**** BEGIN LOGGING AT Fri Jun 29 03:00:00 2018 Jun 29 03:15:12 idk whats happening the script i downloaded seems to be compiling every single lede module ever, jwh Jun 29 03:17:59 yea it was building world Jun 29 04:01:13 helo Jun 29 04:01:47 is there any timetable for the next release (18)? Jun 29 04:02:06 when its ready Jun 29 04:02:08 mo Jun 29 04:02:55 i always hate that answer... Jun 29 04:05:06 Oh i didn't even realize there was an 18 rc Jun 29 04:30:10 apparnetly im still compliling the entirely of the openwrt toolchain Jun 29 05:16:03 rpifan: did you use a config.seed with CONFIG_ALL=y ? Jun 29 05:20:28 nop i jsut used make menuconfig and i followed the wiki Jun 29 05:20:29 in any case Jun 29 05:20:33 it has fnished compiling Jun 29 05:20:37 where od i find my .bin now Jun 29 05:23:45 in bin/ Jun 29 05:24:43 well o/ to you too Jun 29 05:25:25 theres two differnet .bin i imagein its the one that says sysupgrade but what are the other two for Jun 29 05:25:53 probably factory and initramfs or so Jun 29 05:26:35 heres u image and vmliunx Jun 29 07:15:12 My build failed ran out of space. I think I have to get rid of a 64 gig ssd. Jun 29 07:36:31 jow, Ff00b4r0: ping Jun 29 07:36:38 I have information about the slowdown of defconfig that Ff00b4r0 discovered Jun 29 07:36:55 I did a binary search (bisect) to find the commit Jun 29 07:37:03 The culpable is 3ce1158 (2018-06-26, "build: download code from github using archive API") Jun 29 07:37:10 which was committed by Yousong Zhou Jun 29 07:37:31 Since then, the time to execute the following commands has increased by more than 6 times (I have an SSD and 12 GB of RAM) Jun 29 07:37:33 ./scripts/feeds update -a Jun 29 07:37:34 ./scripts/feeds install -a Jun 29 07:37:35 make defconfig Jun 29 07:38:15 The three commands completed in 0m42.178s before, and in 4m39.706s after the commit Jun 29 07:38:28 It seems most of the time is spend on "Collecting package info: " Jun 29 07:38:29 luarenda:Yousong Zhou restored the old behavior https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=e15565a01cea6a61c537f5fb91c4e4d27fb7b290 Jun 29 07:38:47 I've measured each time three times, you can fine more details on https://pastebin.com/MLQL9PR9 Jun 29 07:39:16 Ah, ok Jun 29 07:40:38 I miss that, because it was committed less than two hours ago Jun 29 07:40:39 missed* Jun 29 07:47:38 I just tested, to be sure, and the time is back to normal (fast) Jun 29 07:49:07 is it possible to set a login-cookie with Luci (rember login)? Jun 29 07:51:07 *remember Jun 29 07:52:10 hsp: luci already set a login cookie Jun 29 07:53:15 hsp: the problem is the session timeout on the router side; you can increase that in /etc/config/luci Jun 29 07:53:19 it defaults to 3600 seconds Jun 29 07:54:21 jow, thx Jun 29 07:54:53 i have reject the cookie :( Jun 29 07:56:48 i cannot found the cookie in firefox Jun 29 07:59:00 it is called sysauth Jun 29 08:02:29 not found, any plugin needed? Jun 29 08:49:39 Hi who was it that sent me the [PATCH] mvebu: Switch Linksys WRT series devices to DSA Jun 29 08:49:59 I am sorry but I cant do a build so cant test. Jun 29 08:50:21 So if any one would like to have a crack at it then: Jun 29 08:50:25 https://gist.github.com/neheb/325f44a8e71fa6fe9250f0dbbfadb3f1 Jun 29 08:51:57 I don't meen to be rood but I have forgot the nic of the person that sent it to me. Jun 29 08:52:01 sorry Jun 29 10:16:52 Tapper: it'll be the man of many names mangix, neheb or penev ;-) Jun 29 10:27:11 dwmw2_gone: I still maintain that in a single file compilation, if there's a function labelled inline, the idea that gcc would just go, "nah, I'll assume it's external" is _absurd_ regardless of gcc claiming that it's "works as designed" Jun 29 10:27:37 I can't think of any world where that would be expected or desired in any way. Jun 29 10:28:34 gcc clearly (again) doesn't give a shit about users or sanity though, clearly. Jun 29 10:28:40 karlp: it does *what*? Jun 29 10:28:44 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86360 Jun 29 10:28:57 "not a bug" Jun 29 10:29:16 in the example he has there, it's "fine" it just got inlined as expected, Jun 29 10:29:38 but in rtl-sdr, it decided based on optimization to _not_ inline it, and expect to tbe provided extern, even without an extern decl. Jun 29 10:30:08 karlp: even *I* understand 'inline' Jun 29 10:30:30 nono. you just thought you did. Jun 29 10:30:39 lol Jun 29 10:30:59 clearly there's no use arguing it though, gcc knows best. Jun 29 10:31:04 * karlp goes for a coffee and something different Jun 29 10:31:06 inline behavior changes quite radically depending on which --std option is used Jun 29 10:32:07 dedeckeh: ping Jun 29 10:37:21 ldir: did you had a time for the patch? :d thanks Jun 29 10:39:28 f00b4r0: yeah, because this makes so much sense: https://zerobin.net/?26bfdd68cc85106a#QhUWJhUVGJLiDIYv65iw5y719jjcQC7ANvKMnhlE6uU= Jun 29 10:40:32 gcc version? It works for me ;P Jun 29 10:41:14 IMO, it doesn't matter what version of what gcc docs say, in no world is that expected or rational behaviour :) Jun 29 10:42:04 if you're using a development gcc version, nothing of the sort should raise an eyebrow ;) Jun 29 10:42:09 -std=gnu89 "works" Jun 29 10:42:24 gcc (GCC) 7.3.1 20180303 (Red Hat 7.3.1-5) Jun 29 10:43:02 gnu99/c99/gnu11/c11 all "fail" Jun 29 10:46:56 i suspect gcc devs don't want us codemonkeys to use 'inline'. Ever. Jun 29 10:47:10 compiler knows better. Jun 29 10:47:18 yes, it's meant to be just a hint. Jun 29 10:47:23 I'm ok with that, Jun 29 10:47:48 I'm not ok with the idea of inline not being a hint now, but instead being a flag that says, "throw this code out, I don't actually want it" Jun 29 10:52:02 Tapper: will take a while. which device? Jun 29 10:57:34 jow:pong Jun 29 10:58:45 dedeckeh: there's some ongoing debugging effort in the forum to figure out why on ath79, the br-lan interface does not regain connectivity after chaning the lan ip and calling network reload (uci set network.lan.ipaddr=192.168.2.1; uci commit network; /etc/init.d/network reload) Jun 29 10:58:59 the ip settings are properly applied, ifconfig shows the new ip as seen over serial Jun 29 10:59:06 but no pinging from the outside is possible Jun 29 10:59:12 flushing neighbour cache does not work Jun 29 10:59:20 pinging ipv6 neither Jun 29 10:59:34 ifdown lan; ifup lan neither Jun 29 10:59:48 ifdown lan; sleep 1; ifup lan does work however Jun 29 10:59:59 I was wondering if you have any idea what it could be Jun 29 11:00:09 I suspect some low level ethernet driver mdio things or so Jun 29 11:01:16 jow:is the output of brctl show correct after doing th reload ? Jun 29 11:02:01 dedeckeh: good question, will ask Jun 29 11:08:09 dedeckeh: identical Jun 29 11:09:42 jow:does tcpdump show packets when pinging the interface ? Jun 29 11:41:45 Hi mangix sorry was roped in to doing hose work lol it's for wrt3200acm pleas ad luci-ssl. Jun 29 11:42:44 house* Jun 29 11:43:22 Why I have in feeds http://downloads.lede-project.org/releases/17.01.4/ url