**** BEGIN LOGGING AT Fri Dec 07 03:00:01 2018 Dec 07 03:04:03 instead of rm -rf $safepath, maybe move to a backup location instead, so it's reversible Dec 07 03:23:58 xback: I'll look at the vfp module breakage tomorrow, but I really need to go to bed now :P the variable in the error looks familiar, I suspect it's easily fixable Dec 07 03:46:22 xback: https://git.openwrt.org/a2c0ad34 Dec 07 06:51:34 build #1119 of ipq806x/generic is complete: Failure [failed sourceupload] Build details are at http://phase1.builds.lede-project.org/builders/ipq806x%2Fgeneric/builds/1119 blamelist: Daniel F. Dickinson , Petr ?tetiar , Chuanhong Guo , David Bauer , Mathias Kresin Dec 07 06:51:34 , INAGAKI Hiroshi Dec 07 06:55:31 *yawning* Dec 07 06:55:34 jow: pong Dec 07 08:05:29 greearb: ping Dec 07 08:30:09 blogic: hi, question was about the coldplug patch removing the dev check Dec 07 08:39:37 blogic: morning Dec 07 08:39:50 blogic: I may need your opinion in the [PATCH libubox] hotplug: add hotplug_call() helper Dec 07 09:10:02 rmilecki: i already replied Dec 07 09:10:11 blogic: i saw, thanks Dec 07 09:10:12 jow: what is the issue there ? Dec 07 09:10:29 my initial comment was not "you must" but a "might be worth considering" Dec 07 09:10:34 so as you feel fit Dec 07 09:12:02 blogic: issue was that our coldplug program explicitely skips uevent generation for things that do not have a /dev node which prevents the event triggering for things like usb hosts Dec 07 09:12:56 gabor added this check a few years back, probably as optimization... will simply revert it and see if any chaos ensues Dec 07 09:13:19 jow: boot time quadruples Dec 07 09:13:22 if i recall Dec 07 09:13:33 due to the hotplug-call overhead? Dec 07 09:13:41 i would rather opt-in for subsystems/devicetypes/.... Dec 07 09:13:43 yes Dec 07 09:13:51 there are 1 trizillion uevents Dec 07 09:14:01 i think at least i wrote that code like 8 years ago Dec 07 09:14:09 that would've been my next best idea, simply add a naive usb exception Dec 07 09:22:46 gut feel tells me that explicit opt-in will cause the least side effects Dec 07 09:24:51 any idea how to debug ath10k-ct firmware in relation to dropped packets in a way that will help greearb in fixing it? Dec 07 09:29:52 xback: fyi, now also compile -and runtime tested brcm2708/bcm2708 with 4.9.143 Dec 07 09:30:27 xback: please ping if there's anything else holding you back from pushing the bumps to master now :) Dec 07 09:36:21 Good morning all Dec 07 10:13:51 https://www.3d-lab-av.com/nano-network-player.html Dec 07 10:13:58 such a steal! :P Dec 07 10:30:33 stintel, grab one while stock lasts :p Dec 07 10:31:34 nitroshift: well now I can atleast install OpenWrt on the RPi that's in there :P Dec 07 10:34:13 maybe I should write them once I pushed everything to master "you can now improve your nano player by using openwrt as OS for the pi" :P Dec 07 11:00:11 oh damn, now I broke SDIO Dec 07 11:23:17 blogic: Any luck with the top secret docs? Dec 07 11:39:20 Monkeh: hehe Dec 07 11:39:34 i had someone else ask me a different question that i fwd'ed first Dec 07 11:39:42 Monkeh: i'll ask start of the week Dec 07 11:43:58 sigh, now the overlay doesn't mount. wtf :/ Dec 07 11:45:34 mount_root: failed to mount -t f2fs /dev/loop0 /tmp/overlay: No such file or directory Dec 07 11:47:00 looks like the mkfs.f2fs didn't run for some reason Dec 07 13:02:24 stintel: well, i'm messing with fstools, but I didn't even update package yet Dec 07 13:02:27 i hope it's not me then Dec 07 13:03:53 rmilecki: I think it was related to me disabling the wrong MMC driver in config-4.14 for brcm2708 - cleaned SD, dd'd the factory image -> OK, then sysupgrade -> also OK Dec 07 13:04:16 ok Dec 07 13:04:20 good for me ;) Dec 07 13:04:27 ;-) Dec 07 13:25:44 movi, here Dec 07 13:26:56 greearb: in short. 