**** BEGIN LOGGING AT Wed Jul 03 03:00:28 2019 Jul 03 04:50:22 aparcar[m]: yep, it's here https://git.openwrt.org/?p=openwrt/staging/ynezz.git;a=shortlog;h=refs/heads/wip/python3 Jul 03 04:50:48 aparcar[m]: just waiting for new buildbot images with Python3 and then we're probably going to do the switch Jul 03 04:59:21 rmilecki: BTW do you've any reason for having lua5.3 in the master tree? It seems to me like it should be in the packages feed Jul 03 05:04:32 speaking of lua5.3, it appears to cause widespread build failures Jul 03 05:04:54 since a number of core libraries (libubox etc.) have Lua bindings, these fail to build dragging everything along Jul 03 05:05:49 example: http://downloads.openwrt.org/snapshots/faillogs/mips_24kc/base/libubox/compile.txt Jul 03 05:06:15 I have the feeling that lua5.3 headers are staged in a location where they shouldn't be staged Jul 03 05:08:42 gah Jul 03 05:08:48 https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=package/utils/lua5.3/Makefile;h=61d1d1b48650b20fffcbd211d2ed7941cc7ea8cf;hb=7c23f741e97f6645bb5cd662a4943796a344b26a#l107 Jul 03 05:09:17 rmilecki: you cannot stage 5.3 lua*.h in /usr/include/ ! Jul 03 05:17:41 morning Jul 03 05:26:13 morning blogic Jul 03 05:30:18 jow: heya Jul 03 05:30:27 I plan to do some patchworkery today Jul 03 05:30:35 wanna get back into the swing ;) Jul 03 05:31:15 * jow is still fighting with Dockerizing the buildbot setup Jul 03 05:31:18 so close... Jul 03 05:31:27 business as usual ;) Jul 03 05:31:55 will announce 18.06.3/4 today Jul 03 05:31:57 I started a patch to allow make package/goo/prepare CHECKOUT=1 that essentially does the SRC_TREE_OVERRIDE setup for you Jul 03 05:31:59 its already uploaded Jul 03 05:32:18 we need to talk to woodhouse regarding the announce list Jul 03 05:32:31 its a fairly small one: https://openwrt.org/releases/18.06/changelog-18.06.4 Jul 03 05:33:19 mostly kernel spam ;) Jul 03 05:34:04 and I have the feeling that (lib)curl dominates the CVE list Jul 03 05:54:39 jow: thanks for fiding a problem Jul 03 05:54:47 rmilecki: yw :) Jul 03 05:55:15 rmilecki: for the next step of the transition I was thinking the following: Jul 03 05:55:32 - make lua(5.1) put its headers in usr/include/lua5.1 Jul 03 05:56:51 - create usr/include/lua{,conf,lib}.h headers emitting a "#warning Use of /usr/include/lua.h is deprecated, use /usr/include/lua5.1/lua.h" plus an "#include " Jul 03 05:57:19 jow: nice! Jul 03 05:57:34 - start switching all lua(5.1) using packages to add $(STAGING_DIR)/usr/include/lua5.1 to their CFLAGS Jul 03 05:57:57 (or check that those using the lua.pc pkgconfig still work properly) Jul 03 05:58:10 - rename liblua.so to liblua5.1.so Jul 03 06:04:52 jow: oh, you fixed that include in 5.3 alredy Jul 03 06:04:54 thanks! Jul 03 06:37:47 karlp: thanks for hint on bin/targets/bcm53xx/generic/packages/ Jul 03 06:39:42 LTE-networks, sometimes I want to scream ... I have created a small daemon sending an ICMP request every five seconds, and if three packets are lost then I reboot my LTE-modem. Turns out my operator made a change to state transitions in their network Jul 03 06:40:34 So instead of taking their usual ~1 second, the transitions suddenly took more than three seconds and my device was demoted immediatly to a lower state as I did generate enough traffic Jul 03 06:40:55 As a result my modem was flapping like crazy, resulting in lots of head scratching before I figured out that setting the interval to 1 second