**** BEGIN LOGGING AT Fri May 01 02:59:57 2020 May 01 07:17:07 blogic: ping. do i recall correctly your having a patch for vlan support on DSA somewhere for 5.4 ramips? May 01 07:43:32 Borromini: I think you mean these patches? https://github.com/openwrt/openwrt/pull/2942 May 01 07:54:03 thanks Rene__ ! May 01 07:54:28 Borromini: pong May 01 07:57:28 dengqf6: ping? :) May 01 07:58:31 Borromini: I am the author of that patch May 01 08:00:30 dengqf6: yes :) someone asked about DSA on the forum, and i thought blogic had patches May 01 08:00:37 i've linked to yours :) May 01 08:00:49 https://forum.openwrt.org/t/build-for-edgerouter-x-not-booting-up/61815/18?u=borromini May 01 08:02:58 i'm still fighting with my DIR-878 which somehow keeps using old style interface names instead of lan1 / ... May 01 08:05:23 Borromini: do you also see gpio_keys error in dmesg? error -19? May 01 08:06:09 Rene__: i'm not running 5.4 on a daily basis May 01 08:06:25 ok May 01 08:08:01 i suppose 20.xx is still ways off as well, i just switch my devices once a git branch appears May 01 08:19:17 nbd: could you please check this PR https://github.com/openwrt/openwrt/pull/2724 (backporting your fixes for mac80211/hostapd into 19.07) ? May 01 08:30:23 ynezz: LGTM May 01 08:31:06 nbd: thanks! May 01 08:59:47 useless report: i was just trying to sysupgrade a nanobridge m5 from ar71xx (bullet image) to ath79 (bullet-m-ar7241) and for some reason despite -F'ing the sysupgrade, it wasn't getting written. i don't have a console on it, so no clue why it didn't work (thus useless). i resorted to manual mtd -r write firmware, and that worked. May 01 09:12:18 russell--: sysupgrade has been wonky lately for me as well. May 01 09:22:00 ynezz: you have the wrong @nbd on GH :) May 01 09:27:07 good late morning May 01 09:33:06 rmilecki: hi! May 01 09:33:55 late good morning May 01 09:34:29 what's the name of the shared 'notepad' type thing we have on openwrt.org ? May 01 09:35:18 pirate something? we used sth like that in the past May 01 09:35:51 something like 'ethpad'. May 01 09:37:43 i recall using http://piratepad.net/ in the past only May 01 09:37:58 etherpad! May 01 09:39:35 rats, no longer works May 01 09:41:43 https://pad.riseup.net/? May 01 09:43:12 tks May 01 09:43:41 we used to have our own instance of etherpad but that seems to have gone awol May 01 09:44:14 Am trying to stay focussed on the current task and not be diverted by other rabbit holes :-) May 01 09:50:52 =) May 01 10:00:30 Who knows if I've done any of that right! May 01 10:45:07 what calls "block extroot" during early init? May 01 10:47:48 is that call used at all? May 01 11:00:04 i found it, it's mount_extroot() is fstools in libfstools/extroot.c May 01 11:02:09 rmilecki: glad it's not just me who goes on these code hunts :-) May 01 11:02:47 does anyone know if it's possible to get rrdtool to fill ABOVE a value instead of below it? May 01 11:08:51 rmilecki: i was grep'ing for it, but you beat me to it. i knew it was in a package, but i didn't remember the name May 01 12:11:29 anyone has idea why do we look for extroot twice in fstools? it's in mount_root.c in start() May 01 12:11:33 1. First there is a direct call to block.c::mount_extroot() May 01 12:11:34 2. There is also mount_overlay() call which calls extroot.c::mount_extroot() which executes "block extroot" that is the same as block.c::mount_extroot() May 01 12:31:28 f00b4r0: shell doesnt know how to autocomplete :p May 01 12:31:39 ;) May 01 13:09:24 ah, i think i know what was happening with my sysupgrade ... OOM'ing May 01 13:42:15 russell--: how much RAM? May 01 14:40:34 Anyone able to walk me through the way in which the different kernel .config.xxx files are handled during the kernel build process? May 01 14:40:37 I have a env/kernel-config with some custom options. I can see they're all being put into the .config.target file initially May 01 14:40:52 When I check the final kernel image (extracting the config via the IKCONFIG option), a small subset of these custom options are not actually in the final kernel. Only some of them are. I'm