**** BEGIN LOGGING AT Fri Apr 17 02:59:57 2020 Apr 17 06:01:35 build #333 of kirkwood/generic is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/kirkwood%2Fgeneric/builds/333 Apr 17 08:55:25 Can I use an BLOBMSG_TYPE_INT32 for an BLOBMSG_TYPE_INT16? If I get json from network it is int32, if I get it from interal ubus it is a int16 Apr 17 08:55:41 so, does BLOBMSG_TYPE_INT32 break internal ubus? Apr 17 09:09:09 I now switch between two policies Apr 17 09:28:08 git.openwrt.org seems to be down :_ Apr 17 09:29:03 checking in a minute Apr 17 09:30:00 or at least I'm getting some gateway timeouts... Apr 17 09:32:06 yeah, that vm is simply overlaoded I guess Apr 17 09:32:32 load of 25 atm Apr 17 10:10:05 jow: ping - my problem/question on collectd stats display Apr 17 10:13:05 https://github.com/ldir-EDB0/luci/commit/9a64592aa54313ad1fbf68165efbed4b0ffa983b If I add the overview graph, then all the other graphs only show 1 instance (ie tin 0). if I remove the overlay graph, then all the tins get displayed (in individual instances - great!) Apr 17 10:18:15 This is for today's game of 'bug/feature/silly idea' :-) Apr 17 10:43:49 ldir: hm yeah, sounds like a bug indeed... as if the first overview graph somehow sets a flag/condition which isn't reset for subsequent graphs Apr 17 10:44:41 ldir: can you mail me some .rrd files so that I can play around? Apr 17 10:44:55 you router was an x86 too, wasn't it? Apr 17 11:00:55 jow: apu2 x86_64 Apr 17 11:10:26 jow: you've got mail :-) Apr 17 11:31:25 guys is the mailing list down? Apr 17 11:31:41 hmm. on and off somehow it seems. nevermind me Apr 17 11:43:07 build #70 of bcm47xx/generic is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/bcm47xx%2Fgeneric/builds/70 Apr 17 11:48:35 jow: I'm tempted to split out the 'overview' vs 'by tin' graphs to separate pages. It's a lot of info in 1 page. Apr 17 11:52:44 blocktrron: FYI DEVICE_PACKAGES += rssileds kmod-ath10k-ct-smallbuffers | \ Apr 17 11:53:41 blocktrron: that stray | in 8918c038f330 leads to config warnings Apr 17 12:04:33 'discuss' - shell referencing variables - $foo or ${foo} Apr 17 12:06:14 :P Apr 17 12:10:47 ynezz: will fix that in a minute - thanks for the heads up Apr 17 12:47:30 build #70 of bcm27xx/bcm2711 is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/bcm27xx%2Fbcm2711/builds/70 Apr 17 14:18:02 Is there a way to install a script as part of a package as a specific (non-root) user ? Apr 17 14:18:51 * ldir suddenly has 'post install' pop into head - git grep Apr 17 14:20:37 build #69 of bcm63xx/smp is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/bcm63xx%2Fsmp/builds/69 Apr 17 14:32:22 * ldir urghs Apr 17 14:32:57 * f00b4r0 hands ldir a paper bag :} Apr 17 14:33:03 It's raining here and I've been trapped inside for too long. Apr 17 14:35:08 i can sympathize. How long have you been locked down? Apr 17 14:35:37 4 weeks, maybe more. Apr 17 14:36:19 I've tried to keep myself enthused & happy etc... but it's all wearing a bit thin now. Apr 17 14:36:24 yeah same here. I think we've just completed week 5. Kinda lost track. Apr 17 14:36:31 you bet. Apr 17 14:36:55 we're slowly losing sanity. Apr 17 14:37:07 And today like most days, I've sat in front of the computer fiddling with stuff, some of which is great, some of which is "why did I even contemplate this?" Apr 17 14:37:45 heh Apr 17 14:38:24 I did an on-line 'python for everyone' course, which was interesting. The online bit stops at ch6 so I should get off my backside and do ch7- 'n' Apr 17 14:39:13 But I've been fiddling with 'graphing CAKE sqm' for the past 3ish days. That has stretched me and I'm tired. Apr 17 14:39:52 It is basically done/working, final push to make it vaguely presentable. Apr 17 14:40:03 sounds good Apr 17 14:42:41 None of it uses python of course. 