**** BEGIN LOGGING AT Wed Sep 19 03:00:00 2018 Sep 19 04:29:02 he he so funny by default procps-ng in openwrt show memory in GiB Sep 19 04:30:21 easy calculations :) GiB Mem : 0.058 total, 0.019 free, 0.027 used, 0.012 buff/cache Sep 19 05:29:05 Zero_Chaos: "wep-mixed" was made up by whoever migrated the page, it isn't a valid choice Sep 19 05:37:54 jow: can you bless this? https://github.com/openwrt/packages/pull/6981 Sep 19 07:33:59 * ldir cheers - has got patchwork-apply to work - and applied first contributed patch with it (wireguard) Sep 19 07:36:53 ldir: the automgenerated mail seems to have gone wrong though Sep 19 07:37:51 and it somehow bypassed dmarc mangling Sep 19 07:39:19 it looks ok to me, other than coming from me as opposed to 'ldir' but..... I don't think I care enough today. Sep 19 07:39:29 and I took dmarc stuff off my domain Sep 19 07:39:34 ah okay Sep 19 07:40:05 well I just see a sole "Merged: " without the subject and it appears to not be a reply to an existing mail Sep 19 07:40:11 but that might be a thunderbird issue Sep 19 07:41:32 oh I see, yes subject is poor. Hmmmm Sep 19 07:41:57 ldir: could be that the sed magic in the script failed to extract the subject Sep 19 07:42:24 it's no doubt due to me running on a mac as usual Sep 19 07:42:28 :D Sep 19 07:42:43 should probably rewrite the script in perl to be more platform agnostic Sep 19 07:42:53 but would be interesting to debug it nonetheless Sep 19 07:42:56 I did see some sed warnings but as this is first time I assumed 'normal' Sep 19 07:44:19 ok, well I'm going to have a go with Rosen's patches next as they look sensible enough to me, aren't obviously too invasive should I have made a bad judgement call........ Sep 19 07:46:31 sed: 2: "/^Cc: */ { s/^Cc: *//p; ...": unexpected EOF (pending }'s) Sep 19 07:46:31 sed: 2: "/^Subject: */ { s/^Subj ...": unexpected EOF (pending }'s) Sep 19 07:46:58 I probably relied on GNUism's in sed Sep 19 07:47:06 do you have a gsed or something? Sep 19 07:47:49 let me look Sep 19 07:48:39 yes - I shall hack the script to use that and try another patch Sep 19 07:49:45 maybe put something like SED=$(which gsed 2>/dev/null); SED=${SED:-sed} at the top Sep 19 07:52:57 ahaaa! Right, ok gsed makes it work. So I'll tweak the maintainer script as you suggest. Sep 19 08:00:11 finally got that right as well. Sep 19 08:04:30 lol Sep 19 08:04:57 thanks for testing that with my patches I guess. Sep 19 08:06:21 I'm making progress :-) and your patches will appear shortly once done a compile Sep 19 09:03:35 now either I didn't compile properly, or the sysupgrade didn't stick...again. hmmmmm Sep 19 09:11:53 ok, so that sysupgrade worked. perhaps now it's time for the AP but with a serial console Sep 19 09:12:28 * ldir searches for usb-c to usb converter Sep 19 09:21:13 and that upgrade went through fine. Sep 19 13:58:39 if anyone finds something like a fitlet2 that can hold two full-size mini-pcie cards and can somehow be modified to hold a fan (to cool 4x4 9984 NIC), plz let me know! Sep 19 13:59:05 I did order the latest apu2 model to see if its bios would recognize the 4x4 NIC...but will be a month or so before it is delivered Sep 19 14:00:28 i don't believe the apu2 can physically fit the 4x4 card even without a fan Sep 19 14:00:57 the card fits, but BIOS ignores it Sep 19 14:01:25 and I could cut a hole and mount a fan...but it is sketchy at best Sep 19 14:01:32 does the kernel see it? Sep 19 14:01:54 not in lspci Sep 19 14:02:15 could be a slot power issue. the slots are not 100% wired iirc Sep 19 14:02:43 check the pcengines.info forum Sep 19 14:19:00 jow: respectfully, if it was made up why is it obviously supported in https://git.openwrt.org/?p=project/netifd.git;a=blob_plain;f=scripts/netifd-wireless.sh;hb=HEAD in the *wep* case??? Sep 19 14:22:55 Zero_Chaos: this option never worked, ever Sep 19 14:23:21 you can complain about it if you like but I'd rather drop references to the broken case Sep 19 14:23:30 jow: if you accept my patch, which only modifies the wpa glob to be less greedy, it works perfectly Sep 19 14:24:17 I didn't see such a patch Sep 19 14:24:22 jow: it's documented, and functional with my patch, can you please fix it