**** BEGIN LOGGING AT Tue Jul 09 02:59:57 2019 Jul 09 03:53:10 mangix: will be good if any Python packages will be reviewed by jefferyto and commodo Jul 09 07:10:23 morning Jul 09 07:21:33 morning Jul 09 07:51:05 Is there any issue with putting an out of tree version of a kernel module (cake) into openwrt/packages/kernel ? Jul 09 07:52:27 having dropped the out of tree package in base 'cos cake is now in kernel 4.19+ it turns out not quite all popular targets are on k4.19 yet Jul 09 07:52:57 so is putting an out of tree version in packages a reasonable interim solution? Jul 09 08:04:46 would target/generic/backports-4.9 not be better ? Jul 09 08:05:26 i mean this is an upstream river that you wanna backport, right ? Jul 09 08:14:44 I suppose effectively yes, though I don't *want* to have to backport it to 4.14 seeing as though we're moving forward from that. Jul 09 08:27:08 leave it as if, incentivices people to bump their favorite target to 4.19 ? Jul 09 08:33:11 stintel: agreed. Jul 09 09:05:28 ldir: I would inore kernel 4.9 and just make cake depend on kenrel 4.14 and later Jul 09 09:05:38 fir 4.14 I would add a patch to target/generic/backports-4.14 Jul 09 09:05:41 *for Jul 09 09:05:59 Certainly not doing it for 4.9 Jul 09 09:07:08 I'm really unconvinced that 4.14 is worth it either. owrt 19.07 has a fully functioning cake, master has a fully functioning cake for targets on 4.19 + Jul 09 09:07:34 ldir: oe just make it depend on kernel 4.19+ Jul 09 09:07:36 *or Jul 09 09:24:34 * ldir goes and prints an address label for tapper - and borrows a guide dog to do the parcel wrapping Jul 09 09:47:37 ynezz: ping Jul 09 09:52:13 mangix: Shouldn't we remove open2300? It's no longer maintained and it was updated more than 6 years ago? Jul 09 09:55:44 Pepe: Not that I'm develper, but does it sitll work, compile, etc good? Jul 09 09:59:05 IMHO as long as status quo keps working, do nothing... That will eventually fail obviously, then nuke it :) Jul 09 09:59:08 Would be mine 2 cents Jul 09 10:07:28 It's not in Ubuntu nor Debian feeds. If there will be bug, security issue who will fix it? Jul 09 10:08:20 being there doesn't mean security fixes happen either. Jul 09 10:08:34 Pepe: mark it as DEFAULT:=n, wait for someone to complain that it's missing from snapshots, then forc^Wconvince them to become the new maintainer ;) Jul 09 10:10:10 jow: Hauke do you mind adding this repo as an official openwrt/openwrt-docker.git? It creates docker images for sdk, imagebuilder and rootfs. Usefull for CI or people not running linux but want to develop OpenWrt. Jul 09 10:10:10 https://github.com/aparcar/openwrt-docker Jul 09 10:14:59 KanjiMonster: Great idea! :) Thanks Jul 09 10:17:22 aparcar[m]: you've got a PR ;-) Jul 09 10:19:05 ldir got your mail thanks dude. Jul 09 10:19:40 Tapper: no problem - I borrowed a guide dog to help wrap it back up :-) Jul 09 10:19:50 haha Jul 09 10:19:58 Is it ok? Jul 09 10:20:05 the dog I meen! Jul 09 10:20:53 yeah the dog's great. The blind man I 'borrowed' it off less so... he put up a hell of a fight! Jul 09 10:21:52 ldir: where? Jul 09 10:22:07 aparcar[m]: if it gets renamed to openwrt/docker.git I'm fine with it Jul 09 10:22:23 aparcar[m]: other consideration is that it'll overlap somewhat with buildbot.git/docker Jul 09 10:22:29 no