**** BEGIN LOGGING AT Wed Oct 09 03:01:40 2019 Oct 09 03:54:50 jow can I do some back porting so 19.07 becomes reproducible? https://rebuild.aparcar.org/19.07-SNAPSHOT/ath79/generic/ Oct 09 05:47:52 aparcar[m], Hauke: I've split it into 4 separate commits, hope it's ok now https://git.openwrt.org/?p=openwrt/staging/ynezz.git;a=shortlog;h=refs/heads/staging :) Oct 09 05:54:35 Why would I be getting this error? satisfy_dependencies_for: Cannot satisfy the following dependencies for kmod-: kernel (= 4.14.131-1-082221559376832b2b09c3999edfa01c) Oct 09 05:55:12 I'm using 18.06.4 -- fairly heavily patched, but nothing I would expect to cause this Oct 09 06:03:57 hmm, I deleted my build_dir/target-mipsel_24kc_musl/linux-ramips_mt7620 directory and ran the build again with success. I did disable my "non-reproducible builds: use multiple threads on squashfs compression" option, but I'm not so sure that was the cause. I seem to have had this problem on and off... Oct 09 06:19:57 dansan: There's a kernel hash generated from kernel config and kmod packages checks if it matches the on-board kernel. You'll get that error unless your kernel config is exactly the same as the one for building kmods. Oct 09 06:26:50 ynezz: lgtm Oct 09 06:51:45 gch981213: thanks for the hint. where does it read the kernel .configs for the kmods then? This is currently all in-tree (in-tree kernel, etc.). It makes me wonder if there's some quirk somewhere in the pipeline of generating the kernel .config... Oct 09 06:52:15 I have modified kconfig.pl, but mostly to add documentation -- another change set I need to work on pushing through Oct 09 06:53:20 plus I renamed it to kmungeconfig.pl, because "kconfig.pl" is not a good name -- it's misleading (it's not kconfig) and it only describes the target of the program, not the operation. Oct 09 06:54:45 gch981213: oh! This could be due to a flaw in one of my additions to config/Config-kernel.in that results in one .config, but once we run make oldconfig it changes it perhaps. Oct 09 06:57:10 ynezz: please check the KCFLAGS patch too, it even works with gcc7 Oct 09 06:57:53 And is it possible to compile the entire buildroot with clang? Oct 09 06:58:01 gch981213: See, again this is poorly named. The dependency should probably be renamed from "kernel" to "kernel.config" -- this would have made it more obvious what the problem was, as I was attempting to figure out where it created the hash from in the first palce' Oct 09 07:19:22 dansan: improvements are always welcome :) Oct 09 07:35:28 ynezz: thanks for the merge Oct 09 07:41:17 nbd: good morning. Mind if I backport your mac80211 patch to 19.07? I tested it extensively using that branch Oct 09 08:41:13 what is up with mtk_soc_eth? Oct 09 08:41:14 [33452.017179] mtk_soc_eth 1e100000.ethernet eth0: port 3 link down Oct 09 08:41:15 [33454.942269] mtk_soc_eth 1e100000.ethernet eth0: port 3 link up Oct 09 08:41:24 lots of those in dmesg Oct 09 08:44:17 ynezz: do you mind back porting the initramfs patch from last week to 19.07? Oct 09 08:44:18 https://rebuild.aparcar.org/19.07-SNAPSHOT/ath79/generic/ Oct 09 08:44:20 xback: fine with me. please run a quick test with the version that i pushed to master Oct 09 08:44:34 because i made some minor tweaks Oct 09 08:45:07 nbd: noticed it, currently compiling it. will run-test in the field this afternoon. thanks Oct 09 08:46:21 hmmm, for IIO sensors on the I2C bus, I need to assign the driver to an I2C bus and address like this: `echo bme680 0x76 > /sys/bus/i2c/devices/i2c-1/new_device` Oct 09 08:46:44 does anyone know if there is something similar for sensors on the SPI bus? Oct 09 08:58:07 spi has cs, no addresses Oct 09 08:59:16 https://en.wikipedia.org/wiki/Chip_select Oct 09 08:59:49 * russell-- recalls using device tree overlays for such things Oct 09 08:59:59 russell--: yeah, I've done that for bme280 Oct 09 09:00:14 either that doesn't work for bme680 or I wired it incorrectly Oct 09 09:00:32 scope/logic probe to the rescue :) Oct 09 09:00:42 -ENOHAVE Oct 09 09:00:50 omg, get! Oct 09 09:01:03 $10 on ebay Oct 09 09:01:19 sigrok/pulseview is amazingly useful Oct 09 09:04:18 e.g. https://www.ebay.com/itm/USB-Logic-Analyzer-24MHZ-8-Channel-12C-UART-SPI-IIC-CAN-1Wire-Debug-For-ARM-FPGA/162980270582 Oct 09 09:05:03 any good tutorials on how to use those things. coz I have no clue Oct 09 09:05:53 youtube should be full of tutorials Oct 09 09:06:24 that's why I added "good" in my question ;) Oct 09 09:07:15 i totally wimped out and just used the AppImage downloaded from sigrok.org Oct 09 09:07:52 i don't remember anything specific Oct 09 09:21:59 https://www.youtube.com/watch?v=7x4h7Zq2NNU Oct 09 09:41:06 mangix: I know, so do what we _should_ have done at the time, and said arc can't be in until they work in musl. it's been what, two years of "we're working on it, honest, promise, we just need to get started like this" Oct 09 09:51:00 How does ubusd delete an object when the client process does not exist? Oct 09 09:57:50 After ubus_add_object, it seems that it has not been deleted. Is it something I missed? Oct 09 10:10:48 [ 4079.045168] bme680_spi spi0.0: Wrong chip ID, got ff expected 61 Oct 09 10:10:54 getting at least something now :) Oct 09 12:15:50 ah sweet it's working \o/ Oct 09 12:32:10 is it normal to get very low speeds with ath10k drivers on QCA9886? (around 130mbps) while with stock fw i can get up to 580mbps **** BEGIN LOGGING AT Wed Oct 09 15:03:15 2019 Oct 09 16:23:08 Hi people Oct 09 16:23:26 Is there any news on OpenWrt 19.xx? Oct 09 16:45:09 is there any way to look at the content of the flow offload table? Oct 09 16:45:44 similar to what "conntrack -L" shows for non-offloaded connections Oct 09 16:47:01 I'm trying to debug an issue where after a reboot, a UDP flow is no longer NAT-ed (packets for this flow come out of the WAN port with the LAN address) Oct 09 16:53:03 hmm, this is possibly a flow that sent a packet *before* the openwrt router had a chance to grab an IP address on the WAN side Oct 09 16:58:29 it may be a more general issue, I'll try without offloading Oct 09 16:59:23 zorun: yes, sounds like a very old problem Oct 09 16:59:44 the firewall usually flushes conntrack on startup to avoid that, I guess with flow offloading this workaround is not working anymore Oct 09 17:04:01 jow: ok, thanks Oct 09 17:04:16 jow: from my tests, it only seems to happen with hw flow offloading (on mt7621) Oct 09 17:04:33 with regular flow offloading, I can't reproduce Oct 09 17:24:45 https://bugs.openwrt.org/index.php?do=details&task_id=2541 Oct 09 18:30:07 Hauke: any objections https://github.com/openwrt/openwrt/pull/2429 (mac80211: fix scan when operating on DFS channels in ETSI domains)? **** BEGIN LOGGING AT Thu Oct 10 02:24:45 2019 **** ENDING LOGGING AT Thu Oct 10 02:59:57 2019