**** BEGIN LOGGING AT Wed Jan 23 02:59:57 2019 Jan 23 04:16:32 whoa. more mt7621 stuff made it to stable than i thought Jan 23 04:33:08 i'm hosting a server locally that is on a separate VLAN from my LAN Jan 23 04:33:39 i can access this site perfectly from outside my network, but get a ERR_CONNECTION_REFUSED when trying to access internally Jan 23 04:33:46 any idea how i can fix this? Jan 23 04:34:07 #openwrt Jan 23 04:34:15 not development Jan 23 04:34:17 :D Jan 23 04:34:33 yeah, crickets in there at the moment, i can try back tomorrow Jan 23 04:34:34 sorry Jan 23 04:37:33 _lore_: compiling kernel now. my first build was missing the driver... Jan 23 09:32:15 jwh: pong Jan 23 09:32:30 hello sir Jan 23 09:32:33 cns3xxx ma Jan 23 09:32:34 man* Jan 23 09:32:37 good mornin' Jan 23 09:34:05 hey Jan 23 09:34:31 I have a horrible cns3xxx device and tplink safeloader, just wondered if you'd played with one Jan 23 09:34:54 is it the tp-link vpn gateway by any chance? Jan 23 09:34:58 ye Jan 23 09:35:00 6020 Jan 23 09:35:10 I've taken a look at it after some forum posts Jan 23 09:35:30 it's a lot of work to support that one Jan 23 09:35:37 :( Jan 23 09:35:48 I can't even figure out what magic firmware structure it wants Jan 23 09:36:11 or how all the hardware is interconnected .. Jan 23 09:36:22 well, at least thats achievable Jan 23 09:36:24 maybe Jan 23 09:36:32 but no good if nothing can be booted Jan 23 09:37:50 it seems like a lot of fun to support it. but my main issue is that it's way out of scope at work Jan 23 09:38:08 aw Jan 23 09:38:21 I picked one up by accident yesterday, got it off some girl on facebook - was hoping it was the v2 Jan 23 09:38:31 which is uh, ipq Jan 23 09:38:54 not really got any use for it otherwise, software sucks Jan 23 09:39:59 <\x> malwar3hun73r: a forward rule will do Jan 23 09:48:09 build #101 of ath79/tiny is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/ath79%2Ftiny/builds/101 Jan 23 10:26:37 mkresin: ping Jan 23 10:28:49 mkresin: Hauke: I'm currently bumping the kernels, and following patch fails to apply: https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=target/linux/lantiq/patches-4.14/0400-Revert-MIPS-smp-mt-Use-CPU-interrupt-controller-IPI-.patch;h=7c2deb190e144f914f4a8717ed5fb11971de79a3;hb=76037756d0c97eb2cbb9f227b3d3b11705b5abe0 Jan 23 10:29:08 this is caused due to upstream accepted commit from Hauke: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.14.95&id=a06d94d98d2417510ab2dab39f633d6557b0bb68 Jan 23 10:29:56 looking at the code, It would seem that the patch from mkresin can be dropped in favour of the IRQ fix from Hauke. Is this correct according to you? Jan 23 10:52:20 xback: looks like Haukes fix is the way to go. strange that I missed the patch completely Jan 23 10:53:04 xback: if you have a tree for me, I can runtest the haukes fix vs. the revert Jan 23 10:56:43 mkresin: currently still refreshing the targets. I currently removed your revert-patch in this approach. ill update my staging tree within 30 minutes or so. will trigger you Jan 23 10:56:55 mkresin: thanks for your fast reply :) Jan 23 12:18:03 mkresin: the kernel bumps have been pushed to my staging Jan 23 12:21:36 build #750 of layerscape/armv8_64b is complete: Failure [failed images] Build details are at http://phase1.builds.lede-project.org/builders/layerscape%2Farmv8_64b/builds/750 blamelist: HsiuWen Yen Jan 23 12:37:48 Hi, is there a way to reconstruct the path of an ipk file that was created package/myhelloworld/compile? Would like to automate the step to upload a built package to my router to simplify testing the package on my router Jan 23 12:38:09 *that was created by make … Jan 23 12:43:25 I wanted to write a script that uploads me the ipk to the router and then instructs opkg to install it. Simply hard coding the path of the ipk would be lame since I want to build multiple packages for multiple targets Jan 23 12:45:36 why do you need ipk? can't you just upload the binary? Jan 23 12:47:30 anyway, it's always faster to build/test package outside of openwrt, just using openwrt toolchain Jan 23 12:48:30 for example https://github.com/ynezz/impinj-indy-itk/blob/master/Makefile#L24 and https://github.com/ynezz/impinj-indy-itk/blob/master/cmake/openwrt-toolchain-imx6.cmake Jan 23 12:50:32 you can then add upload target to cmake https://paste.ubuntu.com/p/xWdyQwGmSQ/ Jan 23 12:51:12 so you