'ash', collectd, JSON, javascript. Am no good with any of it 🀣 Apr 17 14:45:57 Yeah March 13th was what I called 'Virus Friday' when I lost all my work bookings. Maybe April 17th is when I lost my sanity πŸ˜‚ Apr 17 14:46:09 ;^P Apr 17 14:47:19 ldir https://www.youtube.com/watch?v=49FB9hhoO6c Apr 17 14:48:08 ldir I love the way the drums are mix on that track. Apr 17 14:48:36 ldir and it's a grate song for how I am feeling rite now. Apr 17 14:48:44 LOL - so I click on link and get a YT advert "meet new sexy models!" and I wonder what Tapper is trying to tell me :-) Apr 17 14:49:19 ldir hahah picks of girls are no good for me mate so I passed them on! Apr 17 14:50:43 Yep he's hitting that low tuned snare in smallish room, with mostly room mics Apr 17 14:51:30 I love the sound. Apr 17 14:52:02 No one records drums like that no more Apr 17 14:53:08 ha. Classic indeed. Apr 17 14:53:18 * f00b4r0 picks up his guitar (: Apr 17 14:53:37 f00b4r0: hey :) can you please take a look at dmesg of a RB device that has no wireless for whatever odd reason https://del.dog/raw/ap_dmesg ? Apr 17 14:53:59 PaulFertser dude he's shredding rite now! lol Apr 17 14:54:09 hopefully I haven't gone this far :-) https://www.youtube.com/watch?v=CKti7QixnJI Apr 17 14:54:11 and banging hi's head Apr 17 14:54:54 And probably tapping too. Apr 17 14:56:07 ldir tune! Apr 17 14:56:44 PaulFertser O yeah tap tap tapit in hahahaha Apr 17 14:57:01 PaulFertser: looks like missing driver in kernel build. That's a mAP lite, i have two of those and there's nothing special about them Apr 17 14:57:18 f00b4r0: the one man band! Apr 17 14:57:46 f00b4r0: but ath9k is in a kmod, and /sys/bus/pci/devices is empty . What might be missing in kernel build then? Apr 17 14:57:58 thank god Thomann still operates. I was able to buy a cheap practice guitar and get it delivered where I am currently locked-in Apr 17 14:58:26 empty pci/devices is expected, there's no pci bus on this device. Apr 17 14:58:34 f00b4r0: there's a fellow on #openwrt that's just lost with this issue and I have no idea how to diagnose, it's just too odd. kmod-ath9k is included in all RB images automatically afaict. Apr 17 14:59:13 is this one of our images or a custom built? I'm not sure looking at the first line of dmesg Apr 17 14:59:20 https://www.youtube.com/watch?v=CKti7QixnJI - David Byrne in full on 'TV Evangelist' mode :-) Apr 17 15:02:11 f00b4r0: official image Apr 17 15:04:04 leir https://www.youtube.com/watch?v=tVCUAXOBF7w Apr 17 15:04:06 board_rev=r2 Apr 17 15:04:11 in kernel command line Apr 17 15:05:32 PaulFertser: mikrotik could be up to something new there. Can you get us a dump of the hard_config mtd segment? Apr 17 15:05:35 PaulFertser: maybe he built without rootfs_per_target? Apr 17 15:06:27 adrianschmutzler: ar71xx is single image, I'm not sure it would matter? Apr 17 15:06:59 more likely something changed in the hardware Apr 17 15:07:23 device packages won't be included then, I'm not sure what happens with default_packages Apr 17 15:07:30 f00b4r0: he says it worked on 18.x release and then sysupgraded to 19.x and now it's not visible with either. Apr 17 15:07:44 oh Apr 17 15:07:47 that's even weirder Apr 17 15:08:09 And if hard_config was damaged I'd expect an error in dmesg... Apr 17 15:08:21 fortunately it's the only spare device I have at the moment, I can boot mine see what gives Apr 17 15:08:54 PaulFertser: i'm not saying hard_config is damaged, I'm saying it would be useful for me to parse and look for differences. Apr 17 15:09:49 f00b4r0: I mean if wmac is not registered by the machine startup code then I can't see how it can happen if data is ok, and if data is not ok then I see pr_err() there so should be visible in dmesg. Apr 17 15:10:08 if you're making questions