9980 on a c2600, trunk, -ct driver and firmware (wmi kind, the default one in trunk). It's dropping packets from all kinds of clients, seemingly randomly. Gets worse with time. What can I do? Dec 07 13:27:36 greearb: it's in the firmware for sure, since i have a second one configured in an identical way. i substitute normal fw for yours, it has the same behaviour Dec 07 13:29:04 so it is not a regression in my firmware as far as you can tell? Dec 07 13:29:53 greearb: it is. stock ath fw works fine, -ct looses packets Dec 07 13:30:43 oh, can you try my 'b' firmware from my site and see if it has similar issues? Dec 07 13:30:52 greearb: already did. same result Dec 07 13:32:22 i also looked at bz, both .11w and .11r disabled Dec 07 13:33:03 those should work in recent ct fw, but either way. Have you tried using ct fw with stock driver and vice versa? Dec 07 13:33:57 i tried ct fw with stock driver, that drops packets. haven't tried the other combo, but no reason why i'd want to either Dec 07 13:34:35 i kinda LIKE the firmware, since i can see it behaves a bit better than stock in relation to speed and power output Dec 07 13:34:44 but the missus keeps getting timeout in the reddit app for example Dec 07 13:37:19 I suggest to try the different combinations to pin down whether it is firmware or driver. Dec 07 13:37:46 so -ct driver+stock firmware should work, and if it creates no timeout it point to the fw, right? Dec 07 13:37:54 *points Dec 07 13:38:08 I can build a series of 'b' fw where first commit is very near stock FW...if it works and later does not, then you can bisect Dec 07 13:38:16 yes Dec 07 13:38:42 please do, i'll be happy to test them out Dec 07 13:39:11 eh, i'll have to reflash again. some fun for the weekend :D Dec 07 13:40:08 should i enable some build option for the driver/drivers so we can get some data for the future? Dec 07 13:40:22 or are the defaults good enough? Dec 07 13:43:51 you could try enabling the fw dbglog messages and send me a 'dmesg' dump when the problem happens, maybe it would show something interesting Dec 07 13:44:36 greearb: is that a CONFIG_ option, or a dbg mask in /sys? Dec 07 13:45:30 runtime dbg mask: https://www.candelatech.com/ath10k-bugs.php Dec 07 13:45:47 0xc0000032 is a good value Dec 07 13:45:47 ah, perfect Dec 07 13:46:19 greearb: throw a tarball at me, and i'll be back on monday with results Dec 07 13:47:23 stintel: do you still maintain the net-snmp package ? Dec 07 13:47:46 what's up with it? Dec 07 13:48:02 movi, can you open a bug for this issue on my ath10k-ct github account? I can post a link there when I get images built Dec 07 13:48:10 greearb: sure Dec 07 13:49:06 karlp: https://sourceforge.net/p/net-snmp/bugs/2636/ and https://sourceforge.net/p/net-snmp/bugs/2426/ - these are causing the snmpv3 engine to fail on every other restart Dec 07 13:49:48 i don't see any official resolution for it, and it's still there in 5.8, so just wanted to know if stijn or you have bumped into this issue, and/or know any resolution for it Dec 07 13:50:54 never seen it myself, but we don't use trapd, just the agentx support Dec 07 13:51:43 greearb: https://github.com/greearb/ath10k-ct/issues/43 Dec 07 13:52:12 karlp: these aren't limited to trapd, but also affect agentx as well when using v3 Dec 07 13:58:31 oh right, yeah, we're not using v3 aint nobody got time for that complexity Dec 07 13:58:46 our users technically could configured it, but chnces of that are ~nil Dec 07 14:00:01 not sure if I ran into that particular problem Dec 07 14:00:05 doesn't look like upstream cares much either :) Dec 07 14:00:20 but I've had a couple of times that net-snmp wasn't running, or not responding to queries at least Dec 07 14:06:31 karlp: yeah, i have the same feeling that (almost) no one cares about v3, apart from some braindead^Wkind guys writing requirements based on it Dec 07 14:07:01 not