**** BEGIN LOGGING AT Mon Jul 29 03:00:13 2019 Jul 29 03:38:54 https://en.wikipedia.org/wiki/Flag_day_(computing) Jul 29 04:36:41 mangix: i think it supports multiple streams (not sure thought) Jul 29 04:36:56 but I think in the past it didn't support more than 20 MHz channel bandwidth Jul 29 04:37:53 mangix: https://wireless.wiki.kernel.org/en/users/drivers/brcm80211#to_be_done_for_softmac_driver Jul 29 07:10:30 Good morning people! Jul 29 07:41:51 ynezz: ping Jul 29 07:50:03 ynezz: there's a bug in commit https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=acf2b6c8881b432530bd98fa86753bf6a3546ff7 Jul 29 07:50:23 Mikrotik boards are not detected properly anymore Jul 29 07:50:30 herei s the before/after on a rb912 board Jul 29 07:50:31 https://pastebin.com/BPNpeYUG Jul 29 07:50:39 reverting that commit fixes it Jul 29 07:55:24 aah .. found the issue .. in ar71xx.sh match is stated as "MikroTik" .. while it should be "Mikrotik" Jul 29 08:02:15 he's back! Unfortunately Greg didn't follow my instructions and did at least 3 kernel bumps without you being around. Jul 29 09:18:37 Hauke: hi, I resumed work on the mikrotik device from battlemesh (hap ac²), but I think I'm stuck at the same place you were Jul 29 09:18:59 boot freezes half-way with SMP, it works fine with "nosmp" on the cmdline Jul 29 09:19:42 I just sent an email to the list with details, if somebody has an idea Jul 29 12:52:28 I get a lot of "/bin/sh: 3: mkhash: not found" messages building opnwrt. Is this a missing dependency? Jul 29 12:54:23 mwarning: wild guess, but try `make tools/mkhash/{clean,install}` Jul 29 12:54:45 or something like that. I keep forgetting these make targets Jul 29 12:55:46 (I don't have mkhash command on my distro and I do have it in staging_dir/host/bin) Jul 29 12:57:30 ok that's not a valid target. apparently it's built in include/prereq-build.mk Jul 29 12:59:31 hm Jul 29 13:00:02 I will try to rebuild from the ground up. Jul 29 13:00:19 make dirclean Jul 29 13:09:55 it seems to work now Jul 29 13:10:14 not sure about the cause really Jul 29 13:10:42 Maybe it is because I checked the files out inside the docker container instead of outside. Jul 29 13:35:48 fwiw, i rm'd staging_dir/host/bin/mkhash and it didn't automagically get rebuilt by the build system Jul 29 13:56:38 mwarning: hm, hexa- had similar issues with mkhash as far as i rmember Jul 29 13:57:06 ok Jul 29 13:57:43 might be a permissions issue Jul 29 13:58:42 anyway, I have a few pull requests floating around that are ready to be merged. How do I find out who to ping? ("device_title" PRs on github.com) Jul 29 14:14:26 * russell-- learning about remake Jul 29 15:27:35 xback: thanks for the fix! Jul 29 15:28:54 jow: would it be possible to either update or kickout that tictex-02 build slave? Build dependency: Please install Python >= 3.5 Jul 29 15:31:17 it seems like 19.07 is somehow in releasable state (all images/packages in green), so maybe about time for 19.07.0-rc1 ? :P Jul 29 15:33:52 will there still be no ath79 images in 19.07? Jul 29 15:37:44 what creates the ipkg-install (PKG_INSTALL_DIR) directory? I had assumed that OpenWRT did that but it seems like we need to do that manually? Jul 29 15:39:09 DonkeyHotei: nope Jul 29 15:39:26 DonkeyHotei: thank you for reminding me, to disable them Jul 29 15:39:37 yw Jul 29 16:12:55 kernel bumps pushed to staging Jul 29 16:20:56 Where can I read about package DEPENDS PROVIDES etc ? Jul 29 16:21:37 i searched: wiki package depends site:openwrt.org Jul 29 16:21:49 may be this> https://openwrt.org/docs/guide-developer/packages Jul 29 16:23:26 broski: thanks Jul 29 16:24:54 just