and answers... Apr 17 15:10:20 curl -k -i -X POST -d '{"jsonrpc": "2.0", "method": "login", "params": { "user": "root", "password":"sdfsdf"}, "id": 1}' https://192.168.1.1/cgi-bin/luci/rpc/auth -- is this still supposed to work on 19.07? Apr 17 15:10:46 I got : {"id":1,"jsonrpc":"2.0","result":null}, did not see any token as it used to be in 18.06 Apr 17 15:12:18 f00b4r0: I know you're the expert and I'm not so I understand my reasoning is likely to be flawed. However I thought it's worth sharing in case my code understanding happened to be right. Apr 17 15:13:03 Could someone click on merge here: https://github.com/openwrt/packages/pull/11870 That is very important change Apr 17 15:13:20 i have an r2 board here too, incidentally. Older bootloader though. I just booted an older master snapshot (built March 28) and I have wifi working Apr 17 15:13:29 Let's see what happens if I netboot current master Apr 17 15:15:02 nick[m]1: done - hope I don't get any comeback on it! Apr 17 15:16:17 @ldir: thank u!!! :) this was fixing a null pointer issue. And someone is trying to use the 802.11k features and I'm struggelin how to enable package source tree override and change config parameters accordingly. Apr 17 15:17:31 the null pointer was not a result of my implementation, more a corner case of a client responding with a wrog repsonse Apr 17 15:18:36 all I know is... don't dereference null pointers, it doesn't end well ;-) Apr 17 15:20:00 For the reference, I do not see wmac in platform devices https://del.dog/raw/platform_devices Apr 17 15:20:30 PaulFertser: wifi working with current master. And I suspect the dmesg you pasted isn't from our images, mine begins with Apr 17 15:20:36 [ 0.000000] Linux version 4.14.125 (buildbot@602ce18687ab) (gcc version 7.4.0 (OpenWrt GCC 7.4.0 r10243-4d11c4c378)) #0 Mon Jun 17 04:53:07 2019 Apr 17 15:20:44 hmm Apr 17 15:20:45 wtf Apr 17 15:20:52 that can't be right Apr 17 15:21:15 oh yes it is. Last built snapshot is june 17 Apr 17 15:21:56 * f00b4r0 tests 19.07.1 for good measure Apr 17 15:24:17 and wifi works there too Apr 17 15:25:45 PaulFertser: since It Works For Me™, I'm going to need a bit more info and once again, a dump of the SPI hard_config would be a good start. Apr 17 15:25:56 f00b4r0: he's on it Apr 17 15:26:15 assuming the hw isn't damaged, of course :) Apr 17 15:28:44 PaulFertser: [ 0.000000] Clocks: CPU:750.000MHz, DDR:600.000MHz, AHB:300.000MHz, Ref:25.000MHz Apr 17 15:28:47 overclocked device. Apr 17 15:28:57 anything can happen. Tell him to revert to stock cpu speed first. Apr 17 15:29:52 it's an especially bad idea to overclock this device which is already prone to overheating due to extremely small form factor. Apr 17 15:30:54 * f00b4r0 resumes shredding :) Apr 17 16:00:58 build #364 of lantiq/ase is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/lantiq%2Fase/builds/364 Apr 17 16:13:38 How can I enable in my config file the CONFIG_WNM_AP flag in hostapd? So make this as dependency? I could specify wpad, but I'm not sure. Apr 17 16:15:04 I mean Makefile Apr 17 16:16:23 I made a change to musl and rebuilt (v18.06.6) and I'm noticing something funny -- all of my files under build_dir/target-mipsel_24kc_musl/root-ramips/bin are dated Jan 5 13:06. What is causing this? Apr 17 16:17:56 I stripped my libc.so (that I know I built) and it doesn't come to the same size -- 599228 (manually stripped) vs 599132 (in root-ramips). Apr 17 16:18:19 Oh, I didn't use the rstrip.sh script. maybe that's it Apr 17 16:22:42 nope, that's not it Apr 17 16:27:47 build #365 of at91/sam9x is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/master/images/builders/at91%2Fsam9x/builds/365 Apr 17 17:00:52 f00b4r0: http://paulfertser.info/mtd1.dump is his dump Apr 17 17:01:04 He said he reverted