**** BEGIN LOGGING AT Sun Aug 12 03:00:00 2018 Aug 12 05:30:26 hi Aug 12 05:33:18 hi Aug 12 08:42:12 quick question: i see tons of pkg-config/dev installs for all kinds of libs, yet (pkg-config/bin --libs) is only used by python3? Do packages use this automatically in there own configure scripts or are we just not using this that often? Aug 12 10:29:09 build #101 of at91/legacy is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/at91%2Flegacy/builds/101 Aug 12 10:46:09 jow: where do I find the faillogs for the 17.01 and 18.06 release? Aug 12 10:46:13 I only found this: https://downloads.lede-project.org/releases/faillogs/ Aug 12 10:48:11 could someone tweet about the 18.06 and 17.01.5 releases: https://twitter.com/lede_project Aug 12 10:59:31 jwh: so you've been following https://github.com/vdorst/openwrt/tree/mt7621-ramips-dsa2 then? Aug 12 11:01:34 I a now Aug 12 11:01:36 am* Aug 12 11:01:59 that's Rene__ 's tree Aug 12 11:02:04 may have to lift some of those bits for my own Aug 12 11:02:05 DSA+++ Aug 12 11:41:36 DonkeyHotei: is there a discussion about current afairs somewhere? Aug 12 11:44:52 abenz: jwh is cataloguing the state of sfp support at https://jwh.me.uk/2018/08/12/openwrt-and-sfp-support/ Aug 12 11:45:12 I'll make it a useful table once I figure out how this wordpress crap works :D Aug 12 11:45:58 ah, I thought we were talking about general DSA switch Aug 12 11:46:29 switch in my msg above as in switching to DSA driver eventually Aug 12 11:46:51 DSA does change how sfp is presented, so hm Aug 12 11:49:13 May i make a suggestion for the 18.06.1 point release? Aug 12 11:52:52 one: CVE-2018-5390, two: why in the name of all that is holy is udhcpc push the local hostname when requesting? Aug 12 11:53:57 CVE-2018-5390 was patched in the 18.06 branch 4 days ago Aug 12 11:57:28 ok, but that won't affect squashfs kernels already out there :). and i'm hoping that the inclusion of the local hostname when doing dhcp request was on oversight? because thats really, really bad :) Aug 12 11:58:14 honestly the old behaviour sucked anyway, having to change the hostname in two sections of uci to identify the device :D Aug 12 11:58:43 what? Aug 12 12:00:14 you set the hostname in config/system but you shouldn't be it telling it to every isp and wifi you come upon Aug 12 12:00:35 that's bad opsec Aug 12 12:00:40 who cares Aug 12 12:00:41 lol Aug 12 12:01:01 its not like people are using actually identifiable hostnames Aug 12 12:01:23 doesn't matter if it's persistent and non-unique, eh? Aug 12 12:01:30 jwh: sfps are suported in mainline kernel 4.15 and later Aug 12 12:01:48 Hauke: mmm, on which boards? Aug 12 12:04:27 /lib/netifd/proto/dhcp.sh, line 65 <--- comment Aug 12 12:05:14 ugly workaround, i know. but gawd i cringed when i saw the new behavior Aug 12 12:05:30 is it actually intentional or a mistake? Aug 12 12:05:39 unknown Aug 12 12:06:50 if intentional, maybe add an option in config/network for dhcp interfaces like "option tellhostname 0/1" Aug 12 12:07:59 yeah Aug 12 12:08:17 at least for marvell SoCs Aug 12 12:08:29 its kinda expected that dhcp client sends the system hostname though, thats what every other platform/OS does Aug 12 12:08:43 jwh: but it adds a framework for SFPs so it should not be so hard to add support for other boards Aug 12 12:08:43 Hauke: ah cool, that will be nice when we move to 4.19 Aug 12 12:08:50 Hauke: orly, you got any urls? Aug 12 12:09:23 jwh: not really search for phylink Aug 12 12:09:36 oh the phylink stuff, yeah I was looking at that Aug 12 12:09:39 jwh: https://lwn.net/Articles/667055/ Aug 12 12:10:05 has/is it all being backported in openwrt? Aug 12 12:11:50 coz that looks really nice Aug 12 12:12:17 jwh: for the marvell SoC some part is backport to kernel 4.14 Aug 12 12:12:40 ah cool Aug 12 12:12:41 ok Aug 12 12:14:06 if someone has any of these boards and can test I can update it, it's mostly for my own reference tbh Aug 12 12:14:16 jwh, it's very often