01:52 <stefanct> so... got a "new" f9k1109v1 today... with working ethernet, woohoo... so before bricking that too i am looking at the boot log of the other one since i dont get a shell in the initramfs

01:52 <stefanct> of_serial: probe of 10000c00.uartlite failed with error -22

01:53 <stefanct> that EINVAL which is returned if there is no compatible device found

01:54 <stefanct> rt3883.dtsi defines that as compatible = "ralink,rt3883-uart", "ralink,rt2880-uart", "ns16550a"; and of_platform_serial_table in 8250_of.c contains a "ns16550a" device so i am not exactly sure why it happens

01:56 <stefanct> on a somewhat related note i am wondering about https://github.com/openwrt/procd/pull/9 and the reason for its stale state. i guess i may have ran into that due to the problem above(?)

03:55 <russell--> i tracked down my problem

08:29 <SnarlingFox> Good morning all

08:32 <SnarlingFox> The WSQ50 has locked up again overnight, these are the last logs that were spat out over the UART; https://pastebin.com/QFrTBu4n

08:35 <robimarko> Basically, ath10k locks up and then kernel locks up

08:35 <SnarlingFox> Hooray for bugs \o/

08:35 <SnarlingFox> (Also good morning robimarko, hows the coffee this morning?)

08:36 <SnarlingFox> (Wouldn't blame you for responding "Not strong enough for this shit" :D)

08:36 <robimarko> Quite nice as I am dealing with a project using modems (5G ones, the worst kind) that is kind of whack a mole

08:37 <SnarlingFox> Ugh. Those are swearwords to me right now. Realise this is a tangent, but hey-ho, ADHD \o/

08:37 <robimarko> :)

08:38 <SnarlingFox> I have a Seeed Studio Odyssey J4105 (https://wiki.seeedstudio.com/ODYSSEY-X86J4105/) that I wanted to set up as a 4G/5G hotspot - I bought a Quectel EM120R-GL (4G LTE, 2x2, 600 MBps) from Amazon because it was only like £27 / 32 euro)

08:39 <SnarlingFox> Advert said "For Lenovo / Thinkpad X1 series" but no mention of any hardware / BIOS locking

08:40 <SnarlingFox> There was a reason that modem was so cheap... XD

08:41 <robimarko> Its probably just rebadged with a Lenovo ID

08:41 <robimarko> So worst case you need to add that ID to the kernel and it will work with ModemManager

08:42 <SnarlingFox> Supposedly you can unlock them with ModemManager and a few super secret AT commands, but mine wasn't even showing up in `lspci` or `lsusb` nor `dmesg` on Ubuntu 22.04, nor in Windows 10. Did some further reading, the modem itself has a secure boot feature where it won't actually allow itself to be enumerated unless the BIOS itself passes its own checks.

08:42 <SnarlingFox> Frankly, I found that digusting.

08:42 <robimarko> Now that sounds like classic Lenovo

08:43 <SnarlingFox> Luckily the retailer were kind enough to take it back for a full refund as I'd kept all the original packaging. I quickly cleaned off my grubby fingerprints with some isopropyl alcohol and posted it back to them.

08:44 <SnarlingFox> So I guess now I'm in the market for a trustable / affordable 4G / 5G LTE modem in M.2 form factor that'll work nicely with OpenWrt :) (recommendations welcome)

08:44 <robimarko> Quectel RM520N

08:45 <robimarko> It works but affordable part is not probably true

08:45 <SnarlingFox> Hah

08:45 <SnarlingFox> You're not wrong.. basically 10x the price of the EM120R XD

08:46 <SnarlingFox> Also, sorry, that's 4x4, I need 2x2.. not sure whether you can run the RM520N in 2x2 by just leaving 2 antennas disconnected

08:47 <SnarlingFox> (That said... could always drill 2 new holes for 2 more antennas)

08:47 <robimarko> Thats the one I have so I know it works (Needs one backport to 1.22 MM)

08:47 <SnarlingFox> Be a shame to butcher this gorgeous case though; https://www.cnx-software.com/wp-content/uploads/2020/07/ODYSSEY-X86J4105-Re_Computer-Case-Review.jpg

08:48 <SnarlingFox> (And now the intrusive thoughts are kicking in.. drilling 2 new holes in that case whilst quoting The Shining "Here's Johnny!" as the drill bit passes through)

08:50 <SnarlingFox> robimarko: Do you find you have to provide heatsinking to the back of the RM520N ?

09:01 <Jenga> How can I build a single package for all architectures ?

09:02 <f00b4r0> nbd: hi, just a friendly ping to ask if there's anything new I can test on mt7915e? :)

09:10 <robimarko> SnarlingFox: Well, I think that the router I use already has it heatsinked

09:28 <owrt-images-builds> Build [#283](https://buildbot.openwrt.org/images/#/builders/231/builds/283) of `master_mpc85xx/p1010` failed.

09:47 <SnarlingFox> robimarko: Cheers, just ordered one.. my bank balance felt that x.x

09:56 <robimarko> :)

10:01 <schmars[m]> Jenga you'll have to build it for each architecture separately. unless it's a package that doesn't need compilation, like shell skripts, luci modules, etc. (architecture = all)

10:15 <rmilecki> Ansuel: there are line breaks broken in the https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=45b3c620e597c56a4eedb864b5e900d89340e986 ("ipq806x: add support for Linksys e8350-v1")

10:15 <rmilecki> in the target/linux/ipq806x/files-6.6/arch/arm/boot/dts/qcom/qcom-ipq8064-e8350-v1.dts

10:17 <Ansuel> .....................................