instead of dropping it as broken? it is such a simple and obviously valid fix https://paste.pound-python.org/show/SJ1SAVjWnuFmVmcSEfA8/ Sep 19 14:26:01 your patch is broken Sep 19 14:26:49 jow: whitespace damage from pastebin or you think the content is poor? Sep 19 14:27:10 I can fix either one, just need to know which Sep 19 14:28:14 jow: assuming whitespace damage, this should fix it https://dev.gentoo.org/~zerochaos/0001-fix-too-greedy-glob-in-encryption.patch Sep 19 14:29:08 I'll leave that to someone more knowledeable Sep 19 14:30:06 jow: can you just confirm that you had an issue with my patch format not nessesarily my patch content? because while I have you if "your patch is broken" was content related I'd love to understand Sep 19 14:32:49 I believe your glob patterns exclude cases that worked previously Sep 19 14:33:50 at least the psk cases all allow for leading text Sep 19 14:34:21 also while dash is mentioned in examples, the previous implementations allowed for a wider range of values Sep 19 14:34:33 should probably be *psk*mixed* Sep 19 14:34:54 wpa*mixed*|*psk*mixed* Sep 19 14:34:57 Zero_Chaos: the issue is in the wep switch case, as the two documented modes are wep-shared and wep-open, not wep-shared and wep-mixed Sep 19 14:36:17 KanjiMonster: I checked the Attitude Adjustment implementation and it indeed (attempted to) support "*wep*mixed*" Sep 19 14:36:34 it also unconditionally did wpa=0 in the wep case Sep 19 14:36:42 it looks to me like a forward porting error Sep 19 14:36:53 still nto convinced that it makes sense to support it Sep 19 14:37:52 https://git.openwrt.org/?p=openwrt/svn-archive/openwrt.git;a=blob;f=package/hostapd/files/hostapd.sh;h=b55b63db7cd9e7c4db1830dbe3f5e5d070561d58;hb=refs/heads/attitude_adjustment#l161 Sep 19 14:38:55 jow: my glob cases specifically covered the documented cases. I would be fine with changing the - to a *, but it seemed unneeded. Sep 19 14:39:08 Zero_Chaos: so apologies, apparently it is supposed to be a thing. I could live with a patch that does wpa*mixed*|*psk*mixed* Sep 19 14:39:21 to remain backwards compatible with wahtever users might have set Sep 19 14:39:48 jow: I don't care about my name in git blame, you want to just make the wpa*mixed*|*psk*mixed* change or shall I make you a patch? Sep 19 14:40:14 alternatively it might be simpler / safer to revert to the old AA logic I linked to simply do wpa=0 for any wep case Sep 19 14:40:31 but not sure if the new script structure allowes for that Sep 19 14:40:32 jow: I see. I used luci as a reference, which only seems to not offer wep-mixed Sep 19 14:40:53 *which doesn't seem to offer wep-mixed Sep 19 14:40:54 jow: as the wpa case is above and seperate (now) from the wep case, that would currently results in wpa=3\nwpa=0 Sep 19 14:41:06 yeah okay, understood Sep 19 14:41:21 jow: my fix is sufficient, although I personally don't mind how you choose to fix it so long as it's fixed :-) Sep 19 14:41:44 jow: want a new patch from me? or just going to make the change? Sep 19 14:41:44 greearb: It seems like there's something of a history of the APU2 not working with some cards due to timing issues Sep 19 14:42:17 Zero_Chaos: would be great if you could quickly respin it, raw paste is enough Sep 19 14:47:02 jow: https://dev.gentoo.org/~zerochaos/0001-make-wpa-less-greedy.patch Sep 19 14:47:09 jow: thanks Sep 19 14:53:17 jow: thank you Sep 19 14:54:36 jow: not to press my luck, but any chance of making openwrt use my fixed version some time soon? :-D Sep 19 14:55:05 jow: doesn't have to be this second, nothing I'm doing is that pressing Sep 19 14:56:51 netifd will likely get bumped at some point during the next weeks Sep 19 14:58:11 jow: is there an easy way to watch for that? or just kinda pay good attention? Sep 19 15:07:41 Monkeh, whatever the reason, I'm still looking for something that is cheap-ish and can support a 9984 NIC and a second NIC. Might end up trying to port my stuff to openwrt and run on a Netgear or whatever is out there...but nice to have an OEM type platform for flexibility. Sep 19 16:21:45 greearb: I'd say just build a little x86 machine, but that's surprisingly uncheap Sep 19 16:22:08 yeah, I need qty 30 or so of them