**** BEGIN LOGGING AT Mon Jul 01 02:59:57 2019 Jul 01 08:18:55 gonna reboot into 5.1.15 on my workstation so we will probably get 5.1.16 (and 4.19.185 and ...) later today Jul 01 08:18:58 :P Jul 01 08:45:32 * ldir laughs at stintel for being behind the times - running net-next here :P Jul 01 08:45:56 when brave people meet... Jul 01 08:45:57 :) Jul 01 08:46:19 ldir: only mainline here. this box pays my bills Jul 01 08:46:25 can't afford instabilities Jul 01 08:46:50 'stable' ???? Jul 01 08:47:33 :P Jul 01 08:49:33 I was only running net-next 'cos I was developing for next kernel and in the network subsystem. Helps to make sure patches apply and the code runs. Jul 01 08:50:21 I did get slightly bitten by developing on net-next, compile testing there but not actually quite run testing... instead run testing on backports Jul 01 08:50:24 makes sense - I'd do it in a vm ;) Jul 01 08:52:20 I don't know if I'm mad (yes) but there's something nice about running bare metal and I've a 'sacrificial' laptop Jul 01 08:53:58 and there's also the story of parallels v debian v recent kernels - it was quicker & easier to go bare metal Jul 01 08:54:54 sacrificial laptop sounds fun :P Jul 01 08:55:03 can I borrow it for the shooting range someday ? Jul 01 08:55:19 it's the goats that are in short supply Jul 01 11:41:50 ldir: just opened the window and first thing is see is goats Jul 01 11:42:01 i wont read the backlog Jul 01 11:48:03 blogic: it's just your usual Satanic goat sacrificing going on, you know, nothing out of the ordinary. Jul 01 11:48:22 Pretty normal stuff when dealing with networking Jul 01 11:48:25 ah that kind of goat stuff Jul 01 11:49:10 😂 Jul 01 11:49:25 yeah, nothing really icky like making cheese Jul 01 11:50:11 stintel: how did you do that ? Jul 01 11:50:28 i upgraded my debian and now i get emojis in irssi Jul 01 11:50:32 :D Jul 01 11:50:44 ehr, I have a thing in irssi Jul 01 11:51:03 ldir has one in goats, you in irssi, whats next Jul 01 11:51:09 you can copy-paste them ofcourse but my irssi replaces : ' ) with 😂 Jul 01 11:51:27 completions is what they are called Jul 01 11:51:30 :') Jul 01 11:51:41 The emoji plague spreads far and wide Jul 01 11:51:56 at least we're still on IRC ;) Jul 01 11:52:02 we should move the channel to a whatsapp group Jul 01 11:52:08 blogic: No no, Discord Jul 01 11:52:12 actually I made that completion for work slack Jul 01 11:52:15 It's apparently the best thing ever. Jul 01 11:52:17 I used that emoji a lot Jul 01 11:52:33 the problem with discord or slack or whatever is that every project now has 100 channels Jul 01 11:52:51 and maintaining that amount of channels in irssi ... meh Jul 01 11:52:51 That and you only get half as much text on screen Jul 01 11:52:58 half? optimist :D Jul 01 11:52:58 IRC's still better. Jul 01 11:53:06 Monkeh: yarp Jul 01 11:53:06 (use slack plugin with weechat) Jul 01 11:53:19 karlp: I use matterircd Jul 01 11:53:22 looks like irc, tastes like irc, just keeps everyone int he slack world happy that you're in their world Jul 01 11:53:32 weeslack is better but I don't use weechat Jul 01 11:53:40 stintel:that sounds suspiciiously like a daemon that I'd need to be running, so no :) Jul 01 11:54:02 I've tried migrating a few times from irssi to weechat but just too much things that I need to "port" Jul 01 11:54:11 well, when slack killed the irc gateway, I needed to replace irssi, and weechat is ~drop in for me at least. Jul 01 11:54:25 changes weren't enough to justify running new services for at least :) Jul 01 11:54:30 I have too much legacy stuff :P Jul 01 11:54:49 stintel: I think in this channel we all have to much legacy stuff of one variety or another :P Jul 01 11:55:29 [kc] rikwetmusegg lavmonwetvij Jul 01 11:55:36 ppfpff fmpmfpmfffmm Jul 01 11:55:56 they're pretty useless but if people use them it's no fun to not know what they