**** BEGIN LOGGING AT Fri Jul 12 02:59:57 2019
Jul 12 04:55:45 <jow> ynezz: ping
Jul 12 07:22:51 <xback> no idea what everybody has against ar71xx which works nicely :-)  I love the tons of ar71xx devices laying around here with the clear readable board files instead of the ugly syntax dts stuff :-)
Jul 12 07:55:35 <ldir> my sarcasm'o'meter has gone full scale
Jul 12 08:56:10 <kristrev> Since I have no experience with the platform, I will pretende that I am asking for a friend :) It seems the answer is yes, but is IPQ4019 fairly well supported by OpenWRT/upstream?
Jul 12 08:57:02 <kristrev> I find some boards, but I am not able to find any comments on how well they are working (but that might be my google-foo)
Jul 12 09:29:02 <Hauke> jow: Why is the build system not complaining about a missing  libjson-c.so.4: dependecy in block-mount
Jul 12 09:29:17 <Hauke> see https://bugs.openwrt.org/index.php?do=details&task_id=2373
Jul 12 09:29:39 <jow> Hauke: because the buildsystem dependency checker follows transient deps
Jul 12 09:29:41 <Hauke> I would just add the dependency, but I would like to understand this
Jul 12 09:29:51 <jow> Hauke: I do not understand it either
Jul 12 09:31:04 <jow> is block-mount part of ubox ?
Jul 12 09:31:31 <jow> nvm
Jul 12 09:31:42 <Hauke> it depends on ubox
Jul 12 09:35:21 <Hauke> ok so it gets the dependecy though that one
Jul 12 09:38:01 <jow> I think it happens like this: block-mount depends on libubox(20170601) which is provided by package/libs/libubox which provides jshn which depends on libjson-c4
Jul 12 09:38:27 <jow> at least thats my working theory that the buildroot library provider checker operates per source package, not per binary package
Jul 12 09:38:53 <jow> but I do have to take a closer look to be sure
Jul 12 09:40:04 <jow> hm no, it operates per .ipk
Jul 12 09:43:14 <jow> Hauke: this is the dependency tree (each dep is only mentioned the first time it appears): https://pastebin.com/7SmVnWRk
Jul 12 09:43:50 <jow> Hauke: libjson-c is satisfied through libjson-c2 << libblobmsg-json << ubusd << ubox << block-mount
Jul 12 09:44:03 <jow> *c4
Jul 12 09:44:38 <jow> Hauke: so my first response was right, this is caused by the build system dependency checker considering transient deps
Jul 12 09:44:48 <jow> we need to fix that
Jul 12 09:45:10 <jow> to now allow transient library depends anymore
Jul 12 09:45:13 <jow> *not
Jul 12 09:51:42 <Hauke> so the problem is that the build system didn't notce that block-mount shoud be rebuild in case of an API change of libjson-c4?
Jul 12 09:51:52 <Hauke> *notice
Jul 12 09:53:25 <Hauke> hmm it depends on the new libjson-c4
Jul 12 09:54:03 <Hauke> jow: the old version is was ABI_VERSION=2
Jul 12 10:14:25 <Hauke> aparcar[m]: whats the url of the fw generator
Jul 12 10:16:52 <jow> Hauke: I think this is a trunk user opkg-upgrading packages
Jul 12 10:17:01 <jow> but - idk
Jul 12 10:17:24 <jow> too little detail in the bug report. I think we already spent more effort thinking about it than the submitter writing it
Jul 12 10:27:32 <Hauke> ok
Jul 12 10:34:18 <andy2244> @nbd quick question why do you bind to loopback by default for your samba3 package? I was wondering what the default should be for samba4/cifsd?
Jul 12 10:35:10 <jow> andy2244: the question is, why not?
Jul 12 10:35:38 <jow> I think it solves some use case - smbd starting when no other configured interfaces are available (up, ip configured)
Jul 12 10:35:42 <andy2244> Was just curious what benefit or problem binding to loopback solves for samba?