for example when I compiled wrt from openwrt-18.06 branch? thanks Jun 29 11:47:47 Do I need to set CONFIG_VERSION_REPO="http://downloads.openwrt.org/releases/18.06.0-rc1" for example? Jun 29 11:57:07 Lukas__: yes Jun 29 11:57:13 dedeckeh: will ask Jun 29 11:57:40 dedeckeh: however the user just wrote that it also works as expectedo n an unbridged interface Jun 29 11:57:51 dedeckeh: so the problem seems related to bridging Jun 29 12:07:00 jow:would be interesting to know if the traffic is seen on the bridge port Jun 29 12:07:56 that would already indicate if the problem is inside the bridge logic or not Jun 29 12:07:58 dedeckeh: I wrote to blogic in parallel and he suspects some issue with adjust_link() Jun 29 12:13:10 yep Jun 29 12:13:19 i simply had not seen the issue Jun 29 12:13:24 someone remind me tomorrow Jun 29 12:13:39 i am abut to go offline, party like i was 18 again Jun 29 12:14:02 prolly crash and burn by 7pm and then have an empty house for 5 days were i can be a happy hacker Jun 29 12:17:08 hey folks, i'm having a strange issue with dhcp on my ar150 and not sure how to debug it Jun 29 12:18:32 i tried to add eth0 ("wan" ethernet port) to the lan bridge Jun 29 12:19:06 router is a standard config gl-ar150 Jun 29 12:19:40 it was working across one reboot but with another lan clients won't get dhcp leases Jun 29 12:20:28 i now removed eth0 from the bridge but still no dhcp leases for clients! Jun 29 12:20:49 pastebin your network and dhcp configs Jun 29 12:21:01 and your /etc/board.json Jun 29 12:24:15 https://hastebin.com/raw/ayipugeyaq Jun 29 12:25:12 looks sane... is dnsmasq complaining about other active servers in logread? Jun 29 12:28:30 cant see nowt like that Jun 29 12:28:51 only error is on boot: Jun 29 12:29:33 Fri Jun 29 11:53:43 2018 daemon.notice procd: /etc/rc.d/S19dnsmasq: Command failed: Not found Jun 29 12:29:36 Fri Jun 29 11:53:43 2018 daemon.notice procd: /etc/rc.d/S19dnsmasq: Failed to parse json data: unexpected end of data Jun 29 12:30:35 sounds like something ldir (?) reported a few days back Jun 29 12:32:08 what's the state of mt76 compared to say ath10k? how well is it working at this time? Jun 29 12:45:30 karlp: the C standards have changed here, and bare 'inline' is the wrong thing to do Jun 29 12:50:25 blogic: happy partying ;) Jun 29 12:54:23 dwmw2_gone: yeah, because it allows having another one defined elsewhere. I still maintain that in a single compilation unit, "assuming that inline means there's an external one" is a terrible choice. Jun 29 12:59:22 if there's not an extern declaration — if GCC would have *warned* about calling an extern had the inline not existed — then definitely Jun 29 12:59:25 as I said in the PR Jun 29 12:59:38 but still, I don't really care. C standard says to use "static inline" Jun 29 13:24:29 so what are we going to do with wpad-mini and 11r? Add 11r support to wpad-mini, or put wpad-full in the builds for dual-radio boxes? Or nothing? Jun 29 13:24:55 ... or make a new wpad-11r which is just mini + 11r :) Jun 29 13:25:08 or make hostapd support loadable modules :) Jun 29 13:29:40 11r is littered across the whole thing; it isn't sanely a loadable feature Jun 29 13:51:15 jow: not me - stintel I think? Jun 29 13:51:33 yeah Jun 29 13:51:40 haven't looked into that yet Jun 29 13:52:00 glad it's not just me :) Jun 29 13:52:40 ldir: if the dust has settled on the pppoa thing... Jun 29 13:52:56 I'm interested in the fact that pskb_expand_head() was having to expand headroom Jun 29 13:53:05 is that happening for routed packets in the general case? Jun 29 13:53:06 ldir:the same issue is reported in https://github.com/openwrt/openwrt/pull/1072 Jun 29 13:53:16 because if so, that's going to hurt performance a lot Jun 29 13:53:38 the kernel has no generic feedback mechanism for a packet source, to say "I'm having to expand every skb you send me; add more headroom to them!" Jun 29 13:54:56 dwmw2_gone: I'd need to put some of our debug back to get a feel for it, which is easy enough. Jun 29 13:56:34 but the reason why llc worked more reliably was that the 6 byte llc header exceeded the headroom so a new skb got allocated with the size bump that everyone knew about, so no size mismatch later on. Jun 29 13:57:00 haha Jun 29 13:57:18 that was for packets from pppd which probably aren't so important Jun 29 13:57:36 I'll take a look next week - currently doing $paid_work which doesn't involved code but more pushing of faders & using ears :-) Jun 29 13:57:38 an Ethernet header probably gives you enough room for the packets destined for ATM Jun 29 13:57:44 ok, have fun. thanks Jun 29 13:57:56 I have hh5a to deploy at some point too, with VDSL to replace the existing tdw8970 Jun 29 13:58:35 I do have a question on this 11r stuff you're so keen to see ;-) Jun 29 13:59:17 go on... I have a few minutes before I head for the train Jun 29 14:00:01 If I have two APs, each with 2 radios 2.4 & 5g - but a 5g SSID and a 2.4g SSID.... are they two difference 'roaming' domains or whatever the 11r terminology is (I forget!) Jun 29 14:00:24 yes Jun 29 14:00:43 if you have the *same* SSID on the 5GHz and 2.4GHz networks, you can 11r roam between them. Jun 29 14:01:18 ok, I don't want that.... but I'd 'like' fast roaming between the 2 APs on the same band. Jun 29 14:01:22 my 5G radios have two SSIDs — the main one *and* a special _5G SSID Jun 29 14:01:29 that's OK, you can do that too Jun 29 14:04:25 ah, so that's my next question... if I have 2 ssids per band, normal and 'guest' are they the same roaming domain? Jun 29 14:05:51 no Jun 29 14:06:02 the idea is that the SSID is the roaming domain Jun 29 14:06:19 which is true already. IT's just *faster* roaming Jun 29 14:06:54 I think stuff is now set up so that this Just Works™ Jun 29 14:07:09 even with multiple SSIDs on each radio etc. Jun 29 14:09:30 understood. thanks :-) Jun 29 14:12:13 can someone either merge those rtl-sdr fixes or give me commit access? Jun 29 14:19:24 dwmw2_gone: psst, we want 11w too. ;) Jun 29 14:21:46 good mornng Jun 29 14:33:18 dwmw2_gone: done, thanks for chasing it. Jun 29 14:55:51 hello everyone, I saw OpenWrt has a firmware for mikrotik 1004kc devices, but not for my router board (RBM33G). Is it possible to use the generic bootloader to load a mips1004kc kernel and install it manually from there? Jun 29 14:57:00 just to know - I still need a DB9 cable to access the serial port. It's on its way Jun 29 14:57:46 jsollie: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=5684d087418d176cfdef4e045e1950ca7ba3b09f Jun 29 14:58:25 f00b4r0: thank you! Jun 29 15:03:45 jsollie: report back how it works, i have been keeping my eye on that little board RMB33G Jun 29 15:04:18 jsollie: two pcie one is for 4G/3G it seems, not sure if it can load two full 11n/11ac radio Jun 29 15:05:23 it seems one pcie for 4G/LTE modem, another for Wifi, kind of a mifi board, it will be interesting to see if it can run two wifi pcie cards in parallel Jun 29 15:06:57 well, I have been installing 2 ath10k chips so