**** BEGIN LOGGING AT Sat May 02 02:59:59 2020 May 02 10:50:46 russell--: on mt7621? May 02 10:51:10 gch981213: old version, new works May 02 10:51:28 turns out i was about a month old May 02 10:51:32 or so May 02 10:52:02 and, yes, on an ubnt erx (mt7621) May 02 10:52:29 russell--: gpio number in sysfs is changed due to the kernel bump, but /etc/board.d/03_gpio_switches isn't changed with the kernel bump. May 02 10:53:04 There's a later PR for ubnt er-x devices. But the other two still aren't fixed. May 02 10:56:42 erx is working for me now (r13136-ad27c133eb) May 02 10:57:17 BTW gpio sysfs is deprecated by libgpiod: https://git.kernel.org/pub/scm/libs/libgpiod/libgpiod.git/tree/README May 02 10:57:46 which means we need to migrate our gpio_switches config at some point. damn it. May 02 14:47:26 gch981213: do you also noticed that gpio_keys has issue with getting IRQ? May 02 14:48:46 gpio-keys keys: Unable to claim irq 25; error -13 May 02 16:24:20 Rene__: mt7621 or mt76x8? I don't remember seeing this when testing the kernel bump. May 02 16:25:11 mt7621 May 02 16:26:09 Same for me, I also can't remember that is what an issue on my kernel development. I remeber that I tested the button. May 02 16:27:52 I checked if it also happens on the v5.4.1 kernel and it does. May 02 16:33:26 Rene__: I don't have my routers with me this week. I'll try to figure it out when I'm back home. May 02 16:33:27 gch981213: I also see the same error message when the I2C GPIO expander tries to claim gpio 8 as an interrupt May 02 16:33:40 ok May 02 16:33:54 https://elixir.bootlin.com/linux/v5.7-rc3/source/drivers/input/keyboard/gpio_keys.c#L524 May 02 16:35:08 This point is fails with error -EACCES https://elixir.bootlin.com/linux/v5.7-rc3/source/drivers/input/keyboard/gpio_keys.c#L632 May 02 19:53:19 Rene__: OpenWrt uses it's own button implementation: https://github.com/openwrt/openwrt/blob/master/package/kernel/gpio-button-hotplug/src/gpio-button-hotplug.c May 02 21:28:51 russell--: How are things in… it’s Portland, isn’t it? May 02 21:44:48 jow: did you see the PR for adding —continguous to wrapping firewall rules? May 02 21:48:20 alright. email from an abuse department. this will be good. May 02 21:50:25 o_O May 02 21:50:48 I still think LEDE or an old kernel should be used for there 4/32MB devices May 02 21:51:00 mangix: what have you done? May 02 21:51:10 dd-wrt actually uses kernel 3.10 May 02 21:51:29 philipp64: your email comes from abuse department May 02 21:53:27 grrr… Mail.app has been buggered since I upgraded to Catalina. May 02 21:54:21 I've heard that Catalina has many bugs, yes. May 02 21:54:35 s/Catalina/OSX/ May 02 21:54:36 scnr May 02 21:54:47 there’s no notion of “preferred” email address and it seems to pick one randomly. sigh. May 02 21:54:50 stintel: it is no longer OSX. It's macOS now May 02 21:55:05 re-sent. May 02 21:55:39 same shit, different name May 02 21:56:47 stintel: stintel: since i have your attention, https://github.com/openwrt/openwrt/pull/2792#issuecomment-623000507 May 02 21:56:49 philipp64: yeah, will review it later May 02 21:58:43 mangix: I'd need to be able to test it but I always end up yak shaving because some other random crap broke, so I haven't found the time May 02 22:18:10 philipp64: yup. probably about the same as most places. May 02 22:22:07 i'm working on building an openwrt-powered network for a homeless camp ... doing a several block hop with nanobridges, an ER-X gateway and a couple meraki mr24's as dumb APs May 02 22:34:18 jow: do you know a better script to get the fingerprint of a usign key? I came up with this May 02 22:34:18 tail -n 1 key-build.pub | base64 -d | dd skip=2 count=8 bs=1 2>/dev/null | hexdump -ve '1/1 "%.2x"' May 02 23:01:37 hi. i'm built master from github but the output bin was fetching packages from the snapshots tree. so I switched to 19.07.2 branch thinking it would solve the problem, but my device is still identifying as a snapshot build and opkg pointing to the snapshots url. How would I build 19.07.2 from source in such a way that the resultant bin is compatible with the packages for that version May 02 23:01:37 so that changes to the snapshots tree aren't going to break future updates for the device in question ? May 02 23:01:48 *i'm = I May 02 23:04:06 the device is a tenda A6 (similar to the supported A5-V11, minimal changes, basically just one LED that is different) May 02 23:04:38 sorry for my ignorance but AFAIK theres no way to use the image builder unless the device is officially supported,right ? so by building from source I am on the right track ? May 02 23:05:37 any pointers would be great, thanks in advance May 02 23:08:00 ok wait I think I'm just blind, I can use the release at https://github.com/openwrt/openwrt/archive/v19.07.2.tar.gz , correct ? May 02 23:20:18 aparcar[m]: I think “hexdump” allows you to specify counts and offsets, so the “dd” might be unnecessary. May 02 23:20:52 philipp64: thank you May 02 23:28:49 of course, it’s a minor optimization… May 02 23:38:22 philipp64: biggest problem is the missing base64 problem, it can be solved via an awk script but that seems a bit messy May 02 23:40:20 “missing’ how? you need to be able to do this without anything from feeds/packages? May 02 23:41:05 doesn’t busybox provide base64? May 02 23:41:49 it is supposed to work on the device as a replacement for usign. base64 is not installed by default and I don't plan to add it for that purpose. So I need to sneak around that. May 02 23:42:04 The build root doesn't matter so much in that case May 03 00:20:13 sorry, then what’s the problem you’re running into? if base64 is provided by busybox and busybox is always present… May 03 00:21:28 base64 is not provided by busybox, at least not in the openwrt standard configuration May 03 00:41:25 is it possible to make for a specific target without running make menuconfig ? May 03 00:42:29 so that it just grabs the defined packages from DEVICE_PACKAGES May 03 00:42:56 something like: make TARGET="RAMIPS/rt305x/A6" ? May 03 00:43:57 rather than editing the .config by hand it should auto-generate it and start the build May 03 00:46:07 docs suggest 'make kernel_menuconfig CONFIG_TARGET=subtarget' , how would one adapt this ? May 03 00:53:41 seeding .config with the basics (target/ device) and running "make defconfig" May 03 00:57:47 pkgadd: thanks. so it needs to be seeded? can't specify the target through commandline ? May 03 00:58:03 not that I'd know, but... **** ENDING LOGGING AT Sun May 03 02:59:57 2020