the default, yes. and shopping malls are much obliged that you provide them with your first name for every dhcp request on their public wifi Aug 12 12:14:26 "rudolfs iphone" Aug 12 12:14:46 if you change it yeah, but most things have a non-identifying default Aug 12 12:15:09 ios has "iPhone" etc, windows defaults to WIN-, android is similar to windows Aug 12 12:15:26 which openwrt also has of course, as the default hostname is "OpenWRT" Aug 12 12:15:43 iOS has uses "%firstnameofregistereduser"-s iPhone" Aug 12 12:16:00 none of my iphones have that as a default Aug 12 12:16:03 but I change it Aug 12 12:16:03 and the winstring is "random", but unique Aug 12 12:17:07 but both windows and ios can also use random mac addresses (arguably more useful than a hostname) Aug 12 12:17:22 ios I think does it by default now, windows is off Aug 12 12:17:24 they do when probing Aug 12 12:17:26 kinda Aug 12 12:17:44 mm, the windows one seems to be after connection too Aug 12 12:17:48 at least when I played with it Aug 12 12:17:51 but once they have connected to an AP, they stick with the same STA MAC for each ESSID Aug 12 12:17:56 oh, yeah Aug 12 12:18:12 note: ESSID, not BSSID Aug 12 12:18:17 that makes sense thuogh if you want predictable ip addresses Aug 12 12:18:58 that is, when my devices connect to my wifi they always get the same address as dnsmasq does some pseudo hashing based on mac address Aug 12 12:19:29 but you're right, the windows way is really, really dumb Aug 12 12:19:48 well it serves its purpose for privacy Aug 12 12:19:53 but its totally random Aug 12 12:20:03 doesn't remember mac addresses ued Aug 12 12:20:05 used* Aug 12 12:20:12 doing probing with the persistent mac, then using a random for each ap? Aug 12 12:20:34 I think its per connection Aug 12 12:20:39 the probing should be random Aug 12 12:20:40 so may be different after reboot Aug 12 12:20:41 isn't Aug 12 12:21:18 oooh Aug 12 12:21:24 you can turn it off for known networks Aug 12 12:21:27 thats not so bad Aug 12 12:21:44 so you can have it on globally, but then turn the random stuff off for your own wifi Aug 12 12:22:26 shame android/ios can't do that properly Aug 12 12:22:44 may want to still have random address everytime for public hotspots Aug 12 12:22:59 which is a good thing Aug 12 12:23:11 yeah Aug 12 12:23:28 but they don't mind telling the AP that this is "Reinhildes iPad" Aug 12 12:24:13 does ios11 change it by default then? I haven't had a new ios11 device yet Aug 12 12:24:26 only upgraded from 10 (which defaults to iPhone) Aug 12 12:24:53 but I guess people also restore from backups everytime Aug 12 12:25:01 so the hostname is probably carried across Aug 12 12:25:33 my opinion is: shouting the local hostname as part of DHCP requests should be off by default. hopefully we'll get a verdict Aug 12 12:25:46 tbh Aug 12 12:26:05 at the very least there should be a flag in config/network for the interface in question to disable it Aug 12 12:26:17 it should just have an option to either disable sending hostname, or send a fixed string Aug 12 12:26:26 if you change the default hostname you should also change the option Aug 12 12:28:00 the new behaviour makes my life a little bit easier though, as I don't need to change the hostname in two places :D Aug 12 12:28:40 I am implementing behaviour similar to android/windows, except its like nte- Aug 12 12:29:12 then ask for an option to explicitly enable it for a dhcp interface in config/network Aug 12 12:29:29 this about protecting people from themselves, not esoteric use cases Aug 12 12:29:46 why though Aug 12 12:29:56 the default is a generic OpenWRT hostname Aug 12 12:30:16 if someone changes that, they can also change the option to disable sending it, or set it to something else Aug 12 12:30:28 the default should probably be what people expect Aug 12 12:31:53 I've always wondered why the hostname wasn't tied to the hostname Aug 12 12:32:20 Wait. Are you saying the choice of dhcp hostname