far Jun 29 15:07:09 and the routeros does not see a pci problem Jun 29 15:07:19 only lacks a driver Jun 29 15:07:31 that's why I was looking at openwrt Jun 29 15:07:44 the kernel of the routeros is too old Jun 29 15:08:00 great, which atk10k module are you using? Jun 29 15:08:03 jsollie: fwiw I know nothing of this particular board, but the RB devices I have played with could all be forced to netboot by holding the reset button for "long enough". Meaning it was technically possible to install without serial Jun 29 15:08:17 two ath10k, gosh it's exciting Jun 29 15:09:17 I am sure I can boot it with netboot, and I have a dhcp and tftp server to do so, but it's my first experience with openwrt Jun 29 15:09:32 so I want to make sure I have an eye on it Jun 29 15:09:38 makes sense Jun 29 15:10:26 for the modules - 2 times a 4x4 board, one for 2.4 and one for 5ghz Jun 29 15:10:56 can you do 2.4G with 11AC, thought it's all 5Ghz Jun 29 15:11:40 yes, but this particular board is especially for 2.4. One sec, I'll look for the chips Jun 29 15:11:42 any brandname for the ath10k pcie card Jun 29 15:12:56 just checked again and yes 11AC is pure 5Ghz Jun 29 15:13:32 Compex WLE1216V2-20 Jun 29 15:13:41 for 2.4 ghz Jun 29 15:14:07 COMPEX WLE1216V5-20 Jun 29 15:14:12 for 5 ghz Jun 29 15:14:17 both are installed Jun 29 15:14:33 great, thanks for the info Jun 29 15:14:36 together with the DB9 I ordered 8 antennas Jun 29 15:14:59 so I can create an AP with 2.4 and 5ghz capabilities Jun 29 15:15:05 next time when i play with multi-radio mesh i will likely use RBM33G+COMPEX2+COMPEX5 then Jun 29 15:15:18 glad I could help Jun 29 15:16:11 but please note that it is only routeros which assigns an IRQ to the PCI device Jun 29 15:16:22 it does not start it because there's no driver Jun 29 15:17:12 * ldir leaves https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commit;h=1284418300c76ab3c4789932e5e96cd751f330fe for muhaha Jun 29 15:18:38 it will only be committed on the basis of a sworn statement on the life of nearest & dearest/firstborn that it has been successfully tested. Jun 29 15:19:17 lol Jun 29 15:22:22 ausjke: do you have a particular forum where I should post my results? I mean, I expect the DB9 cable to arrive wednesday or something, we could have forgotten this Jun 29 15:24:05 ldir: no lovin for 18.06 ? Jun 29 15:24:25 kernel bump :) Jun 29 15:24:53 there's a regression that causes slow speed on some ipq targets.. and its fixed on .51 Jun 29 15:26:04 ldir: why doesn't that module work on any other hardware? Jun 29 15:26:11 I'm terrified enough of pushing to master, let alone pushing to something important like a release!!! Jun 29 15:26:26 I mean, why does it need the check for APU at all? Jun 29 15:26:28 karlp: I do not know, I don't have an APU. Jun 29 15:26:30 jsollie: not sure elsewhere, ping me here, i'm checking around to buy rbm33g Jun 29 15:27:21 the store nearby only has compex WLE600 11ac radios, don't know what 600 means, they have WLE600, 900, etc, you're using 1216 Jun 29 15:28:40 for ath10k, what's the best chip these days? QCA9882 ? Jun 29 15:29:02 https://www.ispsupplies.com/Compex-WLE600VX Jun 29 15:29:52 ldir: I don't see any reason for the check at all, Jun 29 15:30:04 jsut don't include the model for boards that don't have it... Jun 29 15:30:18 it does a probe for the device already to test if it's there. Jun 29 15:31:25 it's not in the code from pcengines either Jun 29 15:31:39 it's something this chrisblake dude added somewhere before submitting it Jun 29 15:33:55 https://www.ispsupplies.com/Compex-WLE900VX this one looks good, 11ac and 11g/11n dual band, at $20 Jun 29 15:34:51 go put it inside your zbt we1326 on the spare slot ifyou like ;) Jun 29 15:35:43 karlp: that's for LTE modem, zbt seems too weak to power wifi cards... Jun 29 15:36:05 besides it will probably take me one week to get dts/pcie-IRQ-routing right Jun 29 15:36:41 * ausjke is still very disappointed about wifi performance with zbt 1326 Jun 29 15:37:35 https://wikidevi.com/wiki/Ath10k claims STA/AP are supported for WLE900VX Jun 29 15:40:46 ausjke: disappointed how? Jun 29 15:42:04 abenz: 2.4G max at 30mbps, basically a 11g, never got 11n speed after many iperfs etc, worst, it disconnects about twice a day(resumes after a few minutes), which never happened with my older ath9k router in the last N years. Jun 29 15:42:34 abenz: 5G/VHT80/11ac max about 100mbps. so i'm getting 11g/11n router instead of a 11n/11ac Jun 29 15:43:28 ausjke: I have the same wifi issues with my wg3526 Jun 29 15:43:34 karlp: I thought I'd try to help out 'cos muhaha has been trying to tweak this for 2 weeks. If you don't think the check is required then patches welcome. Jun 29 15:43:36 abenz: last but not least, signal degrades fast by distance, comparing to both atk9k and little GL.inet portable router with tiny antennas(zbt's atenna is normal size) Jun 29 15:43:47 I think thats a similar device to yours apart from something like sd card or so? Jun 29 15:44:09 but I was under the impression that those were mt76 driver issues that needed ironing out Jun 29 15:44:21 hm, firstboot.d? Jun 29 15:44:29 or something Jun 29 15:44:40 is it specific to the ZBT devices? Jun 29 15:44:54 karlp: I'm perfectly happy to drop & not commit to master. Jun 29 15:45:31 need a sensible way to exec stuff on firstboot, not sure if running the same thing on every boot is good Jun 29 15:45:34 * jwh thinks Jun 29 15:46:05 abenz: good question, i don't have another mt76 board to test, actually my Gl.inet router is mt7620 and it probably uses mt76 driver too, and it performs better than mt7621/zbt/mt76, i would suspect either the antenna design, or zbt board, is not good enough Jun 29 15:46:19 ausjke: I'm contemplating an apu4d4 as main device, together with two wifi modules, 1 ath10k and 1 ath9k Jun 29 15:46:42 ausjke: I see.. Jun 29 15:46:48 * ausjke is thinking about selling 3 zbt1326 on ebay... Jun 29 15:47:12 I was suspecting that as well.. I don't see as many xiomi users complaining.. and they are using the same chipset (and same wifi modules) Jun 29 15:47:12 who is familiar with the boot process enough to answer some questions about ordering and stuff? Jun 29 15:47:44 so it might indeed be an el cheapo device overall. It works quite ok when I disable the wifi completely though.. so as a pure router with sd card and all that, its not bad Jun 29 15:47:50 so like, wanna be able to detect if overlay already exists, or whether its a fresh device thats just booted with initramfs... if not, do sysupgrade to install Jun 29 15:47:57 rather than having to do it manually Jun 29 15:51:12 abenz: agreed, the NAT wired performance is about 700Mbps as I measured by iperf, both udp and tcp Jun 29 15:51:46 so wired speed is fine, it's the wifi design that is not optimal, sadly, otherwise it could be a hot deal Jun 29 15:52:45 ausjke: mt7620 has no relation Jun 29 15:53:08 * ausjke somehow does not like x86 boards, still thinking about buying rbm33g+radios for long-term Jun 29 