that we couldn't crack the required des encryption with our washing machine's SoC Dec 07 14:07:09 😂 Dec 07 14:07:09 anyway, thanks :) Dec 07 14:22:50 movi, test fw uploaded to github Dec 07 14:59:54 are there any other v3 impls in existance? Dec 07 15:00:57 Hi! Who is the maintainer for iptables package? Dec 07 15:10:14 karlp: not that i know of - there might be proprietary ones Dec 07 15:17:26 I've opened a bug about iptables package: https://bugs.openwrt.org/index.php?do=details&task_id=1986 Dec 07 15:21:28 xvifr: -m state --state is deprecated and not compiled in anymore, use -m conntrack --ctstate instead Dec 07 15:22:18 ouch, OK! Dec 07 15:22:21 thanks Dec 07 15:22:38 maybe the makefile should be updated :P Dec 07 15:41:54 greearb: works fine on stock driver, i still need to compile a -ct driver to verify Dec 07 15:50:41 movi, the image I uploaded works fine? Dec 07 15:50:48 yes Dec 07 15:54:02 I'll do top-of-tree build next just to make sure, then can start building images to bisect Dec 07 15:54:28 ok Dec 07 16:12:04 greearb: ath10k-ct just crashed with that firmware? Dec 07 16:12:54 greearb: https://pastebin.com/ApBJEEZ1 Dec 07 16:14:11 if it is that 'first' image, then probably not worth debugging, likely I fixed the crash already Dec 07 16:14:23 I put 'last' image in github just now Dec 07 16:14:55 yeah just got the email, lemme try Dec 07 16:30:35 greearb__: it SEEMS to work fine :/ Dec 07 17:13:40 movi, possibly the 'b' fw on my web page is old...can you retry with it to see if you can still reproduce the issue? Dec 07 17:14:29 greearb__: that would be this one? https://www.candelatech.com/downloads/ath10k-10-4/ath10k-fw-beta/firmware-5-ct-full-community.bin Dec 07 17:15:51 jow: do you know where patch parameters can be changed? I'm looking to make a change to have it be more strict (ie, no fuzz) Dec 07 17:16:07 greearb__: and by the way, does BF-ANTENNA- and SOUNDING-DIMENSION- actually do anything? Dec 07 17:17:39 it is part of MU-MIMO for 9984, but not sure how well that works for 9980 Dec 07 17:18:41 greearb: yeah but i was looking ath ath10k source, and does it actually make the hardware do anything differently, except for transmitting packets about VHT capabilities Dec 07 17:19:16 stock firmware transmits 4 for my hardware, but openwrt always sets it at 1, looks like the driver is asking the firmware for this, and if it doesn't get an answer, it returns 1 Dec 07 17:19:49 stock fw == whatever tplink shipped? Dec 07 17:21:38 greearb__: yeah, the one from the page is bad Dec 07 17:22:03 i'll revert to the second one from github Dec 07 17:22:48 movi, plz show me exactly what URL you are downloading on my page for the bad one Dec 07 17:23:03 greearb__: https://www.candelatech.com/downloads/ath10k-10-4/ath10k-fw-beta/firmware-5-ct-full-community.bin Dec 07 17:23:32 i also tried https://www.candelatech.com/downloads/ath10k-10-4/firmware-5-ct-full-community-10.bin yesterday, and that one was bad for me as well Dec 07 17:25:11 greearb__: yeah, now i'm back at the one from github and it's perfect Dec 07 17:25:25 ok, those are my original FW, not the new 'b' variant that you have been testing, probably my page needs updating to add new links or something Dec 07 17:25:37 so the b doesn't stand for beta? Dec 07 17:25:54 near impossible to bisect my old one since it is based on some very old QCA fw src Dec 07 17:26:10 well, in my case, it means 'second', but I can understand the confusion Dec 07 17:26:27 so it's nowhere to be found on the site? Dec 07 17:26:29 hopefully soon 'b' becomes the main fw Dec 07 17:27:02 i always thought that you mean that beta when refering to 'b' Dec 07 17:27:30 https://www.candelatech.com/downloads/ath10k-10-4b/ Dec 07 17:27:30 what i'm most interested, is will you push this to openwrt master, as your fw and driver will become the default for 19.01 Dec 07 17:27:55 