worked fine Jul 03 06:40:58 End-to-end this is not Jul 03 08:27:58 rmilecki: Thanks for spotting that I had forgot to have a proper CC-list on the relicensing patch. I guess it just slipped my mind to check the log for the files Jul 03 08:28:27 kristrev: ok Jul 03 08:28:48 * ldir has sent in his drop out of tree cake package patch - waits for screams Jul 03 08:32:42 perhaps skipping the relicensing part would be easier, the most important is the restructuring of the dts/dtsi Jul 03 08:37:08 kristrev: maybe try asking Alex directly for sending Acked-by Jul 03 08:37:19 maybe he'll reply quickly and we can be done with that hopefully Jul 03 08:37:59 rmilecki: Yeah, I can do that, but I see that there are three more people that have touched the files ... Jul 03 08:39:41 i checked target/linux/ramips/dts/WE1026-5G-16M.dts only - for that file INAGAKI's change was a tiny one & it was big global rework Jul 03 08:39:47 and sorry for the tone in my email too, not had my morning Coke Jul 03 08:39:51 so for that single file Alex's Ack is the most important Jul 03 08:39:54 I didn't check other files Jul 03 08:40:14 (without sugar, full of coffeine, not an adult yet so no coffee for me) Jul 03 08:40:19 well, I guess my first e-mail tone was worse than yours ;) Jul 03 08:40:46 i don't want OpenWrt to run into licensing problems & it's easy to miss such problems Jul 03 08:40:49 :) Jul 03 08:40:56 Yeah, sure, I agree, great that you spotted it Jul 03 08:40:56 i'm not sure if anyone else than me would pay attentiont to that licensing thing :P Jul 03 08:41:12 and that would mean someone pushing legally tained commit to repo Jul 03 08:41:46 The other file, WE1026-5G.dtsi, has changes from mkresin and ynezz, but they are both global reworks Jul 03 08:41:53 sorry, all global reworks Jul 03 08:42:10 ok Jul 03 08:42:22 also both devs are active, so it won't be a problem to get Acks from them :) Jul 03 08:43:00 the commit from Alex also seems to be a global rework? Jul 03 08:43:16 as well as Hiroshi Jul 03 08:45:33 probably yes, but it touches few more lines, so maybe we should have Ack, just in case Jul 03 08:45:40 i-am-not-a-lawyer Jul 03 08:52:13 mkresin: https://git.openwrt.org/?p=openwrt/staging/mkresin.git;a=commitdiff;h=c41fa1ff6c005f30f68e0af6793936ca31a818f6 is Reviewed-by: Martin Blumenstingl - can you please push it so I can base the kernel 4.19 update on it? Jul 03 08:56:32 What is the best approach to handling a kernel module that is in tree, that I optionally wish to replace with an out of tree version. Jul 03 08:59:27 e.g. kmod-sched-cake is in tree. Great. But some may wish to run an out of tree bleeding edge version. Jul 03 09:02:02 * ldir carries on fiddling with stuff Jul 03 09:03:55 it's sort of a bit like the ath10k ath10k-ct scenario Jul 03 09:04:11 ldir: I'm not sure about mac80211 but it seems there was a similar "issue": it's in mainline, but "newer is better" Jul 03 09:04:40 seems like in both cases the "newer and better" version is the only option Jul 03 09:05:51 I want to drop our out of tree cake package from base and use kernel version (+ patches) but to understand the mechanism for those wishing to carry on using an out of tree version located say in packages feed and not in base. Jul 03 09:07:00 Think I'm 90% there. Jul 03 09:07:27 perhaps following the pattern of ath10k is good enough? kmod-sched-cake would be the in-tree one, kmod-sched-cake-fresh the out of tree one and then you would use dependencies to ensure