**** BEGIN LOGGING AT Wed Oct 24 03:00:00 2018 Oct 24 07:57:05 hauke: I followed current method being used throughout the full tree. Some targets also don't support it, avoiding a symbol which is not required for it. Oct 24 07:57:53 mangix: Compilation for Turris Omnia succeeded here Oct 24 09:11:28 blogic: ping Oct 24 11:41:04 * ldir wanders in Oct 24 11:56:26 hello Oct 24 11:57:07 when multiple packages provide the same `meta` package via `PROVIDE:=`, which one is chosen if another package require it as a dependence? Oct 24 12:11:59 require it with a select? not sure, but should be easy to test. Oct 24 12:32:01 aparcar[m]: undefined currently, needs a fix in opkg Oct 24 13:18:27 jow: what do you think of https://pastebin.com/HZX7M9Pt - any glaring stupid things before I throw it onto the ML? (tested on mvebu/cleafog pro by padjffs2'ing the rootfs in the sdcard image) Oct 24 13:20:31 currently assumes boot partition is the last partition Oct 24 13:24:35 KanjiMonster: i remember someone saying he doesn't like using whole disk space, no idea who was it... jow? nbd? Oct 24 13:24:45 KanjiMonster: i never could understand that argument though Oct 24 13:25:51 rmilecki: we could probably easily offer images with fixes sizes and autosizing ones - I intentionally used a marker to make it opt-in, not automatically Oct 24 13:28:56 the reason why i think fixed size rootfs partitions are better is because of how sysupgrade works Oct 24 13:29:26 having so much space on the rootfs partition can be useless if any data you put on there gets lost on sysupgrade Oct 24 13:29:52 so the idea is to use the remaining space to create a persistent data partition which is unaffected by sysupgrade Oct 24 13:30:10 maybe a better approach would be to auto-create that instead of auto-resizing the rootfs partition Oct 24 13:34:48 KanjiMonster: I like the idea Oct 24 13:35:20 well, having it as a partion unaffected by sysupgrade isn't a bad idea either Oct 24 13:35:39 that introduces further headaches down the road should we ever need to increase the rootfs Oct 24 13:35:48 that's true Oct 24 13:36:38 rootfs size is also a build time configuration value, so nothing we can rely on at any time Oct 24 13:38:18 maybe something semi flexible is feasible Oct 24 13:38:35 *flexible approach Oct 24 13:39:05 like expand rootfs but keep at least X%/MB of the blockdev unused at the end if space allows Oct 24 13:40:02 with the rise of containers and stuff, applications and people tend to place their application data on dedicated mounts points anyway Oct 24 13:40:35 so the rootfs is mainly used to store executables and libraries (which are reinstallable after a sysupgrade) Oct 24 13:50:23 hi all, is there an easy way to determine the OpenWRT version from the build system? Oct 24 14:23:42 aep4: ping Oct 24 14:34:37 jow (IRC): is there an idea for a clean approach? Like marking some "variant" as default? Oct 24 14:35:49 karlp (IRC): I tested it and it seems the last item in alphabetical order is selected... Oct 24 14:36:09 I woudl feel like preferences on defaults should be external, Oct 24 14:36:18 they all PROVIDE Oct 24 14:36:44 just don't do +VIRTUAL_PACKAGE and you're fine. Oct 24 14:36:52 if you're building your own things, name what you want explicitly, Oct 24 14:37:11 if you're a package with a depends, just use VIRTUALPACKAGE without the + so it is a proper depends not a select Oct 24 14:47:24 KanjiMonster: regarding your paste, line 79. Any reason for the -1 in the buffsize? snprintf reserves space for the 0-term in the size provided :) Oct 24 14:48:24 xback: it does? I very rarily code userspace, so I try to be as defensive as possible ;) Oct 24 14:49:06 KanjiMonster: https://linux.die.net/man/3/snprintf Oct 24 14:49:16 "The functions snprintf() and vsnprintf() write at most size bytes (including the terminating null byte ('\0')) to str. " Oct 24 14:54:06 KanjuMonster: line 128 Oct 24 14:54:26 KanjiMonster: line 128 (typo :) ) Oct 