can just do `make all upload` and it will just compile, upload, test... Jan 23 12:59:14 ah I see, so your recommendation here is to include a rule in the (c)makefile for the package/software to include a rule that builds the package for the platform in question and upload it afterwards Jan 23 13:03:15 no, I use packages for final deployment as it's slow, for development where I need faster iteration during build/deploy/test cycle, I just build binaries outside of openwrt and upload them to target and test it Jan 23 13:05:56 that might work as well, I need to figure out how I can cross compile that piece of software with gccgo now though Jan 23 13:07:54 as a start, you can use the same commands used in the compile step `package/myhelloworld/compile V=s` Jan 23 13:08:12 make package/compile V=s Jan 23 13:14:30 I don't exactly think this is an easy task either https://github.com/openwrt/packages/blob/master/lang/golang/golang-package.mk#L218 Jan 23 13:22:24 well, it's just a shell script embedded in the makefile Jan 23 13:30:38 mhh, not really sure anymore if it is really the way to go for me, I would like to test the binary with an erx and an ar71xx router, that means I need to differentiate between these two toolchains somehow … also duplicating the shell script doesnt sound right to me Jan 23 13:33:25 xback: your staging tree works like a charm. SMP up and running with 4.14.95 on lantiq/xrx200 Jan 23 13:37:17 mkresin: what conkrete hardware do you use? Jan 23 13:37:38 mkresin: i have a 3370 and this one works fine Jan 23 13:38:00 but i needed to extract the dsl-firmware from a avm image to get DSL+Vectoring to work Jan 23 13:40:12 rubberduck: bt homehub 5a Jan 23 13:41:40 mkresin: is it possible that you re-work my fritzbox 7320/30 Patch for second usb so that it can be commited? Jan 23 13:42:10 rubberduck: -ENOTIME Jan 23 13:42:30 okay Jan 23 13:42:44 this are only 8 additional lines in the .dts file Jan 23 13:43:32 so what's the problem? :) Jan 23 13:44:23 i'm not able to create a suitable github-environment so that i cannot provide the patch in the form as wanted by the project Jan 23 13:46:44 can you pastebin those 8 lines? Jan 23 13:47:55 https://slim.webnmail.de/7320_2ndUSB.patch Jan 23 13:48:43 same as paste: https://sky.webnmail.de/privatebin/?062b2e68b3b62a20#EnDeFMhT7uxoIZ1wP/WmW52ftbLRG+Xda7Du5p+U4zk= Jan 23 13:49:15 and you're the original author? can you give me your 'Signed-off-by: real name ' ? Jan 23 13:51:01 i sent this on the openwrt-devel-mailinglist already Jan 23 13:51:07 my name is Robert Resch Jan 23 13:52:00 I can't find it in patchwork https://patchwork.ozlabs.org/project/openwrt/list/ Jan 23 13:52:27 ok, this one? http://lists.infradead.org/pipermail/openwrt-devel/2019-January/015461.html Jan 23 13:53:18 yes - this one Jan 23 13:54:05 mkresin: thanks for testing! mind if I add your tested-by tag? Jan 23 13:57:09 ynezz: in the meantime i managed to test the patch on both 7330 and 7320 and works on both as expected Jan 23 13:58:03 xback: feel free to do so Jan 23 13:58:58 xback: so uh, reckon this er6020 is a lost cause? Jan 23 14:02:04 xback: should i test, too on a avm 3370 box? Jan 23 14:02:18 jwh: no. but I will not provide support for it myself :) if someone else can develop the patches, I'm happy to review them Jan 23 14:02:49 hmmm, wonder who knows enough about them and is willing to do it Jan 23 14:02:57 rubberduck: which arch/target is that? Jan 23 14:03:04 xrx200, too Jan 23 14:03:10 rubberduck: is it ok https://github.com/ynezz/openwrt/commit/bde90e03bf95a4c44d9619f794acdb792c3a159e ? Jan 23 14:03:23 rubberduck: feel free to do so Jan 23 14:04:01 [Wed 2019-01-16 05:52:49 AM PST] has anyone seen the following in dmesg on lantiq devices? Jan 23 14:04:01 [Wed 2019-01-16 05:52:50 AM PST] [709739.855248] ltq_etop 1e180000.etop eth0: tx ring full Jan 23 14:04:01 [Wed 2019-01-16 06:18:41 AM PST] it seems to happen under load and breaks all connectivity Jan 23 14:04:03 rubberduck: feel free to send me en e-mail with the results and your Tested-by tag Jan 23 14:04:53 via the developer-ml? Jan 23 14:06:20 rubberduck: to me directly is sufficient. I'm not going to send the kernel bump patches to the ML Jan 23 14:07:30 the changes were pretty minor, excluding the IRQ stuff for lantiq Jan 23 15:16:58 Hi all, I'm trying to figure out what (and when) the boot functions are called in /etc/init.d/? I