15:54:11 Monkeh: mt76 is not shared between mt7620 and mt7621 then? thought about upgrading Gl.inet to 1806 for comparsion, seems like no need then Jun 29 15:55:13 MT7621 has nothing to do with wifi at all Jun 29 15:55:21 MT7620 has an older wifi chipset integrated Jun 29 15:55:28 Monkeh: oops you're right Jun 29 15:55:38 ldir: I don't have one either, I just think that hacking on the name detection is fixing the wrong end of it. I just don't know why it was ever needed. Jun 29 15:55:56 i'm referring to 7603/76x2 to 7620's SOC wifi Jun 29 15:56:23 No relation. Jun 29 15:56:24 I'm very happy with my zbt we1326. I get much faster wifi than auskje gets Jun 29 15:57:09 maybe the pretty lady who solder your wifi connector pads was in good mood Jun 29 15:57:34 karlp: what's your 2.4G speed Jun 29 15:58:12 i have 3 zbt1326, they are the same bad by the way Jun 29 15:58:20 even tried their OEM firmware Jun 29 15:59:48 my 2.4gig was similar to yours, not great, btu that's just standard rt28xx driver problems. Jun 29 16:00:16 Eeeeeexcept there's no rt28xx here Jun 29 16:00:56 it's almost like the namign of the drivers is impossible for _anyone_ to follow. Jun 29 16:01:24 I manage Jun 29 16:01:25 :P Jun 29 16:01:38 I get ~similar shitty perf on 2.4gig on the zbt we1326 as I got on a mt7620 board (miwifi) I get much better 5gig speeds Jun 29 16:05:40 zbt1326's OEM firmware does not use mt76, still 2.4G is sluggish Jun 29 16:06:17 anyway living with it for now while searching for replacements, not all PC has 5G STA so 2.4G solid performance is a must Jun 29 16:06:45 I'm still rocking an unifi uap for 2.4g Jun 29 16:07:14 reasonably fast Jun 29 16:09:30 jwh: with openwrt in full AP(non dumpAP) mode? Jun 29 16:09:39 you're using Enterprise Gears! :) Jun 29 16:10:37 ausjke: ye Jun 29 16:10:37 hostapd, couple of ssids Jun 29 16:10:45 what's inside(chip,wifi radio) Jun 29 16:10:56 its an ath5k Jun 29 16:11:08 can't remember cpu, its on the openwrt toh :D Jun 29 16:11:34 you should know anyway Jun 29 16:11:36 :) Jun 29 16:11:48 they're pretty old, so Jun 29 16:12:10 https://openwrt.org/toh/ubiquiti/unifi Jun 29 16:12:15 7240 Jun 29 16:12:51 yeah, pretty old indeed Jun 29 16:13:16 I "borrowed" it from an old employer Jun 29 16:13:38 they still have about 14k or so of unifi aps I think Jun 29 16:14:25 I actually wrote a reverse engineered controller for them to replace the horrendous unifi controller Jun 29 16:14:37 but then ubnt changed the encryption on newer firmare and I haven't figured it out yet Jun 29 16:44:08 ldir: what if i told you i've been running 18.06-rc1+the 4.14.51 on ipq8064 for 4+ days and it's rock solid ? :3 Jun 29 16:44:18 would you reconsider merging that then? Jun 29 16:44:52 stintel: can we talk about dsa on ipq8064 for 5 minutes? Jun 29 16:46:09 hey guys Jun 29 16:46:25 are mails to lede-dev still processed/forwarded to openwrt-devel? Jun 29 16:46:25 movi: I don't know anything about dsa nor ipq* Jun 29 16:46:45 stintel: oh, sorry, i got the patch from your staging repo Jun 29 16:47:13 huh Jun 29 16:51:22 stintel: do you know anything about lede-dev@ ? Jun 29 16:55:06 Borromini: haven't been following lately, I know that it was forwarded, no idea if things changed Jun 29 16:55:39 no it seems there's quite a bit of lag Jun 29 16:55:46 but they're popping up now :) Jun 29 16:55:50 Borromini: your patches are missing a commit message and SoB Jun 29 16:56:26 mkresin: bah. i will re-send, sorry :( Jun 29 16:56:43 so... no comments on how to reliably detect if overlay exists? :( Jun 29 16:56:52 or whether its a first boot Jun 29 16:57:25 I mean the obvious fix is to only inject commands into initramfs images but thats probably just as messy Jun 29 16:58:21 ideally want something like: if (firstboot || !overlay) do_sysupgrade Jun 29 17:01:52 oh what would it be nice if patchwork had an option "reject patches without sob" Jun 29 17:01:56 and "reject HTML mail" Jun 29 17:02:05 and "reject top posting mails" Jun 29 17:02:06 heh Jun 29 17:03:26 it seems top posting is just fine on openwrt-adm :P Jun 29 17:04:04 stintel: if you get around to it, please make it gentle. I am a frail soul and rejections always make me a bit depressed QQ Jun 29 17:04:26 so a nice e-mail 'unfortunately, we are unable to accomodate your request blabla' or sth :P Jun 29 17:05:37 Borromini: haha yeah, it's easy to write a nice response once Jun 29 17:05:43 but for topposting ... Jun 29 17:05:48 I mean for the 1002382438234th time Jun 29 17:05:52 it's hard to stay nice Jun 29 17:05:55 hehehe Jun 29 17:05:57 maybe you're just wrong then ;) Jun 29 17:05:59 so I just don't read those mails anymore Jun 29 17:06:46 the only fatal error is mixing top and bottom posting. Jun 29 17:11:06 movi: koen has some patches in the works. Jun 29 17:11:36 hello mr ldir Jun 29 17:11:59 * ldir hides Jun 29 17:13:35 :( Jun 29 17:13:38 karlp: I had a business contact me 'cos they thought there was something wrong with my email client 'cos I bottom posted. Jun 29 17:13:46 lol Jun 29 17:14:17 No, I'm doing it right, you're all muppets :-) Jun 29 17:15:00 So business it seems can only cope with top posting. Jun 29 17:28:05 let's blame microsoft Jun 29 17:30:00 Borromini: anyway, how may I be of assistance? Jun 29 17:30:34 karlp: ping Jun 29 17:35:33 ldir: nothing good sir just a delight to speak to you. congratulations on advancing your station. Jun 29 17:35:58 i am truly honoured to be in the company of such enlightened folk like yourself. Jun 29 17:35:58 ha - it's bloody scary! :-) Jun 29 17:36:03 hehehe Jun 29 17:36:08 more things to break huh :P Jun 29 17:36:34 if you find the enlightened person can you have them shot, stuffed & put in a display case? :-) Jun 29 17:38:29 surely you don't have that much power Jun 29 17:38:36 let alone a taste for overreach Jun 29 18:47:09 ldir: yes? Jun 29 18:59:56 ldir: ping Jun 29 19:04:03 bah, forgotten how to just make the toolchain (can't do kernel_menuconfig without it) Jun 29 19:12:36 oh hm, no its just broken Jun 29 19:13:21 toolchain-mips64_octeonplus_64_gcc-7.3.0_musl Jun 29 19:13:22 umm Jun 29 19:13:50 did it change? I swear it was just octeon Jun 29 19:14:25 ah it did 9 days ago Jun 29 19:18:13 not sure there is any need for the cpu type being set anyway, theres runtime detection (and the optimisations don't even matter in the open source version anyway, all the goodies are in the NDA code) Jun 29 19:28:36 blogic: https://gist.github.com/joeholden/d5d1ecd3aae39a39287b3c54764b59a2 Jun 29 19:28:48 oh Jun 29 19:28:57 nm its just not reset args Jun 29 19:30:26 should probably reset it on target change even if its been modified Jun 29 19:32:58 karlp: was driving home - what do you think are the implications of removing the board checks? Jun 29 19:34:25 muhaha: pong Jun 29 19:37:50 hello Jun 29 19:42:23 ldir Jun 29 19:42:23 ldir: did you look at the patch? would be great if it can make it to 18.06 release Jun 29 19:42:49 we are running openwrt on gateworks sbc. the image was provided by our vendor (Epiq). I need to