24 14:54:46 I think it's missing a check for fd over there Oct 24 14:58:50 xback: yeah, still a bit of proof-of-concept code Oct 24 15:45:51 greearb: hi! do you have any tips about what I can do to try to cause a crash other than wait for it to randomly occur? Oct 24 15:47:03 huaracheguarache, maybe hope the other reporters can hit it. If it is power-save issue, having lots of different buggy stations connected and going in and out of range and in an out of sleep might hit it. Oct 24 15:47:22 The original test case was something like 20 systems doing download tests. Oct 24 15:47:40 but we don't see this bug internally (often enough to debug, at least) Oct 24 15:48:13 afk for a bit Oct 24 15:51:17 greearb: hmm, ok. Oct 24 15:55:04 karlp (IRC): I'm not sure if I understand you corrently. There is a main package with require a profile, one and only one profile should be installable. If none is selected, fallback to a default Oct 24 15:55:36 I don't follow Oct 24 15:55:49 Using it without the + would hide the main package until a profile is selected Oct 24 15:56:56 yes. Oct 24 15:57:14 what's the point of alterantives if you're not choosing them? Oct 24 15:57:28 if you'er doing this automatically, just selct them yourself. Oct 24 15:58:33 okay, then I rename the default package to aa-profile-name so it's selected as default if not other profile is explicitly selected Oct 24 16:00:42 you mean make an extra package that selects a specific dependency? sure that works, just see whether peopl elike having the "normal" package and preconfigured pacakges. Oct 24 16:18:17 ldir ping Oct 24 16:44:23 Hi any one want to push this patch for me? Oct 24 16:44:24 https://patchwork.ozlabs.org/patch/987769/ Oct 24 16:44:40 Bilt and tested and working on my wrt3200acm Oct 24 16:56:45 Tapper: done Oct 24 16:56:54 Thanks Oct 24 16:58:31 no problem, thanks for testing it ;-) Oct 24 17:16:30 xback: hi Oct 24 17:24:53 What is the proper way to enable CONFIG_DEVMEM? I want to use util 'io' for development. Oct 24 17:26:36 global build settings->kernel-build options->"/dev/mem virtual device support" Oct 24 17:27:20 karlp: thanks Oct 24 17:39:19 some one on the forums is giving away hardwair Oct 24 17:39:22 https://forum.openwrt.org/t/giving-hardware/23889/2 Oct 24 17:41:53 whats with the sucky flash on even modern devices Oct 24 17:45:43 Yeah 8 meg is not grate in 2018 Oct 24 20:27:32 stintel, you around? Oct 24 20:27:52 Or, anyone else know the details of how to tftp to the netgear r7800? Oct 24 20:28:05 using tftp from cmd-line on my linux box I assume Oct 24 20:28:25 no experience with that device Oct 24 20:28:42 hmm, someone was telling me about it, must have been someone else Oct 24 20:29:38 pkgadd, ahh, it was you :) Oct 24 20:30:39 greearb_: https://openwrt.org/toh/netgear/r7800#debricking should be pretty simple, power on the r7800 with the reset button pressed - and upload the firmware image with a tftp client at the right time (in case of problems, put a simple unmanaged switch between the router and your workstation, to avoid the link loss/ link training) Oct 24 20:31:07 yeah, but I don't use tftp that often, so I wish that page had some very precise examples Oct 24 20:32:06 I only have the nbg6817, which needs a tftp server on the workstation instead, while the r7800 needs a tftp client (e.g. atftp) Oct 24 20:33:49 different device, but this should basically work with the r7800 as well (using the correct image names and addresses): atftp --trace --option "timeout 1" --option "mode octet" --put --local-file openwrt-ar71xx-generic-wzr-hp-ag300h-squashfs-tftp.bin 192.168.1.1 Oct 24 20:36:05 I might have it working.... Oct 24 20:38:43 serial-port is dumping mostly garbage Oct 24 20:38:48 for the nbg6817 I need to install tftpd-hpa, dump the firmware to /srv/tftp/ras.bin and configure the workstation to 192.168.1.99; or just install a recent OpenWrt master snapshot (didn't make it for 18.06.x) from the OEM gui Oct 24 