**** BEGIN LOGGING AT Thu Aug 02 03:00:01 2018 Aug 02 03:55:27 tmn505: i made/tested the changes https://github.com/tofurky/openwrt/commits/routerstation Aug 02 03:56:12 i think it's about good to go, maybe it should all be squashed into a single commit for PR Aug 02 05:53:27 Hauke: ping Aug 02 05:53:33 let me know when you are awake Aug 02 06:32:23 Hey, is the maintainer of the zynq target active? according to the Makefile is Jason Wu . The last interaction (e-mail) I could find is from 2016 Aug 02 06:32:39 Does that means the target is considered abandoned? Aug 02 06:38:30 luaraneda: probably Aug 02 06:41:32 luaraneda: if you wnna step up, go for it Aug 02 06:43:22 Ok, so is there any requirements to be a maintainer? I'm considering it in the near future Aug 02 06:43:43 to be active Aug 02 06:47:14 That sounds easy. I will consider it after I finish porting a board Aug 02 06:51:48 blogic: Should I send an e-mail to the current maintainer asking if he still is interested on being the maintainer? (assuming that I am interested) Aug 02 06:52:33 just send a patch to the ML and put him on Cc: Aug 02 06:52:35 blogic: short pong Aug 02 06:53:42 blogic: Ok, thanks :) Aug 02 06:56:41 Hauke: can i call ? only need 2 minutes Aug 02 06:56:47 quick update regarding wifi Aug 02 06:59:37 ok Aug 02 06:59:42 blogic: ok Aug 02 07:06:26 morning Aug 02 07:19:51 wigyori: hi Aug 02 08:26:21 muhahah Aug 02 08:26:35 i managed to bump compat-wireless tot he v4.18-rc7 drivers Aug 02 08:26:56 cool Aug 02 08:27:09 indeed Aug 02 08:27:29 i am thinking of how to organize patches Aug 02 08:28:10 blogic: please ping me if I can find it in your staging tree Aug 02 08:28:19 I would like to test if it improves my ath10k horror Aug 02 08:28:45 i think i'll introduce patches-{build,backports,subsys,ath9k,ath10k,brcm,mrvl,hacks} Aug 02 08:29:05 any comments ? Aug 02 08:29:17 s:-:/: maybe? Aug 02 08:29:23 if there are going to be that many Aug 02 08:29:29 true Aug 02 08:29:41 hmmmm Aug 02 08:29:49 but sounds good to me Aug 02 08:31:12 Hello .. I have a router with two ports and using OpenWRT CC .. eth0 and eth1 .. I am configuring my router in Dumb AP mode. I followed this page: https://wiki.openwrt.org/doc/recipes/dumbap .. The problem is my SSID is working fine and getting IP + internet. But if i connect a wire to the port and connect it to my Laptop, I get the IP but no internet .. Any help?? Aug 02 08:31:28 mangix_: pong Aug 02 08:32:16 blogic: I'd start with flat patch directory hierarchy first, then see if we really need to split them Aug 02 09:37:01 anyone tried to compile lxc 3.0.0 ? Aug 02 09:42:02 nfs-kernel-server produces errors when i attempt to compile it to include it in my build Aug 02 09:42:13 i get: format not a string literal and no format arguments Aug 02 09:43:15 what's the correct way to patch this? should i attempt to patch it by inserting "%s" into all f_print where it's missing? Aug 02 09:50:08 jow thanks for fixing r71xx Aug 02 10:25:29 jow: figured out my last css quirks after updating to recent base theme. was the "treewide: convert tables to div" :) Aug 02 10:31:41 hehe Aug 02 10:32:13 * karlp is _nto_ a css/html person normally. Aug 02 10:48:55 hey all Aug 02 10:49:24 Having recently bought a GL ar150 running OpenWRT I must say I'm really impressed Aug 02 10:50:05 I've previously used dd-wrt and my experience with both has been fantastic Aug 02 10:50:19 gninrom Aug 02 10:51:16 On 18.06 I've installed package isc-dhcp-server-ipv4, but I can't find dhcpd.conf. Someone knows where I can find that? Aug 02 10:55:54 TCMSLP: glad to hear that. Aug 02 10:56:07 usually we just get bug reports :) Aug 02 11:00:23 m4t: looks good and squashing to single or two commits is a way to go. Aug 02 11:07:57 some idea? Aug 02 11:11:47 ultito: for starters, is there something particularyou need isc-dhcp for? Aug 02 11:12:58 karlp: yes Aug 02 11:13:51 karlp: it means I need isc dhcp server in 18.06 too Aug 02 11:15:41 ultito: so... where did you look for dhcpd.conf? Aug 02 11:16:14 it looks like it doens't install any