trying to work out why this is the case May 01 14:48:21 oh noes. it seems mt7621 ethernet driver has broken TSO. May 01 14:49:39 oh never mind. It's with TX checksum offload. May 01 14:55:23 i finally reviewed that old [PATCH] fstools: block: make extroot mount preparation more robust May 01 14:55:28 hmmm May 01 14:55:49 ethtool tx off disables TSO too. May 01 14:56:37 gch981213: m4t: jow: ^^ May 01 14:56:50 cool May 01 14:57:14 well, i'm a bit annoyed, it's another change that handles a single specific use case while breaking others May 01 14:57:47 i understand people will keep making/suggesting such changes, but it takes so much time to review things properly and avoid breaking code May 01 15:17:37 which use cases does it break? May 01 15:21:40 m4t: /etc/config/fstab on "rootfs" May 01 15:26:50 i don't really understand that code tbh May 01 15:30:06 uhm, would disabling CONFIG_UBIFS for x86 work instead? May 01 15:30:39 er, UBIFS_EXTROOT i mean May 01 15:33:21 it would help with the check_extroot() part and would be another case specific trick May 01 15:33:39 it would not help with finding /etc/config/fstab on overlay May 01 15:33:49 m4t: what's the "mount" output on x86? May 01 15:34:19 this is with the patch: https://paste.ee/p/AAKDy#0orEvGmX0l1iiyoLe6QKbPngtdXmqtB3 May 01 15:36:14 that pakes makes fstools read /tmp/overlay/etc/config/fstab during init, but I'm not sure what's mounted at that path May 01 15:36:42 my guess is /dev/sda3 May 01 15:37:09 m4t: do you have /etc/config/fstab with "overlay" UCI entry on /dev/sda3 ? May 01 15:37:11 isn't it the loop partition May 01 15:37:15 yes 1 sec May 01 15:37:36 need to remount the /dev/loop0 or whatever May 01 15:38:26 https://paste.ee/p/UQFa1#mcBKA9M0paWREPlg2iot5i5gG56yNkXC May 01 15:38:58 also, here's the uuids as referenced in fstab: https://paste.ee/p/t3rFT#vJ48trIG8iPEQC8riL83PoVtFbLTplEL May 01 15:39:07 ok, so it's about /dev/sda2 that uses f2fs May 01 15:39:26 the proper solution is to make main_extroot() f2fs aware then May 01 16:21:29 nbd: ping. this might be an issue with mt76 but it's unclear whether the user is running the latest mt76 you pushed a few days ago - https://forum.openwrt.org/t/support-for-dir-878/18160/49 May 01 16:38:55 hi, who is the maintainer of the netifd project? May 01 16:55:45 dedeckeh: ^^ May 01 16:59:39 guifipedro:I'm reviewing the patch May 01 17:02:04 I don't like the usage of the linux struct ifla_vlan_qos_mapping in vlan_devconfig as this is linux specific May 01 17:03:14 Better to use a new defined struct in netifd and use struct ifla_vlan_qos_mapping only in system-linux.c May 01 17:04:32 I will add the comments to the patch together with possible other comments May 01 17:04:56 when I've completely reviewed the patch May 01 17:08:45 dedeckeh: hi! May 01 17:09:20 dedeckeh: I've took a brief look at https://bugs.openwrt.org/index.php?do=details&task_id=3056 and wonder if there's an easy way to fix it May 01 17:10:56 dedeckeh: do you have any idea what might have been the motivation to include these options in the first place? May 01 17:15:24 jow: no idea why this option was included as this added by cyrus May 01 17:15:36 *was added May 01 17:16:04 I was also planning to have a look at it when I find some time May 01 17:19:28 allright. I thought it might be something we can easily #ifdef out, but that send_router_advert() is a kind of god function... May 01 17:22:54 jow:I'm wondering if we just should disable the automatic inclusion of this option and offer an uci option allowing to configure routes which should be included as RIO RA option May 01 17:24:34 Then it's up to the user to configure routes if they're needed to send them as RIO option May 01 17:33:10 dedeckeh: yep, I had the same thoughts May 01 17:35:06 If a “.keep” file is present in a directory, the directory is kept across sysupgrades? What’s it for? May 01 17:37:05 ok, thanks dedeckeh I will forward this comments to pespin May 01 17:44:41 jow: can /usr/share/xt_geoip (which contains a downloaded and unpacked database) be added to “conffiles” or would that be inappropriate? I