**** BEGIN LOGGING AT Mon Oct 14 02:59:57 2019 Oct 14 05:25:54 for mwarning: 2019-08-14 03:06 russell--: ooh, i found a kludge. i can add an interface in /etc/config/network for eth0 with proto none and i get hotplug events Oct 14 06:00:04 jow: thanks for the repo! any idea about that missing IRC buildbot output? Oct 14 07:05:10 ynezz: what is your plan with that repo? Oct 14 07:08:02 aparcar[m]: 08:45:46 < ynezz> background: https://github.com/openwrt/openwrt/pull/2490#issuecomment-541391542 Oct 14 07:11:07 ynezz: thanks, maybe this matrix bit misses some messages... Oct 14 07:21:27 very unlikely :) Oct 14 07:44:19 *yawn* Oct 14 07:48:08 aparcar[m]: thanks to you & ynezz I now have a backup build plan on macos using docker docker run --rm -it --tty -v "$(pwd)/wrt":/home/build/my-openwrt:delegated openwrtorg/sdk Oct 14 07:48:45 blogic: morning Oct 14 07:49:15 ynezz: strange you posted that here at 8:45? i really can see it Oct 14 07:50:01 ldir: nice! we could also offer images with precompiled tools if thats of use Oct 14 07:52:35 doesn't really help my bizarre usecase - in essence I'm using the sdk image as a source of 'debian + build dependencies' Oct 14 07:54:03 and then pointing it at a local directory and saying "build what you find there" Oct 14 07:57:33 so the precompiled tools would be in the wrong directory, ie in build/openwrt as opposed to build/my_openwrt Oct 14 07:58:58 aparcar[m]: those precompiled tools would be probably usable for CI only, otherwise one would need to mount bunch of volumes, probably triggering rebuild of some packages etc. Oct 14 08:00:16 ynezz: yea for CI it could be usfull Oct 14 08:01:06 ynezz: I made something like this for rebuild and it safes about 10 minutes per build 😛 https://gitlab.com/aparcar/rebuild/blob/master/.gitlab-ci.yml Oct 14 08:05:38 ynezz: please have a look at https://github.com/openwrt/openwrt/pull/2492 Oct 14 12:13:29 stintel: o/ Oct 14 12:13:59 stintel: I'm busy bumping the kernels, and see a delta in brcm2708 Oct 14 12:14:11 would you agree that https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/brcm2708/patches-4.19/950-0726-can-mcp251x-Allow-more-time-after-a-reset.patch;h=a5863dff3bd5eb469418c9e6521a1b14f62ace09;hb=HEAD Oct 14 12:14:24 is deprecated by this upstream one: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.19.79&id=4aaea17d3c31464ecd783c9a0f554080d2298c09 Oct 14 12:25:38 xback: I had a look at that yesterday and came to the conclusion that our local patch has been deprecated by upstream (different) changes Oct 14 12:26:14 xback: ack Oct 14 12:27:07 thanks both :-) Oct 14 12:27:13 xback: https://github.com/raspberrypi/linux/commit/c7f07c53723609bf4c0e99b5ff6cd3e9bfc019ff Oct 14 12:27:27 that patch was reverted in the raspberry pi kernel repo :) Oct 14 12:27:34 so definitely ack Oct 14 12:27:41 excellent. thanks! Oct 14 12:31:40 xback: I would say "best ask an expert though" ... and you have :-) Oct 14 12:32:33 I even got a reply from 2 experts :-) beat that Oct 14 12:33:57 within 15 minutes even :P Oct 14 12:38:14 stintel: yeah, there is some room for improvement there ;-) lol Oct 14 12:38:54 haha Oct 14 12:40:11 ah well, I have work slack in irssi so whenever I'm at work I will also see hilights on IRC. the other way around as well, I sometimes catch myself doing something for day job late in the evening when folks from US are asking something Oct 14 12:44:48 fyi, bumps pushed to staging Oct 14 12:45:43 I also added a bump for musl while at it Oct 14 13:04:22 xback: yeah sorry Sir, must try harder - pokes tongue out :-) Oct 14 13:04:40 lol Oct 14 13:06:50 If anyone remembers my 'ath10k high latency wierd packet drops on 5Ghz' of a couple of days ago... well... I think it's an Apple-ism 'cos I've just turned an MIR3G into an AP...and guess what... I see the same thing. Oct 14 13:08:51 of course I now have another unrelated question. It seems simple - but I warn you it has one of my https://pbs.twimg.com/media/DJyAs4rXoAIqyMb.jpg moments written all over it. Oct 14 13:10:43 In ash I wish to extract a setting stored in uci. For added fun I wish to extract uci items that are stored as lists as separate lines. Oct 14 13:11:13 uci get ucisection.foo.bar would seem to answer the first. Oct 14 13:14:30 ldir: one of the reasons I ditched crapple, wifi experience got worse with every OS X upgrade Oct 14 13:20:24 https://pastebin.com/3Ux3sBPM Oct 14 13:20:54 note the difference between show vs get output and the output of ' on the first item in the list. Oct 14 13:21:07 Why? Oct 14 13:39:20 Aha!! strpbrk(e->name, " \t\r\n"). ok and there's a space in the latter two strings. ok, that