at reasonable price Sep 19 16:23:00 I don't think you'd manage to build an mATX or ITX machine for less than $500-600 a piece Sep 19 16:23:12 indeed Sep 19 16:23:43 yes, size would be important too, so it needs to be small. Basically, a working apu2 with slightly bigger case to add cooling fan would be perfect :) Sep 19 16:24:02 You can get some sort of cubish mATX cases which aren't too horrible Sep 19 16:24:32 Some of the Atom based boards aren't too terrifyingly priced lately.. but you still need case, power supply, RAM, adapter cards.. Sep 19 16:26:17 lots of little systems around, but haven't found something just right yet...amazon is full of small firewall-type boxes from the likes of Jetway, etc. Sep 19 16:27:00 There's things like this for a case: https://www.newegg.com/Product/Product.aspx?Item=N82E16811352058 Sep 19 16:27:43 I've had poor experience with mini-itx, anything cheap enough is built like crap it seems Sep 19 16:27:51 Indeed Sep 19 16:28:16 And finding something that size with more than one PCIe slot is hard Sep 19 16:28:36 Seeing as ITX only offers one, you're reliant on finding something bending the spec Sep 19 16:29:20 greearb: Do you have any M.2 options for wifi cards? Sep 19 16:29:35 no, no M.2 qca cards that work, afaik Sep 19 16:29:45 That's lame Sep 19 16:35:10 greearb: Still, you might be able to do something with an ITX boad and one of these: https://mikrotik.com/product/RB14eU Sep 19 16:35:45 There's a cheaper version without the USB and such Sep 19 16:40:26 greearb: Still probably looking at ~$300-350 a piece that route, and a little bulky Sep 19 16:41:37 Surprised by how cheap these little Atom boards are getting though Sep 19 17:00:49 Indeed, I just found out about the fitlet2 yesterday and that seems like it would make a great router or HTPC and is cheaper than the NUCs I have been using for the latter. Sep 19 17:04:30 can we also enjoy uefi backdoo^H weaknesses on our routers now? :) Sep 19 17:25:40 [Wed 2018-09-19 09:29:18 AM PDT] greearb: Do you have any M.2 options for wifi cards? <----- what about that marvell card i linked earlier? Sep 19 17:26:42 given that they're working on ath10k, I don't think that would be very relevant? Sep 19 17:28:11 it's weird how compex is pushing qca6174 if it doesn't work Sep 19 19:04:11 DonkeyHotei: what's wrong with it? Sep 19 19:04:51 everybody seems to say qca6174 does not work. i do not have one. Sep 19 19:11:35 hi, how can I remove usb support by default via a patch Sep 19 19:11:49 what files needs to be edited? Sep 19 19:13:12 mwarning: the packages included in your device's profile Sep 19 19:13:15 in the Makefile i think. Sep 19 19:13:42 hm, what specific Makefile? Sep 19 19:16:51 mwarning: DEVICE_PACKAGES in your target's image Makefile Sep 19 19:17:22 or you use the -package notation in the per device rootfs Sep 19 19:17:27 config Sep 19 19:18:28 I see, thanks Sep 19 19:23:33 pkgadd: where is the device rootfs config? Sep 19 19:24:30 I want to remove all usb items from all builds. Sep 19 19:26:34 often in image/ below your targets' root directories Sep 19 19:42:45 DonkeyHotei: client mode works, AP mode works (but not 11w) Sep 19 19:42:53 Monitor mode does not Sep 19 19:43:26 11r? Sep 19 19:43:31 doubt it Sep 19 19:45:57 hrm just had a though. monitor mode does not work because of firmware. wonder if older firmware would make it work Sep 19 19:56:25 is there documentation anywhere on how to select which antenna to use with ath9k? Sep 19 19:56:42 in the case of 3x3, that is Sep 19 19:58:52 DonkeyHotei, iw can set the antenna mask Sep 19 19:59:10 to what? Sep 19 20:00:18 which value(s) select 1x1? 2x2? 3x3? 4x4? Sep 19 20:01:03 it's used a bitmask Sep 19 20:02:38 so would 4x4 be 0xF, then? Sep 19 20:12:48 DonkeyHotei: is there any 4x4 hardware? Sep 19 21:39:35 is there a way to build all images for all target/models? Sep 19 21:47:58 you may want to go look at https://git.openwrt.org/?p=buildbot.git;a=summary Sep 19 21:50:18 ok, thanks! Sep 19 21:53:01 hm, I have no idea how to use the buildbot.. Sep 19 21:54:32 I have this script, but it does not seem to be working alright: https://paste.ubuntu.com/p/RxzPgccWps/ **** ENDING LOGGING AT Thu Sep 20 03:00:01 2018