clock to stock. Apr 17 17:05:17 dmesg with fixed clocks https://del.dog/raw/new_dmesg2 Apr 17 17:05:27 Have there been build dependency issues related to the the libc fixed in v19 or HEAD since v18? I don't think my musl libc.so is being properly propagated after changing and re-building. I'm about to verify this. Apr 17 17:06:20 ok, nevermind, it is Apr 17 17:06:37 the build is just stripping it in a way that I can't reproduce. Apr 17 17:06:47 and changing the date to January 5th, 2020! Apr 17 17:29:43 PaulFertser: now I know what the problem is Apr 17 17:30:20 PaulFertser: his device uses a new scheme (for this device) for storing wlan calib. I suspect it throws off wlan detection code Apr 17 17:30:47 * f00b4r0 double checks Apr 17 17:33:01 yep Apr 17 17:33:03 that's exactly it Apr 17 17:34:05 PaulFertser: mikrotik is introducing changes in new hardware that are going to break an increasing number of devices in ar71xx. Another evidence of a similar problem is here: https://github.com/openwrt/openwrt/pull/2729 Apr 17 17:35:17 I'm working on cleanly addressing this issue (see my two patchsets currently under review by xback). Fixing this problem in ath79 should be fairly trivial, fixing ar71xx is going to require quite a bit more work but the plan is to backport to 19.07 bits I wrote for that. Apr 17 17:35:59 PaulFertser: the key problem here is that these changed are introduced without any change in SKU, so you can only figure this out once you bought the device. Apr 17 17:37:26 I suspect the routerboot version is a good clue however: my working device is on 6.42.x, his is on 6.44.x Apr 17 17:38:06 PaulFertser: however I doubt that his device ever worked in 18.06. That would invalidate my reasoning. Unless Apr 17 17:38:10 * f00b4r0 checks git log Apr 17 17:40:10 Hauke: I booted ebin from UART with Armbian U-Boot and USB with SATA work without problems with the patch I posted here Yesterday. Apr 17 17:47:51 PaulFertser: nothing in git suggests a software-side breakage. Either this /particular/ device never worked with owrt before; or the user upgraded routeros on it and then flashed owrt again; or the wlan hw is fried; or i need to dig more. In that order :) Apr 17 17:49:23 tmn505: nice Apr 17 17:49:44 tmn505: I thought there is some sort of autodetection and if that fails it falls back to the old behaviro Apr 17 17:49:55 at least I understood the error message like this Apr 17 17:52:26 f00b4r0: thank you so much! Apr 17 17:52:50 PaulFertser: if my assumption is right, he'll have to be patient before he can use owrt on this one. Apr 17 17:53:53 f00b4r0: understood. Converting from new format to old would be a waste of time and it'd be problematic to overwrite etc. Apr 17 17:54:08 yes. Calibration data is device dependent. Apr 17 17:54:13 Hauke: I'll send that patch to ML this weekend. Apr 17 17:54:43 PaulFertser: support will come. Just not today :) Apr 17 17:56:40 f00b4r0: I understand. I meant probably using rbextract on this very dump and then somehow pack it back in the old format. Apr 17 17:57:06 unclear it would work. It's probably packed that way because it's too big. Apr 17 17:57:58 just be patient. I've got code in flight with xback that's going to set the first step toward addressing that problem, and I'm presumably locked in (with more time than usual) for another month. I'll fix this. Apr 17 17:59:50 I know you're working on proper ath79 support for mikrotik boards and that involves partition parsing and exposing calibration data in a meaningful way to kernel modules etc. Apr 17 18:00:10 Thank you very much Apr 17 18:01:06 It's unfortunate dmesg doesn't show any obvious signs of being unable to parse it. Apr 17 18:01:46 PaulFertser: there is a dirty workaround that could probably work immediately but I'm