makes sense now. Oct 14 14:02:52 Can someone suggest a less ugly version of https://pastebin.com/NcHUXBdC Oct 14 14:04:12 I would have thought delim="$(printf '\n')" would be more portable but it simply doesn't work. Oct 14 14:06:29 stintel: iOS too, apparently. My dad's work iPhone recently got an update that broke connecting to the WPA3-enabled network and I had to switch it back to the WPA2-only one. :( Oct 14 14:06:51 mamarley: I'm never touching apple products again Oct 14 14:07:08 overpriced scrap metal Oct 14 14:07:10 * mamarley never has and never will. Oct 14 14:10:01 each to his/her/its own eh? Oct 14 14:10:13 ldir: IFS=$'\n' Oct 14 14:11:27 from https://patchwork.ozlabs.org/patch/1094476/#2172218 Oct 14 14:13:31 * ldir goes and tests - I won't tell you how long I've been hitting my head against that Oct 14 14:26:50 ynezz: will check tomorrow. I thikn irc status reporting is simply disabled for the docker based buildmaster deployments Oct 14 14:29:33 jow: it's not working for a few months already, so no need to hurry :) Oct 14 14:29:53 sorry for that Oct 14 14:30:26 no worries, I didn't noticed anyway until bump of gcc to 8.3.0 Oct 14 14:51:12 ldir: are you on catalina? Oct 14 14:52:52 Affecting HS just as much, so not a Catalina specific Oct 14 14:53:35 heh ldir what isnt apples fault when it comes to wifi it seems Oct 14 14:55:39 [Wed 2019-10-09 01:23:22 PM PDT] DonkeyHotei: there was some mac80211 regression that broke apple devices Oct 14 14:55:39 [Wed 2019-10-09 01:23:38 PM PDT] it apparently fixed in upstream but i dont have details yet Oct 14 14:55:43 ldir: ^ Oct 14 14:56:43 So not Apple's fault ;-) Oct 14 14:57:39 that's funny cause when something breaks or doesnt work its a apple device while everything else around it works fine haha Oct 14 14:58:28 ldir: most likely a regression in a workaround for apple brokenness Oct 14 14:58:45 you mean an apple "optimization" :) Oct 14 14:58:56 Christ such hatred Oct 14 14:59:15 some of their optimizations were clever and we should have all done them years ago Oct 14 14:59:32 Well... I do see why generally some folks hates apple... but not everyting is their fault either Oct 14 14:59:55 my main issue is the fanboism, and people claiming "nothing ever breaks" Oct 14 15:00:20 everything breaks, some more than others :) Oct 14 15:00:24 ldir: ;) Oct 14 15:02:47 rmilecki: ping Oct 14 15:04:33 stintel: indeed, it's the fanboy'ism that I hate most too, and "only" second then some things apple does that I'm incompatible with, but I don't hate apple for everything still =) Oct 14 15:05:24 I gave it a chance, had an MBA 11" for 2 years and then an MBA 13" for 2 years, and I wanted to smash that thing into the wall on multiple occasions Oct 14 15:05:30 let's just say we're not compatible :) Oct 14 15:15:37 yn Oct 14 15:15:39 ynezz: Oct 14 15:25:23 rmilecki: https://paste.ubuntu.com/p/665T5YSk9T/ Oct 14 15:26:19 this is on r11215+5-bba6646b5cba Oct 14 15:34:24 ynezz: so according to the validate_firmware_image this firmware image is valid as we can see "valid": true Oct 14 15:34:30 for some reason procd doesn't think so Oct 14 15:34:50 ynezz: this is from procd: valid = validation[VALIDATION_VALID] && blobmsg_get_bool(validation[VALIDATION_VALID]); Oct 14 15:34:55 yeah, just found that string there Oct 14 15:35:08 ynezz: for some reason that "valid" is false in procd Oct 14 15:35:26 I am still trying to get this Marvell 88E1116R PHY to work. This is what I added to the dts https://bpaste.net/show/2z4- Oct 14 15:35:26 According to the kernel log, it sees the phy (or at least something) at address 2, but it doesn't show anything when I plug or unplug the ethernet cable in that port Oct 14 15:35:27 Here is some of the log https://bpaste.net/show/JzR8 Oct 14 15:35:31 maybe there is a problem with procd calling validate_firmware_image binary? Oct 14 15:37:59 ynezz: can you add debugging to the validate_firmware_image_call() ? to print whatever it got from validate_firmware_image call Oct 14 15:38:12 ynezz: there is that loop: while ((len = read(fds[0], buf, sizeof(buf)))) Oct 14 15:38:24 try printf("GOT: %s\n", buf); Oct 14 15:38:28 inside the loop Oct 14 16:00:12 "valid": false Oct 14 16:01:41 https://paste.ubuntu.com/p/pGr7HC9kZz/ Oct 14 16:06:39 rmilecki: the problem is, that you're calling /usr/libexec/validate_firmware_image from two places Oct 14 16:06:45 isn't it? Oct 14 16:06:58 once in sysupgrade and 2nd time in procd Oct 14 16:07:18 the call in sysupgrade removes the metadata in the image, so the 2nd check in procd fails Oct 14 16:08:28 ynezz: what? I had no idea a call in sysupgrade may