both can't be selected Jul 03 09:07:47 or at least hide the out of tree one if the in tree one is selected, or something like that Jul 03 09:17:49 xdarklight: done Jul 03 09:18:55 mkresin: could you have a look at https://git.openwrt.org/?p=openwrt/staging/ldir.git;a=commit;h=95469eb0fc817c32d86a16623fa9587730c0dd3c Jul 03 09:19:25 It's been in my tree for ages - I can't even remember where it came from - it appears sensible Jul 03 09:21:14 rmilecki: anytime :) Jul 03 09:28:47 ldir: I do remember this change. the change looks correct. AFAIK it isn't possible to trigger this code at the moment Jul 03 09:29:21 ldir: hence, no idea if it works at all or more register manipulation is required Jul 03 09:30:29 ok - I think I'll drop it from my tree then - an anonymous drive-by patch that somehow I thought "hmm that's interesting" and then.... :-) Jul 03 10:06:42 ynezz: please keep me updated once thats done. I'd be eager to test it Jul 03 10:40:58 lynxis: ping Jul 03 10:47:16 Tapper's router lives! Jul 03 11:02:06 mkresin: please send Acked-by if you approve some patch :) Jul 03 11:02:20 hint: [PATCH 1/2] ramips: Update ZBT WE1026 DTS-files Jul 03 11:11:00 jow: does getver.sh somehow support to get the current tag as well? Jul 03 12:37:00 ynezz: thanks for pulling that openvpn init fix Jul 03 12:41:21 ldir: what turned out to be the original issue? Jul 03 12:56:40 DonkeyHotei: I don't know - the bootenv was corrupt so I just restored it to defaults and it booted. I plan to build firmware for it and check out the 'luci dual boot' toy and go from there, see if I fall into any traps. Jul 03 12:57:44 Was quite worried - it initially came up with something like 'command bootnand not found' Jul 03 13:01:24 Hauke: When were you planning to upload the v5.2-rc6 mac80211? Jul 03 13:04:49 rmilecki: You can in fact be done with that quickly Jul 03 13:30:59 Monkeh: thank you for the serial leads Jul 03 13:34:31 ldir: Hope they help :) Jul 03 13:34:48 oh it's back & booting. Jul 03 13:35:11 haven't investigated (-ENOTIME) any further than that Jul 03 13:36:17 ldir cool! Jul 03 13:36:21 why are nbd's solutions to macos problems always so much more elegant than mine? :-) Jul 03 13:37:34 Tapper: the uboot environment was corrupt, restored to default and away it went. Would like to investigate the dual boot situation before I send it back. Jul 03 13:38:08 That's ok mate hang on to it as long as you like. Jul 03 13:38:13 That won't be until Sunday at earliest 'cos some work came in last night for Thurs->Sat Jul 03 13:38:22 NP mate Jul 03 13:42:10 ha - now working Sunday - Am very popular all of a sudden Jul 03 13:54:28 Using a btrfs subvolume with compression enabled for the openwrt source is a bad idea -> Not working Jul 03 13:55:24 I'm doing that for many years and it works just fine, never noticed any issues with that Jul 03 13:55:54 stintel: Patching libtool failed Jul 03 14:00:24 stintel: I did discard the logs but the following error meassage should be the right one Jul 03 14:00:24 patch: setting attribute btrfs.compression for btrfs.compression: Permission denied Jul 03 14:01:26 ah you enabled compression via xattr. have not tried that Jul 03 14:02:01 jow: could you help me out here (https://github.com/openwrt/openwrt/pull/2192) with the release_version? Jul 03 14:02:07 Yes, that is right. How did you enable the compression? Jul 03 14:02:55 mount -o compress=zstd Jul 03 14:03:34 seems like a Gentoo user ran into the same problem with