compile another kernel module. I dont want to do out of tree compiling (cross compile). I've been looking at guides on the net but I cant find anything about on board compiling Jun 29 19:44:16 yes I looked at the patch and came up with https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commit;h=f6c9c8b2d32bde39cb578ba09569c01f16848579 Jun 29 19:44:19 howudodat_: on board compiling is not supported at all Jun 29 19:44:25 you'll need their packaged image builder Jun 29 19:44:40 howudodat_: the entire infrastructure is lacking for that. No gcc, no kernel headers, no proper make etc. Jun 29 19:45:57 ok, thank you. I was afraid of that :( Jun 29 19:46:30 however karlp has raised the very valid point of why we're checking the board ID at all - and at this stage I'm inclined to agree. I'd like to speak to the PKG_MAINTAINER for their opinion. The 'don't bother checking' patch looks like https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commit;h=02dea758bc9e2ae6ce2fa97928779df1b6ffbf39 Jun 29 19:47:06 muhaha: which board do you have ? Jun 29 19:47:41 apu2c4 Jun 29 19:48:13 and this returns 'PCEngines....' as opposed to 'PC Engines...' ? Jun 29 19:54:00 teh only reason I can think of is if the nct5104d_find() stuff somehow crashes stuff, but again, I'd say just don't install the module if you don't have the hardware. Jun 29 19:57:09 I guess #lede-dev is no longer in use? Jun 29 19:58:52 ldir: https://pcengines.github.io/ Known issues: wrong names in SMBIOS causes some modules not being loaded on certain OSes Jun 29 19:59:55 karlp: agreed Jun 29 20:01:08 muhaha: do you have any issues with your APU's LEDs ? Jun 29 20:02:14 ldir: leds module is already fixed https://github.com/openwrt/openwrt/blob/openwrt-18.06/package/kernel/leds-apu2/src/leds-apu2.c#L335 Jun 29 20:09:24 * ldir gives up throwing popcorn at the dog, keeps on missing! Jun 29 20:10:21 lols Jun 29 20:11:12 keeps on bouncing off teeth/nose Jun 29 20:11:21 need a bigger dog Jun 29 20:17:51 how can i coax one of you to help me with the rtl8812au Jun 29 20:18:51 that code in leds-apu also looks pointless. Jun 29 20:18:56 just don't load leds-apu on non-apu. Jun 29 20:19:09 and stop hacking on string matching Jun 29 20:23:26 it never _uses_ the boardname afterwards anyway. Jun 29 20:25:16 karlp: problem is that there are 4 options for boardname based on version of coreboot. Legacy before change apu2. Legacy after chagne PCEngines apu2. Mainline before change apu2. Mainline after change PC Engines apu2 Jun 29 20:26:12 It should be fixed in newest coreboot, but it can hits you if you load a specific version of coreboot Jun 29 20:26:25 muhaha: yes, but the thing is, the module's dont' _care_ what teh board name is. Jun 29 20:26:50 the only thing that string checking is doign is getting in your way. Jun 29 20:27:21 you can try this yourself. Just drop it all, like ldir linked to, and do it for the leds one too if you like. Jun 29 20:29:02 i see Jun 29 20:29:21 sorry - was called away from keyboard. Jun 29 20:36:23 muhaha: top 2 commits https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=shortlog;h=refs/heads/exp Jun 29 20:39:41 ldir: thanks. I dont understand to wrt gerrit/git, but it is currently in staging branch? This branch for codereview? Jun 29 20:41:47 it is currently in my 'personal' experimental staging branch as somewhere to put it for you to look at. Jun 29 20:43:01 I still want to email the pkg_maintainer with regard to the sanity or otherwise of checking DMI info before permitting module load. Jun 29 20:44:12 It seems like a waste of effort in people having to update the permitted boardname/vendor name strings on BIOS update, when in reality 'if you don't need the functionality on your board...don't load the driver' would appear a better long term solution. Jun 29 20:46:51 but I ain't gonna merge it/them into master (& backport) until I've heard from the pkg_maintainer & you swear on the life of your firstborn that you've tested them ;-) Jun 29 21:00:49 ldir: good work though. Jun 29 21:01:38 enably tui is a nice change too. Jun 29 21:01:59 is that stuff you just carry permanently? Jun 29 21:02:17 man I wish cgit's default behaviour was to take you to commitdiff view instead of commit view. Jun 29 21:04:57 heh, yes Jun 29 21:05:28 email sent - and the 2 relevant commits in own branch https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=shortlog;h=refs/heads/apunocheck Jun 29 21:06:58 sirs, how do you manage your signing keys, do you commit the build-key to the repo or just move it around to your builders? Jun 29 21:07:08 need more automation Jun 29 21:09:14 karlp: yes - there's a lot in there at the moment, occasionally I go on a purge to get it upstream if I think useful. Jun 29 21:20:40 I'd like to get cake sorted out - that would get rid of a lot of rubbish in my queue. Jun 29 22:01:12 jsollie: just re-read RBM33G, the manual repeatedly said only one wifi card should be used, second slot is for LTE modem Jun 29 22:01:52 either a power issue(wifi radio is more power hungry), or routeros does not support two wifi cards, if the former is true then we got a problem Jun 29 22:01:54 I think thats the one I was looking at the other day Jun 29 22:02:02 the second slot is wired for usb sim Jun 29 22:02:05 so wifi won't operate Jun 29 22:03:24 there is a M.2 slot pcie for SSD or MicroSD storage expansion Jun 29 22:03:32 yeah Jun 29 22:03:35 they're both "pcie" Jun 29 22:03:46 but the spec allows multiple modes (pcie, usb etc) Jun 29 22:04:03 however from what I read, the second slot is fixed for usb operation Jun 29 22:04:03 no fish then, I added them to the shopping card and just got back to double check, not a good fit, must need two wifi cards Jun 29 22:04:04 and sim Jun 29 22:04:20 I was looking at it for a compact lte router Jun 29 22:04:21 jwh: i think you're right Jun 29 22:04:31 for lte router it should be a nice fit Jun 29 22:04:51 yeah, something I can throw in my bag with battery when I'm travelling Jun 29 22:05:12 you got tplink and gl.inet for that already i think? Jun 29 22:05:33 portable router with 3G/4G have a few options and are cheap Jun 29 22:05:44 mmm, I ruled them out for some reason Jun 29 22:06:16 this board recommends 18V input, not good fit for battery Jun 29 22:06:19 I think it was the lack of ethernet on the travel 3g boards Jun 29 22:06:42 mm, yeah but its 11-32 Jun 29 22:06:50 usb-ethernet done can do 300mbps easily and cheap, i have two of them and they plug and play Jun 29 22:06:57 s/done/dongle/ Jun 29 22:07:14 ah, the ones I was looking at required a usb modem (none built in, just basically a shit wifi ap) Jun 29 22:07:16 when you have wifi cards you pretty much need go more than 11v Jun 29 22:07:28 well I can step up to 12V Jun 29 22:07:32 ooorr give it 16V Jun 29 22:07:43 my bigger pack will output 5/12/16/19 Jun 29 22:07:58 at a few amps Jun 29 22:08:03 i c Jun 29 22:08:30 really want something I can just leave in the bag Jun 29 22:08:31 * ausjke misses RB532 and Routerstation Pro