remove anything from firmware image! Oct 14 16:09:27 ynezz: so validate_firmware_image calls fwtool_check_signature and fwtool_check_image Oct 14 16:09:32 does any of those MODIFY a firmware image? Oct 14 16:09:45 you're calling fwtool_check_image which under the hood does `fwtool -q -t -i /tmp/sysupgrade.meta` Oct 14 16:10:15 -t removes the metadata Oct 14 16:10:25 from image Oct 14 16:10:38 this fwtool_check_image() is sometihng I would never expect Oct 14 16:10:48 thanks for debugging that problem! Oct 14 16:11:09 i don't know how to fix it at this point Oct 14 16:18:55 probably by renaming those files and that call to validate_*and_modify*_firmware_image :) Oct 14 16:23:33 confusing I would say https://paste.ubuntu.com/p/snDxc27mX3/ Oct 14 16:26:46 two calls on the same image return different results Oct 14 16:28:05 ynezz: i fully agree Oct 14 16:54:25 Do I need to do anything to enable or configure a rgmii phy? Oct 14 20:01:02 Why is the espeak: sinth in the openwrt packages? Oct 14 20:01:09 https://github.com/openwrt/packages/commit/21e195f65792caa2323ece7831f84318633de7c4 Oct 14 20:01:18 How can you use a sinth on a router. Oct 14 20:02:01 My screen reader uses it as a fall back sinth if the windos wan core voices don't work. Oct 14 20:02:29 wan = one Oct 14 20:04:32 Tapper: well a router can be computer or I guess any method of producing sound (even like via network audiosink or so), tough I would say rather niche on router still Oct 14 20:04:46 * Tapper nods Oct 14 20:05:51 I can see the how, but not really the why either, but hey, someone has made it possible, why not =) Oct 14 20:16:27 for a while making an audio streaming client out of OpenWrt routers seems to have been quite popular among users. so USB soundcard, active speakers and some kind of USB hid device for the controls, using mpd on the router as backend. espeak might have been a side effect of that Oct 14 20:16:57 hmm.. indeed :) Oct 14 20:17:46 I can see that, usually you have router already, so why not make it simple internet radio listener, if nothing else Oct 14 20:18:15 the Linksys RE6500 has an audio jack on which you can plug a speaker Oct 14 20:18:51 although I have no idea how it works and if it even works on openwrt Oct 14 20:19:16 I just noticed there was a jack :) Oct 14 20:19:35 so apparently some vendors think this is a use-case Oct 14 20:20:14 hey, I'm actually considering something like that myself (dedicated device, not doing routing tasks as well), I just haven't really found a good solution - read something that can be operated by non computer savvy users (which means physical knobs and controls) and which doesn't end up in a bowl of different PSUs and gets more expensive than it would be worth (even assuming an old router to be Oct 14 20:20:20 'free') Oct 14 20:20:20 anything to boost sales :) Oct 14 20:20:56 pkgadd: I think there is N+1 projects for raspberry pi for such Oct 14 20:21:29 olmari: sure, but as soon as you tally up the components, it's far from 'cheap' Oct 14 20:22:08 I'm somehow not willing to spend 60+ EUR on something like that Oct 14 20:22:52 you can hack buttons and controls however you want, for router or rpi... if you want anything that doesn't look "first prototype" then stuff cost some moneyz Oct 14 20:25:16 first prototype isn't a problem, but it should be fit as a kitchen radio to mere mortals (so phyiscal on/off, volume +/- is a hard requirement, channel switch not even strictly necessary) - and while I don't own a RPi (or any other ARM SBC) yet, my understanding is that its native analogue audio quality isn't sufficient for this task, so a dedicated sound hat for that, ... Oct 14 20:26:20 starting from zero just has a significant lead up cost (well, considering the relation to what it's supposed to replace) Oct 14 20:26:42 RPi native analog is just fine for "kitchen radio" =) while it definately is no hifi I'll give you that =) Oct 14 20:28:16 I mean no matter you buy, it will cost some 10's of euros/dollars... unless you make everythign yourself etc.. does not matter is if rpi or some old router or whatnot, it's the buttons and maybe display that will cost Oct 14 20:28:28 yep Oct 14 20:40:46 I'm still not getting anywhere with the phy on the device I am trying to port, does anyone have any ideas? Oct 14 21:06:44 what is snapshot considered? 18.0x or 19? Oct 14 21:07:07 "master" Oct 14 21:24:15 DonkeyHotei yes! you called? Oct 14 21:24:31 lol Oct 14 21:24:34 called what? Oct 14 21:24:42 me master! haha Oct 14 21:25:44 Slimey it's 19 + a kernel bump from 4.14 to 4.19 and about 1000 changes **** ENDING LOGGING AT Tue Oct 15 02:59:57 2019