portage trying to apply a patch, but no reaction Jul 03 14:04:02 you should probably report this, but I'm not sure where. (patch or kernel bugtracker) Jul 03 14:04:17 Yes, I read the posting before a few minutes. ;) Jul 03 14:04:40 what does a btrfs subvolume with compression for openwrt source mean? Jul 03 14:05:09 you have a btrfs volume, and then put the openwrt source inside some logical block inside that with autocompression? what's that for, what does that help wtih? Jul 03 14:06:35 karlp: You can simple create snapshots of the openwrt source with a very low disk space usage Jul 03 14:06:59 karlp: Compression, saving space on the ssd Jul 03 14:08:25 the talkback from our git server is getting excessive https://pastebin.com/Nkx5x8ke Jul 03 14:09:58 I'm simple put all the projects in a subvolume. It is very simple to send them for backup to your nas Jul 03 14:19:32 does a make dirclean ; make to see if it all still works Jul 03 14:20:33 ldir: Host git.openwrt.org github.com Jul 03 14:20:34 IdentityFile ~/.ssh/id_ed25519_lede Jul 03 14:20:34 SetEnv LC_MEASUREMENT=en_US.UTF-8 LC_MONETARY=en_US.UTF-8 LC_NUMERIC=en_US.UTF-8 LC_TIME=en_US.UTF-8 Jul 03 14:21:01 in .ssh/config - in your case do that for LC_CTYPE Jul 03 14:21:34 it silences some of it :) Jul 03 14:21:59 do what eh? Can you translate that into idiot ? Jul 03 14:23:09 ldir: write this in your .ssh/config: https://pastebin.com/7MXwh7Nt Jul 03 14:23:21 and then git push again Jul 03 14:24:24 ahhh! suitably translated - thank you! Jul 03 14:28:47 * ldir reads man ssh_config Jul 03 14:29:47 stintel: when are you going to fix your processors ? Jul 03 14:29:51 * ldir hides Jul 03 14:37:08 ldir: eh? Jul 03 14:38:52 Hallo32: actually you could ask in #btrfs if someone knows a solution / workaround Jul 03 14:44:39 stintel: Quick fix, disable compression. I have placed it on my list, but there are also a lot of other points on it. The time ... Jul 03 17:34:40 mamarley: I plan to do an offical backports 5.2-rc7 release tomorrow and then send a patch to OpenWrt Jul 03 17:34:54 Awesome, thanks! Jul 03 17:34:57 mamarley: does backports 5.2 work for you or have you found any regression? Jul 03 17:35:45 Hauke: 5.1 works fine for me with ath9k and ath10k-ct but I haven't been able to try 5.2 because you haven't uploaded it yet. :P Jul 03 17:39:50 mamarley: I uploaded it some days ago Jul 03 17:40:00 and I updated the link in the commit Jul 03 17:40:22 Oh, OK, I must be running an old version of that. Let me pull it again. Jul 03 17:54:39 Hauke: I just got it compiled. It works well so far, thanks! Jul 03 19:56:29 for the power (ie, 23) limit shown in 'iw reg get', does that include the antenna gain? Jul 03 20:20:00 Tapper: ping Jul 03 20:20:50 ldir pong Jul 03 20:21:14 what firmware would you like on your 3200acm? Jul 03 20:22:04 ldir just stick a snapshot on there if you like, install luci tho pleas. Jul 03 20:22:11 :) Jul 03 20:22:52 Sure, ok, I'll pick a snapshot build in a couple of days when I'm back and put that on, as well as luci of course. Jul 03 20:22:56 ldir Do you know what went rong? Jul 03 20:23:01 Thanks Jul 03 20:23:43 Beyond 'the uboot environment was wrong in some way' no - I just did the default procedure and it all came good. Jul 03 20:24:18 I managed to flash my own screaming edge build 4.19.57(!) on to both partitions and have booted from both, so the dual boot appears to work. Jul 03 20:24:19 at this point, there is little difference between a snapshot and