**** BEGIN LOGGING AT Mon Jan 25 02:59:57 2021
Jan 25 08:01:14 <reiffert> mangix: thank you
Jan 25 09:26:27 <mangix> aparcar[m]: it's only a compilation fix...
Jan 25 09:36:51 <ynezz> as always I've hard time with your "compilation fixes" as you don't include the error
Jan 25 09:37:11 <ynezz> how could anyone review that?
Jan 25 09:38:33 <mangix> the buildbots have one for you https://downloads.openwrt.org/snapshots/faillogs/arc_archs/base/umdns/compile.txt
Jan 25 09:39:37 <mangix> that's actually laughable. i get criticized when including the error and not including it
Jan 25 09:39:45 <mangix> whatever
Jan 25 09:40:03 <ynezz> but you see, it's not even musl related
Jan 25 09:41:25 <mangix> it actually is. it just so happens that glibc with ARC fails in the same way
Jan 25 09:42:50 <aparcar[m]> what is the status of arc anyway? Will it be removed anytime soon? If so we could remove whole bunch of errors as it seems
Jan 25 09:43:02 <aparcar[m]> If not I may request a test device at some point
Jan 25 09:43:42 <mangix> doesn't seem like it
Jan 25 09:44:05 <mangix> arc_hs works. arc700 crashes on boot with glibc
Jan 25 09:44:30 <mangix> synopsys doesn't know why yet.
Jan 25 09:44:56 <mangix> I recommended marking arc700 as BROKEN. synopsys agreed on the mailing list
Jan 25 09:47:11 <aparcar[m]> patch?
Jan 25 09:47:22 <mangix>  haven't made one
Jan 25 09:56:40 <karlp> did we ever get any actual platforms on arc? it kinda feels like we spent years fucking aroudn with keeping glibc so they could run openwrt on a couple of dev boards internally.
Jan 25 09:59:53 <mangix> you mean uClibc-ng
Jan 25 10:00:19 <mangix> glibc is used by at least Hauke and dedeckeh for develpment purposes
Jan 25 10:36:07 <Pepe> Hauke: Hi, I'm so sorry for disturbing you, I noticed that you have in a stagging - commit, which adds ath11k. Is there anything new about ath11k in OpenWrt?
Jan 25 10:42:07 <blogic> Pepe: last I tried it did not work
Jan 25 10:42:18 <blogic> Pepe: the upstream kernel support is pretty much broken
Jan 25 10:42:23 <blogic> Pepe: what HW do you have ?
Jan 25 10:50:58 <Pepe> I'm asking rather if there is some Wi-Fi 6 card - m.2/minipcie which is running with OpenWrt as many commits in upstream kernel is included in newer kernel than 5.4 LTS. It would be helpful to know if you know if chipset QCA6391 works. I know that it is somehow complicated regarding Intel Wi-Fi cards as it supports client mode right now and it is a million dolar question if it will run under AP as
Jan 25 10:51:00 <Pepe> well.
Jan 25 10:51:37 <Pepe> Also if something from Marvel like 88W9098 will work
Jan 25 10:53:15 <Pepe> and if there are plans for support DBDC in LuCI as currently can be done manually or via UCI
Jan 25 11:07:54 <Borromini> Pepe: there's MT7915 I think
Jan 25 11:11:40 <Borromini> Pepe: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=fb4d7a9680117a00721936c98ce41eeb2dea95c9
Jan 25 11:20:01 <Pepe> Thanks! I'll try to find some sellers or distributors, where I can buy it. I found it on Asiarf, but is is available on backorder, so lets see.
Jan 25 12:11:38 <blogic> Pepe: none yet
Jan 25 12:11:49 <blogic> the first pci/e m.2 cards are just popping up
Jan 25 12:47:37 <johnf> I'm trying to build firmware for a device using a custom built image-builder
Jan 25 12:47:50 <johnf> something seems to be trying to select libustream-wolfssl20200215
Jan 25 12:48:04 <johnf> but I already need openssl on here, and I don't have enough space for both
Jan 25 12:48:16 <johnf> how can I figure out what package is the source of this dependancy?
Jan 25 12:48:40 <ynezz> in menuconfig
Jan 25 12:50:16 <johnf> you mean in openwrt, not the resulting imagebuilder package?
Jan 25 12:50:23 <johnf> I do see that it's selected by the platform
Jan 25 12:50:23 <ynezz> hit / to search for symbol, then enter libustream-wolfssl and look at Selected by
Jan 25 12:50:29 <johnf> ah, ok
Jan 25 12:51:36 <johnf> https://gist.githubusercontent.com/johnfzc/8ea89389dc29753c8340039937bde8f3/raw/96514472d5b8e58e2017f2b02c72301294059466/gistfile1.txt
Jan 25 12:52:31 <johnf> hmm, ok, I've moved it from M to < > state
Jan 25 12:52:53 <johnf> I'll see what that does
Jan 25 12:53:00 <johnf> maybe I just needed one SSL lib
Jan 25 12:53:34 <johnf> I'm not certain this is going to work, but maybe it'll clarify the cause of the issue
Jan 25 12:54:23 <Borromini> ynezz: hi! did you get testing confirmations on the zyxel realtek factory images? if so i don't need to test them anymore.
Jan 25 12:56:59 <ynezz> Borromini: already paged that out
Jan 25 12:58:06 <Borromini> sorry don't understand?