Jul 12 10:35:58 <jow> it also allows to use tools locally against the server (testing things with smbclient etc.)
Jul 12 10:36:22 <jow> but thats only relevant for scripting things I suppose
Jul 12 10:37:04 <andy2244> In the latest cifsd version, they create a separate process per interface binding, so its one extra process running for loopback if i make this a default, which seemed weird if i dont actually know why i should bind to loopback.
Jul 12 10:37:56 <andy2244> so you still recommend binding to LAN + loopback by default for samba4/cifsd?
Jul 12 10:38:19 <jow> doesn't hurt I suppose
Jul 12 10:38:34 <andy2244> or should i just bind to LAN and leave extra to manual UCI configs
Jul 12 10:39:01 <andy2244> can you think of any reasonable scenario where loopback binding is needed? I had a hard time to come up with one
Jul 12 10:39:42 <karlp> it's possibly a hangover from not having hotpluyg actions for interfaces in the old days?
Jul 12 10:47:32 <aparcar[m]> Hauke: chef.libremesh.org?
Jul 12 11:49:37 <rmilecki> anyone with mikrotik rb750gr3/rbm11g/rbm33g here?
Jul 12 11:50:51 <rmilecki> or maybe someone with one of: rb-750-r2 rb-750p-pbr2 rb-750up-r2 rb-911-2hn rb-911-5hn rb-931-2nd rb-941-2nd rb-951ui-2nd rb-952ui-5ac2nd rb-962uigs-5hact2hnt rb-lhg-5nd rb-map-2nd rb-mapl-2nd rb-sxt-2nd-r3 rb-wap-2nd rb-wapg-5hact2hnd rb-wapr-2nd
Jul 12 11:51:11 <rmilecki> i need to test sysupgrade change
Jul 12 12:00:12 <aparcar[m]> rmilecki: I got a rbm33g. How can I help?
Jul 12 12:00:24 <rmilecki> aparcar[m]: hooray! i'm just commiting my change locally
Jul 12 12:00:30 <rmilecki> aparcar[m]: i need 5-10 minutes
Jul 12 12:00:57 <aparcar[m]> I need 2-3 hours before being back at my device, however let me know what I should try
Jul 12 12:10:14 <rmilecki> aparcar[m]: could you test a patch I just sent? https://patchwork.ozlabs.org/patch/1131345/ [PATCH 3/3] ramips: get rid of platform_pre_upgrade()
Jul 12 12:10:50 <rmilecki> aparcar[m]: you need to apply that patch, build firmware, install it & THEN try the sysupgrade to install yet another change
Jul 12 12:10:54 <rmilecki> *another firmware
Jul 12 12:11:38 <rmilecki> aparcar[m]: or... it may be easier actually - you may apply my changes directly to /lib/upgrade/platform.sh on your device
Jul 12 12:11:44 <aparcar[m]> So install your modified firmware twice?
Jul 12 12:11:45 <rmilecki> and then try sysupgrade
Jul 12 12:12:02 <rmilecki> aparcar[m]: if you want to build firmware - yes - install twice
Jul 12 12:12:03 <aparcar[m]> And only 3/3 or all patches?
Jul 12 12:12:09 <rmilecki> aparcar[m]: only 3/3 is enough
Jul 12 12:12:30 <rmilecki> aparcar[m]: 1/3 and 2/3 are for other targets
Jul 12 12:12:40 <aparcar[m]> rmilecki: will do as soon as I'm home
Jul 12 12:13:30 <rmilecki> aparcar[m]: if I were you, i'd simply do "vim /lib/upgrade/platform.sh" carefully, i'd modify it the same way as in the patch and try tried to sysupgrade to some whatever firmware version
Jul 12 12:13:32 <rmilecki> thanks aparcar[m]
Jul 12 12:17:28 <Neighbor11111117> Does every 802.11 frame require an acknowledgement. I've assume it does but I feel like I read somewhere that it is not always the case. Does anyone know if there are exceptions to the rule?