file by default Aug 02 11:16:17 you have to make your own Aug 02 11:17:11 karlp: in /etc where it was in previous version of openwrt Aug 02 11:17:29 though the v6 version covers both v4 and v6 and installed dhcpdv6.conf Aug 02 11:17:32 but I could not find in in whole /etc including subdirectories Aug 02 11:17:36 dhcpd6.conf sorry Aug 02 11:18:27 it's also been converted to use uci as far as I can tell Aug 02 11:18:32 ok, will try Aug 02 11:18:41 what means uci? Aug 02 11:18:46 unless the /etc/dhcpd.conf file has been created Aug 02 11:20:43 after installing package isc-dhcp-server-ipv6 i see created /etc/dhcpd6.conf Aug 02 11:21:07 is this only for ipv6 confing, or also for ipv4 config for dhcp server? Aug 02 11:22:07 * ultito is melting down due to hot weather Aug 02 11:25:04 or should I handly create dhcpd.conf? Aug 02 11:38:44 jow: any tips on converting colspan into your new div style? Aug 02 12:10:00 karlp: sadly this is not really possible. You can define a colspan over the entire with if that is what you need Aug 02 12:10:09 *entire width Aug 02 12:10:23 but not over, say, 2 of 5 cells Aug 02 12:10:40 if you need the former, check how the "placeholder" class is defined Aug 02 12:12:36 * karlp might just add some css padding back in for my "legacy" tables Aug 02 12:20:46 karlp: that might be the simplest, yeah. I mean if you have actual tables it might make sense to keep tham es such Aug 02 13:02:13 https://ibb.co/io2Wke color 669966 https://www.artlebedev.ru/colors/ Aug 02 13:53:50 greearb_: ping Aug 02 14:00:59 jow: ping Aug 02 14:01:19 can you help me start of next week to setup a wireless backports buildbot instance ? Aug 02 14:01:39 the idea would be to have rolling builds uploaded to snapshots Aug 02 14:01:50 and new tags ending up inside a release folder Aug 02 14:04:56 Monkeh: sooo sanity check ? https://patchwork.ozlabs.org/patch/952384/ Aug 02 14:05:04 did you do that by hand ? Aug 02 14:05:18 i did a few random checks on that patch Aug 02 14:05:22 and they looked good Aug 02 14:13:43 blogic: pong Aug 02 14:13:46 yes, can try Aug 02 14:13:53 note that we're very short on resources Aug 02 14:14:04 but I can work something out Aug 02 14:14:21 question is what you want to build exactly, and how Aug 02 14:18:17 I should finally order that 64GB RAM kit for this box so I can add a buildbot slave again Aug 02 14:19:04 blogic: Yes, that was hand done - it's a bit mind numbing to review Aug 02 14:21:45 personally I'd suggest reviewing the current snapshot buildbot procedure (switch to e.g. a daily build instead of building as soon as a new commit is detected), which could free some CPU cycles to build other things in-between Aug 02 14:22:35 blogic: mkresin's planning on pushing that with some other cleanups later, so let it marinate a little? :) Aug 02 14:24:38 it would also improve consistency across the snapshot repo: it would be possible to have all targets at the same commit revision at any point in time Aug 02 15:03:47 blogic, here Aug 02 16:18:08 jow: would it make sense to drop mipsel74k as a build target? Aug 02 16:18:18 only two platforms use that Aug 02 17:11:11 is zbt we3526 locked to usa+china channels for 5 Ghz ? http://zbt.en.alibaba.com/product/60656692792-0/we3526_sim_card_wireless_modem_rj45_wireless_router_open_wrt_4g_lte_wifi.html Aug 02 17:11:49 vendor provided us with a list of channels 149 to 165 Aug 02 17:12:13 nbd, IIRC you were using this model to dev the hw NAT acceleration Aug 02 18:45:24 congrats on 1806 release, well done! (suggestion: update topic) Aug 02 18:59:01 Hello All Aug 02 18:59:08 Please help me with DS-lite Aug 02 18:59:12 After upgrade to 18.06 Aug 02 18:59:16 it's broken for me Aug 02 18:59:33 Firstly please help me with the basics Aug 02 19:00:06 I can reach Ipv6 addresses but not ipv4 addresses Aug 02 19:00:37 Anyone can help me? Aug 02 19:03:03 Nobody? Aug 02 19:45:00 Anyone know how to get the AFTR number? Aug 02 20:03:00 hello Aug 02 20:03:32 Anyone running Archer c7 with openwrt 18.06? Aug 02 20:09:19 wigyori: can you update the