blind men or guide dogs were harmed during the wrapping of tapper's linksys router - in case anyone is worried :-) Jul 09 10:23:03 I got band from my co-op for swinging mine around my head. When asked what I was doing I said I am trying to find the god dam baked beans! Jul 09 10:23:24 my very own spaniel came to 'assist' by putting its nose where it wasn't helpful :-) Jul 09 10:23:46 lol you have got love nosy doggys Jul 09 10:23:46 lol Jul 09 10:25:32 ldir: can't seem to find anything Jul 09 10:25:43 let me check Jul 09 10:26:16 https://github.com/aparcar/openwrt-docker/pull/3 Jul 09 10:34:23 Can you believe this crap? BIOS ROM Size Limitations Almost Derail AMD's Zen2 Backwards Compatibility https://www.techpowerup.com/257201/bios-rom-size-limitations-almost-derail-amds-zen2-backwards-compatibility-promise Jul 09 10:35:29 TLDR they could not fit the AGESA ComboAM4 1.0.0.3a in to 16 meg flash chips Jul 09 10:35:56 That is mad when you think what OpenWrt can fit on a 8 meg one! Jul 09 10:38:23 and on a 4MB one when you squish it hard enough :P Jul 09 10:39:19 blame UEFI, and also the new AGESA might as well be cursed, it might be the old style agesa, dirtwrapped with UEFI interfaces Jul 09 10:43:50 ldir: ty Jul 09 10:44:02 np Jul 09 10:48:00 jow: can you grant me acces to the openwrtorg docker? Jul 09 11:12:54 aparcar[m]: should be done Jul 09 11:17:46 ldir: can you reproduce this https://pastebin.com/3pea1YNU ? I'm not sure what I'm doing wrong here, I already rm -rf'ed build_dir/target* and staging_dir/target* Jul 09 11:20:42 xdarklight: do you have you own local package of cake somewhere in your tree ? Jul 09 11:21:18 ldir: find says no: https://pastebin.com/bLEdSaL3 Jul 09 11:22:39 how about cd package; find . -name "*cake*" Jul 09 11:23:16 ldir: nothing. even with -iname "*cake*" Jul 09 11:23:30 aparcar[m]: pong Jul 09 11:24:33 erm! oh :-( let me see if I can repro off master 'cos this was all just working 2 days ago. Jul 09 11:28:36 xdarklight: what CAKE things are enabled in .config ? Jul 09 11:29:33 ldir: https://pastebin.com/Z0QPAZgW Jul 09 11:30:06 (I haven't changed these in a *long* time, so let me know if I have to update one of these) Jul 09 11:42:09 ynezz: I'm having lunch, update you in 30 minutes Jul 09 11:49:07 bon apetit! Jul 09 11:59:40 Hi.. I started as an embedded dev in a company. We have new device and i want to add new files, patches and dts into system. But i cannot find any way to do it automatically like packages. I want to fetch files and put them their location. Is there a way to do it? I have searching days but nothing found useful. Jul 09 12:01:42 Co_ops: which target? Jul 09 12:05:59 Device mips 7621 based. I can put files locations by hand but i wanna do it automatically somehow Jul 09 12:06:10 xdarklight: Just completely rebuilt my x86 target - no error. So am like you, confused. Jul 09 12:08:34 Co_ops: just put the dts there once and leave it Jul 09 12:08:48 what else could you need? Jul 09 12:10:33 I know i know no need to do what i want but my boss want to this done. I ask him do it with a script but he refuse. I would like to know what he is thinking. Jul 09 12:13:42 xdarklight: any customisation to package/kernel/linux/modules/netsupport.mk ? Jul 09 12:17:51 ldir: latest commit on that file is 5dc7d63d0a, so: no :/ Jul 09 12:18:00 xdarklight: 'cos sched-core shouldn't have SCH_CAKE included in it. Jul 09 12:22:24 xdarklight: trash bin/ packages & targets - and rebuild again? I'm grabbing at straws really. Jul 09 12:22:39 it doesn't make sense. Jul 09 12:26:00 oh hold on....just checking something! Jul 09 12:30:01 xdarklight: right, yes I can - I thought I had kmod_sched enabled like you... but I don't. So, for the moment, if you can, de-select kmod-sched Jul 09 12:33:10 This is the same trap that act_ctinfo fell into - will try to fix. Jul 09 12:45:43 aparcar[m]: I thought the openwrt/docker repository was supposed to be hosted on git.o.o and only mirrored to github Jul 09 12:45:53 imho it is not acceptable have any non-mirror repos solely on gh Jul 09 12:46:16 but maybe II can mirror it from gh to g.o.o later Jul 09 12:52:32 Hauke: FYI just wanted to test your `mac80211: Update to version 5.2-rc7` http://paste.ubuntu.com/p/72mxGzxdzV/ Jul 09 12:54:12 ynezz: are there any further issues with PR 1359? Jul 09 12:55:47 xdarklight: fix pushed to master Jul 09 12:57:32 jow: I have no git.o.o commit access Jul 09 12:58:24 feel free to create a g.o.o/docker.git and mirror it to gh, I can't do anything of that Jul 09 12:59:12 ynezz: thanks Jul 09 12:59:33 I only tested this on my BT home hub 5a (ath9k + ath10k) yet Jul 09 13:01:34 Hauke: probably some patchwork/pwclient hiccup, merged it from your staging tree Jul 09 13:01:58 ynezz: ah yes I had to remove some parts because it was too big for the mailling list Jul 09 13:02:33 The patch is 950KB and there is a max of 512KB on the mailling list Jul 09 13:02:48 jow: do we have "unlimited" circleci minutes? Or should we ask to get them? Building all these docker containers and PRs consumes some time Jul 09 13:02:56 ouch! Jul 09 13:03:24 ldir: ouch? Jul 09 13:04:01 ah bad timing - ouch re Hauke 950KB patch Jul 09 13:06:36 ldir: removing 20K lines Jul 09 13:07:56 Hauke: wow! Jul 09 13:08:09 Hauke: where? Jul 09 13:08:27 in the mac80211 update to 5.2-rc7 Jul 09 13:08:44 many patches backporting stuff can be removed Jul 09 13:11:14 https://cloud.docker.com/repository/registry-1.docker.io/openwrtorg/sdk/tags the first docker images appear Jul 09 13:33:56 aparcar[m]: uhm, I thought you're the circleci expert :) Jul 09 13:34:06 I think ted hess might know more Jul 09 13:34:36 jow: wait what? Everything works as expected from my side... Jul 09 13:35:20 I was referring to your question "do we have "unlimited" circleci minutes?" Jul 09 13:38:04 interesting ci usage stats https://paste.debian.net/1090927/ Jul 09 13:39:03 117h 54m 50s is slighly more then the free 250h, looks like somethings works in our favor Jul 09 13:39:24 It's unlimited for opensource :) Jul 09 13:42:34 Ycarus: sweet, thanks! Jul 09 13:47:03 117h << 250h ? Jul 09 13:48:58 jow: 250 minutes, sorry Jul 09 13:54:45 jow: do you know if it is possible to call this function directly for an image? https://github.com/openwrt/openwrt/blob/master/include/image.mk#L198 Jul 09 13:57:17 not without hacks Jul 09 14:00:39 stintel: ping Jul 09 14:04:07 blogic: pong Jul 09 14:06:13 jow: these hacks would be cool to CI dt changes Jul 09 14:07:17 stintel: hi Jul 09 14:08:13 jow: actually calling define Device/Build/kernel should be enough for testing Jul 09 14:50:04 how can