20:43:16 for posterity, here is how I did it. If someone has permission to update the wiki, please feel free: Oct 24 20:43:22 cd to dir that holds the factory image Oct 24 20:43:27 tftp 192.168.1.1 Oct 24 20:43:30 > bin Oct 24 20:43:43 > put [factory-image] Oct 24 20:52:17 https://openwrt.org/toh/netgear/r7800#debricking /done Oct 24 20:52:50 thanks Oct 24 21:05:51 greearb_: what was the garbage dumping about? Oct 24 21:06:30 like it has wrong serial rate, but I'm at 115200 8 N 1, no hw or sw flow-ctrl Oct 24 21:06:58 it mostly worked until I upgraded (ie, factory image showed useful output on serial) Oct 24 21:07:53 sounds like a baud mismatch Oct 24 21:08:26 hmm, https://github.com/openwrt/openwrt/blob/master/target/linux/ipq806x/files-4.14/arch/arm/boot/dts/qcom-ipq8065-r7800.dts#L40 suggests a correct/ expected setting there Oct 24 21:12:34 is there some way to query current serial port settings from ssh login? Oct 24 21:13:12 cat /proc/cmdline should show it Oct 24 21:13:29 blank Oct 24 21:15:56 hmm, what happens with dmesg | grep tty, I'm looking for something like Bootloader command line (ignored): board=NBG6817 root=/dev/mmcblk0p8 rootwait zld_ver=2.04 console=ttyHSL1,115200n8 mtdparts=m25p80:0xC0000(SBL)ro,0x40000(TZ)ro,0x40000(RPM)ro,0x80000(u-boot)ro,0x10000(env)ro,0x10000(ART)ro,0x10000(dualflag),0x210000(reserved) Oct 24 21:16:38 you may have to re-add console=ttyHSL1,115200n8 via the dts Oct 24 21:17:20 is it OK to run sysupgrade from ssh terminal? First two times it worked, last one seems to have bricked it Oct 24 21:17:28 yes Oct 24 21:17:30 nevermind, it came back Oct 24 21:18:03 root@OpenWrt:~# dmesg|grep tty Oct 24 21:18:03 [ 0.000000] Bootloader command line (ignored): console=ttyHSL1,115200n8 Oct 24 21:18:03 [ 0.302068] 16340000.serial: ttyMSM0 at MMIO 0x16340000 (irq = 101, base_baud = 460800) is a MSM Oct 24 21:18:03 [ 1.023395] console [ttyMSM0] enabled Oct 24 21:19:29 ah, yep, [ 0.308042] 16340000.serial: ttyMSM0 at MMIO 0x16340000 (irq = 101, base_baud = 460800) is a MSM Oct 24 21:22:05 so, er, what does that mean? Oct 24 21:22:42 well, trying 460800 could be fun... (I don't really consider myself to be a dts specialist) Oct 24 21:29:41 anyone know if there is a package that will include 'dhclient' ? Oct 24 21:30:30 I've been using an old dhclient for years, but it is a pain to cross-compile Oct 24 21:31:25 probably isc-dhcp-client-ipv4 isc-dhcp-client-ipv6 Oct 24 21:32:50 yes, thanks Oct 24 21:39:51 what about a full version of iproute? Oct 24 21:40:33 ip-full I guess? Oct 24 21:40:43 yes Oct 24 21:51:48 what about full 'ps' support? ie, ps -auxwww Oct 24 21:52:13 my bet would be on procps-ng-ps Oct 24 22:29:03 w/regard to serial, a new usb ttl cable made it work better, I guess the old one had issues Oct 24 22:46:02 so with regard to the default network config on the r7800, there is a br-lan that contains only eth1.1. Any reason not to ditch the br-lan and just deal with the eth1.1 directly? Oct 24 22:47:01 the reason for that is bridging LAN and wlan together in the default config Oct 24 22:47:25 ahh, ok. I'm planning to use this more like a PC than an AP though Oct 24 22:47:50 I guess the vlans are needed to talk to the switch hardware? Oct 24 22:48:10 yes, two CPU ports going to the same switch Oct 24 22:54:05 I guess eth0.2 would be the WAN ethernet port? Oct 24 22:57:09 according to the bootlog (https://openwrt.org/toh/netgear/r7800#openwrt_bootlog), eth0.2 seems to be WAN and eth1.1 LAN, but you can freely configure the switch as needed Oct 24 22:57:21 nod Oct 24 22:57:27 looking at the network config now Oct 24 22:58:38 so I could put vlan 1 on port 1, vlan 2 on port 2, vlan 3 on port 3, etc? I'm going to give that a try Oct 24 22:58:52 yes Oct 24 23:00:10 you know what the 0t means in this, or documentation link? option ports '5 0t' Oct 24 23:00:41 ahh, think I found docs Oct 24 23:01:29 tagged vs untagged Oct 24 23:01:43 greearb_: port 0 is the one connected