topic Aug 02 20:12:46 blogic, you pinged me? Aug 02 20:15:29 greearb_: yes Aug 02 20:15:36 some wif changes coming up Aug 02 20:15:41 some wifi changes coming up Aug 02 20:15:54 had a question but forgot what it was :-) Aug 02 20:16:00 heh, ok Aug 02 20:16:07 what sort of changes are you talking about? Aug 02 20:16:15 i've split the ath10k-ct-firmware out into its own package Aug 02 20:16:26 we will split compat-wireless into 2 Aug 02 20:16:40 one based on stable kernels with very small patch delta Aug 02 20:16:55 and one only for rt2x00/mt76 based on -testing Aug 02 20:17:14 so that we will in future use regularly rebasable mac80211 based on latest stable.git Aug 02 20:17:27 ok. What stable kernel version are you using? Aug 02 20:17:30 and use the FW blobs that come with said stabels Aug 02 20:17:42 right now i am testing with v4.18-rc7 Aug 02 20:17:52 fw blobs == ath10k-fw etc? Aug 02 20:18:01 yeah Aug 02 20:18:07 whatever fw a driver might need Aug 02 20:18:21 well, those are from external git trees, not directly related to the kernel, eh? Aug 02 20:18:41 so its up to you to decide if you want youe ath10k-ct to depend on mac80211-stable or mac80211-testing Aug 02 20:18:53 we will use linux-firmware.git Aug 02 20:19:13 not the kvalo tree Aug 02 20:19:46 basically move all drivers as close to upstream stable as possible and then send folks with their wifi troubles to linux-wireless if issues arise Aug 02 20:19:47 I guess stable is best for me. I'm just now working on getting 4.16 stable, can update openwrt to use that directory of ath10k-ct instead of the current 4.13 version sometime soon. Aug 02 20:20:12 i'll push my stuff to my staging tree the following days Aug 02 20:20:28 good luck with that! Aug 02 20:20:33 i've already bumped my production AP to the rc7 drivers Aug 02 20:20:40 and am runtime testing them Aug 02 20:21:06 nod Aug 02 20:21:37 upstream qca firmware and driver is generally OK these days in your testing? Aug 02 20:22:22 its been running in my production setup all day Aug 02 20:22:27 not sure really Aug 02 20:22:32 i'll know more in a few days Aug 02 20:22:36 nod Aug 02 20:22:40 once i've watched a few streams over it Aug 02 20:22:50 ath10k normally blows up after 2-3 days Aug 02 20:22:59 stintel: has composed a poem and a song about ti Aug 02 20:23:05 heh Aug 02 20:23:17 Some of us got bored with that and went to -ct. :P Aug 02 20:23:44 I'm guessing if it ever gets to 2^32 miliseconds up uptime, it will go south as well :) Aug 02 20:23:57 just based on looking at code Aug 02 20:24:42 that is option 2 Aug 02 20:24:53 just drop ath10k from the mac80211 package Aug 02 20:25:08 so have mt76/rt2x00 sit ontop of mac80211-testing Aug 02 20:25:36 ath10k-ct sitting ontop of mac80211-stable and then have mac80211-stable provide all other drivers Aug 02 20:25:49 does ath10k-ct run stable ? Aug 02 20:25:55 stintel: did you test it ? Aug 02 20:26:12 honestly i'd love to ship only one option per driver Aug 02 20:27:16 What do you mean by 'run stable'? The firmware is stable in our testing scenarios, at least mostly. And ath10k-ct driver as well. Aug 02 20:27:26 I had big problems with certain devices with ath10k, but -ct and -ct firmware appear to be solid for me Aug 02 20:27:36 Experiences seem to vary. Aug 02 20:27:47 but, we don't really test in normal AP mode much Aug 02 20:28:04 Pulling up on the drive was enough to make upstream firmware choke.. Aug 02 20:28:49 I just pushed my latest ath10k-ct driver, if someone wants to try making OpenWRT use the 4.16 instead of 4.13 driver version, please do so. I'm too busy to muck with it right now. Aug 02 20:28:54 greearb_: that is what i am worried about really Aug 02 20:29:12 greearb_: that is the other thing i worry about Aug 02 20:29:15 That we don't test AP mode? Aug 02 20:29:19 with upstream i can bump myself Aug 02 20:29:20 much Aug 02 20:29:31 yes and that you are too busy Aug 02 20:29:48 we'll see how it goes Aug 02 20:30:22 I spend 90% of my time hacking on wifi and ath10k, just we don't use openwrt much. Aug 02 20:30:28 greearb_: Isn't 2^32ms uh.. about 50 days? Aug 02 20:31:27 sounds right, it is more of justa symptom of sloppy coding in the FW that bugs me :P Aug 02 20:31:50 If the stock fw really craps out after that someone needs shooting. :) Aug 02 20:32:20 I actually will be happy enough if any FW runs 50 days :) Aug 02 20:32:34 I.. think -ct did for me Aug 02 20:32:40 Right up until I molested the power strip Aug 02 20:32:59 I'll let you know in.. 49 days? Assuming I don't fondle it further Aug 02 20:33:20 sounds good! Aug 02 20:34:14 Well, unit #2 should be safe from bad touches in any case Aug 02 20:34:53 Anyone using any wave-2 ath10k-ct firmware? Aug 02 20:35:44 Not currently - these are those 9880s everyone knows and hates Aug 02 20:35:50 nod Aug 02 20:38:04 dumb question… is there a good mPCI-e card that's 4G that has a good API for both passing IP packets, voice calls, and getting 4G time source? Aug 02 20:38:22 or 5G or LTE… Aug 02 20:38:25 You don't ask for much, do you? Aug 02 20:38:26 :) Aug 02 20:38:39 oh, and comes with a solid gold toilet. Aug 02 20:38:48 for under $200. Aug 02 20:38:59 That's still probably easier than finding a good card which works right Aug 02 20:42:30 ath10k pretty stable for me in 18.06 on hap-ac lite Aug 02 20:43:23 Monkeh: Okay, forget the toilet. This is for the US market, probably AT&T as the carrier. Aug 02 20:49:24 btw, anyone have any insight on this hack-attack vector mentioned in this bug? https://github.com/greearb/ath10k-ct/issues/27 Aug 02 20:53:48 problem is that ath10k-ct is unusably slow for me Aug 02 20:53:57 I tried all possible combinations Aug 02 20:54:21 if we can get that resolved, that would be amazing Aug 02 20:54:40 but with the current performance, neither ath10k nor ath10k-ct is suitable for anything Aug 02 20:54:41 stintel: how slow do you mean? Aug 02 20:54:58 like 400Mbps with ath10k vs 20 with ath10k-ct or so Aug 02 20:55:03 can you open a bug on that with some details about what you tried and how you tested and such, on my ath10k-ct github bug tracker? Aug 02 20:55:08 hurk Aug 02 20:55:12 wow. indeed Aug 02 20:55:13 stintel: With what chipsets? Aug 02 20:55:33 00:00.0 Network controller: Qualcomm Atheros QCA986x/988x 802.11ac Wireless Network Adapter Aug 02 20:55:46 let me reboot the thing (dmesg is filled with garbage again) Aug 02 20:55:51 in particular, my firmware should work with stock driver, and my driver should work with stock fw, would be good to know if problem is FW or driver Aug 02 20:55:52 greearb_: I will Aug 02 20:56:12 I'm certainly constrained in no such way on 9880 Aug 02 20:56:28 greearb_: I tried ath10k with your firmware, ath10k-ct with your firmware, and ath10k-ct with your htt firmware Aug 02 20:56:34 all 3 were equally slow Aug 02 20:56:56 well, that seems, er, unlikely :P Aug 02 20:57:07 and ath10k + stock firmware farts itself to death after a short time, if you have the right combination of clients Aug 02 20:57:58 we routinely get 700+Mbps OTA perf in our testing on our hardware, etc.... Aug 02 20:58:01 Now I've said that. Aug 02 20:58:17 My phone doesn't want to pass more than about 40Mbps Aug 02 20:58:26 Please hold while I get something which doesn't suck.. Aug 02 20:58:47 greearb_: I'll retry all scenarios soon'ish Aug 02 20:59:00 thanks Aug 02 21:01:28 ~140Mbps with the laptop Aug 02 21:01:39 Which is not exciting Aug 02 21:02:22 how are you testing? Aug 02 21:02:35 iperf3 Aug 02 21:02:42 where to where? Aug 02 21:02:47 Between the laptop and my workstation Aug 02 21:02:52 Gigabit backend Aug 02 21:03:10 ok. What MCS do you see on average? Aug 02 21:03:52 and does your AP have a wired port where you plug in the client (assuming server is on upstream WAN port?) Aug 02 21:04:36 I can go wired, yes Aug 02 21:04:44 It will be bottlenecked but not that bad iirc Aug 02 21:05:19 Monkeh: should be easy to monitor CPU usage during test? Aug 02 21:05:47 Monkeh: I was testing 10G with iperf, and I found I needed to do a LOT of tweaking to get optimal performance. Aug 02 21:06:02 philipp64: For 10G, sure, but rarely