reluctant on "publicizing" it: misused it's going to trigger more false positive bugreports Apr 17 18:02:14 that's because the code that's failing isn't verbose. Apr 17 18:21:38 jow: do you have an advise on how to do more caching when running buildbot locally for testing? so that it does not rebuild tools etc every time Apr 17 18:21:51 f00b4r0: if the wireless driver didn't need platform device registered and would request calibration data as "firmware" from filesystem I can imagine a dirty workaround (extracting from a dump on host with rbextract). But since it's not the case, hm, you got me really puzzled. Apr 17 18:22:30 ynezz: how did you fix the signify VERIFY_ONLY? It break compilation on my device. Apr 17 18:25:03 Yes, this dirty trick should work with ath10k ac devices. Apr 17 18:34:57 PaulFertser: no, there are two lines of code that need to be changed to support his device. However doing so will break _all_ other map-lite devices that don't have the same "feature" as his. If someone puts an image with this mod in the wild saying it works on map-lite devices we're screwed. Apr 17 18:35:15 and by support I really mean "support", quote-unquote Apr 17 18:35:54 f00b4r0: so requiring kernel rebuilding, not that instant. I see. Thank you for all your work on OpenWrt! Apr 17 18:36:14 i'm in the process of cleaning all that for ath79 but it takes time, as you can imagine. Apr 17 18:36:29 yes, kernel rebuild required Apr 17 18:37:29 rmilecki: ping Apr 17 19:19:06 f00b4r0: pong Apr 17 19:25:53 Ping Apr 17 19:59:45 rmilecki: is there any development on the broadcom ax devices ongoing? Apr 17 20:00:44 Hauke: not that I'm aware of Apr 17 20:00:57 rmilecki: ok thanks Apr 17 20:14:15 aparcar[m]: unfortunately no, I simply threw hardware at the problem Apr 17 20:15:27 jow: I think I got the staging implementation working. I'll send you more patches :S Apr 17 22:50:04 So the first (at least the first I'm aware of) MediaTek ax stuff seems to be coming from D-Link: https://fccid.io/KA2IRX1860A1 Apr 17 22:50:24 not available in EU yet as it seems Apr 17 22:52:30 and no drivers, i'm sure Apr 17 22:58:45 that perception would be incorrect, at least in terms of the wireless drivers, https://lkml.kernel.org/r/cover.1586895593.git.ryder.lee@mediatek.com Apr 17 23:02:21 wow Apr 17 23:08:15 what does "HE" mean in 802.11ax terms? Apr 17 23:12:21 high efficiency Apr 17 23:12:37 karlp: 802.11ax == High-Efficiency wireless Apr 17 23:12:44 == WIFI 6 Apr 17 23:12:47 thanks. Apr 17 23:14:00 marketing. a.k.a the art of obfuscating facts. Apr 17 23:14:15 nice only 9K line of code Apr 17 23:17:23 I am considering buying a Xiaomi AX3600, it uses a IPQ8071A, are other IPQ807X devices easy hackable or do they use secure boot where the SoC verifies some signatures Apr 17 23:17:29 I asume the SoC support this Apr 17 23:19:11 can that be queried easily from dmesg/ sysfs? Apr 17 23:20:26 I haven't found any logs and would like to know this before investing some money Apr 17 23:20:44 or at least I would like to make an informed guess Apr 17 23:27:39 Hauke: xiaomi seems to lock bootloaders quite often Apr 17 23:31:19 during build of 19.07 it stucks at mkhash: https://pb.nanl.de/show.php?id=44982&mode=raw Apr 17 23:32:05 `mkhash sha256` without any further argument (path to file) being called - probably waiting for STDIN forever that way Apr 17 23:39:19 I'm trying to put openwrt on linksys mr8300, but in the step where it should allow me to upgrade manually, there is no manual section. Any ideas? Apr 17 23:47:37 stuck at "sh -c mkhash sha256 > '/home/mirko/openwrt.git/dl/gcc-7.5.0.tar.xz.hash'" again.. weird Apr 18 01:52:42 Hauke: I expect xiaomi to at least disable uart rx like they've done on all recent routers. **** ENDING LOGGING AT Sat Apr 18 02:59:57 2020