Jul 12 12:17:47 <KanjiMonster> rmilecki: you should be able to replace the "mtd erase firmware" call with setting MTD_ARGS to "-e firmware" in do_upgrade
Jul 12 12:18:16 <rmilecki> Neighbor11111117: i don't think beacon frames get acked
Jul 12 12:19:15 <Neighbor11111117> thanks rmilecki
Jul 12 12:19:47 <Neighbor11111117> so all l3 traffic encapsulated in 802.11 frames requires acks?
Jul 12 12:24:01 <KanjiMonster> rmilecki: also why not merge the switch blocks and call default_do_upgrade "$ARGV" at the end in the microtik cases? similar to the tp-link one
Jul 12 15:07:39 <aparcar[m]> rmilecki: ping
Jul 12 15:25:25 <neonknight77> hello, does anyone know how I can ignore warnings from image-builder regarding PACKAGE SIZE MISMATCH, unable to install?
Jul 12 15:27:18 <jow> try  rm packages/Packages*
Jul 12 16:41:18 <aparcar[m]> anyone experience with flashing a mikrotik router board 33g? It's not posible via the web interface, serial console won't fire up and tftp transfers the file, but does not boot
Jul 12 19:27:15 <rmilecki> KanjiMonster: i can do that, no opinion on this
Jul 12 19:27:31 <rmilecki> KanjiMonster: thanks for pointing MTD_ARGS
Jul 12 19:27:34 <rmilecki> aparcar[m]: pong
Jul 12 19:30:05 <KanjiMonster> rmilecki: I thought "hey, fully erasing the firmware partition sounds like something more than target might need, let's check the code how hard it would be to implement it in a genric way", and then I found the MTD_ARGS - and when I grepped for it it was ramips itself that used it (and even the only user)
Jul 12 19:30:26 <rmilecki> :)
Jul 12 19:34:36 <neonknight77> hello, when creating a new sdk build am I suppose to flash the img or bin?
Jul 12 19:34:53 <neonknight77> i already have openwrt luci installed
Jul 12 19:36:02 <Borromini> sysupgrade neonknight77. never factory.
Jul 12 19:54:34 <neonknight77> Borromini but I do factory if I'm in stock x branded firmware?
Jul 12 19:55:02 <Borromini> factory images are meant to be used when coming from the vendor's official firmwar
Jul 12 19:55:05 <Borromini> * firmware
Jul 12 19:55:26 <Borromini> this kind of questions belongs in #openwrt btw, not in the dev channel.
Jul 12 19:55:34 <neonknight77> well hang on
Jul 12 19:55:39 <neonknight77> but thx for answering
Jul 12 19:55:49 <Borromini> yw hope that gets you on your way
Jul 12 19:55:51 <neonknight77> ok and last question, where can I get the list of recommended packages to compliment my profile
Jul 12 19:56:24 <Borromini> there's no recommended packages to my knowledge, openwrt has everything for expected router functionality onboard
Jul 12 19:56:35 <Borromini> unless you are maybe running an old flash and ram constrained device
Jul 12 19:57:00 <neonknight77> so what you're saying is, the profile already comes with what I flash the first time i ever did open wrt
Jul 12 19:57:08 <neonknight77> and all we do with SDK is just ADD on top of that
Jul 12 19:57:25 <Borromini> or remove
Jul 12 19:59:40 <neonknight77> oh shoot no wonder I was having so much trouble
Jul 12 19:59:47 <neonknight77> I kept adding luci to the packages
Jul 12 20:00:44 <neonknight77> and pretty much a ton of other packages I thought I had to add to get close to stock openWRT firm
Jul 12 20:00:47 <neonknight77> LOL
Jul 12 20:03:00 <Borromini> sounds like you're using the master sdk
Jul 12 20:04:18 <neonknight77> well the only way I was able to get things to work was get the master sdk and then the specific one to my chipset, and copy contents of master sdk in to the specific one and only overwrite whatever was missing
Jul 12 20:04:29 <neonknight77> write whatever did not need to be overwritten*
Jul 12 20:05:00 <Borromini> sounds fishy, but OK.