a problem this slow Aug 02 21:06:29 best I ever saw was 9.86GB… and I had to use 8 connections with a window of 2MB each, as I remember. Aug 02 21:06:54 thats pretty much the most you'll see annyway Aug 02 21:06:56 ayway Aug 02 21:06:57 kjddhjhdj Aug 02 21:06:59 I'm just saying don't assume the first set of numbers you get are your actual limits. Aug 02 21:07:27 just running iperf3 with no parameters, I hit 3.88 Gb/s. Aug 02 21:08:26 ldir: ping? Aug 02 21:08:49 hello Aug 02 21:09:06 iperf isn't that useful at 10G anyway, you're better off with a proper packet generator and counting sent/received Aug 02 21:09:22 Wired via the AP while I herd some cats is.. 946Mbps Aug 02 21:09:48 Monkeh, ok, so not CPU bound. What MCS do you see reported? 'iw foo station dump' ? Aug 02 21:09:59 during the test Aug 02 21:10:01 jwh: if you have a better way, mind writing a short, easy (for the dumb people like me) procedure to do it Aug 02 21:10:20 coz iperf is just too easy Aug 02 21:10:20 there isn't really, thats why people buy proper equipment heh Aug 02 21:10:29 greearb_: Well wired was via the switch, so there's no CPU use there Aug 02 21:10:37 ixia or w/e Aug 02 21:10:42 There will be some CPU limits but I recall them being rather higher than this Aug 02 21:10:51 oh, no way to put one upstream and one downstream? Aug 02 21:11:06 jwh, or lanforge ;) Aug 02 21:11:15 It's.. not impossible Aug 02 21:11:29 Monkeh, ok, so assume CPU is OK I guess, what MCS do you see? Aug 02 21:11:41 greearb_: meh, people trust ixia tests more heh Aug 02 21:12:07 jwh, yeah, I hear that too sometimes :) Aug 02 21:12:41 greearb_: VHT-MCS 6 presently Aug 02 21:13:16 1x1, 2x2, 20Mhz, 80Mhz, etc? Aug 02 21:13:39 2x2 80MHz Aug 02 21:14:07 you have external antenna? Aug 02 21:14:14 Nope Aug 02 21:14:36 Phone's getting 30-40Mbps with MCS 9, 80MHz, 1x1 Aug 02 21:14:37 maybe try rotating AP a bit or laptop? Aug 02 21:15:04 that would seem a phone limitation since the other thing runs a lot faster Aug 02 21:15:08 Yeah Aug 02 21:15:12 This phone's always been faster on LTE Aug 02 21:15:16 is it android? :D Aug 02 21:15:26 jwh: iperf3 counts dropped/retransmitted packets, too. Aug 02 21:15:30 wigyori: can you put a link to http://logs.nslu2-linux.org/livelogs/openwrt-devel/ in the /topic too, per freenode's requirements? Aug 02 21:15:31 No, I decided to spend £800 on a banana Aug 02 21:15:33 you'd hope for decent MCS at 3x3, assuming your laptop can do 3x3? Aug 02 21:15:34 my android phone performs better than last gen iphone Aug 02 21:15:36 Of course it's Android Aug 02 21:15:40 philipp64: only at layer3 Aug 02 21:15:45 greearb_: No, limited to 2x2, this is reality. :) Aug 02 21:16:18 jwh: uh… if it's dropped at layer 3 it's dropped at layer 2, also… Aug 02 21:16:21 ok, mcs 6 isn't great, but isn't awful either Aug 02 21:16:26 philipp64: umm, no Aug 02 21:16:27 lol Aug 02 21:16:40 greearb_: That was with it sat on the desk - I could ease out another 10-15Mbps by moving it around a little Aug 02 21:16:54 it could well make it all the way through the stack and then not be delivered to the application Aug 02 21:17:10 or make it to processing but be dropped in the kernel Aug 02 21:17:14 it queries the kernel for statistics on the TCB like duplicates received, out-of-window received, etc and adjusts the statistics accordingly. Aug 02 21:17:19 o.. kay, that's not helpful Aug 02 21:17:21 * Monkeh kills luci Aug 02 21:17:22 Monkeh, you might be getting a good as your environment can handle. Aug 02 21:17:55 theres also what the nic received vs what it passed on Aug 02 21:17:58 greearb_: Well, I've done >250Mbps through block walls to an MT7612 with this laptop Aug 02 21:18:04 Around 400 without walls Aug 02 21:18:20 Monkeh: linux typically allocates twice the receiver window to buffering, so it's pretty hard to get a received packet to be dropped. Aug 02 21:18:36 but i don't really care about end to end tests, only forwardnig Aug 02 21:18:38 forwarding Aug 02 21:18:54 if you want to put a WLE900VX in a power-ful-ish PC and make an AP of it, then