yes, will try to get that done soon Dec 07 17:28:13 or I can post patches at least, no push permissions Dec 07 17:28:38 is there some kind of versioning inside of /sys that i can check what i'm running? Dec 07 17:29:02 dmesg will show FW version id string, and 'ethtool -i wlan0' if you have ethtool enabled Dec 07 17:29:55 firmware ver 10.4b-ct-9980-fW-011-3ee5715a Dec 07 17:29:58 i see Dec 07 17:30:38 that is latest, that last bit is my git commit ID so I can be sure Dec 07 17:32:46 ah i see, the one on the page is firmware ver 10.4-ct-9980-fW-011-fdcdb73 Dec 07 17:32:53 that is... confusing Dec 07 18:32:36 rmilecki: thanks for all this hard work on blockd Dec 07 19:54:11 man that brcm2708 target Dec 07 19:54:29 tested both zero w and 3b+ several times, no issues Dec 07 19:54:36 upgrade a headless device: doesn't come back Dec 07 19:55:07 problem: when there's no HDMI device plugged in, I get vmalloc errors and cfg80211/brcmfmac can't be loaded Dec 07 19:55:37 ouch, that sounds lovely Dec 07 19:55:56 I'm not amused after all the time spent on this :/ Dec 07 19:57:12 reverted the firmware bump I did earlier this week but the problem remains Dec 07 19:57:37 stintel: I had a similar crazy problem with my RasPi, there was some kind of stupid race condition that would prevent the Ethernet interface from coming up successfully unless an HDMI cable was plugged in. Dec 07 19:57:46 mamarley: bah Dec 07 19:58:17 I ended up solving it by switching from ifupdown to systemd-networkd, not that it has any applicability here. Dec 07 19:59:06 I remember a dummy hdmi adapter is sold for some devices who wouldn't boot without hdmi Dec 07 19:59:44 Mister_X: sure but 4.9 (master) works fine, and 4.14.84 didn't have the problem either Dec 07 19:59:55 guess I'll be doing some annoying manual bisecting Dec 07 20:00:08 that's gonna be a pain Dec 07 20:00:11 good luck Dec 07 20:02:32 Hauke: I would love to try out the mac80211 update patch you just posted to the mailing list, but are you planning on pushing it to your staging tree? That would make applying it for testing easier. Dec 07 20:03:24 mamarley: https://patchwork.ozlabs.org/patch/1009640/ Dec 07 20:04:56 mamarley: is the link pkgadd posted sufficent? Dec 07 20:05:11 Sure, I can apply it manually. Dec 07 20:05:37 Mister_X: thanks Dec 07 20:06:08 you should look into git-pw Dec 07 20:18:40 blogic: sure thing :) I just happen to want to have USB storage working nicely Dec 07 20:42:12 stintel: Ahh, the excitement of Broadcom chips. Dec 07 20:45:29 Monkeh: I knew I should have stayed away from rpi :P Dec 07 20:46:14 man I really hope someone comes up with a 3rd party system for the intel compute cards, perhaps with more pcie Dec 07 20:46:23 but mostly, multiple cards Dec 07 20:48:27 rmilecki: go go go Dec 07 20:48:41 stintel: rpi, its like a black hoe Dec 07 20:49:08 blogic: never tried black hoes Dec 07 20:49:13 stintel: make kodi work ontop of owrt ad i'll jump on the wagon Dec 07 20:49:19 lol Dec 07 20:49:27 blogic: it already does ;) Dec 07 20:49:34 does it ? :-D Dec 07 20:49:36 yes Dec 07 20:49:44 oh dear ... Dec 07 20:49:54 i need to order a b2 unit Dec 07 20:49:58 or was it b2+ Dec 07 20:50:59 they're at b3+ these days Dec 07 20:51:07 3b+ :P Dec 07 20:51:14 with optional PoE HAT Dec 07 20:51:20 ok Dec 07 20:51:20 I just had to get one Dec 07 20:51:23 so be it Dec 07 20:51:37 i have not watched TV in 15+ years Dec 07 20:51:45 the 3 needs much more power than the older ones Dec 07 20:51:49 i got a fire stick to stream kids show recently Dec 07 20:52:23 and my dad has some M$ tool to rip prime vids to disk Dec 07 20:52:23 oh, and they still ban people on their forums who dare to mention that you can't get reasonable ethernet out of it Dec 07 20:52:25 honestly since my new tv runs kodi and plex .. I'm not running kodi on openwrt anymore Dec 07 20:52:37 