Jul 12 20:19:32 <neonknight77> Borromini i created the image and just flashed with the sysupgrade.bin for my router and it took me to LEDE without any of my packages I added
Jul 12 20:19:51 <neonknight77>  LEDE Reboot 17.01.5 r3919-38e704be71 / LuCI Master (git-19.193.26639-d01cdd9)
Jul 12 20:21:04 <Borromini> i have never used the sdk myself. try #openwrt and see if someone can help you there. sounds like maybe you tried adding too many packages and there's no new image created.
Jul 12 20:21:07 <Borromini> good luck.
Jul 12 20:22:04 <neonknight77> it was actually just 2 lol
Jul 12 20:28:42 <FishesBeCrazy> Hey there - I'm sorta new to this. I just started using OpenWRT and I'm trying to familiarize myself with the build process.
Jul 12 20:29:27 <neonknight77> hello
Jul 12 20:29:36 <neonknight77> i spent last 4 days, a lot of hours getting it down
Jul 12 20:29:41 <neonknight77> what kind of questions do you have
Jul 12 20:30:00 <FishesBeCrazy> I've got an Archer C7, and I'm trying to build the binaries for it. The tricky bit is that the resulting SHA256 sums don't match up with what's available on the download page
Jul 12 20:30:59 <neonknight77> are you talking about different file size errors in cmd?
Jul 12 20:31:10 <olmari> You don't compare self built sums onto download site ones
Jul 12 20:31:41 <FishesBeCrazy> aha - that's good to know
Jul 12 20:32:08 <neonknight77> olmari how do you ignore opkg unable to install because of mismatched size?
Jul 12 20:32:49 <olmari> Install what? You either basically compile everything yourself, image and packages, or use site offered ones
Jul 12 20:33:42 <olmari> Sure, all sorts of mismatching could be achieved, but those are usually highly special cases
Jul 12 20:33:54 <neonknight77> Ok. 1) How do I check the version of OpenWRT in an image I build? 2) How do I check the contents/packages that were compiled
Jul 12 20:34:23 <neonknight77> built*
Jul 12 20:34:45 <olmari> Wiki has info, buildroot I think has the info
Jul 12 20:35:06 <olmari> 2, whatever you choose in menuconfig
Jul 12 20:35:39 <neonknight77> Right, but I didn't have any option to select Target and I want to make sure that it built the firmware for my model
Jul 12 20:36:17 <olmari> How do you mean?
Jul 12 20:36:51 <olmari> Target selections is first menuitems in menuconfig pretty much
Jul 12 20:37:04 <neonknight77> Where you select the chipset -> router etc etc
Jul 12 20:37:07 <neonknight77> Yes, I've seen the guide
Jul 12 20:37:10 <neonknight77> I don't have that lol
Jul 12 20:37:59 <neonknight77> I am using this sdk builder by the way http://downloads.openwrt.org/releases/18.06.4/targets/ipq806x/generic/
Jul 12 20:38:03 <olmari> What DO you have then, because that sure does not sound like openwrt buildroot to me
Jul 12 20:38:39 <olmari> Try openwrt buildroot instead?
Jul 12 20:40:43 <neonknight77> this is my make menuconfig : https://i.imgur.com/xao4MxH.png
Jul 12 20:44:28 <FishesBeCrazy> You have the relevant target architecture in target/linux?
Jul 12 20:46:18 <neonknight77> FishesBeCrazy I see about 15 different chipset folders but I do see mine as well, I'm just curious how it knows which one to build
Jul 12 20:48:43 <FishesBeCrazy> So I haven't tried the sdk yet. For me, so far, I see the target at the top of the menu - https://imgur.com/a/4VHBc6q
Jul 12 20:50:04 <neonknight77> You know what, I apologize. I was in the master folder. I am building from the specific folder for my chipset and yes, target/linux only has 2 folders, a generic, and my chipset = ipq806x
Jul 12 20:50:14 <neonknight77> So that means it automatically builds with the right firmware?