you can test and be sure it is not CPU of some small embedded processor thing causing trouble Aug 02 21:19:16 greearb_: Eh, looks like it is the CPU limit - I recall it being closer to 200Mbps but it's pretty pegged right now Aug 02 21:19:22 That was a couple kernel bumps ago Aug 02 21:19:52 So cannot reproduce stintel's magic. Aug 02 21:19:57 DonkeyHotei: umm, archives ? Aug 02 21:20:03 if there's a discrepancy in CPU usage between -ct and non-ct, isn't that worth looking into? Aug 02 21:20:11 Monkeh: what do you mean? 400Mbps+ ? Aug 02 21:20:19 stintel: No, 20Mbps Aug 02 21:20:31 stintel: I have some Lantiq limitations on going really fast with ath10k. Aug 02 21:20:43 wigyori: freenode has a requirement that any public logs must be disclosed in either the onjoin msg or in the topic Aug 02 21:21:08 or better yet, stop logging :D Aug 02 21:21:13 if there is cpu usage difference, I would like to know about it Aug 02 21:21:53 greearb_: Reloading with upstream firmware for a test Aug 02 21:22:17 DonkeyHotei: mkay, i saw that only in one channel i'm on on freenode, but sure, one sec Aug 02 21:22:42 greearb_: I'm pretty sure I'm not environmentally limited at any rate Aug 02 21:23:00 https://gist.github.com/stintel/66e3b9e49321d94ed9f0dc2a032b77ba Aug 02 21:23:03 ty Aug 02 21:23:23 As I said, I've had some stellar performances around here even with walls, I know these APs are a little restricted Aug 02 21:23:42 * Monkeh waits patiently for DFS Aug 02 21:24:10 new sofa? Aug 02 21:25:13 greearb_: Upstream fw is actually 20-30Mbps slower. Aug 02 21:26:13 With about 10% less CPU usage Aug 02 21:26:25 ok so I guess at least -ct FW works well in some cases. Aug 02 21:26:59 Stellar for me Aug 02 21:27:06 hm, sysupgrade time soon Aug 02 21:27:14 100% expect this to go horribly wrong Aug 02 21:28:26 stintel: Exactly what hardware are you using? Aug 02 21:35:50 * f00b4r0 notes that Luci disagrees with his laptop on what the current wireless bitrate is Aug 02 21:36:38 f00b4r0: they're probably both wrong Aug 02 21:36:53 ldir: heh ;) Aug 02 21:38:43 ldir: Probably? :) Aug 02 21:39:26 ok, they're both lying to you - and so is the cake ;-) Aug 02 21:40:15 The cake isn't lying Aug 02 21:40:28 * ldir sends a few radar pulses in Monkeh's direction just to make sure. Aug 02 21:40:30 Being a lie and lying are not the same :) Aug 02 21:40:44 Oi, don't anger the DFS Aug 02 21:40:49 You'll ruin my test Aug 02 21:41:27 Monkeh: btw, remember my mt76 reset woes and suspicion on bad cabling? Aug 02 21:41:53 ooh, that new traverse board looks nice Aug 02 21:41:57 well, cabling isn't the problem. Devices are more stable since I up'd them to 18.06. And they're even more stable with 5GHz disabled, apparently Aug 02 21:41:59 f00b4r0: you were going to knock a wall down? ;-) Aug 02 21:42:02 wonder how outrageously expensive it is like their geos boards Aug 02 21:42:08 ldir: turns out I don't have to ;) Aug 02 21:42:12 f00b4r0: Still might be the cabling. ;) Aug 02 21:42:17 heh :) Aug 02 21:42:23 RFI's a thing Aug 02 21:42:26 true Aug 02 21:42:33 but that seems a bit farfetched Aug 02 21:42:45 You still need to fix the cabling Aug 02 21:42:46 :P Aug 02 21:43:29 well. I *should*. Whether I *need* is another matter ;^) Aug 02 21:43:44 We can fix that.. Aug 02 21:43:57 Wear a tin foil hat to deflect the RFI Aug 02 21:44:13 maybe if I wrap my APs in tinfoil they'll work better ;D Aug 02 21:44:17 https://traverse.com.au/docs/ls1043s/ Aug 02 21:44:21 reaaaally Aug 02 21:44:48 f00b4r0: yeah, put your APs in metal boxes to be really sure ;-) Aug 02 21:45:12 ldir: lead sounds like a good idea ;) Aug 02 21:45:25 know how i have been saying there is a lack of useful sfp boards Aug 02 21:45:32 traverse to the rescue? Aug 02 21:46:19 jwh: Can we use them usefully? :P Aug 02 21:46:44 they've upstreamed their openwrt fork Aug 02 21:46:46 so guess so Aug 02 21:46:57 And how many zeros are involved? Aug 02 21:46:57 * ldir knows all about RFI from faulty switch mode PSUs and the effects of ADSL. 3 bloody times I've had to fix/replace stuff