to the cpu, probably Oct 24 23:01:49 so it needs to see the tag Oct 24 23:01:54 ok Oct 24 23:01:57 (it == cpu) Oct 24 23:02:07 russell--: 0 and 6 Oct 24 23:02:50 two CPU ports, both connected to the same switch Oct 24 23:03:12 that options line says only 5 and 0 though Oct 24 23:04:10 "1:lan" "2:lan" "3:lan" "4:lan" "6@eth1" "5:wan" "0@eth0" from /rom/etc/board.d/02_network Oct 24 23:05:22 so can I do something like this? Oct 24 23:05:25 https://pastebin.com/9iVwJGmW Oct 24 23:05:47 for my use, it would be nice to have individual control over each Ethernet port Oct 24 23:06:48 that should work Oct 24 23:07:15 I'm using a different vlan for all ports on my nbg6817, except for two serving br-lan Oct 24 23:08:12 i once built a device with seperate networks on each lan ports (for parallel flashing ath25 devices) Oct 24 23:12:15 I don't have a managed switch, and ones with enough 1 GBit/s ports (>>=24) are a bit beyond my budget at the moment - so I need to split up my independent networks right at the router (with unmanaged switches from there) Oct 24 23:14:07 hmm, no vlans were built on eth1 after making those changes Oct 24 23:15:43 how can I attempt to manually (re)load that /etc/config/network logic? Oct 24 23:15:48 maybe it will spit out errors Oct 24 23:17:44 greearb_: /etc/init.d/network restart Oct 24 23:18:16 heh, no errors Oct 24 23:18:38 you should have an eth1.N in your config somewhere? Oct 24 23:19:37 don't yet, or I removed the thing that creates the bridge, didn't add anything back Oct 24 23:22:47 "option ifname eth1.1" or whatever Oct 24 23:22:48 ok, that is working Oct 24 23:26:23 i wonder if anyone will give me a time-domain reflectometer for xmas Oct 24 23:53:34 i heard that uboot will automatically load an image named root_uImage from the tftp server if it exists, anyone know if that's true? Oct 25 00:00:55 where is a good place to put files to save through sysupgrade? I notice /home/foo is purged Oct 25 00:01:28 whatever you define to be kept via /etc/sysupgrade.conf Oct 25 00:02:18 heh, so easy! Oct 25 00:07:04 I notice all vlans have the same MAC addr, technically this should be OK, but some (cisco) switches get confused...just FYI, not sure it is worth changing Oct 25 00:09:15 config device 'jail_dev' \n option name 'eth1.12' \n option macaddr '60:31:97:XX:XX:XX' \n Oct 25 00:11:03 nice, sending from port 1 to port2 works, just like real ports Oct 25 00:13:57 ok, next thing....is udev supported? Reason being, if I reload the ath10k driver, the wiphy will come back named phy2, phy3 (instead of original phy0, phy1). That confuses my application. Previously, I've used udev rules to set their name based on the radio MAC addr Oct 25 00:16:38 I guess the answer to that would be along the lines of "not really) Oct 25 00:17:33 I can write scripts to deal with it manually..already have to do that for some things since normal udev is a pain anyway Oct 25 00:17:47 greearb_: just rmmod mac80211 cfg80211 too Oct 25 00:18:25 heh, that messes up other things of mine...I can fix it to work how I want w/out much trouble Oct 25 00:19:09 I'm still recompiling every 2 minutes when I find another perl module I'm missing :) Oct 25 00:31:56 anyone up for fixing this? Seems to still be there: https://github.com/zyxmon/qnapware/issues/2 Oct 25 00:34:22 probably I don't need the diagnostics feature though, so not worth the effort Oct 25 01:05:00 I have some photos of the inside of a Zyxel NBG6817, but no rights to upload them to the wiki. Can anyone help? Oct 25 01:17:32 spacewrench: /* How to add images ========> http://openwrt.org/meta/adding_images_to_openwrt_wiki */ Oct 25 01:21:02 disclaimer, I haven't tried that yet - last time I had to add an image, I wrote a forum pm to tmomas Oct 25 01:22:34 It's working better than just trying to upload from the text editor. Thanks! Oct 25 01:23:06 noice! looks like it worked! Oct 25 01:23:40 did you connect a serial console to yours? Oct 25 01:28:24 feel