dearly Jun 29 22:08:41 heh Jun 29 22:09:09 hm actually you might be right though, 19W is pretty high Jun 29 22:09:46 bye RMB33G for now Jun 29 22:10:38 yeah Jun 29 22:11:03 Whether RouterOS supports it or not the hardware clearly supports PCIe on both slots Jun 29 22:11:13 its hardware Jun 29 22:11:53 either the way the bootloader configures it, or just designed to be fixed in usb sim mode Jun 29 22:12:14 can't find the page I was looking at now Jun 29 22:12:22 Both slots have PCIe lanes. Jun 29 22:12:23 since its manual said twice no two wifi radios, i guess it's fixed by hardware for usb on the second slot Jun 29 22:12:31 Stop guessing - look at the board.. Jun 29 22:12:40 Monkeh: pcie slot doesn't imply pcie operation though Jun 29 22:12:46 it could be pcie, usb or ahci Jun 29 22:12:47 etc Jun 29 22:12:54 jwh: .. the differential pair routed to the pins used only for PCIe do. Jun 29 22:13:05 are they actually connected, though? Jun 29 22:13:36 .. why would the pairs and coupling caps be present if they're not connected? Jun 29 22:13:54 why do mikrotik do any of the things they do Jun 29 22:14:05 The SoC has three lanes Jun 29 22:14:10 I have been following this wiki to install a xcompile environment. I installed debian 9 in a vm and then installed sudo since debian doesn't do that by default. However after following this wiki step by step, make menuconfig followed by make results in errors. additionally there appear to be libraries that are used in the build that aren't getting installed or aren't the correct versions. https://pastebin.com/thHthBjD Jun 29 22:14:26 hm Jun 29 22:14:30 http://www.alphanetworks.com/en/product_detail/f705281a19052673 first 5Gbps Ethernet AP i saw Jun 29 22:14:32 oops wiki I'm following:https://wiki.openwrt.org/doc/howto/buildroot.exigence Jun 29 22:14:34 maybe its disabled in software then Jun 29 22:14:43 but that seems silly, so there must be a reason Jun 29 22:14:51 Seems doubtful as they say several times you can use either slot for wifi Jun 29 22:15:27 from the product brief: one miniPCIe and one USB 3G/LTE modem leaving other Jun 29 22:15:30 miniPCIe slot for 802.11 wireless card (consult product Quick Jun 29 22:15:30 Guide for details). Jun 29 22:15:30 or could be a routeros selling pitch, that they can but choose not to enable two wifi on this board Jun 29 22:15:44 which echos what I saw some guy who bought it say Jun 29 22:16:23 From the product brief: You can use the device with e.g.: ... two miniPCIe wireless cards Jun 29 22:16:34 ausjke: that alpha networks looks like a beast... but i can only imagine they'd need 100's of them in a single walmart to keep up with all the music video streaming. Jun 29 22:16:35 hm Jun 29 22:16:40 The lanes are connected. Jun 29 22:16:43 I'm pretty sure they'll work. Jun 29 22:16:54 https://img.routerboard.com/mimg/1365_hi_res.png Jun 29 22:16:54 https://www.ispsupplies.com/MikroTik-RouterBOARD-RBM33G user-guid and data-sheet in the page Jun 29 22:16:55 perhaps its a software/loader limitation thats been fixed then Jun 29 22:16:58 Examine yourself Jun 29 22:18:05 neither of those docs say you can use two wifi cards though Jun 29 22:18:16 https://i.mt.lv/routerboard/files/M33G-171219112521.pdf Jun 29 22:18:23 not only that, both of them advise you do not use two wifis Jun 29 22:18:42 yeah, 3rd bullet point Jun 29 22:18:47 and similar on the datasheet Jun 29 22:18:58 Two miniPCIe slots for either a 802.11 wireless card, or a 3G/LTE modem. Mini SIM 2FF slots available for Jun 29 22:19:01 miniPCIe 3G/LTE modems and have no functionality when a modem is not installed. [Important! Two Jun 29 22:19:04 802.11 wireless radio cards cannot be used at the same time on this device. It is suggested to use J8 slot Jun 29 22:19:07 for a LTE modem, and the J10 slot for a wireless card.] Jun 29 22:19:10 but you're right re the board Jun 29 22:19:21 so the question remains *why* they say it can't be used Jun 29 22:19:23 The current version of the product brief on their own site clearly says two wifi cards. Jun 29 22:19:25 is it software configurable? Jun 29 22:19:37 where? Jun 29 22:19:53 both the docs you pasted say 2 modems, one wifi slot Jun 29 22:19:59 Right here: https://i.mt.lv/routerboard/files/M33G-171219112521.pdf Jun 29 22:20:04 yes, I'm looking at it Jun 29 22:20:18 I don't think its right Jun 29 22:20:31 ... why? Jun 29 22:20:38 because the current version of the datasheet doesn't say that Jun 29 22:20:51 and someone who bought one said it doesn't work either Jun 29 22:20:56 Define 'current version' Jun 29 22:21:00 buuuut, like I said it could be a limitation thats been fixed Jun 29 22:21:14 whatever the version is that they're providing Jun 29 22:21:38 well, the "quick guide" not really a datasheet Jun 29 22:21:41 It says 'for either' Jun 29 22:21:47 It also says 'you can't use two at once' Jun 29 22:21:53 But it does NOT say 'use only this slot' Jun 29 22:22:03 mmm Jun 29 22:22:08 Logically therefore both slots have functional PCIe Jun 29 22:22:16 Which can be clearly seen routed on the board Jun 29 22:22:16 ha, theres a wording change between the two versions Jun 29 22:22:23 must be software then Jun 29 22:22:33 or loader/firmware thats been since updated Jun 29 22:22:36 They suggest using J8 not J10 for a modem because J10 DOESN'T normally have USB Jun 29 22:23:01 if it's firmware restriction that it's more challenging Jun 29 22:23:02 You have to lose the external port for that. Jun 29 22:23:09 ah yes Jun 29 22:23:28 If their firmware is crippled, replace it? Jun 29 22:23:47 well obviously its fixed Jun 29 22:23:57 but still doesn't answer *why* it was originally not working Jun 29 22:24:02 don't have a jtag here... Jun 29 22:25:01 get one and try :D Jun 29 22:25:34 worked two years full time on u-boot but kind of getting bored with it...not sexy there Jun 29 22:25:49 don't even need to interact with it though Jun 29 22:25:55 https://github.com/openwrt/openwrt/commit/5684d087418d176cfdef4e045e1950ca7ba3b09f Jun 29 22:26:04 I doubt the bootloader will be an issue. Jun 29 22:26:37 interesting Jun 29 22:26:43 so if it is routeros, why would it even care Jun 29 22:27:12 Monkeh: that's all kernel stuff Jun 29 22:27:13 its just linux (albeit a pretty ancient one) Jun 29 22:27:21 ausjke: ... Jun 29 22:28:41 more importantly, I'm taking bets on the sfp port of this rb2011 not working properly Jun 29 22:28:57 I can't actually find any concrete evidence either way Jun 29 22:29:27 gonna look like a bit of a bellend if it doesn't :D Jun 29 22:30:35 http://www.kepfeltoltes.eu/images/hdd1/2018/05/31/817M33G_1.jpg Jun 29 22:30:47 Shocking. Jun 29 22:32:42 heh Jun 29 22:35:15 Monkeh: running at 3/1 btw Jun 29 22:35:31 sweet, two raiods with wifi Jun 29 22:35:38 better odds than the world cup Jun 29 22:36:01 Monkeh: winbox, is this the routeros? Jun 29 22:36:10 ausjke: Yes. Jun 29 22:36:22 ausjke: let me know if 12V is enough :D Jun 29 22:36:28 12V will be fine, too.. Jun 29 22:36:38 its obviously got a half decent regulator Jun 29 22:36:55 i