10:17 <Ansuel> damn github that doesn't show those....

10:19 <robimarko> Dont we have .gitconfig to make sure they are UNIX?

10:21 <Ansuel> think we can't due to some patches that have those line break (packages that supports both windows and linux)

10:23 <robimarko> OK, I will convert the endings and remove all of those excess newlines

10:23 <robimarko> *indentations

10:23 <Ansuel> just fixed it

10:23 <robimarko> Yeah, you were fast

10:24 <robimarko> Also, those debounce-interval and wakeup-source properties?

10:24 <robimarko> Those are not polled keys

11:09 <robimarko> Anybody knows how is zlib getting pulled in?

11:11 <robimarko> Cause, its not being set in .config at all

11:11 <robimarko> But building mtd-utils somehow pulls it in despite it not depending on zlib

11:12 <robimarko> Only PKG_BUILD_DEPENDS make sense

11:38 <Jenga> schmars[m]: thanks, looking at https://github.com/openwrt/docker?tab=readme-ov-file#sdk seems like the path I should take. but I still don't understand how I should specify targets.

11:38 <Jenga> like would it be sufficient to only use openwrt/sdk:aarch64_cortex-a53 for all mediatek qualcommax etc targets ? or I should explicitly use openwrt/sdk:mediatek-mt7622 openwrt/sdk:mediatek-filogic for each of them

11:50 <robimarko> Ok, so mtd-utils compiles fine purely by accident because libncurses-dev somehow pull-in zlib (But it is not selected in the end .config at all)

11:51 <robimarko> I will update to 2.2.0 which doesnt require zlib as we are not packaging mkfs.* tools for target devices anyway

12:54 <schmars[m]> for jenga so i can copy it later: for packages only the architecture matters, which means you can use either target's sdk as long as it's the same architecture. run 'openwrt.git/scripts/dump-target-info.pl architectures' to get a full list of architectures and their targets

12:54 <schmars[m]> (i actually use that script's output to generate buildbot jobs for a custom firmware project)

17:23 <robimarko> Mangix: You around?

17:23 <Mangix> I am

17:24 <robimarko> Well, I dont think we can drop e2fsprogs

17:24 <robimarko> Cause even though it doesnt look like it, it is indeed installing its tools

17:25 <Mangix> Can its libuuid be disabled?

17:25 <robimarko> Yes, as its currently being manually enabled

17:27 <Mangix> Hmm? Where is uuid.pc in staging_Dir from?

17:27 <robimarko> I mean, we are passing --enable-libuuid to e2fsprogs

17:27 <robimarko> So we can also disable it

17:28 <Mangix> Right. Should switch to util-linux now that it's present

17:28 <robimarko> Yeah, I will just do that instead of dropping e2fsprogs

17:29 <Mangix> Bte

17:30 <Mangix> There's both a host build of package/e2fsprogs and tools/e2fsprogs. Always seemed weird to me

17:31 <robimarko> We do similar for mtd-utils

17:31 <robimarko> One is meant for image building while other runs on the device

17:33 <Mangix> There's no host build for package/mtd-utils

17:34 <robimarko> Oh, you mean in package/e2fsprogs there is also a host version?

17:34 <robimarko> Now that is weird

17:34 <robimarko> Oh, indeed

17:35 <robimarko> Hm, happened 6 years ago in this commit: https://github.com/openwrt/openwrt/commit/20d63ebc942012040e08bb815a94f40e535b9cf9

17:36 <robimarko> e2fsprogs depends on e2fsprogs/host via PKG_BUILD_DEPENDS

17:36 <dwfreed> Mangix: necessary to avoid system-provided e2fsprogs making a filesystem that openwrt e2fsprogs is incapable of reading

17:36 <Mangix> But tools/e2fsprogs is avaikable

17:36 <robimarko> dwfreed: The confusion is that we provide e2fsprogs built in tools

17:36 <robimarko> But also build a host version in package/e2fsprogs

17:37 <dwfreed> ah

17:37 <dwfreed> yeah, *that's* dumb

17:37 <robimarko> Its even weirder that package/e2fsprogs has PKG_BUILD_DEPENDS:=e2fsprogs/host

17:40 <Mangix> Sounds like tools/e2fsprogs should be made to provide the same as e2fsprogs/host. I'm away from a computer right now so I can't look.

17:40 <robimarko> Well, they added host build to get compile_et and mk_cmds tools

18:50 <owrt-images-builds> Build [#284](https://buildbot.openwrt.org/images/#/builders/231/builds/284) of `master_mpc85xx/p1010` completed successfully.

19:20 <Ansuel> robimarko should we add autoconf as a dependency of that tool ?

19:20 <Ansuel> just to handle corner case of -j1000000 ?

19:37 <dwfreed> I mean, your dependencies should be correct, regardless of whether you think something is an edge case

19:37 <dwfreed> make --shuffle is a thing these days

20:12 <robimarko> Ansuel: Yeah, I added automake

20:13 <robimarko> Now I just wish damn macOS did not fail like always