wrote :P Jul 01 11:57:03 i need to buy a new laptop Jul 01 11:57:11 my hdd is too small Jul 01 11:57:33 I replaced my SSD not so long ago, from 512GB to 1TB Jul 01 11:57:52 still waiting for ultrabooks with 32GB to become available before investing in a new laptop Jul 01 11:58:01 Moar RAM Jul 01 11:58:24 I'd settle for a laptop that doesn't thermally throttle the cpu all the time.... Jul 01 11:58:36 karlp: run thermald Jul 01 11:58:48 how's that going to help? Jul 01 11:58:52 it helps for me Jul 01 11:59:18 it limits max frequencies based on all kinds of parameters, which is by far better than throttle due to overheat Jul 01 11:59:20 how, by allowing it to run hotter? Jul 01 11:59:43 or is the goal to cap perf earlier, so it doesn't overheat? Jul 01 11:59:47 yeah that Jul 01 11:59:48 try and stay closer, but not over? Jul 01 11:59:52 ok, that makes sense I guess. Jul 01 12:00:43 what I did before if it annoyed me too much was setting the clock frequency just below the max non-turbo frequency, which effectively disabled turbo and solved the whole throttling too Jul 01 12:00:58 ... but that annoyed you too much? Jul 01 12:01:00 but with thermald you can still have bursts, but shorter Jul 01 12:01:39 feel free to have a look at it, or not ;) Jul 01 12:01:45 well I just installed it, Jul 01 12:01:50 and .... Jul 01 12:01:56 Flames? Jul 01 12:02:01 I'll try and read a bit, it's not given a great out of box experience at least. Jul 01 12:02:20 yeah, there's this flag I use that makes it magically do stuff Jul 01 12:02:33 but on my workstation now and laptop is sleeping Jul 01 12:04:06 --exclusive-control I think Jul 01 12:12:05 jow are you aware of this project? https://github.com/zhaojh329/oui Jul 01 13:08:49 aparcar[m]: no. can you give it a try? Jul 01 13:36:58 jow: will do. Maybe it is possible to suggest some "simple" views in addition. He already implemented the RPC in C Jul 01 14:09:43 jow: https://patchwork.ozlabs.org/patch/1116473/ looks legit. comments ? Jul 01 14:10:06 actually the sleep 1 is quirky Jul 01 14:10:25 why dont they just kill batman ? Jul 01 15:03:44 ap I know, looks like my code Jul 01 15:04:10 aparcar[m]: Jul 01 15:05:15 assuming you mean the stuff in oui-rpc-core Jul 01 15:11:46 blogic: you sound like a dc villain ;) Jul 01 15:13:31 I would imagine that if you tried to kill the batman, the batman would kill you first. Jul 01 15:14:27 KanjiMonster: :-) Jul 01 15:17:23 Just seen this tweet on twitter. "OpenWrt 18.06.4 released: Linux operating system targeting embedded devices https://meterpreter.org/openwrt/ #info #news #tech " Jul 01 15:17:51 What the heck is that site? Jul 01 15:18:38 blogic: stopping wifi makes sense in general Jul 01 15:18:54 likely frees quite some amount of ram Jul 01 15:20:35 They are telling people that 18.06.4 is the same as snapshots by macking the download link go to the snapshots folder Jul 01 15:33:16 its an automatic news aggregator cobbling together random pieces of info Jul 01 15:38:12 for curisoity sake, I've always wondered, if you are to write a bootloader what information do you need? The SoC datasheet and also the datasheet and programming guide of RAM Jul 01 15:38:19 is tht correct? Is there anything else? Jul 01 15:53:48 a way to recover the board Jul 01 15:54:05 because most likely you'll be flashing a bad bootloader or 2 on the way t success Jul 01 15:59:42 stintel: well, thermald reports nothing, and ThermalMonitor shows it still well below thresholds, but kernel log still has "mce: CPUx above threshold, clock throttled" type messages Jul 01 16:00:06 unless it's just that it's out of the box limits ar too high. Jul 01 16:00:16 Thanks blogic Jul 01 16:02:46 and intel decided to use google+ to document things, so that's all gone... Jul 01 16:07:18 karlp: did you try with that --exclusive-control option ? Jul 01 16:09:20 bah, need to order