free to upload higher resolution images - and looking at your serial.jpg, the answer seems to be yes ;) if you still have it open or connected, please catch the bootloader messages from u-boot Oct 25 01:31:22 (I'm kind of hoping for hints to toggle the active boot partition set in the bricked state, like keep button x pressed for y seconds while powering on, etc. tftp recovery is easy, but there must be some way to toggle the bootflag using just the buttons and the bootloader) Oct 25 01:35:02 I ended up adding a wire to the outside of the case for serial, then I put the lid back on. I can get you bootloader messages in a sec. Oct 25 01:35:17 that would be great, thanks! Oct 25 01:35:42 (no need to hurry, I'll be around for a while) Oct 25 01:36:50 FWIW, the stock configuration appears to be incorrect. It chooses a wifi driver ... QCA9xx0? But the router has QCA9984. I've been trying to build a system that has the appropriate firmwere in a place that can be located and installed. No luck yet...wifi doesn't work for me yet. Oct 25 01:38:56 is selects ath10k-firmware-qca9984-ct in current master, ath10k-firmware-qca9984 in 18.06 and earlier, both are correct. https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/ipq806x/image/Makefile;h=b82d54b0f6ac658e41c8f60a96742d167b28cd2d;hb=HEAD#l254 Oct 25 01:39:39 and it's working on my nbg6817 Oct 25 01:39:57 I cloned master then checked out branch v18.??.1 (the latest non-bleeding-edge) Oct 25 01:40:53 I kept rebuilding, adding random little utilities, and something was keeping the sysupgrade from being the most recent (I use the random utilities to see whether I'm getting the stuff I expect) Oct 25 01:40:54 https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/ipq806x/image/Makefile;h=2902af3231a87592e9c040e7e7e1e4e1958a01a5;hb=70255e3d624cd393612069aae0a859d1acbbeeae#l250 Oct 25 01:41:12 Anyway, I just did a clean build, compiling now. Hopefully it will work better. Oct 25 01:41:43 check ./scripts/diffconfig.sh Oct 25 01:43:10 I'm following master, but need to revert the mac80211 to the 4.14 based older variant for the time being (as there is an issue with ath10k and ath10k-ct using the newer mac80211 with WDS/ 4addr) Oct 25 01:43:43 OK, I have a log from power-up...where did you want me to put it? Oct 25 01:44:15 you can dump it to http://paste.debian.net/ or just add it on top of the bootloag in the wiki Oct 25 01:46:13 http://paste.debian.net/1048903/ Oct 25 01:47:26 thanks a lot :) Oct 25 01:48:05 np Oct 25 01:48:39 zyxel really isn't good at spelling ;) Oct 25 01:49:00 "Multiboot clinent version: 2.0" the firmware upgrade code uses "daulflag" Oct 25 01:53:28 Is the ath10k firmware somewhere visible in the filesystem? How can I tell whether the right thing is included in the image so the boot-up process can send it to the HW? Oct 25 01:54:02 opkg list_installed | grep ath10k --> ath10k-firmware-qca9984 and kmod-ath10k Oct 25 01:54:43 http://paste.debian.net/1048904/ that's what should be in the file system Oct 25 01:56:55 that's my problem...I have /lib/firmware/ath10k/QCA99X0 Oct 25 01:57:22 from the source selection that shouldn't happen Oct 25 01:57:43 ./scripts/diffconfig.sh might provide a clue Oct 25 01:57:43 I've been building from make menuconfig Oct 25 01:58:22 selected the device, but didn't fiddle with the firmware (until later). But after the first build, it seems like make clean doesn't really delete everything. Oct 25 01:59:03 it doesn't (there are multiple stages of cleaning) Oct 25 02:08:20 for a quick test, you can opkg update && opkg install ath10k-firmware-qca9984 Oct 25 02:08:54 (of course your buildconfig should include the correct packages, but testing is easier this way) Oct 25 02:10:28 good idea...should've thought of that. I try to get everything I want in the base build and just go with that. This router has a lot of space, tho, so I could reasonably add packages... **** ENDING LOGGING AT Thu Oct 25 03:00:00 2018