**** BEGIN LOGGING AT Mon Aug 12 02:59:58 2019 Aug 12 08:06:34 mangix: Thanks for the details. I'm seeing similar things on ar71xx, specifically gl.mifi device Aug 12 08:07:28 mangix: the 2nd ethernet port is 1Gbit capable, but connecting it to a gigabit switch shows the glmifi reporting 1Gbit/s connection, while the remote switch indicates the link is actually 100Mbit/s (and I trust the switch a lot more here) Aug 12 10:37:36 kernel bumps pushed to staging Aug 12 16:21:45 where are the mappings of KEY_xxxx names to the etc/rc.button filenames kept? there's KEY_RESTART in ~most of the dts files, but there's reboot and reset handlers, no restart handler? Aug 12 16:24:48 karlp: https://github.com/openwrt/openwrt/blob/master/package/kernel/gpio-button-hotplug/src/gpio-button-hotplug.c#L75 Aug 12 16:26:39 thanks. so it's ~just restart that's not consistent :) hooray for legacy compat I guess :) Aug 12 16:28:11 karlp: probably because the button is usually labeld Aug 12 16:28:32 "reset", not "restart", but there is no KEY_RESET (I assume) Aug 12 16:29:27 jupp, there isn't one Aug 12 16:31:13 the button has two actions anyway, factory reset and reboot Aug 12 16:31:41 neither of which match the KEY name in the dts file when you're trying to track things down :) Aug 12 16:32:40 KEY_FACTORY_RESET_OR_REBOOT Aug 12 16:33:03 I doubt it would be upstream material :p Aug 12 16:33:39 about as easy as dropping the _BUTTON suffix from KEY_WPS_BUTTON I'd imagine :) Aug 12 16:33:57 we could move our script to restart though.... (but it's just churn for the most part) Aug 12 16:36:37 ynezz: karlp: the KEY_* defines are part of uapi, so there will never be any renames, at most new aliases Aug 12 18:05:22 nbd: if you are around, trying to get feedback on https://patchwork.ozlabs.org/patch/1143363/ or https://patchwork.ozlabs.org/patch/1137710/ Aug 12 18:07:39 champtar: took a quick look. is your main intent to fix processing of EAP frames that are addressed to a multicast address? Aug 12 18:08:11 xback: in my case, traffic stops Aug 12 18:09:10 nbd: I want to be able to let the EAP frame pass through the bridge when using group_fwd_mask Aug 12 18:09:33 only for group addressed EAP frames, or also for unicast ones? Aug 12 18:10:01 nbd: both Aug 12 18:10:27 it's for https://github.com/nccgroup/phantap Aug 12 18:12:17 in that case, it doesn't make sense to me to configure it via group_fwd_mask Aug 12 18:12:22 and it would be better to have a separate sysctl Aug 12 18:12:51 nbd: first, do you think your hack is still needed ? Aug 12 18:13:00 i don't know for sure Aug 12 18:13:13 i don't remember the exact conditions that triggered this issue Aug 12 18:13:19 too many years ago Aug 12 18:13:36 yup Aug 12 18:14:13 In that case https://patchwork.ozlabs.org/patch/1143363/ can be a first step Aug 12 18:14:51 yes Aug 12 18:15:15 or I can rewrite it as enable_eap_hack, and add a hotplug file Aug 12 18:16:12 in any case having a sysctl would make it easier to test Aug 12 18:16:59 (and would unlock me) Aug 12 18:29:40 let's stick with disable for now Aug 12 18:31:22 nbd: you want me to resend the same patch as non RFC ? Aug 12 18:33:29 yes Aug 12 18:33:52 btw. val ? true : false can be written as just 'val' when assigning a bool Aug 12 18:34:12 or if you want to make conversion explicit, you could also use !!val Aug 12 18:37:07 Didn't knew about the implicit conversion when using bool, and I'm not the only one as it's a copy paste from the kernel code :P Aug 12 18:37:17 I'll resend with !!val Aug 12 18:58:28 Do y'all experience sysupgrade not working without physically disconnecting usb drives from the device? Aug 12 20:26:33 ynezz: could you please have a look at this? https://lists.infradead.org/pipermail/openwrt-devel/2019-August/018439.html Aug 12 20:52:30 !! is just a way to make a bool that's assured of being 0/1, rather than just 0/non-zero **** ENDING LOGGING AT Tue Aug 13 02:59:56 2019