and vendor id made it into a CVE because it's "bad opsec"? Aug 12 12:32:43 lol what? is that actually what this is about? Aug 12 12:34:30 salcedo, no, the CVE and response was a seq question :) Aug 12 12:34:44 separate question even Aug 12 12:37:09 Oh. My bad. Lol Aug 12 12:39:12 ntd I recently did that though. Modded some script to make udhcpc mimick an Android device. Aug 12 12:40:19 I didn't see a way to do it through uci. Aug 12 12:40:50 you can set the dhcp hostname on the interface section Aug 12 12:41:03 or could, anyway :D Aug 12 12:46:01 jwh, ? Aug 12 12:54:17 build #93 of ath25/generic is complete: Success [build successful] Build details are at http://release-builds.openwrt.org/18.06/images/builders/ath25%2Fgeneric/builds/93 Aug 12 12:54:34 https://www.amazon.com/NETGEAR-Multi-Gigabit-Ethernet-lifetime-technical/dp/B075Q6NPM2 Aug 12 12:54:53 Will this successfully interface with a 10Gbps NIC at 2.5/5Gbps speed? Aug 12 12:57:53 also, the new way openwrt does dhcp request, which vendor id is used when doing mac spoofing? Aug 12 13:11:22 anyone know of a powerline/plc+wifi plug that can run owrt? Aug 12 13:12:43 I remember one. Aug 12 13:12:48 wait Aug 12 13:12:51 will try find Aug 12 13:13:32 https://www.globalscaletechnologies.com/t-products.aspx Aug 12 13:13:37 One of these I think. Aug 12 13:14:56 None of these are on the supported list though, but that doesn't mean you can't cram openwrt on it. Aug 12 13:15:07 If Linux runs on them, you probably can. Aug 12 13:15:39 I wanted to use a HP thin client for OpenWRT. It's not on the list either. Aug 12 13:15:59 The device may or may not be able to something useful depending on drivers though. Aug 12 13:16:17 Look if the Wi-Fi chip is supported. Aug 12 13:16:31 Also how are you going to connect the uplink to it? Aug 12 13:16:35 ntd Aug 12 13:17:19 the QCA PLC stuff is on their github Aug 12 13:18:13 I could use one as a travel router too. Aug 12 13:18:18 But how fast are they? Aug 12 13:19:57 the wr710n works pretty well for travel Aug 12 13:20:24 i'd prefer dual band/radio, unfortunately the wr902ac can only be powered by usb and has only one wired port Aug 12 13:20:28 I need Wireguard VPN. Aug 12 13:20:37 DonkeyHotei: jwh: Great that you following me about mt7621, dsa and SFP work. But I am a bit stuck. Blogic is also working on the DSA code for mt7621. Aug 12 13:20:47 Sheevaplug will work regarding your question. Aug 12 13:21:01 https://wiki.openwrt.org/doc/howtobuild/sheevaplug Aug 12 13:25:51 I have issues to get 2nd GMAC talk to external phy. I do something wrong. Possable that the current ethernet code overwrites a default which setups the external ephy pins. Aug 12 13:31:03 That is one issue. 2nd is that external phy at8033 is not handled right by the kernel. The status registers of the phy are right but driver/kernel doesn't pick them up right. Aug 12 13:33:38 But SFP module is detected also while replugging the SFP module. Current only detection works, SFP kernel module doesn't setup/controlles the ethernet phy at the moment. Aug 12 13:58:34 Rene__: good work though sir Aug 12 13:58:43 is your DSA working? Aug 12 14:01:05 koops: none of these appear to be powerline/plc capable tho Aug 12 14:02:35 I don't think there's any that are f00b4r0 Aug 12 14:02:54 ntd asked for plc-capable devices Aug 12 14:03:01 afaik none are supported by openwrt. Aug 12 14:04:01 Rene__: take a look at some of the recent commits and pr's for the 7620 for ideas Aug 12 14:06:07 power lines should never be used for networking Aug 12 14:06:52 -- The Holy Hackbible III.2 Aug 12 14:08:14 to be fair, the plc stuff actually works remarkable well given the crappy environments they work in Aug 12 14:08:19 ^ Aug 12 14:09:01 couldn't make high voltsge distribution work though, which is a shame Aug 12 14:09:04 voltage* Aug 12 14:51:23 https://openwrt.org/toh/netgear/wnr2000 Aug 12 14:51:41 why has v5 support been so elusive? Aug 12 14:51:57 https://wikidevi.com/wiki/Netgear_WNR2000v5 <-- doesn't look all that special Aug 12 14:52:52 https://github.com/lede-project/source/pull/1256 Aug 12 14:53:07 yet there is no specific image for it? Aug 12 15:01:40 DonkeyHotei: +1 for not using power lines Aug 12 15:01:57 It's like free PoE though. Aug 12 15:54:45 hrm, there's the dlan device iirc, it has PLC Aug 12 15:55:23 https://openwrt.org/toh/devolo/devolo_dlan_pro_wireless_500_plus Aug 12 15:55:50 there's not much support for it's plc directly, but you can (aiui) load the stock binaries and do things by hand. Aug 12 16:07:38 Can someone having trouble with 802.11w with ath10k-ct open a bug in the ath10k-ct github project with details including the supplicant/hostapd config files used? Aug 12 16:17:19 jwh: Unfortunately it makes other, licensed environments crappier. Aug 12 16:31:02 greearb Hi is there a way to see what speed a ath10k radio is working at my radio just says 6 mbits Aug 12 16:31:31 Bitrate: 6 Mbit/s Aug 12 16:32:55 Monkeh: heh Aug 12 16:33:16 Monkeh: they got acknowledged by all the moaning and PLC was notched accordingly :D Aug 12 16:33:36 s/by/after/ Aug 12 16:38:04 jwh: That won't fix any of the existing hardware or the fact that it's still non-compliant with basic regs, by its own function. Aug 12 17:00:42 Monkeh: Does that mean powerline is illegal? Aug 12 17:56:03 Anyone in here familiar with ucert's interaction with usign while trying to build an OpenWRT package? Aug 12 17:56:05 Rene__: is there a at8033 ? Aug 12 17:56:15 Rene__: i have sfp support patches for at8033 Aug 12 17:58:58 My build is failing and I've tracked it down to an odd piece of apparently unexpected behavior. Aug 12 17:59:30 ucert runs usign and reads its output over a pipe, expecting apparently 16 characters followed by a newline. Aug 12 17:59:57 However, for me, usign is writing 15 characters followed by a newline, causing ucert to exit with an error status. Aug 12 18:00:01 Any ideas? Aug 12 18:10:59 *facepalm* alert: ucert+usign interaction is broken, one or both need fixed Aug 12 18:11:33 I found the problem: usign's fingerprint() function doesn't ensure leading zeros are printed. Aug 12 18:12:19 That would be the easiest fix: change "%"PRIx64"\n" to "%016"PRIx64"\n" in usign's main.c:fingerprint() Aug 12 18:12:32 mophie: can you send a patch ? Aug 12 18:12:57 and/or dump that info on the ML and Cc: Daniel Golle Aug 12 18:13:26 but a fix in ucert's usign-exec.c:usign_f() to allow non-leading-zeroed output wouldn't be a bad idea Aug 12 18:14:35 blogic: I'm not good with computers but I'll see what I can do Aug 12 18:15:04 (kidding/trolling haha) Aug 12 18:15:08 mophie: send an email to the MailingList please Aug 12 18:15:36 jaja Aug 12 18:16:00 mophie: computers are similar to your casio calculator, just with more buttons Aug 12 18:18:58 assuming there's a 1-in-16 chance that this problem would occur, it's interesting that nobody else has found it Aug 12 18:19:17 ucert has been a trainwreck since it was added Aug 12 18:19:18 it's only been merged for a day or two Aug 12 18:19:36 ah, thanks, that helps Aug 12 18:19:36 (that was a reply to mophie) Aug 12 18:19:49 karlp: So it's a very recent trainwreck. :P Aug 12 18:20:07 I built OpenWRT around 12 years ago for an original WRT54GS and haven't checked on the project again, until today Aug 12 18:20:22 was not making any comment on that, it appears to be as much wreckage as timeing build steps... Aug 12 18:20:31 but was jsut saying why perhaps noone else had seen it yet :) Aug 12 18:20:42 karlp: That was a trainwreck too Aug 12 18:21:07 ...that's why I mentioned it :) Aug 12 18:21:32 This one I think may keep rolling.. :P Aug 12 18:22:11 hrm. is rt2860 _meant_ to work or not? I just get strings of errors and "failed to enter state" and "WPDMA tx/rx busy" whenever I try and enable it. Aug 12 18:22:48 jwh: As far I can tell yes, Last I made a change in Luci and that seems to work. dmesg: https://pastebin.com/g2pyd33t Aug 12 18:23:19 