2.0.2/` Apr 26 19:14:05 have been added to the list in $(INSTALL_BIN) $(PKG_INSTALL_DIR/usr/sbin/{}) $(1)/usr/sbin/ Apr 26 19:14:14 but aren't in the ipk for that install Apr 26 19:14:59 I didn't see anything in `patches/` -- is it likely something in the OpenWrt build system, or in the upstream Makefile? Apr 26 19:15:50 (typo a couple lines up, the closing paren is after _DIR) Apr 26 19:18:51 It's as if I'm missing a "packing list" for `./ipkg-install/` somewhere Apr 26 19:43:01 jeffsf: Are the files you're trying to copy, already installed in ipkg-install? Apr 26 19:43:26 No, they're not getting copied into ipkg-install -- just "hanging out" in the build dir Apr 26 19:44:26 There's a huge list of files (past those specified in the package Makefile) that get passed to `/bin/bash ./libtool --mode=install /usr/bin/install -c` Apr 26 19:44:49 But I haven't figured out where that list of files comes from yet Apr 26 19:44:54 Then you can't use $(PKG_INSTALL_DIR), as it points to ipkg-install. You must get them from $(PKG_BUILD_DIR) Apr 26 19:46:03 OK, that's a good workaround -- probably not a good idea to release a utility designed to burn out flash to the masses ;) Apr 26 19:46:30 (and most of the useful ones probably only matter to devs) Apr 26 20:29:19 cotequeiroz: Thanks, looks like that worked Apr 26 20:47:52 :) Apr 26 20:59:41 my WG is set to forward all traffic thru VPN. VPN server is using dynamic dns. A cold restart will have no connectivity because WG adjusts routing table before resolving the dynamic dns and establishing a connection Apr 26 21:00:19 I will disable wireguard autostart, and put a script in /e/hotplug.d/iface to watch for when my wan is up, start WG Apr 26 21:00:45 I was wondering if there's a more elegant solution or if I'm missing something obvious Apr 26 21:02:52 abenz: I'd say that would even be an elegant spot-solution to that problem, ought WG have no other means detecting changes and/or configurations (not that I know about WG much at all) Apr 26 21:04:41 olmari: will do that then Apr 26 21:05:30 Someone else might have better answer.. mine point was just that your suggestion didn't sound that bad in general :) Apr 26 23:25:40 <_jf> Earlier today we were testing openwrt on the new TPLink WR940N v6.1. We noticed that a small bug in the network config of the failsafe mode: the ip is configured on the wan interface instead of the lan interface. Apr 26 23:26:49 <_jf> To fix this, we had to append "tl-wr940n-v6" to the list at the start of "target/linux/ar71xx/base-files/lib/preinit/05_set_preinit_iface_ar71xx" Apr 26 23:27:20 <_jf> Could someone bring this fix upstream? Apr 26 23:27:51 <_jf> (tested with WR940N v6 and v6.1) Apr 26 23:30:11 _jf: it's not too likely that anyone in charge notices that at this time of day, considering the usual time zones of the persons involved. Ideally you submit a patch https://openwrt.org/submitting-patches or at least file it as a bug https://bugs.openwrt.org/ so it both gets documented and receives attention of the relevant developers Apr 26 23:52:47 upstream has responded. missing feature **** ENDING LOGGING AT Sat Apr 27 02:59:56 2019