and i need a solution to view the pay per view stuff Dec 07 20:52:51 so that would be kodi Dec 07 20:52:54 stintel: how's kodi on that? does it switch refresh rates? Dec 07 20:53:01 sound better than chromecast + tablet Dec 07 20:53:14 movi: I've never bothered to check that Dec 07 20:53:19 oh :/ Dec 07 20:53:34 movi: but I could watch 720p video on an rpi zero w Dec 07 20:54:00 which I though was kind of impressive Dec 07 20:54:07 especially knowing that the wifi sucks Dec 07 20:54:13 stintel: not really, they have dedicated silicon to make that work Dec 07 20:54:36 silicon black box Dec 07 20:54:44 impress or be scared Dec 07 20:55:07 right i am off to hit the sack Dec 07 20:55:13 stintel: you have to remember, the pi is really videocore+arm core that's more of a parasite on it than a cpu Dec 07 20:57:37 I found my RasPi too slow to do anything with direct user involvement, so instead I just use it to download and hold off-site backups of my server. Dec 07 20:58:09 odroid c2 is quite a bit faster than rpi3 Dec 07 20:58:20 but it managed to fry itself within a year or so Dec 07 20:58:27 i refuse to give the rpi people any money. they're shady as f*ck Dec 07 20:59:16 I don't intend to buy any more of them either; they just suck too much, especially with regards to USB and Ethernet. Dec 07 20:59:18 movi: oh well, I used to not buy them because of it being broadcom Dec 07 20:59:43 but then the Zero W appeared and ... well let's just say it can be handy Dec 07 20:59:45 if it works :P Dec 07 20:59:48 stintel: oh that too. they're basically broadcom discount[TM], selling off chips for settop boxes Dec 07 21:00:35 I have an ESP8266 running MySensors, which I can read with domoticz Dec 07 21:00:46 but the WiFi of that thing Dec 07 21:00:48 ugh Dec 07 21:00:51 but let's not forget how they stonewalled a project that wanted to make a better one (i think they were called lemonpie or something like that), but the foundation bullied broadcom to rescind the chip orders - they wanted to use the same SOC as rpi at the timer Dec 07 21:01:15 and it's just so limited compared to an rpi Dec 07 21:01:38 or how they lied about having opensource GPU drivers, and then argued with a kernel maintainer saying to the contrary Dec 07 21:01:54 and all of this under the guise of "education" Dec 07 21:17:07 stintel: Optional USB overcurrenting HAT, you mean Dec 07 21:19:46 movi: And you can't really buy the best models they've ever made. Dec 07 21:20:12 Monkeh: define best Dec 07 21:20:15 The 0W is IMO the most useful Pi. Dec 07 21:20:22 "Look, they're only £5!" Dec 07 21:20:53 Rest of story: You can only buy them from three places, nobody lets you order more than 1, delivery costs more than the Pi, and only one of those places will even sell you one without a £20 pack of useless crap you didn't want Dec 07 21:22:24 All of this under the excuse of "We can't manufacture enough of them for the demand" Dec 07 21:22:34 .. they moved production to China. Pull the other one. Dec 07 21:23:27 well i'm seeing $20 on aliexpress Dec 07 21:23:33 far away from 5 quid Dec 07 21:24:01 Shady sources of shady units Dec 07 23:36:26 build #301 of mediatek/mt7623 is complete: Failure [failed targetupload] Build details are at http://phase1.builds.lede-project.org/builders/mediatek%2Fmt7623/builds/301 blamelist: Romain MARIADASSOU Dec 08 01:05:15 build #1149 of brcm2708/bcm2710 is complete: Failure [failed kmodupload] Build details are at http://phase1.builds.lede-project.org/builders/brcm2708%2Fbcm2710/builds/1149 blamelist: Romain MARIADASSOU Dec 08 01:05:42 build #1138 of omap/generic is complete: Failure [failed kmodupload] Build details are at http://phase1.builds.lede-project.org/builders/omap%2Fgeneric/builds/1138 blamelist: Romain MARIADASSOU **** BEGIN LOGGING AT Sat Dec 08 01:45:08 2018 **** ENDING LOGGING AT Sat Dec 08 03:00:00 2018