have some passive PoE here at least Jun 29 22:37:18 need to find my 12V step up Jun 29 22:37:25 I'm sure i've got a spare somewhere Jun 29 22:37:29 Even with a couple radios it'll be fine. Jun 29 22:37:30 alright, let me spend the $80 for one board + two radios, comparing to my kid's cavity filling which caused me $600 today this is nothing :( Jun 29 22:37:39 Hell, you could give it 3. Jun 29 22:37:45 Monkeh: heh Jun 29 22:38:25 are the 3 gig ports switched? Jun 29 22:38:32 where to add the 3rd radio Jun 29 22:38:34 ah yes Jun 29 22:38:48 All three feed to the same internal switch, yes. Jun 29 22:38:50 maybe a 900Mhz someday, for the forest... Jun 29 22:38:55 ausjke: In the third slot. Jun 29 22:39:11 is 19W gonna be enough for 3? Jun 29 22:39:19 mikrotik is a liar Jun 29 22:39:26 jwh: Easily. Jun 29 22:39:36 I have no idea what modern wifi power draw is like Jun 29 22:39:56 maybe those 'high power' ones pull more current Jun 29 22:39:56 If you budget 2W a piece you're probably overestimating. Jun 29 22:40:05 heh thats pretty good Jun 29 22:40:08 jwh: 0.2747664784 watts Jun 29 22:40:22 expected more especially for the ac stuff Jun 29 22:40:34 Peak can be a little high on transmit but you're never going to be transmitting at high duty cycles. Jun 29 22:41:07 mmm Jun 29 22:41:08 not so bad Jun 29 22:41:14 if this turns out to be working well, it can be used for place with a large crowd at reasonable cost Jun 29 22:41:31 yeah its pretty cheap Jun 29 22:41:43 and it's a good fit for mesh, though ath10k ibss mode is not mature for use Jun 29 22:41:46 does it have DSA? :DD Jun 29 22:41:50 but ath9k radio should be good Jun 29 22:42:28 jwh: You're stuck to 100mW at 2.4GHz, 200mW or 1W at 5GHz, and again very short duty cycles - and few cards have the amplifiers for 1W Jun 29 22:42:37 trusay Jun 29 22:43:10 that's are the Tx power for wifi signal strength, different from the pcie module itself Jun 29 22:43:10 not really a candidate for FWA anyway tbf Jun 29 22:43:15 iirc enabling the 2T2R 2.4GHz radio on one of my boards made for about 300mW Jun 29 22:43:18 ausjke: .. No, really? Jun 29 22:43:47 at least that's my understanding, corret me if i am wrong Jun 29 22:44:22 how did you measure its power, check the difference between on/off wifi radios? Jun 29 22:44:31 Again, 2W average is probably overestimating Jun 29 22:44:38 if the chip is also a matter of a couple of hundred mW then not even a concern Jun 29 22:44:49 How else would I measure it? Jun 29 22:45:23 anyway, better get back to writing these dumbass docs on how to deliver services right the first time Jun 29 22:45:34 I'm sure I need an ITIL qualification or something for this Jun 29 22:45:44 Honestly, 3G/4G modules probably consume more Jun 29 22:45:49 yeah Jun 29 22:45:59 some of the LTE modems are pretty heavy Jun 29 22:46:03 Never measured one but multi-watt transmit is pretty common in that arena.. Jun 29 22:46:09 especially when they're struggling or signal Jun 29 22:46:15 for* Jun 29 22:47:08 the module is only CE certified, not sure why US dealers are selling it, i assume FCC is mandatory Jun 29 22:47:35 as long as it conforms to regulations Jun 29 22:47:37 heh Jun 29 22:47:40 The RBM33G? Jun 29 22:47:43 It has no radios.. Jun 29 22:47:46 yes the weaker the signal, the more power it needs Jun 29 22:48:07 "hm, why am I not getting a signal?" Jun 29 22:48:18 "oh, right its broken because usb port can't deliver enough power" Jun 29 22:48:20 :D Jun 29 22:50:03 on the other hand, jow should be crowd-funded to upgrade luci/lua :) Jun 29 22:50:07 heh TIL sierra actually sell a "global" version of the lte modem in my laptop Jun 29 22:50:16 interesting Jun 29 22:50:48 but apparently without any 2G support Jun 29 22:50:50 wonder why that is Jun 29 22:50:55 talking about lte modem, what's the practice real speed these days Jun 29 22:51:00 pratical Jun 29 22:51:04 mmm Jun 29 22:51:11 I get 70meg on my cheap huawei Jun 29 22:51:27 way better than my 11ac wifi then Jun 29 22:51:32 not sure if limited by crappy band support, signal, network or chip Jun 29 22:52:05 but I think its only a cat3 anyway which puts it at 100meg Jun 29 22:52:06 so its reasonable Jun 29 22:54:11 its total garbage outside of the UK though heh Jun 29 22:54:17 I think it only has 2 bands Jun 29 22:54:29 and only one of those are actually in use here Jun 29 22:57:29 ugh, I suck at creative writing Jun 29 22:57:49 take a udemy course :) Jun 29 22:58:00 you can't teach creativity! Jun 29 22:58:38 false, the greatest artist copy/stole first before they reach the point to do their own thing Jun 29 22:59:08 some people just aren't creative, or good at numbers Jun 29 22:59:17 can't really teach those things if they're just not wired for it Jun 29 22:59:27 well you can teach them, but it won't make them any better Jun 29 23:02:13 compex may have its similar rbm33g board, checking Jun 29 23:03:05 https://www.compex.com.sg/embedded-board/ Jun 29 23:21:21 some reasonable looking boards there Jun 29 23:22:06 no pricing info though Jun 29 23:22:10 yeah Jun 29 23:23:40 google showed $210 retail bare board price Jun 29 23:23:59 the rest boards only have 1 minipcie, none hasd two minipci Jun 29 23:25:25 WPQ864 looks pretty decent Jun 29 23:27:17 https://wiki.compex.com.sg/wiki/Using_LEDE_with_WPQ864 Jun 29 23:27:36 compex has its own openwrt version, no where to see its github so i guess it's closed Jun 29 23:27:56 openwrt does have a page for WPQ864/865 but no support info Jun 29 23:31:20 1806 make menuconfig does have a wpq864 profile, seems like supported Jun 29 23:53:08 hm, gonna have to make multiple profiles for dm200 I think... can I just define a subtarget and then match on that in config_generate etc? Jun 29 23:53:12 or.. Jun 29 23:53:33 trying to detect dsl type and reconfigure isn't reliable enough Jun 29 23:59:57 jwh: this is abandoned patch I've been using in the past, http://paste.debian.net/hidden/63c2f39e/ Jun 30 00:00:58 I was just looking at doing that Jun 30 00:01:43 could just define 2 subtargets that are functionally the same, just with different ids Jun 30 00:01:50 so I could have a different default config Jun 30 00:02:04 unless theres a more elegant way? Jun 30 00:04:31 the above is the most elegant way I could come up with, but it's a pain to maintain long term Jun 30 00:04:35 yup Jun 30 00:07:29 need some way of generating an image that looks the same (same board ids etc), but just has some sort of hint that it should generate a different config Jun 30 00:08:40 the patch above does that, well, it uses different -but compatible- boardname, so you can sysupgrade in every direction Jun 30 00:08:55 ah hmmmmm Jun 30 00:09:33 I guess in that case, one could just sysupgrade without preserving the config to the alternate profile Jun 30 00:09:49 hey guys I could use a bit of help getting a xcompile environment setup. I'm using https://wiki.openwrt.org/doc/howto/buildvm and then https://wiki.openwrt.org/doc/howto/buildroot.exigence and trying to set up on debian 9. We are using