having a nightmare with packages & dependencies Jul 29 16:32:06 i believe you, share what you got? Jul 29 16:46:21 ynezz: will mail the admin Jul 29 16:53:05 Some will already understand context. I have a kernel module that only appears in kernel >4.18. I have an out of tree build of that same module. Jul 29 16:53:36 the out of tree build is for <= 4.18 ? Jul 29 16:54:01 ideally I'd like to be able to use OOT for any kernel version. Jul 29 16:54:52 I'd try DEPENDS:=@(LINUX_4_9||LINUX_4_14) Jul 29 16:55:20 So the approach was to have a package called say 'OOT-kmod-foo' whilst certain kernel has 'kmod-foo'. The OOT package PROVIDES=kmod-foo Jul 29 16:58:07 jow: Is that in the in-tree or OOT package? Jul 29 16:59:00 the oot one Jul 29 17:03:52 Can I do anything about the in-tree version? Remember it doesn't exist in < K4.19 - so the build system (apparently) builds an empty package..which is better than failing Jul 29 17:04:42 I would have thought DEPENDS:=@!(LINUX_4_9||LINUX_4_14) if that's legal. Jul 29 17:05:05 or maybe it has to be @(!LINUX_4_9&&!LINUX_4_14) Jul 29 17:05:13 let me grep for examples Jul 29 17:05:45 none found Jul 29 17:05:57 so uhm, iirc anything after "@" is passed to Kconfig literally Jul 29 17:06:29 so both @!(LINUX_4_9||LINUX_4_14) or @(!LINUX_4_9&&!LINUX_4_14) should be legal Jul 29 17:07:04 you can even skip the parens as long as thhere's no spaces in the expression Jul 29 17:08:12 ok, so I'm back where I'd got to... which leads to recursive dependency hell Jul 29 17:08:29 interesting, what recursive deps do you get? Jul 29 17:09:20 https://paste.ubuntu.com/p/tvgGnyJFnR/ Jul 29 17:10:36 can you pastebin the relevant sections of "./scripts/package-metadata.pl config tmp/.packageinfo " as well? Jul 29 17:10:45 https://paste.ubuntu.com/p/cTr4YvwTPg/ Jul 29 17:10:48 or maybe the complete file if it does not exceed pastebin limits Jul 29 17:10:57 you're too fast Jul 29 17:11:00 that's not it - hold on Jul 29 17:14:25 * ldir was desperately trying to remember how to copy text into the paste buffer from macos command line - pbcopy Jul 29 17:14:33 https://paste.ubuntu.com/p/qJTXwGr5r3/ Jul 29 17:15:01 Waiting for paste.ubuntu.com... Jul 29 17:15:07 ah now it rendered. Jul 29 17:17:35 ldir: can you provide the package definition of kmod-sched-cake-oot as well? Jul 29 17:20:17 https://paste.ubuntu.com/p/RFNNVP9Rqf/ Jul 29 17:22:57 ldir: seems you triggered a bug in the kconfig generator, will look into it Jul 29 17:23:03 It's something to do with the provides in kmod-sched-cake-oot Jul 29 17:24:16 * ldir finding corner cases so you don have to(tm) Jul 29 17:24:28 don't Jul 29 17:24:32 * broski chuckles Jul 29 17:34:52 ldir: I can only offer this dirty hack: https://pastebin.com/3uibjT1K Jul 29 17:35:09 a real fix would entail that entire part of the code as its a hideous mess Jul 29 17:35:19 *would entail rewriting that part Jul 29 17:35:26 I can take a stab at this tomorrow Jul 29 17:37:59 the mconf_depends() procedure sort of generates kconfig syntax from our annotations, then it uses string checks and fuzzy regex matches to skip recursive expressions but it fails to catch more complex instances Jul 29 17:38:41 so a proper fix would refactor that function to use some kind of AST internally which can be properly traversed to skip circular depends or false references Jul 29 17:42:49 I'm not in any screaming rush. If anything I'm more relieved that I wasn't doing something illegal, the frustration factor has gone to 1/10 the level it was before. Jul 29 18:40:56 ldir: your iptables CONNMARK patch needs an extra header Jul 29 18:41:10 #include Jul 29 19:25:19 i bumped