Jan 25 12:58:58 <ynezz> multitasking, have small cache so once I finish the task I flush the cache
Jan 25 12:59:50 <Borromini> :P ok
Jan 25 13:00:20 <ynezz> IIRC I've registered your additional tested-by and included it in that commit
Jan 25 13:01:12 <Borromini> ah. i didn't test the factory image though
Jan 25 13:01:23 <Borromini> but bmork said the code was rather trivial
Jan 25 13:01:31 <Borromini> i'll give it a run and if there's issues i'll report back.
Jan 25 13:01:49 <ynezz> well, there is no such Tested-by-only-with-sysupgrade :p
Jan 25 13:02:35 <Borromini> :P
Jan 25 13:02:48 <ynezz> trivial = huge problems :p
Jan 25 13:02:56 <ynezz> (once merged, that's it)
Jan 25 13:03:25 <Borromini> :P :P
Jan 25 13:32:32 <johnf> ynezz: ok, so now when I create my image in imagebuilder, I get
Jan 25 13:32:33 <johnf>  * opkg_install_cmd: Cannot install package libustream-wolfssl.
Jan 25 13:32:49 <johnf> but I can't tell which package is demanding the wolfssl version of libustream
Jan 25 13:33:29 <johnf> I'm adding a lot of packages in (and removing some)
Jan 25 16:14:08 >KGB-1< https://tests.reproducible-builds.org/openwrt/openwrt_tegra.html has been updated. (0% images and 98.3% packages reproducible in our current test framework.)
Jan 25 17:26:23 <Grommish> Pepe: https://github.com/openwrt/packages/runs/1760030654
Jan 25 17:26:33 <Grommish> Now, it runs out of disk space :d
Jan 25 18:57:24 <johnf> I really don't understand what dependancy is causing this build to require libustream-wolfssl and, consequently, fail
Jan 25 18:57:29 <johnf> https://gist.github.com/johnfzc/81f169084f2c75812c42c3451dacf6db
Jan 25 19:02:00 <johnf> omg, I managed to do it, I've removed all the real dependancies and it seems to now come from the hardware platform itself
Jan 25 19:02:28 <johnf> Selected by [y]
Jan 25 19:02:37 <johnf> TARGET_ath79_generic_DEVICE_tplink_tl-wr842n-v2 [=y] && <choice> && TARGET_ath79_generic [=y]
Jan 25 19:03:02 <johnf> so, with all the other deps out, I was able to add -libustream-wolfssl
Jan 25 19:03:10 <johnf> and the firmware was small enough to package
Jan 25 19:07:34 <Hauke> Pepe: I added ath11k, to build test it. I haven't tested the hardware
Jan 25 19:10:12 <Hauke> karlp: I used glibc to use the sanitizer
Jan 25 19:10:23 <Hauke> this is not supported by musl, only glibc
Jan 25 19:44:10 <nick[m]1> someone here who has a good and working alternative to getifaddrs?
Jan 25 20:21:10 <karlp> Hauke: yeha, I did mean uclibc-ng as manginx reminded me
Jan 25 20:31:54 <grift> i just want to know how ya'll leverage openwrt. i know me openwrt assembly is awesome but i can't help but think i am just scratching the surface. what are some of your hidden gems? not pretenting i can produce all of it but i just want some hints
Jan 25 20:46:14 <Hauke> karlp: I am happy uclibc-ng is gone ;-)
Jan 25 22:13:36 <Pepe> Grommish: Are you so sure about that? You can download logs there and check what was wrong. I take a look at cortex-a9_vfpv3-d16 and see there: collect2: error: ld returned 1 exit status , error: aborting due to previous error , error: could not compile `std` ,
Jan 25 22:13:45 <Pepe> and so on. :)
Jan 25 22:13:56 <Pepe> Hauke: Thanks for reply, though!
Jan 26 00:11:16 <Grommish> Cortex-a9 should never be attempting to compile rust as the HOST, so I'd hope it would error..
Jan 26 00:11:48 <Grommish> Pepe: but yes, look at x86.. Fails because it runs out of disk space
Jan 26 00:13:13 <Grommish> Unless yu can reasonably expect to compile OpenWrt on it, the package will fail more than likely.  It's the fact that x86_64 and powerpc work (and other platforms used to actually build OpenWrt) that I'm concerned with
Jan 26 00:13:35 <Grommish> No one would be compiling Rust on their router :D
Jan 26 00:13:59 <Grommish> /would/should/s
Jan 26 00:14:31 <Grommish> I'd be crazy enough to try.. and in fact I do have rustc/cargo for mips64 so I COULD
Jan 26 00:16:11 <Grommish> The only artifact would be the compied toolchain, which for myh x86_64 to mips64 is 530MB in a tar.xz file
Jan 26 00:17:47 <Grommish> Pepe: Unless OpenWrt wanted to host the "safely" compiled toolchain installation stuff, the user would have to generate their own at least once and then it'll use that resulting installation blob to deal with rust in the future (stashed in dl/*)..
Jan 26 00:18:10 <Grommish> and then again each time the toolchain updates
Jan 26 02:07:25 <mangix> Grommish: I've compiled OpenWrt on an mvebu host. AMA.
Jan 26 02:08:24 <Grommish>  mangix :D Nah, Pepe was saying the packages were failing on the CI.. and they are, but the ones I actually care about fail due to out of space
Jan 26 02:10:22 <Grommish> mangix: Would you want/need a rust toolchain that was cortex_a9 or whatever as the host?
Jan 26 02:13:36 <Grommish> or failing because rust doesn't support tier 3 for standard install, just cross compile. which is why I don't mind those
**** ENDING LOGGING AT Tue Jan 26 02:59:57 2021