blogic: yes it is a at8033. Aug 12 18:23:37 Monkeh: timing got added in ff6e62b288c in april 30, last patch on it (so far) was a692e4e3d on aug 5 Aug 12 18:23:44 Rene__: i'll test your patches tomorrow Aug 12 18:23:48 ucert's got a long way to go to reach the timing trains :) Aug 12 18:24:03 and I am planning to rework at803x driver upstream this week anyhow Aug 12 18:24:05 blogic: great to hear Aug 12 18:24:43 at8033 has the right bootstrap resistors but the driver handles it well. Aug 12 18:24:44 s linux-wireless the only place to try and trakc down rt28xx issues? Aug 12 18:25:01 Rene__: the upstream driver does not handle the fiber registers at all Aug 12 18:25:10 karlp: yarp Aug 12 18:33:07 ok. will try there a bit. don't see any point trying to submit this board support anwyhwere if it's jsut a 5 port ethernet device. Aug 12 18:33:34 was expecting ralink wireless to at least -work_ even if it wasn't high performance. Aug 12 18:59:19 dependency failure: gdbserver (=y) didn't lead menuconfig to change my zlib from (=m) to (=y) Aug 12 18:59:28 I'll note that in my message to the mailing list as well Aug 12 19:03:06 other than that my build succeeded, should I trust it? ;-) Aug 12 19:05:01 hmm, on second thought I have *-sysupgrade.bin files but no *-factory.bin files Aug 12 19:05:26 shouldn't a *-factory.bin file have been produced (this is for DIR-825 B1)? Aug 12 19:38:10 mophie: only if things can be done from the factory sw perhaps? Aug 12 21:44:17 Hi would it be OK if i make a twitter called OpenWrt-help to help out noobies with getting started with OpenWrt? Aug 12 22:07:56 18.06: it tried commenting out line 65 in /overlay/upper/lib/netifd/proto/dhcp.sh to avoid disclosing the hostname during dhcp requests Aug 12 22:08:03 works fine, except for on tagged interfaces? Aug 12 22:08:55 as in dhcp no longer works on tagged interfaces with that line commented out. wifi interfaces, non-tagged interfaces? no problem Aug 12 22:09:04 /lib/netifd/proto/dhcp.sh Aug 12 22:09:24 que Aug 12 22:10:01 Monkeh: well yeah Aug 12 22:10:44 the interface (in this case a trunk uplink) works fine, traffic is tagged/untagged Aug 12 22:11:04 but dhcp on a specific VID on that interface? nope Aug 12 22:12:40 weird Aug 12 22:12:50 don't see requests going out? Aug 12 22:13:19 they go out Aug 12 22:13:40 wait, no Aug 12 22:13:51 i wasn't able to download tcpdump to check Aug 12 22:14:23 heh Aug 12 22:14:40 not some broken firewall rules? Aug 12 22:15:12 no. commenting/uncommenting that line solves/causes it Aug 12 22:15:21 but only affects tagged interfaces Aug 12 22:15:43 is that line literally just the hostname option? Aug 12 22:15:57 dnsmasq facing clients ofc works fine, issue is router doing dhcp request upstream on tagged interfaces Aug 12 22:16:34 looking at the wiki, are the options for "proto dhcp" specified anywhere? new wiki seems limitied Aug 12 22:16:36 limited Aug 12 22:16:55 they should be in the network portion of the old one Aug 12 22:18:11 link? Aug 12 22:18:32 https://openwrt.org/docs/guide-user/network/ipv4/start#protocol_dhcp Aug 12 22:19:01 https://wiki.openwrt.org/doc/networking/network.interfaces Aug 12 22:19:05 heh Aug 12 22:19:14 is there an option for "hostname" that *doesn't* send it instead of spoofin? Aug 12 22:19:45 doubt it Aug 12 22:19:54 maybe set it to an empty string? Aug 12 22:20:17 tried setting a blank one in config/system, defaulted to openwrt Aug 12 22:20:25 i'm gonna try empty later Aug 12 22:20:27 ah Aug 12 22:20:37 in config/network Aug 12 22:21:21 but this seems like a weird issue. hw is archer c7 v2, sysupgraded from 170105 Aug 12 22:21:50 if anyone can try to reproduce? tagged on wan port Aug 12 22:54:50 New twitter for help with openwrt is @openwrt-H Aug 12 23:00:57 Do I need Cisco support package to get firmware updates to Cisco switches? Aug 12 23:14:47 is there a reason OpenWrt builds wpa_supplicant with bgscan support disabled? **** ENDING LOGGING AT Mon Aug 13 03:00:00 2018