openwrt-networking package to development branch to evaluate and support new features. besides the faillogs/ path, what other resources are avail for seeing if a specific package fails to build. Jul 29 19:45:36 broski: make -j1 Jul 29 19:46:34 or, here's my one-liner: for i in $(find logs -name compile.txt) ; do tail $i | grep -H --label=$i 'Error ' ; done | grep -v ignored | awk -F: '{ print $1 }' | sort -u Jul 29 19:52:26 i can work with this.. i poked around ah, have you seen https://github.com/openwrt/docker Jul 29 19:53:50 if crawling the webserver for compile.txt is OK, id rather work with a 1 or 2 liner =) Jul 29 19:54:07 thanks m8 Jul 29 19:57:30 mangix: ty - I think I fixed it. Am assuming this is for ffs() definition. Not sure I understand how it worked without. Jul 29 20:09:50 ldir: yes. it's required for ffs Jul 29 20:10:47 I still don't understand how it managed to build without the include Jul 29 20:11:14 that's because implicit function declarations are a GCC extension Jul 29 20:11:40 pass -Werror=implicit-function-declaration and it will fail Jul 29 20:12:31 I think it also fails if you pass -std=cXX Jul 29 20:12:54 so iptables must be built without those warnings. Jul 29 20:13:37 well, -Werror=implicit-function-declaration is what I pass in my local builds to catch erroras Jul 29 20:13:41 *errors Jul 29 20:14:19 ok so this wasn't found by a buildbot Jul 29 20:14:26 no. Jul 29 20:14:53 implicit function declarations can also cause bad code compilation as GCC assumes implicit functions to return int Jul 29 20:14:58 which is not always true Jul 29 20:15:33 I get why it's a bad idea - I didn't understand why I didn't see the error when building locally. Jul 29 20:18:20 anyway, the warning breaks way too many packages. Can't enable by default Jul 29 20:18:58 especially ncurses ones **** BEGIN LOGGING AT Mon Jul 29 21:59:51 2019 Jul 29 22:14:44 mwarning: https://github.com/openwrt-routing/packages/pull/476 could you hit the green button please Jul 29 22:14:52 sbyx is not active anymore Jul 29 22:15:07 somehow i am not an admin of that repo otherwise i;d do it myself Jul 29 22:20:01 I see blogic is as well pedant about indentation.. ynezz told me that I should be more relaxed about that and he was wondering where are the rules for that... Looking at this commit: https://github.com/openwrt/openwrt/commit/ef3f868da0d78adf2490a762ff567cf5b636c213 ! :P Jul 29 22:23:41 ynezz: ^^ please dont Jul 29 22:23:54 orthographic correctness is really important Jul 29 22:24:09 ynezz: we have a code style that we try to adhere to Jul 29 22:24:44 that actually makes me shiver Jul 29 22:26:27 *twitch* Jul 30 00:49:44 https://github.com/openwrt/openwrt/commit/2b074654b0f259518aa56e0975ca8e26c0c12bc9 Jul 30 00:50:32 How come ath79 was marked as source only for 19.07? Jul 30 00:51:36 While I know it's a replacement for ar71xx, and 19.07 will be the last release to support ar71xx, wouldn't it make sense to have both enabled? Jul 30 00:51:55 (Unless i'm missing something?) Jul 30 01:32:47 hi Jul 30 01:33:18 rmilecki: you there? Jul 30 02:07:13 foxtrot: twice the automated builds and the space they take up, as well as twice the confusion Jul 30 02:07:59 people can still build ath79 from the 19.07 branch and tags Jul 30 02:21:50 Is this good to merge? Jul 30 02:21:52 https://www.mail-archive.com/openwrt-devel@lists.openwrt.org/msg47483.html Jul 30 02:23:04 https://patchwork.ozlabs.org/patch/1137192/ Jul 30 02:25:39 looks like it's missing the aliases section in the dts Jul 30 02:26:35 DonkeyHotei: thanks, I will try and relay to the supplier **** ENDING LOGGING AT Tue Jul 30 03:01:12 2019