**** BEGIN LOGGING AT Tue Oct 16 02:59:59 2018 Oct 16 07:03:08 jwh: both of those are symptoms of a process running as a daemon/backgrounding whereas as procd expects processes to run in foreground...ie. not 'fork & exit' Oct 16 08:11:00 kernel bumps pushed, nightly tests were OK Oct 16 08:32:33 \o/ Oct 16 08:39:09 * ldir pulls \o/ Oct 16 09:04:38 dedeckeh: I wonder if we should move the ip-tiny etc. exceutables into /usr/libexec/ Oct 16 09:24:40 jow:in order to avoid confusion ? Oct 16 09:25:20 yes Oct 16 09:26:32 jow:and do that for every package which uses the ALTERNATIVES logic ? Oct 16 09:27:15 although for busybox that would be difficult Oct 16 09:27:23 I think that would make sense, yes. Usually the suffixed variants are not meant to be called directly Oct 16 09:27:35 correct Oct 16 09:27:38 a busybox exception would be fine imho Oct 16 09:28:59 jow:will have a look at it Oct 16 09:33:58 I'll deal with the dnsmasq jail issue - ooops! :-) Oct 16 10:16:15 I'll look into cherryp icking hostapd usptream fixes for the mac acl logspam Oct 16 11:05:18 Hauke: ldir: regarding the hostapd variants; I kind of like the idea to only offer wpad-tiny, wpad-basic, wpad-full variants and only offer full ssl-enabled wpa_supplicant and hostapd bianries Oct 16 11:06:07 and in a 2nd step we could consider turning wpad-full in a dummy package merely selecting hostapd + wpa_supplicant Oct 16 11:08:10 I hope someone knows *how* :-) At the moment it is beyond my understanding/time etc. maybe the conference will make a lot of things clearer to me Oct 16 11:08:47 yeah, this was not a specific change request, just something to think about Oct 16 11:09:29 I suppose when you're at the point that you've got the space to install hostapd-everything-included you don't need to save a few K by having a hostapd + wpa_supplicant multicall executable Oct 16 11:13:13 say one is not using wds atm, but might in the future. When setting up a wifi are there any disadvantages for using mode Access Point (WDS) as opposed to normal AP mode? Oct 16 11:13:35 * ldir has to admit he doesn't really understand what all the processes/options do anyway :-) Oct 16 11:13:54 ldir: hmmm perhaps, but the script looks ok... Oct 16 11:13:59 * jwh checks maintainer Oct 16 11:14:42 abenz: it might have slight, theoretical security implications if you have to deal with unstrusted stations Oct 16 11:14:53 so umm Oct 16 11:14:57 just a heads up I guess Oct 16 11:15:02 ebtables still causes panics :D Oct 16 11:15:12 (maybe still smp related) Oct 16 11:16:06 abenz: an authenticated sta could request wds and gain the ability to do layer2 things which would normally not be possible or be harder at least Oct 16 11:16:14 like arp spoofing, mitm etc. Oct 16 11:16:20 jow: for said implications to affect it the nefarious users need to know the password no? Oct 16 11:16:25 I see Oct 16 11:16:29 yes, exactly Oct 16 11:16:39 many thanks Oct 16 11:17:15 ldir: you mean what hostapd and wpa_supplicant do? Oct 16 11:17:58 jow: yes - and like Hauke says... there are sooo many different options :-) Oct 16 11:18:27 hostapd is the deamon for providing access points, wpa_supplicant the one for being a client Oct 16 11:18:43 wpad is simplyy both fused together to save some space for code which is shared among those two Oct 16 11:19:39 the variants are different fixed compile time options, the many choices stem from all the permutatios of {wpad, hostapd, wpa_supplicant} + {variants} Oct 16 11:24:02 ahhh - ok, that conceptually has clarified things! Oct 16 11:25:13 so actually my wifi routery AP things only need hostapd....as they're never going to be wifi clients. Oct 16 11:27:52 * ldir notices that nbd is obviously lurking... I see commits ;-) Oct 16 11:28:52 I'm seeing stations with high inactivity time in station dump on mt76 phys, up to 30 minutes. who's supposed to manage that? firmware? driver? Oct 16 11:30:20 ldir: yes, correct - if all you ever