more USB cables Jul 01 16:09:29 it looks like it's set to trip at 95 based on something, but something else is throttling at 75 Jul 01 16:09:36 out of time for this today. Jul 01 16:19:51 Where is the make rule for the canonical kernel+squashfs sysupgrade.bin defined? Jul 01 16:20:50 (when not explicitly defined in the `define Device/mfgr_board` stanza) Jul 01 16:24:16 * russell-- gets on plane for berlin this evening (pacific coast time) Jul 01 16:24:48 russell--: do you stay for longer? Jul 01 16:25:50 i'm going to battlemesh in paris on july 8 Jul 01 16:26:13 ahh, right Jul 01 16:26:19 totally forgot Jul 01 16:27:10 cool Jul 01 16:27:36 very pleased the weather forecast has cooled off, fwiw Jul 01 16:32:19 @jow: Checked adding `usleep` to busybox some time ago and it surprisingly didn't add anything to the busybox binary size. My need for it is deferred, but something to tuck away if you run across it again Jul 01 17:49:34 I just saw the patch on the mailing list for improving reliability of sysupgrade by adding wifi down + sleep 1 Jul 01 17:49:41 yay... success. Just completed the first build using a fully dockerized build cluster Jul 01 17:49:48 I solved the problem a bit differently and have been carrying a local patch for a long time Jul 01 17:50:48 Spacing out the loop iterations by sleeping 1 second between each iteration has been a reliable work-around for me Jul 01 17:51:26 kristrev: sleeping to wait for what exactly though? Jul 01 17:51:27 (I agree thay any kind of sleep-based solution is not very nice :)) Jul 01 17:51:37 to wait for wifi to shut down? Jul 01 17:51:44 In my case it also wifi that causes sysupgrade to fail Jul 01 17:52:17 I have not had a chance to take a look at exactly which part of wifi that fails Jul 01 17:52:22 kristrev: wifi down has been quick and robust for me. It doesn't address some of the reported issues around swap, but I couldn't find any "steps to reproduce" for those Jul 01 17:52:47 But I just looked through my notes, and it seems that, at least on my devices, a single sleep 1 has been enough Jul 01 17:53:12 However, I guess I had the you can never be too careful-approach when adding this work-around Jul 01 17:53:20 Problem seems to be respawning of hostap-related processes. I forget if just hostapd or if wpa-supplicant was involved as well Jul 01 17:53:23 I wonder if we could at least refactor the loop into something like while [ -e /sys/class/net/wlan0 ]; do sleep 1; done Jul 01 17:54:08 but anyhow, whats actually needed is some proper shutdown / "init 1" mode in procd Jul 01 17:54:16 that killall loop is not nice Jul 01 17:54:36 That should work I guess. Perhaps extended to check for any wlan-interfaces Jul 01 17:56:40 if for some reason it fails to bring the interface(s) down, it'd loop forever... Jul 01 17:57:14 jeffsf: I see that initially also added a running wifi down, but then generalized to "sleep 1" to cover other applications too (not that I have seen any on my devices) Jul 01 17:58:35 I added the "sleep 1" there, if `wifi` exists, just to give it a tiny bit of time to gracefully exit, before the killall, no, really killall loops Jul 01 18:14:35 OpenWrt's tcpdump does not seem to support monitor mode on my WLAN hardware. What is the recommended method to capture WiFi traffic for analysis in the absence of tcpdump's monitor mode? Where should I start? Jul 01 18:15:35 `tcpdump-mini` has worked for me, at least with "option mode 'monitor'" Jul 01 18:18:08 I have the full (unabridged) tcpdump version. It is the one that does not seem to support the monitor mode. My WLAN hardware is ath10k, and the driver is ath10k-firmware-qca99x0-ct, Jul 01 18:42:02 jeffsf: have you seen sysupgrade performing an 'upgrade' but coming back up with the previous firmware? Jul 01 18:42:05 with your wifi issue Jul 01 18:42:28 i have defaulted to bringing wifi down in my