I force reload the contents of target/linux/x86/64/profiles? Jul 09 14:50:11 changes don't appear in the menuconfig Jul 09 14:51:55 mwarning: ^ this should be the easiest solution for the problem Jul 09 14:56:22 touch target/linux/x86/64/Makefile Jul 09 15:41:35 jow: thanks, I changed 000-Generic.mk but it's still not reloaded Jul 09 16:12:18 aparcar[m]: what is your plan? :P Jul 09 16:14:17 mwarning: changing the x86/64 profile name from Generic to something like "Generic x86/64", so it is possible to distinguish in an overview Jul 09 16:15:31 ok Jul 09 16:21:29 hey, i need to temporarily disable a PKG_CONFIG option, but want to keep it in the makefile. Is there a way to disable a option and throw a error and fail a build, if its found via a old imported .config? Jul 09 16:27:40 I guess doing a check in define Build/Configure and exit 1 with a error message should work? Is this the way to handle this or is there some macro for this case? like the @broken for packages? Jul 09 16:39:58 in Build/Compile Jul 09 16:40:48 \t $(if $(CONFIG_BROKEN_SYMBOL),@(echo CONFIG_BROKEN_SYMBOL is broken; exit1)) Jul 09 16:40:59 *exit 1 Jul 09 16:51:54 Pepe: ehhhhhhh. I don't really know who uses it or if there is a proper replacement. Jul 09 16:52:36 Probably need download statistics to figure that out Jul 09 16:52:42 Hauke: you had a mac80211 update ? Jul 09 16:53:14 oh thia reminds me. I need to see if the new MMC driver works on ramips Jul 09 16:57:03 works. cool. Jul 09 17:11:48 blogic: yes Jul 09 17:12:15 blogic: https://git.openwrt.org/?p=openwrt/staging/hauke.git;a=shortlog;h=refs/heads/mac80211-5.2 Jul 09 17:15:28 blogic: I would like to get this into OpenWrt soon, and want to do an upgrade to 5.3-rcX in about 4 weeks Jul 09 17:15:47 if we are lucky 5.3 will be an LTS kernel Jul 09 17:47:42 @jow thanks will try this Jul 09 17:59:28 Hauke: thanks Jul 09 17:59:56 O my days I have 6 kids in my front room dancing around like nutters to this. lol Jul 09 17:59:57 https://www.youtube.com/watch?v=bN8vY8qkgVs Jul 09 18:00:34 mangix: It was just a question, but what proposed KanjiMonster doesn't sound bad at all for me. Jul 09 18:14:58 Pepe: for reference, this is still in the tree even though it is abandoned and alternatives exist: https://github.com/openwrt/packages/tree/master/multimedia/crtmpserver Jul 09 18:15:09 I really think nobody uses it Jul 09 18:16:35 blogic: if you find any regressions in the mac80211 update please report or fix them Jul 09 18:17:06 are you ieee8021ax updates accepted for 5.3? Jul 09 18:29:03 is anyone else getting similar to "WARNING: previous mirror push of repo 'openwrt/staging/ldir' to host 'git-02' failed, status is:' Jul 09 18:29:25 yes, its harmless Jul 09 18:30:11 is there a way to persuade it to stop? Jul 09 18:30:59 we need to disable the mirroring Jul 09 18:31:50 ok, I'll carry on in ignore mode Jul 09 18:35:06 should be fixed now Jul 09 18:35:14 I disabled the mirroring to git-02 Jul 09 18:37:11 confirmed - thanks Jul 09 18:53:28 jow: can I create a firewall3 repo on gh? Jul 09 18:55:53 but please disable issues and PRs there Jul 09 18:58:48 sure! Jul 09 18:59:06 jow: oh wait, why no PRs? Jul 09 18:59:34 because it is tedious to deal with them when the repo on github is not the canonical source Jul 09 19:00:11 also, wasn't the