the 19.07 branch Jul 03 20:24:27 OK thanks for all your work mate I appreciate it Jul 03 20:24:58 You can just install luci on your build if you like. Jul 03 20:25:24 DonkeyHotei: he wants to be bleeding edge though Jul 03 20:25:28 can't take that from a man Jul 03 20:25:29 What's screaming edge about it? Jul 03 20:25:46 when bleeding edge hurts :-) Jul 03 20:25:50 haha Jul 03 20:25:56 papper cuts? Jul 03 20:26:04 that was just your wife yelling in your ear ldir Jul 03 20:27:21 I've left the serial JST to standard dupont style pins lead that Monkeh made up connected and in the case wrapped with a bit of insulating tape. Jul 03 20:27:57 OK cool that's good. Just encase ! Jul 03 20:28:03 You can undo the front 2 screws and gently pull the front off to gain access to that lead. Jul 03 20:28:26 Yeah what is on the end of it? Jul 03 20:28:50 You'll need a sighted person to tell you the colours, black is of course ground. Yellow & orange TX/RX (but might have that the wrong way around) Jul 03 20:28:58 I can do the frunt bit simples! Jul 03 20:28:58 2.54mm pins, "dupont style pins" Jul 03 20:29:21 O cool my lede i have will fit that then I think Jul 03 20:29:42 2.54mm pins - in theory ready to go into your normal usb serial lead that often terminate in dupont style sockets. Jul 03 20:29:50 Yeah and my wife will tell me what is what when it comes to cullers Jul 03 20:30:12 good to know thanks to you and muncky Jul 03 20:30:34 I assume your serial adapter is a 3.3 volt type and not 5 volt TTL ? Jul 03 20:31:03 I don't know I think it is 3.3 Jul 03 20:31:23 That's perhaps the best misspelling of my name I've ever seen Jul 03 20:31:28 if you don't want fireworks and magic smoke, you better know for sure ;) Jul 03 20:31:29 they usually are Jul 03 20:31:33 hahah Jul 03 20:32:12 monke sorry Jul 03 20:32:16 :) Jul 03 20:32:59 pkgadd is there a way to tell? My wife looked on the usb plug and it does not say. Jul 03 20:33:08 Anyway, good news it lives :-) Jul 03 20:33:22 Yes I am happy. Jul 03 20:33:48 I'll leave my own snapshot build on there then, it has luci, so you can get to either an official release or your own easily. Jul 03 20:34:03 Cool. Jul 03 20:34:09 Tapper: not that easily, unless specified by the manufacturer/ seller and/ or written on the PCB (further complicated by some allowing both, depending on a jumper setting or solder bridge) Jul 03 20:34:21 The less messing about you have to do the better. Jul 03 20:34:23 sure, a scope would tell for sure, but... Jul 03 20:34:47 I will look on amazon where I got it from Jul 03 20:36:25 great that it could be recovered (relatively easily) Jul 03 20:37:18 if it was sold as being for raspberry pi, it's likely 5V Jul 03 20:38:07 DonkeyHotei r pi yes Jul 03 20:38:11 nothing except raspberry pi has used 5V in ages Jul 03 20:38:27 hmm, is the RPi really 5 volt based, not 3.3V? I think only (some arduinos used 5V?) <-- no experience with RPi though Jul 03 20:38:56 This one says it's 3.3 V Jul 03 20:39:25 It says not for usage with arduinos Jul 03 20:39:26 idk, but i have a serial console adapter sold as being for the pi, and it only does 5V Jul 03 20:39:51 yes, RPi is using 3.3V serial and GPIOs (5V only used for USB and USB power, in front of the voltage regulator) Jul 03 20:41:38 https://www.raspberrypi.org/documentation/configuration/uart.md Jul 03 20:42:35 It's this one http://Amazon.co.uk Adafruit 954 USB TTL Serial Black, Blue Interface Cable and Adapter Jul 03 20:43:58 