Jul 12 20:50:17 <FishesBeCrazy> my understanding is that the makefile you define with your architecture is read by the scripts/feeds file. So you add your (properly defined) architecture, then run ./scripts/feeds update <pkgname>; ./scripts/feeds install <pkgname> and it should become available in the menu
Jul 12 20:51:05 <FishesBeCrazy> If somebody more experienced would like to correct me, I'd love to learn a bit more.
Jul 12 20:54:11 <olmari> Like said, please look wiki for buildroot, and use it instead SDK in target folder
Jul 12 22:29:37 <mangix> joy. more fallout from the lua 5.3 update
Jul 12 22:43:29 <karlp> this surprises you? :)
Jul 12 23:19:24 <neonknight77> anyone encounter .stamp issues in the build_dir?
Jul 12 23:41:39 <mangix> jesus. I took a look at ag71xx with pahole and found massive problems
Jul 12 23:42:03 <mangix> no wonder it's slow
Jul 12 23:42:09 <neonknight77> can someone explain why -j8 causes issues compiling?
Jul 12 23:42:32 <jeffsf> neonknight77: Such as dying part way though a build?
Jul 12 23:42:51 <neonknight77> Yes
Jul 12 23:42:55 <jeffsf> Try `make -jN toolchain/install` if a fresh target
Jul 12 23:43:09 <jeffsf> The `make -jN clean download world`
Jul 12 23:43:30 <jeffsf> clean as to your option, though with `ccache` enabled I use it most every time
Jul 12 23:43:33 <neonknight77> Is there a way I can find the last .conf I saved tho?
Jul 12 23:43:50 <neonknight77> and by ccache, we are speaking of ccache in the make menuconfig selection?
Jul 12 23:44:03 <jeffsf> should be in opwnwrt/.config
Jul 12 23:44:29 <jeffsf> Or use `./scripts/env` to manage it with git
Jul 12 23:44:52 <jeffsf> Developer options > use ccache or something like that, yes
Jul 12 23:45:08 <neonknight77> kk, I am curious why any time target selection is done, it doesn't automatically choose luci GUI
Jul 12 23:45:10 <jeffsf> CONFIG_CCACHE=y
Jul 12 23:45:17 <neonknight77> ok
Jul 12 23:45:27 <jeffsf> Because LuCI is optional
Jul 12 23:45:45 <neonknight77> without it my GUI refuses connection, SSH only?
Jul 12 23:46:04 <jeffsf> That is correct, it is explicitly added to "release" build configs
Jul 12 23:46:49 <neonknight77> Ok, I have been studying the SDK package for many hours past few days. I am still confused as to why there is a SDK package for specific chipsets empty, and a master
Jul 12 23:47:02 <neonknight77> Is the single 60MB builds for updates or what
Jul 12 23:48:11 <jeffsf> At least as I understand it, the SDK is to build single packages
Jul 12 23:49:08 <jeffsf> Given that the full build system, SDK, and image builder all have about the same system requirements, I just use the full build systtem
Jul 12 23:49:14 <neonknight77> Are you able to attach a single package to a "release" OpenWRT image that general population use to flash?
Jul 12 23:49:36 <jeffsf> No clue -- I can't remember the time I ran a "release" build for anything but testing
Jul 12 23:50:44 <neonknight77> Ok, and can I report a bug in here if it's not getting any views on the bug tracker?
Jul 12 23:50:50 <neonknight77> I found it yesterday, very annoying. Lol
Jul 12 23:51:56 <jeffsf> Posting it here isn't, in my opinion, going to attract much more attention.
Jul 12 23:52:26 <jeffsf> A clear post to the mailing list, including what you did, what happened, what you expected, and how to replicate (in detail) would probably be a better path
Jul 12 23:52:49 <neonknight77> Well, what's your opinion on extra protocols being added to OpenVPN config, and every time the VPN restarts or a change is made
Jul 12 23:52:59 <neonknight77> I have to SSH and remove those lines from the config file. Lol
Jul 12 23:53:06 <neonknight77> Just a bad install?