gateworks ventana (BTW, has 4 mini PCI slots) Jun 30 00:10:42 just need to either clone the openwrt repo, or use an image builder bundle, build-essential and all that jazz on debian, should be enough Jun 30 00:11:26 do_page_fault(): sending SIGSEGV to pineapd for invalid read access from 0000005c Jun 30 00:11:28 Sat Jun 30 00:05:18 2018 kern.info kernel: [ 711.598148] epc = 774768a1 in libpcap.so.1[7746e000+2b000] Jun 30 00:11:31 Sat Jun 30 00:05:18 2018 kern.info kernel: [ 711.603528] ra = 004053f8 in pineapd[400000+e000] Jun 30 00:11:33 im getting that erro Jun 30 00:12:24 pkgadd: think that will do, thanks Jun 30 00:12:53 after following the wiki, I still dont have a clean build environment. first errors are here (https://pastebin.com/taHthBjD) but that's even after installing those libs Jun 30 00:13:37 you missed the bit about feeds, most likely Jun 30 00:13:45 aaaand whatever debian idiocy is going on with time Jun 30 00:17:04 yeah the feeds took care of the make warnings, but I'm not sure how to fix the time issue. that prevents building Jun 30 00:17:09 time: invalid option -- 'f' Jun 30 00:17:35 ls -la `which time` Jun 30 00:19:11 repo suggests apt-get install time should install gnu version Jun 30 00:22:47 time must have been busybox linked. after apt install time ls -al which time is more sane Jun 30 00:22:58 -rwxr-xr-x 1 root root 15088 Jul 21 2015 /usr/bin/time Jun 30 00:23:08 why am i getting a segfault when trying to enable the AP Jun 30 00:24:02 but make appears to still be calling the bsybox version. even afte exit and restart of the shell Jun 30 00:24:30 is there a preferred distro for xcompile? Jun 30 00:26:17 i compiled the whole damn toolchain Jun 30 00:26:18 on debian Jun 30 00:27:38 howudodat_: might be path order, just checked ubuntu vm and time appears to be gnu, maybe use that if you are used to debian Jun 30 00:27:48 that's exactly what I am trying to do :) and then to get the rtl8812au drivers running. I got this running already for espressobin running debian and gateworks running debian, now I am trying to get it to work with openwrt Jun 30 00:27:58 personally i tend to use arch or alpine, each have their own issues though heh Jun 30 00:29:30 if that is the default debian install and time is useless then its clear the idiocy is getting worse in that project Jun 30 00:29:53 howudodat_: quick fix is find the symlink and nuke it Jun 30 00:30:45 personally I preferr to get a base image working and then in tree (or on board) compiling, but I understand that wont work with OpenWRT. So just looking for the "cleanest" way to get going. I can fire up an ubuntu vm in about 5 minutes. I'll go see if I can find the link as 'which time' now reports sane Jun 30 00:31:15 find / -type l -name time Jun 30 00:31:22 maybe Jun 30 00:35:12 did that and /usr/bin/time shows up, it's not a link, it's a binary that got installed just a few minutes ago, still calling busybox version. I'm gonna fire up an ubuntu vm to see if the idiocy is any better Jun 30 00:35:30 rpifan: what did you do to get it running on debian? Jun 30 00:39:14 if you get a result for -type l then its a link Jun 30 00:39:38 -type f will likely reveal a different path for the gnu version Jun 30 00:41:01 -type l only shows: /home/peter/openwrt/staging_dir/host/bin/time Jun 30 00:41:09 -type f shows usr/bin/time Jun 30 00:41:33 on Debian, GNU time gets installed to /usr/bin/time, if you actually installed the "time" package and don't rely on the shell builtin Jun 30 00:41:42 ah Jun 30 00:42:02 echo $PATH /usr/local/bin:/usr/bin:/bin:/usr/local/games:/usr/games Jun 30 00:42:25 and yes, building on Debian works just fine, the buildbots use Debian 7 or 8, I use current Debian unstable Jun 30 00:42:42 so.. what did you break howudodat_? :D Jun 30 00:42:47 LOL Jun 30 00:42:57 I'm good at breaking things Jun 30 00:43:02 ii time 1.7-25.1+b1 amd64 GNU time program for measuring CPU resource Jun 30 00:44:36 I'm not familiar with ii. file /usr/bin/time Jun 30 00:44:36 /usr/bin/time: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.32, BuildID[sha1]=cddf99ed1cbf87fbecfa55bc1b83499f0ed4b9fb, stripped Jun 30 01:06:21 ok, I fired up an ubuntu vm and it's now compiling (make defconfig, make) Jun 30 01:10:54 so if I already have a running openwrt on gateworks, can I just build an additional kernel module and then just copy the .ko file to my running board, or do I need to build a whole new openwrt image and copy the whole image over? Jun 30 01:12:04 needs to match, so likely the latter Jun 30 01:14:27 cant I get away with it if it's a build version difference (major.minor.build), or is the versioning that sensitive? Jun 30 01:16:09 the vendor image I have is kernel version 4.4, what is current openwrt kernel Jun 30 01:18:45 what target is it Jun 30 01:18:48 the generic answer is no - and in most cases the specific answer would also be no Jun 30 01:19:16 heh Jun 30 01:19:17 and when vendor kernels are involved, the answer is a resounding no Jun 30 01:19:21 yup Jun 30 01:19:41 speaking of vendor kernels Jun 30 01:19:43 come on ubnt you clowns Jun 30 01:19:44 :D Jun 30 01:19:49 wheres the gpl archive Jun 30 01:28:56 hm Jun 30 01:29:02 QCA8337N is supported, right? Jun 30 01:29:51 may have to get an ER-4 as well as the 6P Jun 30 01:30:36 howudodat_, i didnt do anything special Jun 30 01:30:38 just followed the guid Jun 30 01:30:51 yes, QCA8337 and its premium QCA8337N variant are pretty common Jun 30 01:31:12 mm Jun 30 01:31:16 (QCA8337N has hardware offloading capabilities, even if those aren't supported yet) Jun 30 01:31:18 https://wikidevi.com/wiki/Ubiquiti_Networks_EdgeRouter_4_(ER-4) Jun 30 01:31:34 suspect it'll be similar problem as the ER-POE though Jun 30 01:31:51 switch supported but nobody has figured out how to talk to it yet Jun 30 01:36:28 wish I knew more about hardware things :( Jun 30 01:41:57 I mean its not a huge deal now I guess, ubnt *should* release their 4.9 firmware soon and the gpl archive Jun 30 01:44:47 buying ubnt products and putting openwrt on them, that doesn't seem to make much sense? Especially the ones with hw offloading with anemic CPUs? Jun 30 01:46:01 depends on the product Jun 30 01:47:00 ER-X is generic enough, the octeon stuff is actually possible now as cavium have actually published their code (but nobody has imported to mainline and cavium don't even know what version control is, so they're unlikely to help) Jun 30 01:47:37 but... since the beta edgemax firmware is based on 4.9 it *should* be possible to build with openwrt without too much hassle Jun 30 01:48:45 only actually need ubnt code so that their modules can be loaded and the switch enabled, the offloading can be obtained from caviums generic kernel Jun 30 01:50:16 SwedeMike: https://github.com/Cavium-Open-Source-Distributions Jun 30 02:16:13 and anyway lets face it, edgeos sucks a bit and will suck even more when they finally make it only manageable with UNMS or whatever it is Jun 30 02:16:16 :D **** ENDING LOGGING AT Sat Jun 30 03:00:03 2018