do is be an ap then theoretically hostapd alone would suffice Oct 16 11:30:44 ldir: openwrt also shipped with only hostapd in the distant past, I think in the pre-AA or BB times Oct 16 11:31:17 ldir: many people were confused by not being able to setup a repeater or client so nbd made wpad Oct 16 11:31:32 which has been the default ever since Oct 16 11:34:25 using wpad-full instead of hostapd + wpa_supplicant saves about 118k gzip compressed Oct 16 11:34:33 which is more than 'just a few K', in my opinion Oct 16 11:44:44 mrs ldir defeats the dog's intelligence by (in sight of the dog) removing worming tablet from box, then opening packet with tablet, placing tablet in the 'treats' jar, removing tablet from 'treats jar'...where it has miraculously turned into a treat. Yum! Dog is wormed. Oct 16 13:40:13 Hello Oct 16 13:40:24 Where is defined the build process for initramfs-kernel.bin? Oct 16 13:42:39 it is in buildmenu under image options i think> Oct 16 13:45:35 vk496: the process is just building a kernel with the initramfs symbols pointing to the rootfs Oct 16 14:06:50 * ldir is going to push those hostapd-basic patches. I guess I can always revert if the screaming gets too intense. Oct 16 16:46:35 I hate webcrap... I took me less than 30s to find an ms edge bug that seems rather... serious Oct 16 16:48:14 Thnks jwh and Slimey! Oct 16 16:48:41 jow: I mean, it's MS - horrifyingly bad browsers are their specialty Oct 16 16:49:00 Who I need to contact to create new Brand in the Table of Hardware? Oct 16 16:49:58 With whom I have to contact* Oct 16 17:00:40 oh and after filling out the bug report: "At this moment we cannot verify if you are human or not." Oct 16 17:00:43 w-t-f Oct 16 17:01:04 needless to say that my entire text is gone Oct 16 17:21:32 nbd: Any progress with mt7603e lately? Oct 16 17:32:13 hmm, wonders happen - "qtnfmac: add support for QSR1000/QSR2000 (aka Topaz) chipsets" (linux-wireless), thatÄs pretty interesting (no, I don't have any devices using it, but a couple of interesting ones do) Oct 16 17:52:02 ldir: my comment was not meant as blocking I like the idea Oct 16 17:55:13 pkgadd: the qtnfmac driver is already in the backports we use Oct 16 17:55:33 but this is the version based on kernel 4.9, I haven't checked which cjipset it supports Oct 16 18:03:08 Hauke: and I agree it's quite confusing and needs to be simplified - or ideally not made any more complicated - jow's explanation earlier has added a lot of light to the subject for me. anyway, looking forward to the conference :-) Oct 16 18:03:25 Hauke: that was only for the newer ones, while QT3840BC (QSR1000) has been quite popular in th early 802.11ac wave2 devices (e.g. netgear r7500v1, zyxel nbg6816, several Asus ones, etc.) Oct 16 18:04:01 a bit cheaper to obtain used than the current crop of fully ath10k devices Oct 16 18:10:21 it just adds the potential to support an interesting assortment of rather powerful devices that weren't usuable (well, at least not their 5 GHz wlan) before Oct 16 18:13:34 ldir: I aggre with jow that we should probbaly only ship wpad Oct 16 18:17:48 pkgadd: I was thinking about if we upgrade backport to 4.20-rc1 when it was released or stay with the 4.19 stable branch Oct 16 18:22:25 Hauke: I'm still on the 4.14 based backport package, because of my problems with ath10k/ ath10k-ct (both the 4.13 and 4.16 based ath10k-ct) and WDS/ 4addr Oct 16 18:23:40 WDS with ath9k seems to be fine though Oct 16 18:42:21 libssh.org Oct 16 18:43:23 https://www.libssh.org/security/advisories/CVE-2018-10933.txt Oct 16 20:07:02 mkresin: did you create a uart u-boot image for the bt home hub 4A (AR10) ? Oct 16 20:47:38 Monkeh: what's wrong with mt7603e again? Oct 16 22:17:25 mangix: afaik, the mt7688 has trouble in station mode. of the 10+ networks i have in my testbed, my mt7688's can only connect without major problems with one of them. **** ENDING LOGGING AT Wed Oct 17 02:59:59 2018