Jul 12 23:53:33 <jeffsf> No idea. I don't run non-essential services under OpenWrt.
Jul 12 23:53:55 <jeffsf> Link the bug?
Jul 12 23:54:43 <neonknight77> Yes
Jul 12 23:54:45 <neonknight77> https://bugs.openwrt.org/index.php?do=details&task_id=2382
Jul 12 23:56:07 <jeffsf> Reading it, I would have no idea what you did or how to go about repllicating it
Jul 12 23:56:50 <neonknight77> Install OpenVPN with the packages I listed and run a VPN that uses user_pass authentication in the OpenVPN folder in /
Jul 12 23:57:02 <neonknight77> I'll update with the settings I use, I reckon that would help
Jul 12 23:58:07 <jeffsf> Search for "example of a good bug report" or the like
Jul 12 23:59:02 <jeffsf> Steps to reproduce, actual result, expected result need to be super clear -- assume that whoever is triaging or testing it has basic familiarity with the system, but no more.
Jul 12 23:59:20 <neonknight77> There's so many bug reports that I doubt anyone is even interested in this bug
Jul 13 00:01:22 <jeffsf> First thing is to check if it exists on master. If not, then you're done.
Jul 13 00:01:50 <jeffsf> I'm not a core dev, but if I were, I wouldn't be putting any further energy into v18
Jul 13 00:02:28 <neonknight77> That's where I found it. Lol. Which got me trying to resolve it via building sdk because it was so annoying. Although I've been trying to build it for past 3 days about 6 hours a day I spend and not yet :D
Jul 13 00:02:39 <neonknight77> Is it bad?
Jul 13 00:02:50 <neonknight77> Or nothing more to be done
Jul 13 00:03:51 <jeffsf> Shouldn't take more than an hour or so to build a "normal" device from source, the first time. A half hour once the tool chain is built -- two-core VM running on a MacBook Pro, Debian 9
Jul 13 00:04:06 <jeffsf> Incremental builds are 5-15 min.
Jul 13 00:04:56 <neonknight77> I like to think my gaming laptop is quite strong. But every time it gets right about done, some error pops up
Jul 13 00:06:27 <neonknight77> Nice. It's finally on install
Jul 13 00:07:08 <neonknight77> First time ever. W00T
Jul 13 00:07:12 <jeffsf> :+1:
Jul 13 00:17:08 <neonknight77> LOL. root@linux-x3lf:
Jul 13 00:17:14 <neonknight77> is it done?
Jul 13 00:17:33 <jeffsf> Poke it a couple times, if it rolls over, it's not dead
Jul 13 00:17:50 <jeffsf> If it bites you, well, then. you shouldn't have poked it
Jul 13 00:27:41 <neonknight77> Hmm
Jul 13 00:30:59 <neonknight77> jeffsf so I flash and it takes me to the same place, the LEDE Luci landing page
Jul 13 00:31:11 <neonknight77> None of my packages are there either
Jul 13 00:31:49 <neonknight77> Let me try from stock
Jul 13 01:08:58 <ssn> anything preventing this from getting merged? https://patchwork.ozlabs.org/patch/1122661/ [PATCH] ath79: add support for gl-ar750
Jul 13 01:19:42 <jeffsf> ssn: Probably needs a couple tweaks related to DEVICE_TITLE is split up into DEVICE_VENDOR, DEVICE_MODEL and DEVICE_VARIANT
Jul 13 01:20:08 <jeffsf> as well as https://github.com/openwrt/openwrt/pull/2159
Jul 13 01:21:32 <jeffsf> First one is commit 8c3dbaf8b4
Jul 13 01:22:13 <ssn> jeffsf: thanks for clarifying
Jul 13 01:23:10 <jeffsf> The lead devs are likely sleeping, so they may have more to say
Jul 13 02:13:29 <mangix> hrm
Jul 13 02:13:41 <mangix> i guess pahole doesn't run on cross compiled .o files
Jul 13 02:41:37 <guest7896554> johnf hi
**** ENDING LOGGING AT Sat Jul 13 02:59:57 2019