in near neighbours houses Aug 02 21:46:59 https://traverse.com.au/docs/ls1043s/sfp/#sfp-gpios Aug 02 21:47:04 Monkeh: heh! Aug 02 21:47:10 australia, so everything is expensive Aug 02 21:47:31 long time ago I used their geos boards Aug 02 21:47:32 ldir: chinese wallwarts. That plague. Aug 02 21:47:47 was about 500 quid to get a dual port board from there to the uk Aug 02 21:48:03 with a couple of hundred for shipping and customs Aug 02 21:48:04 fuckers Aug 02 21:48:21 maybe they suck less at distribution now Aug 02 21:48:28 2 faulty satellite RXs. 1 faulty wallwart......itself the PSU for the neighbours ADSL router! Aug 02 21:48:50 ooh train arrives /me goes home. byeee Aug 02 21:48:51 heh Aug 02 21:48:52 box is well overspeced though Aug 02 21:49:07 needs a lite version :d Aug 02 21:49:17 * Monkeh sneaks the bluray player downstairs into ldirs presence Aug 02 21:49:24 https://traverse.com.au/distributors/ Aug 02 21:49:27 nope i lied Aug 02 21:49:28 still suck Aug 02 21:50:29 the french one listed doesn't even have the board Aug 02 21:51:00 Antéor. Never heard of them Aug 02 21:51:03 same sort of nonsense as it as a decade ago Aug 02 21:51:13 nice kit but unobtainable Aug 02 21:51:31 Make your own Aug 02 21:51:36 because nobody imports them and sod getting them shipped from australia Aug 02 21:52:09 sure, teach me how to do board design and electronics Aug 02 21:52:14 and i'll make you all some nice boards Aug 02 21:52:15 :D Aug 02 21:52:24 I already have some of the skills, fund me. ;P Aug 02 21:52:39 lol Aug 02 21:52:39 same here ;) Aug 02 21:52:45 fix my rb2011 first Aug 02 21:52:55 well, fix everyones rb2011 Aug 02 21:52:59 More importantly: Sweet-talk the dickbags at vendors to give me documentation which is useful Aug 02 21:53:14 all the code for that board is in git at least Aug 02 21:53:18 mainline u-boot too Aug 02 21:53:35 Sounds great, import 'em :P Aug 02 21:53:36 looks like they upstreamed all of it too Aug 02 21:53:52 probably a grand a board Aug 02 21:53:56 they have ridiculous prices Aug 02 21:54:10 So they're like the Australian AVM, with upstreaming Aug 02 21:54:27 yeah I guess Aug 02 21:54:57 https://traverse.com.au/products/five64-arm64-platform/ Aug 02 21:54:58 lol Aug 02 21:55:00 ridiculous spec Aug 02 21:55:23 nice, right? I keep trying to get one. Aug 02 21:55:36 gonna be well expensive Aug 02 21:55:58 I used to maintain Traverse's Geos2 platform on OpenWrt. Aug 02 21:56:21 heh Aug 02 21:56:28 geos was a good idea Aug 02 21:56:38 shame the solos modems never performed well enough to be useful Aug 02 21:57:37 had some test lines where the solos would get a max sync rate of 2 or 3 meg Aug 02 21:57:44 broadcom on the same line happy at 13-14 Aug 02 21:58:50 here (.fr) that's because all DSLAMs are bcm Aug 02 21:59:19 heh, lantiq was much better than solos too Aug 02 22:00:44 obviously nothing beats bcm though Aug 02 22:00:47 they've always been better at it Aug 02 22:00:56 even with non-bcm dslams Aug 02 22:00:59 What, industrial espionage? Aug 02 22:01:13 so bcm modems have the fancy stuff that talks to bcm dslams and get the best perf. That's typically why the TD-W9970 is pretty popular here Aug 02 22:01:28 Monkeh: heh Aug 02 22:01:57 right, better do one final build and sysupgrade some kit Aug 02 22:02:10 fully expect to still be here in 10 hours trying to unbreak it Aug 02 22:03:55 need to back out a bunch of changes to my tree now as well Aug 02 22:04:00 since pkgadd showed me the light Aug 02 22:04:28 maybe with a couple more things I may actually be able to use a vanilla tree Aug 02 22:10:39 fuuuuuuuuuuu conflicts Aug 02 22:11:14 oh, lol Aug 02 22:11:27 looks like someone committed almost exactly the same thing I was meaning to upstream Aug 02 22:12:18 https://github.com/openwrt/openwrt/commit/8492ad0cc183550137ea5afadcbebbfb4fd68fed#diff-93a47ed48038855e7c8ee34bcc356855 Aug 02 22:58:41 k, before I fuck this up again, any git wizards? :D Aug 02 22:59:27 when editing conflicts, I'm obviously doing something wrong as git can't figure out whats changed Aug 02 