https://www.adafruit.com/product/954 --> The power pin provides the 5V @ 500mA direct from the USB port and the RX/TX pins are 3.3V level for interfacing with the most common 3.3V logic level chipsets. Jul 03 20:44:39 in other words, safe, as long as you don't connect Vcc (and you should never do that anyways, unless you know exactly why you really need to) Jul 03 20:44:51 pkgadd so I cant use that? Jul 03 20:45:08 O I can thanks Jul 03 20:45:12 you can, just never connect Vcc - only RX, TX and GND Jul 03 20:46:12 I think it's fine Tapper but whatever you do, tie up or hide the red wire. It's carrying 5V from USB and you don't want that near your linksys Jul 03 20:46:23 Will do Jul 03 20:46:26 I'd suggest to glue the red wire back against the cable with some insulating tape, you'll 'never' need it Jul 03 20:46:57 Good point! I am going to crack out the tape rite now! :) Jul 03 20:47:22 Be gon red wire! Jul 03 20:47:41 the black wire is ground... and black is used as ground on the internal cable made up by Monkeh - so you've only 2 others left :-) Jul 03 20:48:00 It's like count down! Jul 03 20:48:12 Tick tick tick boooooommmmmmm! Jul 03 20:48:13 50 50 - no you can't phone a friend Jul 03 20:48:27 * Tapper grins Jul 03 20:48:54 * Tapper thinks hang on who wood I call? Jul 03 20:48:58 as a rule of thumb (almost) nothing you'll 'ever' do with OpenWrt will require anything else but 3.3 V Jul 03 20:49:09 OK Jul 03 20:49:46 * ldir decides to call it a night - early ish start Jul 03 20:51:22 pkgadd: iirc the archer c7 uses 1.8V instead Jul 03 20:51:54 DonkeyHotei: I know that some highend ARM SBC (sometimes) use 1.8V, but the Archer C7 would surprise me Jul 03 20:52:11 ...but I don't have one, so haven't followed it Jul 03 20:52:31 i never connected serial to mine, but i think i remember seeing that on its wiki page Jul 03 20:53:12 at least wikidevi suggests 3.3 volts Jul 03 20:54:24 OK red is taped up I have black wite and green. Jul 03 20:54:56 ldir later mate. Jul 03 20:56:13 https://openwrt.org/toh/tp-link/archer-c5-c7-wdr7500#recovery_using_serial_connection also just suggests some shenanigans with an additional pull-up resistor, but still 3.3V Jul 03 21:00:35 maybe i'm thinking of a different device Jul 04 00:59:37 hi Jul 04 00:59:54 no news Jul 04 01:00:38 nope :( Jul 04 01:04:47 anyway to dump the spi flash without using programmer Jul 04 01:05:04 when i do it powers up the board and doesnt work Jul 04 01:06:43 no bootloader? Jul 04 01:07:42 no everything is fine just want a backup in case things do go south Jul 04 01:08:52 i meant when the rpi and everything is connected and the soic clip is on the power going to the chip actually powers on the board and you cant get a read of the flash Jul 04 01:09:05 correct Jul 04 01:09:31 any options other then removing the flash Jul 04 01:10:04 but with access to the bootloader, you can potentially boot an initramfs that can read the flash Jul 04 01:10:18 barring that, jtag Jul 04 01:10:49 yeah like an image that has spi access enabled Jul 04 01:11:40 ie /dev/spidev0.0 Jul 04 01:12:22 what board is this? Jul 04 01:13:22 the 1925 we have been messing with Jul 04 01:15:37 dumping its flash is dead simple with the image i built, because the whole flash is exposed Jul 04 01:16:35 if you concatenate all the mtd blocks, you should have exactly 32M Jul 04 02:57:22 * mangix is having second thoughts about this local tarball thing. too many failed hashes. **** ENDING LOGGING AT Thu Jul 04 02:59:57 2019