upgrade scripts prior to sysupgrading Jul 01 18:43:19 I actually see that occasionally with my UAP-AC-PRO. I had never tried turning WiFi off to fix it; I have just tried again and it usually then works. Jul 01 18:44:05 Borromini: Yes, those are the symptoms. I'll need to figure out why my serial isn't working on my EA8300 to address John Crispin's questions, but I believe it is that the hostap processes are set to respawn. Jul 01 18:44:05 ok Jul 01 18:44:55 mamarley: i have a sysupgrade wrapper that runs locally, so there's no issues turning off wifi prior to sysupgrading Jul 01 18:44:59 jeffsf: ok :) Jul 01 18:46:22 I poked around for a while on showing "error codes" through the LED, but the refactoring required was too invasive to tackle in the ath79 work I was doing. Jul 01 18:46:23 Borromini: It wouldn't be an issue for me either, my device is just a WAP and I always connect over the Ethernet interface to upgrade it anyway. Jul 01 18:50:19 yeah :) Jul 01 18:53:38 Hauke: it looks like some compiler optimisation flags made my dir-860l choke at some point. from what i can tell others have been running dir-860l on master/19.07 builds without too many issues, so it looks like a rather isolated case Jul 01 18:54:05 referring to FS #2297 that is =) Jul 01 19:01:03 * Slimey grinds his Netgear R7000 into small granular bits. Jul 01 19:02:11 broadcom? =) Jul 01 19:05:18 yesss Jul 01 19:05:37 Can someone suggest what I should use to capture WiFi traffice please? I have one WiFi client, an older iPod, that refuses to connect to my WiFi (for no obvious reason) Jul 01 19:06:38 got another pc that can boot kali? Jul 01 19:07:24 802.11ac "forced" with no fall-back to 802.11n? -- At least the iPhone 5s and prior don't support 802.11ac Jul 01 19:08:03 no, I have other clients connecting with 802.11n and even 802.11b without problems Jul 01 19:08:32 this is 2.4GHz radio Jul 01 19:09:12 i'm looking to replace BCM43142 in a laptop, that is 802.11bgn with BT 4.0 Jul 01 19:09:19 i plan to replace it with AR9463 Jul 01 19:09:33 that chipset is also 802.11bgn with BT 4.0 Jul 01 19:09:43 should AR9463 work well with ath9k? Jul 01 19:10:02 valku: right you can use airodump-ng with kali to log packets to file for review with wireshark later Jul 01 19:10:45 rmilecki yes Jul 01 19:11:49 Slimey : Looks like aircrack-ng is available in Ubuntu repos... Thank you for the suggestion, I am going to try it out with my Ubuntu 18.04 box Jul 01 19:11:53 nbd: any arguments against AR9463? Jul 01 19:12:05 valku: yes that should work also Jul 01 19:23:33 Hi people how I build a image with a difrent IP? Jul 01 19:24:01 I have a dumb AP I want to reset, but I have it in a cuby Jul 01 19:24:31 I would like to firstboot it but can I build a IP in to the bin file I flash? Jul 01 19:25:02 I want it to come up on 192.168.1.2 and not 1.1 Jul 01 19:25:51 Can I do that in make menuconfig? Jul 01 19:26:05 Tapper : in make menuconfig, make sure "Image configuration" is checked, then in "Image configuration" go to "Preinit configuration options" and set your IP Jul 01 19:26:20 valku thanks Jul 01 19:27:27 In buildroots menuconfig you can set default ip (I've never tested), or you can include an etc/config/network file that has correct settings in buildroot /files/ folder before make Jul 01 19:29:22 Tapper: I handle that kind of thing with openwrt/files/etc/config/* Jul 01 19:30:08 Handles my wireless-connected APs as well Jul 01 19:31:45 preinit ip != lan ip Jul 01 19:32:08 AFAIK you can't set default lan ip via menuconfig Jul 01 19:35:03 Guys, can somebody look at this issue https://bugs.openwrt.org/index.php?do=details&task_id=2304 ? This is not fixed at all. The bug report is about OpenWrt 18.06 release. There is still wireguard version 0.0.20190123 - https://github.com/openwrt/openwrt/tree/openwrt-18.06/package/network/services/wireguard could