thought it was procd that would do that but greping the source doesn't reveal anything Jan 23 15:17:01 build #396 of ipq40xx/generic is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/ipq40xx%2Fgeneric/builds/396 Jan 23 15:48:08 build #1203 of x86/legacy is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/x86%2Flegacy/builds/1203 Jan 23 15:48:46 grep harder? Jan 23 15:50:06 xback: how can i update my buildroot for building kernel 4.14.94? Jan 23 15:50:35 master 'only' has 4.14.94 (not 95 - typo) - this is working fine Jan 23 15:51:04 OpenWrt SNAPSHOT r9104-fe7d965ea9 / LuCI Master (git-19.022.57088-61b9f7a) is working fine on my avm 3370 Jan 23 15:55:25 rubberduck: clone my staging tree and use that one as buildroot: https://git.openwrt.org/?p=openwrt/staging/xback.git;a=summary Jan 23 15:55:43 dont know how Jan 23 15:55:54 okay Jan 23 15:56:11 rubberduck: git clone git://git.openwrt.org/openwrt/staging/xback.git Jan 23 15:56:41 rubberduck: cd xback make menuconfig select target exit & save start the build Jan 23 15:58:51 okay - will take a while. Jan 23 15:58:58 hope you are still here then Jan 23 15:59:32 probably not :) leaving the office within 30 minutes or so Jan 23 15:59:56 i hope i have your mail-address? You may mail me so that i have it... Jan 23 16:00:18 koen.vandeputte@ncentric.com Jan 23 16:14:54 compile is running - but this will take a while. Jan 23 16:33:23 Neighbor11111113: /etc/inittab contains ::sysinit:/etc/init.d/rcS S boot Jan 23 16:34:52 Neighbor11111113: procd parses /etc/inittab and maps the sysinit line to runrc() here: https://git.openwrt.org/?p=project/procd.git;a=blob;f=inittab.c#l214 Jan 23 16:35:09 hey jow: need some quick help with a new kernel module, having trouble understanding where the kernel config defines come from (you have a sec) ? Jan 23 16:35:31 Neighbor11111113: runrc() in turn calls the builtin rcS() with the arguments from inittab ("S", "boot") Jan 23 16:37:41 Neighbor11111113: rcS() is declared here: https://git.openwrt.org/?p=project/procd.git;a=blob;f=rcS.c#l181, it scans all /etc/rc.d/S* and adds it to the runqueue along with the param ("boot") Jan 23 16:38:09 Neighbor11111113: the runqueue will then process all entries (all found /etc/rc.d/S*) and execute them one-by-one using the "boot" param Jan 23 16:39:13 andy2244: sure Jan 23 16:40:02 in my kernel module makefile i had to define these "EXTRA_CFLAGS+= -DCONFIG_KEYS=y Jan 23 16:40:02 EXTRA_CFLAGS+= -DCONFIG_ASSOCIATIVE_ARRAY=y" Jan 23 16:40:02 Yet i already changed those in the generic/config-4.14, so something is wrong. How do kernel modules pick up those config defines normally? Jan 23 16:42:43 andy2244: usually there's a process in place which merges target/linux/generic/config-*, target/linux/$arch/config-* and the various KCONFIG:=... symbols from define KernelPackage/foo sections Jan 23 16:43:12 andy2244: a "make target/linux/prepare" should take care of that Jan 23 16:43:39 build #294 of at91/sama5d2 is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/at91%2Fsama5d2/builds/294 Jan 23 16:45:32 mhh its not working in my case, since i still got errors that clearly point to a missing CONFIG_ASSOCIATIVE_ARRAY=y yet this is actually defined already for all targets.. How are those defines normally handed over to the Build/Compile ? Jan 23 16:46:44 include/kernel.mk does -include $(LINUX_DIR)/.config Jan 23 16:47:06 which usually maps to something like build_dir/target-*/linux-*/linux-*/.config Jan 23 16:48:10 the include imports .config as makefile, effectively setting all the CONFIG_whatever=... assignments into make variables Jan 23 16:48:52 however, in the case of kbuild, its not the openwrt build system but kbuild itself which sources .config Jan 23 16:51:32 mhh ok, guess if i cant find the problem i will just do the PR and have you take a closer look, why i need those extra defines Jan 23 16:52:30 fair enough Jan 23 16:52:45 i'm also not sure, should i add a cifs server as kernel module option or via static build option? It either sits in packages than or openwrt base? Jan 23 16:53:21 not sure what you mean with this question Jan 23 16:53:55 is it about packaging some cifs.ko ? Jan 23 16:54:32 then it should be a kmod addition to either package/kernel/linux/modules/fs.mk or package/kernel/linux/modules/netsupport.mk, not sure which of both is more appropriate Jan 23 16:55:16 its about a new cifs