23:01:32 jwh, git pull --rebase ? Aug 02 23:01:39 yeah Aug 02 23:01:42 so like, I have a conflict Aug 02 23:01:50 >>>>, ==== etc Aug 02 23:01:51 usual Aug 02 23:01:54 3 ay merge Aug 02 23:01:55 way* Aug 02 23:02:10 question is, how do I edit it properly to not upset git Aug 02 23:02:33 edit it to be correct, check with: git diff HEAD or maybe HEAD~1 ? Aug 02 23:03:25 I don't really get 3-way merges thats the problem I think, however I edit it to look like what it should be, git gets upset Aug 02 23:03:46 how is it upset? Aug 02 23:04:03 usually says no changes when doing --continue Aug 02 23:04:08 hm, maybe I got it this time Aug 02 23:04:16 nope Aug 02 23:04:21 dumbass vcs Aug 02 23:04:21 :D Aug 02 23:04:30 if there no no changes, just use --skip ? Aug 02 23:04:35 there are Aug 02 23:04:44 git add [file-that-changed] ? Aug 02 23:04:49 this is retarded, whichever way I do it, its just garbage Aug 02 23:05:06 if I edit the conflicting file so it looks like it should (so without merge pointers), it says no changes Aug 02 23:05:26 if I just edit the conflicting bit but leave the merge pointers intact, it leaves the actual file in the repo as that Aug 02 23:05:39 why is this hard, tools have been doing 3-way merges for decades without a problem Aug 02 23:05:52 only git gets it wrong Aug 02 23:05:58 what exact '--continue' command do you run? Aug 02 23:06:11 git rebase --continue, obviously :D Aug 02 23:06:49 you try 'git add' on the file you changed? Aug 02 23:06:56 obviously I did that Aug 02 23:07:02 but "no changes" Aug 02 23:07:05 because there are no merge pointers Aug 02 23:07:10 as I removed them... Aug 02 23:07:20 if I leave them in git will commit the pointers Aug 02 23:07:29 but without them in, git doesn't know there are any changes Aug 02 23:07:36 even though there clearly are Aug 02 23:07:41 dunno, I haven't hit that yet I guess...maybe add a small comment and commit that? Aug 02 23:07:53 and then can go back with 'stg' or whatever and fix it up later? Aug 02 23:08:07 heh Aug 02 23:08:15 git is nice and all but cvs does a better job at this bit Aug 02 23:08:23 *shudder* Aug 02 23:08:26 :D Aug 02 23:08:32 everything else about it sucks, but merges are ok Aug 02 23:09:04 I haven't used it in a decade....ever since I could import my cvs tree into git ages ago :) Aug 02 23:09:34 see now if it was a normal merge conflict its ok Aug 02 23:09:50 but when its a rebase it just gets it wrong Aug 02 23:09:53 I don't see why Aug 02 23:10:39 maybe I need some of those magical barely documented config options Aug 02 23:10:48 you can manually rebase with 'git format-patch' and then git am Aug 02 23:11:01 I do that when really paranoid Aug 02 23:11:19 and make lots of copies of your tree first Aug 02 23:11:25 well in this case I'm tracking master, so pretty typical workflow: fetch upstream, rebase upstream/master Aug 02 23:11:30 heh Aug 02 23:11:49 well I *do* have two copies of the essentially the same thing as branches Aug 02 23:11:58 so I can just rename one if git rebase fucks it up (it does a lot) Aug 02 23:12:19 but I need to figure out *why* and change my workflow so it doesn't keep happening Aug 02 23:13:47 since its obviously me doing it wrong (git is perfect, ofc), I need to see a before and after example of a merge conflict Aug 02 23:18:56 shame git isn't just a tiny bit more intelligent, like being able to figure out that actually rewritten histories are the same and not making me force push every time Aug 02 23:22:30 k, its totally broken the entire tree now :D Aug 02 23:24:19 about time I started with a fresh tree anyway, since I have been shown the light and I don't need so many changes Aug 02 23:29:47 lol, exact same commands on 2 branches, worked on one but not the other Aug 03 01:31:15 ah this is so much easier, wish I'd discovered this earlier Aug 03 01:32:43 The manual? Aug 03 01:36:17 fu Aug 03 01:36:22 magic config symbols Aug 03 01:36:38 don't need no stinkin' commits Aug 03 02:05:26 is patchwork down?\ Aug 03 02:06:43 Looks it. **** ENDING LOGGING AT Fri Aug 03 03:00:00 2018