it be updated to the latest version? Jul 01 19:38:49 valku Hi does that chang the IP when the devices is fully booted? Jul 01 19:38:56 change** Jul 01 19:39:28 It should, but I have never tested it myself (no need so far) Jul 01 19:40:18 OK thanks ynezz Jul 01 19:40:40 I don't have a /files/ in my build root do I have to make one? Jul 01 19:42:48 OK jeffsf so it seems that this is a thing you have working can you explain a bit more for me pleas. Jul 01 19:43:30 High level, whatever is in openwrt/files/ appears in the same location in your ROM Jul 01 19:43:59 so openwrt/files/etc/config/network is the ROM's notion of /etc/config/network Jul 01 19:44:00 I don't have a dir called openwrt/files do I have to make it? Jul 01 19:44:17 $(TOPDIR)/files Jul 01 19:44:24 not there by defauylt Jul 01 19:44:49 OK I will have a go. Jul 01 19:45:00 can also be a symlink to, for example $(TOPDIR)/env/files -- if you're using the `env` script or managing it yourself with git Jul 01 19:45:04 Tapper : you may want to have a look at package/base-files/files/bin/config_generate Jul 01 19:45:35 https://openwrt.org/docs/guide-developer/build-system/use-buildsystem#custom_files Jul 01 19:45:44 valku I am crap at working stuff out on my jack sorry lol Jul 01 19:45:55 thanks Jul 01 19:46:34 because /etc/config/network is generated on the fly during install. This config file is then retained for all future sysupgrades Jul 01 19:46:43 I get a config working on a router, then copy it to my files directory Jul 01 19:47:16 Take a look at jeff@deb-devel:~/devel/openwrt$ ./scripts/env --help Jul 01 19:47:49 It is a nice tool for managing multiple configs like that without worrying about the git commands themselves Jul 01 19:48:10 $ ./scripts/env --help Jul 01 19:59:28 Borromini: were you talking about the kernel crashes you first tought were related to the copy user hardening? Jul 01 20:15:55 Pepe: https://git.openwrt.org/aced9de9a4c2cc19e3e7a27cf1af40b9748aeccd Jul 01 20:23:45 Pepe: anyway, wireguard is one of the packages which should be moved to packages feed repository and maintained there Jul 01 20:24:10 so would be nice if you could take it over :) Jul 01 20:30:04 ynezz: why move it? Jul 01 20:30:08 seems better where it is now Jul 01 20:34:31 there was a discussion in Hamburg about this topic, conclusion was, that we should clearly define what packages should be in the master tree, and move the rest to the packages feed Jul 01 20:36:49 main reason is probably the maintenance burden Jul 01 20:40:02 the question is, why should be wireguard, openvpn (and probably few others) maintained in the master tree, if none of the devices depend on this packages, in other words need this packages in order to boot the device properly Jul 01 20:40:53 this is in short what I still remember from the discussion :) Jul 01 20:51:38 Hauke: yes i was Jul 01 21:03:28 I've been zapped by a mini-migraine so the following may not make sense. Jul 01 21:05:05 from what I vaguely remember (and probably misunderstand) there was pressure to have certain modules such as cake & wireguard in the master tree because at the time they were 'fast moving' and had unstable kernel/user space APIs. Jul 01 21:06:38 This certainly isn't the case for cake anymore, nor wireguard. cake is upstream too. I've been tempted for a while to drop the cake package but import the module as a 'backported' kernel patch. Jul 01 21:07:06 a bit like how I've done for act_ctinfo Jul 01 21:08:05 now that we're on K4.19 backporting multiple versions of cake isn't too much of a burden. Jul 01 21:09:06 in fact, 4.19 has cake natively. Hmmm. Jul 01 22:37:24 ldir: ACK Jul 01 22:39:55 ynezz: are there any further issues with PR 1359? Jul 01 23:58:59 Is no one else running 4.19? Jul 02 00:02:28 Been running 4.19 on ath79 and ipq4019 for a long time **** ENDING LOGGING AT Tue Jul 02 02:59:56 2019