agreement in hamburg to switch to a hosted gitlab solutioln? Why do we transfer even more stuff to github? Jul 09 19:01:06 jow: fair point, I'll work on gitlab Jul 09 19:02:04 jow: me an lynxis evaluated the mail support of gitlab and it is currently not git send mail compatible (wtf). therefore I though to postpone it for a little while Jul 09 19:25:35 someoner: KanjiMonster: SSB_IDHIGH allows reading ID of a given core, if scan_read32(bus, 0, SSB_IDHIGH) gives you 0, it means there is problem with accessing a core Jul 09 19:25:38 ssb assumes that core 0 can be always accessed Jul 09 19:25:40 in 99% cases core 0 is a ChipCommon which contains SoC version data Jul 09 19:25:41 i'm wondering if this device could have SSB bus mapped at different address than SSB_ENUM_BASE (0x18000000U), see: Jul 09 19:25:42 https://elixir.bootlin.com/linux/latest/source/arch/mips/bcm47xx/setup.c#L111 Jul 09 19:26:11 rmilecki: serial at 0x18000300 is working fine Jul 09 19:26:23 ok, that's weird :/ Jul 09 19:26:44 maybe that SoC doesn Jul 09 19:26:55 doesn't have SPROM/EEPROM/whatever Jul 09 19:27:00 and can't read list of cores Jul 09 19:27:16 maybe some weird neither ssb nor bcma Jul 09 19:27:19 but that doesn't explain why IDHIGH read is empty Jul 09 19:27:21 uh Jul 09 19:27:29 there was something between... Jul 09 19:27:32 some another bus Jul 09 19:27:40 UI? UA? Jul 09 19:27:44 I can't remember Jul 09 19:29:35 ub(us), but that was bcm63xx Jul 09 19:30:10 SOCI_SB / SOCI_AI / SOCI_NAI Jul 09 19:30:19 SB is ssb, AI is bcma I think Jul 09 19:30:21 maybe it's NAI SoC? Jul 09 19:30:32 ah, right, there was also SOCI_UBUS Jul 09 19:31:02 a look into the gpl whatever-ball should help Jul 09 19:31:02 I think it has a ssb bus. I took quick a look at the ecos sources from netgear: It appears to use the HAL package for a "bcm953710". This package contains files which begin with sb. These file contain have comments which mention "SiliconBackplane-based Broadcom chips" Jul 09 19:33:17 sih->socitype = (w & CID_TYPE_MASK) >> CID_TYPE_SHIFT; <-- SOCI_UBUS is 2 and SOCI_NAI is 3, while we get 0 Jul 09 19:33:24 so it's something else... Jul 09 19:33:30 maybe we don't do some required setup Jul 09 20:04:00 aparcar[m]: FYI https://gitlab.freedesktop.org/patchwork-fdo/patchwork-fdo ("The missing link between mailing lists and CI.") Jul 09 20:07:12 aparcar[m]: patchwork has REST API, so I believe, that cooking some kind of bridge between patchwork and GitLab might be possible Jul 09 20:10:03 ynezz: sweet! Fixing my bike right now but will check it out tonight Jul 09 20:17:44 aparcar[m]: found https://github.com/ruscur/snowpatch as well, but its written in Rust :) Jul 09 21:50:50 Interesting insight from the ecos sources: For my device the code isn't scanning the ssb bus, but assigning the cores statically (https://github.com/fvollmer/ecos-2.0-GS108Tv2/blob/master/packages/hal/mips/bcm953710/v2_0/src/sbutils.c#L210). For other devices there is a sb_scan function (https://github.com/fvollmer/ecos-2.0-GS108Tv2/blob/master/pac Jul 09 21:50:50 kages/hal/mips/bcm947xx/v2_0/src/sbutils.c#L354). Jul 09 21:55:20 and there is the static map https://github.com/fvollmer/ecos-2.0-GS108Tv2/blob/master/packages/hal/mips/bcm953710/v2_0/src/sbutils.c#L118 **** ENDING LOGGING AT Wed Jul 10 03:01:22 2019