want it preserved across “sysupgrades”… May 01 17:46:26 philipp64: depends on the size I guess May 01 17:47:05 7MB about... May 01 17:47:14 then its not appropriate May 01 17:47:27 it will exceed the available backup space ofr many targets May 01 17:48:18 it’s 500 files, most averaging about 1400 bytes… the median is 400 bytes. May 01 17:49:23 okay. if you use it in /etc/firewall.user (as I do) and then sysupgrade, when the system reboots, it will lose the database and not be able to reapply it until you manually run the scripts to rebuild it… May 01 17:49:45 not ideal from a security perspectivel. May 01 18:01:28 ldir: what was that config clean command for the buildroot again? May 01 18:11:55 found it :-) make -C scripts/config clean May 01 18:30:07 jow: btw, where is the config processing for firewall? I wanted to add logic to automatically append --contiguous to a time-based rule when stop_time < start_time … May 01 18:42:45 actually, easier than I thought: http://paste.ubuntu.com/p/sFkppqXb63/ May 01 18:42:48 Hauke: I'll discuss the Season of Docs thing with Thomas from the wiki and post it to the public list. Will keep you updated May 01 19:04:12 jow: just tested the —continguous patch… running it locally. May 01 20:13:11 hi. What's the correct way to flash a stock TP-link image using 'mtd'? (I have skipped 512 bytes as in the docs and then tried to flash it using mtd) May 01 20:13:50 but I get the following after a couple of seconds: May 01 20:13:50 root@OpenWrt:/# mtd -r write /tmp/tplink.bin firmware May 01 20:13:51 Unlocking firmware ... May 01 20:13:51 Writing from /tmp/tplink.bin to firmware ... [e]Failed to erase block May 01 20:15:42 and if I ignore erase with -n I get: May 01 20:15:43 Writing from /tmp/tplink.bin to firmware ... [w]Error writing image. May 01 20:16:07 I'm wondering if that's not the correct way to flash the stock image and it's too large to fit in the flash, for instance... May 01 20:19:08 mac80211 is failing building in MODPOST stage (undefined references) because kernel's Module.symvers only contains vmlinux symbols, not symbol from other modules May 01 20:20:26 nbd: I see that you applied some patch regarding Module.symvers, do you know what is the issue? May 01 20:41:24 who is responsible for administering the buildbot? May 01 20:44:50 it needs gcc-multilib installed on it. the host component build of luajit fails since it does not have the 32 bit libraries installed May 01 21:05:57 blogic: can you take a look at my fstools / mounting question, please? May 01 21:05:58 http://lists.infradead.org/pipermail/openwrt-devel/2019-December/020887.html May 01 21:06:39 the problem is that samba.hotplug doesn't work [1] https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=package/network/services/samba36/files/samba.hotplug May 01 21:07:38 ubusd and blockd aren't up when first mounting happens and so mounting scripts can't read devices info using "ubus call block info" May 01 22:24:53 Hauke: I sent an email regarding the google docs program, please try to respond asap due to the tight deadline May 01 23:08:10 nbd: so in kernel 5.4 scripts/Makefile.modpost changed: when modpost is being done for vmlinux.o, it ignores the already built modules and throws away the information already stored in Module.symvers from make modules May 01 23:09:34 nbd: so if vmlinux.o is built after modules (which is being done in my case on openwrt), Module.symvers contains only symbols from vmlinux.o, and subsequenty mac80211 and other packages compiling external kernel modules fail to build May 01 23:11:00 nbd: this happens when package/kernel/linux/compile is being done after target/linux/compile May 01 23:15:46 nbd: I sent a RFC patch to linux-kbuild maling list May 01 23:58:21 anyone ever worked with https://google.github.io/oss-fuzz/? May 02 00:20:39 ynezz: maybe that could be cool instead of CI integration May 02 01:48:15 there is a gpio declaration in /etc/config/system that refers to the poe passthrough, but it doesn't show up in sysfs anywhere i can find **** ENDING LOGGING AT Sat May 02 02:59:58 2020