server inside the kernel, i can try the exfat-nofuse route inside packages or like the other kernel modules inside the openwrt base? Jan 23 16:56:46 it has a kconfig, so can be build either way as static module or .ko file Jan 23 16:56:49 kmod in base I'd say Jan 23 16:57:52 Thanks jow Jan 23 16:58:19 ok, will try this, still in early stages, but maybe it can be a simpler/faster replacement for samba3/4. Jan 23 17:17:19 ah seems the wireguard package in base is a good example for me, has a module and userspace-tools similar to the package i try to build Jan 23 17:29:08 jow I'm trying to understand wireless setup and how it interacts with netifd. Your explanation at the bottom of this wiki is very informative (https://openwrt.org/docs/techref/netifd). And so is your post here (https://forum.openwrt.org/t/netifd-interaction-with-various-driver/12522) about /sbin/wifi Jan 23 17:29:41 What confuses me is the realtionship between netifd wireless handlers and the /sbin/wifi command. They seem to have overlapping jobs. Jan 23 17:29:52 could you help please when you get a chance? Jan 23 17:35:39 yes, they do overlap Jan 23 17:36:04 the shell stuff explained in the forum is for the old legacy code, the chat log explains the new netifd wireless integration Jan 23 17:50:46 ahh I see Jan 23 17:50:54 I must be playing around on old legacy code Jan 23 17:51:13 I notice that wifi still uses /lib/wifi/mac80211.sh for detecting the config file Jan 23 17:51:18 I assume the rest is done through the handler? Jan 23 18:05:49 andy2244: what is this cifs server you speak of? Jan 23 18:07:40 mkresin: xback yes the upstream commit should replace the revert Jan 23 18:08:01 I only tested this on kernel 5.0, but can do a testrun with 4.14 today Jan 23 18:18:25 Hauke: already tested Jan 23 18:33:37 build #279 of gemini/generic is complete: Failure [failed images] Build details are at http://phase1.builds.lede-project.org/builders/gemini%2Fgeneric/builds/279 blamelist: Deng Qingfang , HsiuWen Yen , Jason A. Donenfeld Jan 23 18:36:02 DonkeyHotei: thanks for the hint about kmod-sdhci-mt7620. it works until I add gpio_export{} to the dts. I'm guessing because that prevents mkt_sd from doing the initialisation that allows exporting the gpio via sysfs to work. to get this actually fixed... what's the usual workflow for doing kernel development? is there an alternative to injecting .patch files during development? Jan 23 18:39:29 mnemoc: https://openwrt.org/docs/guide-developer/patches#adding_or_editing_kernel_patches Jan 23 18:40:00 DonkeyHotei: thanks Jan 23 18:40:05 yw Jan 23 19:45:45 xback: AVM 3370 works with 4.14.95 Jan 23 20:01:14 rubberduck: is it ok https://github.com/ynezz/openwrt/commit/bde90e03bf95a4c44d9619f794acdb792c3a159e ? Jan 23 20:02:54 ynezz: yes - that is fine Jan 23 20:05:31 ynezz: thank you very much! Jan 23 20:09:35 i finally found out what to do if putty terminal and make menuconfig are shitty: export TERM=putty Jan 23 20:09:50 doing make menuconfig after that everything will be fine. Jan 23 21:07:52 jow have you seen this? Jan 23 21:07:53 https://neonsea.uk/blog/2018/12/26/firewall-includes.html Jan 23 21:08:10 I think it's old news but just want to check! Jan 23 21:09:31 Tapper: doesn't affect openwrt Jan 23 21:09:39 and I'd argue that its not cve worthy either Jan 23 21:09:56 a config that is designed to include script can include scripts Jan 23 21:11:16 mnemoc: there's upstream work on that driver. There might be a useful commit to backport. But then again there's also a different GPIO driver there... Jan 23 21:36:03 OK jow thanks. I just saw it on twitter and thorght I would check. Jan 23 22:28:13 mangix: I'm new to openwrt, how can I find the git source of the kernel I'm building? Jan 23 22:28:57 mangix: or it's quilt-only? Jan 23 22:31:23 upstream means mainline in this case or vendor? Jan 23 22:36:17 mainline Jan 23 22:37:10 mnemoc: kernels come from kernel.org , which are patched using patches from target/linux/generic and then further patched by target/w/e target you're building. Jan 23 22:39:52 mangix: clear, thanks a lot. I'll try to backport upstream's gpio driver to your 4.14.94 Jan 23 22:42:06 it's highly reassuring you aren't using vendor's hacked kernels as base :) Jan 23 23:42:19 mnemoc: lol given the amount of platforms OpenWrt supports, that would be a horrible idea **** ENDING LOGGING AT Thu Jan 24 02:59:56 2019