**** BEGIN LOGGING AT Wed Aug 21 02:59:58 2019 Aug 21 04:18:38 KanjiMonster: -nopad has been in include/image.mk since 2006 Aug 21 05:54:09 Hauke: can you backport this to 19.07? https://github.com/openwrt/openwrt/commit/ebbec2fdc6cec4119fb4fefc3080146aa00bc07d Aug 21 07:36:24 mangix: ok will do so Aug 21 07:45:24 finally got around to unwrap my xiaomai mir3g and to install openwrt on it Aug 21 07:45:33 was surprised to see that it ships with a modified luci Aug 21 08:07:54 russell--: even earlier than that Aug 21 08:08:31 it was added with the helpful commit message "add some changes from whiterussian to head" in 2005 Aug 21 09:34:54 lol Aug 21 09:37:47 nopad is probably desirable in nor contexts Aug 21 09:53:11 russell--: yupp, since we don't align the start of the squashfs on nor, aligning the end is completely ineffective. At best it does nothing, and at worst pushes the start of the jffs2 overlay into the next erase block Aug 21 09:54:59 what determines the squashfs block size? Aug 21 09:55:38 seems like /lib/upgrade/nand.sh is the right place to do the padding? Aug 21 10:08:50 nbd: is it required to backport your upstream minstrel fixes? thanks Aug 21 10:13:51 i will backport them to master soon. already have queued up commits in my staging tree Aug 21 10:30:44 russell--: https://elixir.bootlin.com/linux/latest/source/fs/squashfs/squashfs_fs.h#L21 Aug 21 10:31:32 yeah, so 4k seems the safe bet Aug 21 10:31:42 patching nand.sh won't fix existing setups/sysupgrades, only the sysupgrade after that Aug 21 10:32:18 i think the best place is probably include/image.mk, since the padding infrastructure is already there. Aug 21 10:32:50 mangix: Mister_X: Just ordered an APU4C4 with 2x WLE600VX, 2 pairs of antennas (2.4 and 5), case and so on Aug 21 10:33:21 and I'll probably need to drill an additional SMA holes in the standard case Aug 21 10:34:23 also planning to put an LTE card with 2x2MIMO antenna as a backup connection Aug 21 11:11:31 looks like mr24 uses scripts/sysupgrade-tar.sh Aug 21 11:21:28 KanjiMonster: how about something like this? https://dpaste.de/BJ4Z Aug 21 11:28:30 hang on, what's going on here? IMAGE/sysupgrade.bin/squashfs := append-rootfs | pad-rootfs | sysupgrade-tar rootfs=$$$$@ | append-metadata (that's for e.g. Device/openmesh_a62) Aug 21 11:38:30 russell--: something like this, though I'm not 100% sure if 'k' is understood by every version of dd (like otheros's), and maybe a test if root is a squashfs by its magic Aug 21 12:02:15 KanjiMonster: or maybe something like this: https://dpaste.de/jcHs Aug 21 12:05:27 nope Aug 21 12:05:43 the landing place isn't right Aug 21 12:12:32 this works a bit better: https://dpaste.de/NUoQ Aug 21 13:11:05 that breaks the ext4 image for mr24, not sure why, will sleep on it Aug 21 14:18:27 hmm, firewalld on centos leaves a lot to be desired Aug 21 14:18:37 or I am too stupid to understand it Aug 21 15:13:53 Hi all, what is the recommended method for enabling a kernel configuration? I know I shouldn't just modify .config in the kernel build_dir as it will get overwritten. Am I supposed to use kernel_menuconfig? Aug 21 15:29:05 Neighbor11111112: yes, kernel_menuconfig is the right way (which will then update target/linux//config-* with the new value(s)) Aug 21 15:33:40 awesome, thanks! Aug 21 16:24:51 jow: using the USB3 port wrecks 2.4ghz band on the mir3g from what I've heard. Aug 21 16:26:19 mangix: yeah, the chinese original firmware also offers an option to disable usb3 with a description text that usb3 might affect 2.4 ghz Aug 21 16:26:43 not a big issue to me though, I have no use for the usb anyway. just wanted some mt76 thing to test luci on Aug 21 16:32:13 That sounds acceptable Aug 21 17:04:01 jow: any chance this can get reviewed? https://patchwork.ozlabs.org/patch/1133914/ Aug 21 18:27:26 hey guys Aug 21 18:27:36 jow: is there anything amiss with luci on mt76? Aug 21 18:49:54 can someone point me at the code that parses the piped strings in the IMAGE/foo := bar | baz | boing-boing-wobble-wobble ? Aug 21 18:53:46 hang on, i think i found it Aug 21 18:56:32 Is anyone else having issues with connecting to hidden networks? I firstly noticed this by seeing there are no probe requests being sent by a openwrt station. Then I set the AP to hide it's ssid and sure enough, my openwrt station doesn't connect. Scan ssid is set to 1, as expected. Using latest git. Aug 21 19:02:20 95a945a7cb83b830b241931651cae2d2d44c8140 Aug 21 19:47:57 Turns out it is because of DFS. But still it should send probe reqs if it detects APs on the frequency, right? Aug 21 19:51:28 I don't know well the techdetails DFS involves, but if device thinks DFS restriction is in effect on said channel then I'd say device won't touch it, but this is only mine toughtprocess Aug 21 20:48:18 so, CONFIG_TARGET_PER_DEVICE_ROOTFS=y works only with CONFIG_TARGET_MULTI_PROFILE=y ?? how do i build all devices from one target together with these two options then without specifying each device ? Aug 21 20:49:57 maybe ... sed -i "s|# CONFIG_TARGET_ath79_generic_DEVICE|CONFIG_TARGET_ath79_generic_DEVICE|" .config ? Aug 21 21:02:19 rotanid: CONFIG_TARGET_ALL_PROFILES Aug 21 21:35:47 KanjiMonster: thanks, works! Aug 22 02:57:09 what's the way to disable ext4 builds for meraki mr24? for one thing, it's building a 128MiB ext4 rootfs, but the entire flash is only 32MiB. Also, I'm not sure puting an ext4 filesystem on NAND even makes sense, even if the size problem was worked out. so, it seems most reasonable to just disable ext4 for the device. But how? **** ENDING LOGGING AT Thu Aug 22 02:59:56 2019