Tapper has quit [Read error: Connection reset by peer]
nitroshift has joined #openwrt-devel
rmilecki has joined #openwrt-devel
<aparcar[m]>
I'm trying to build something via CI but it fails since a new kernel symbol is required and unset, how do I fix that?
<aparcar[m]>
specifically SECURITY_NETWORK_XFRM which is added when enabling SELinux
<rsalvaterra>
Grommish: Pong
<aparcar[m]>
rsalvaterra: hey can you help me with the kernel question?
<rsalvaterra>
hauke: Oooh, not the first time I've seen it, I totally forgot the crystal ball existed! :) So, my prediction isn't bad at all, 5.15 for the next LTS. ;)
<rsalvaterra>
aparcar[m]: I hope so! It's the SECURITY_NETWORK_XFRM symbol, right?
indy has quit [Read error: Connection reset by peer]
indy_ has joined #openwrt-devel
<rsalvaterra>
It's not obvious to me why the build should fail. SECURITY_NETWORK_XFRM shouldn't be required unless you enable both SECURTY and IPSec, right?
<russell-->
fwiw, my initial problem was in self-ventilating my foot, i had an erroneous command in a uci-defaults script (trying to uci commit to a non-existing file, apparently). the tftpboot problem is separate and not new.
<rsalvaterra>
aparcar[m]: You could try and add CONFIG_SECURITY_NETWORK_XFRM=y to the generic kconfig. It doesn't feel very correct to me, though.
<aparcar[m]>
rsalvaterra: what do you mean by not very correct?
<aparcar[m]>
the config only enabled SELINUX... mhh
<rsalvaterra>
aparcar[m]: I mean, if both SECURITY and XFRM are enabled, CONFIG_SECURITY_NETWORK_XFRM will also be enabled. But it's a "depends", which means it should be optional.
<rsalvaterra>
Unless, of course, something "selects" it… I haven't checked that.
<rsalvaterra>
Nope. Nothing selects it (on 5.10.61, at least).
<rsalvaterra>
aparcar[m]: Oh, wait! That's another story. It's just complaining about the missing symbol, that's all. :)
<aparcar[m]>
help
<rsalvaterra>
Let me fire up the dev machine, I just woke up. :P
<rsalvaterra>
Like I suspected, the symbol is missing from the generic kconfigs. Easy peasy: just add "# CONFIG_SECURITY_NETWORK_XFRM is not set" to both generic kconfigs.
<rsalvaterra>
I can cook a patch, if you prefer.
<rsalvaterra>
aparcar[m]: I think you're seeing that build failure because of this: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=eaa9c94c75748265b9260691d5f59c9d3711f7d4
<rsalvaterra>
Which is a really nice change, to be honest. ;)
<aparcar[m]>
yea this should go upstream
<aparcar[m]>
how to do the kernel refresh?
<aparcar[m]>
like, I want the config to be sorted
<rsalvaterra>
aparcar[m]: LOL! I was searching for the same. I haven't memorized the command for that.
<rsalvaterra>
aparcar[m]: I have to disagree about sending the exit-on-unset-symbol change upstream… When building a new kernel version, the usual procedure is taking your current kconfig and doing a make oldconfig, on which you'll be prompted for the missing symbols.
<aparcar[m]>
sure, but it's only enabled via a env right?
danitool has joined #openwrt-devel
<aparcar[m]>
please feel free to provide a patch, I've tried the suggested command but it doesn't do the trick, acutally just resetting the config...
<rsalvaterra>
aparcar[m]: No problem, give me a minute. I'll send it to the mailing list and cc you. ;)
<rsalvaterra>
If so, I can mention it in the commit message.
<aparcar[m]>
uhm yea I guess it is
<aparcar[m]>
thanks
<rsalvaterra>
aparcar[m]: There you go. :)
<aparcar[m]>
thanks
<aparcar[m]>
lgtm
<aparcar[m]>
do you have merge access yet?
<rsalvaterra>
aparcar[m]: Yes, but still sorting out some key issues… I just noticed now I can't push to my repos on Git{Hub,Lab}. Meh. I need to configure the ssh hosts and identities.
<aparcar[m]>
ugh why is patchwork so slow..
<rsalvaterra>
Yeah, takes quite a bit of time to react… :/
rejoicetreat has joined #openwrt-devel
<aparcar[m]>
mhh seems broken
<rsalvaterra>
Yeah, it's taking too long.
<rsalvaterra>
Ok, I can push it, but I've never done it before… what's the procedure? Just cherry-pick the commit and push to master?
f12 has quit [Ping timeout: 480 seconds]
<aparcar[m]>
sorry I just did
<aparcar[m]>
sudo git push -f
<rsalvaterra>
sudo…? o_O
<rsalvaterra>
Why do you need to run git as root?
<stintel>
you shouldn't
<stintel>
might be a joke like "sudo make me a sandwich"
<aparcar[m]>
it's a joke
<rsalvaterra>
Yeah, that's… scary.
<stintel>
but not sure ;)
<aparcar[m]>
also never use -f on oringin
<aparcar[m]>
*origin
<rsalvaterra>
Right, I was going to say that.
<rsalvaterra>
-f shouldn't be needed at all, you're just adding commits on top of existing ones.
<rsalvaterra>
So, on master, I'd just cherry-pick the commit I want to push and do git push origin. This should be it, right?
<stintel>
-f is probably disabled on openwrt.git
<stintel>
but don't remember actually all the safeguards that are in place
<aparcar[m]>
rsalvaterra: yea I usually cherry-pick whatever i want to merge
<rsalvaterra>
stintel: Oh, nice! Didn't know permissions could be so fine-grained, that's great.
<rsalvaterra>
aparcar[m]: Cool, let see if the CI complains this time. :)
<rsalvaterra>
aparcar[m]: Wow. Now the patch showed up on patchwork. :P
decke has joined #openwrt-devel
rejoicetreat has quit [Quit: Leaving]
f12 has joined #openwrt-devel
<aparcar[m]>
yea not great
aleksander has joined #openwrt-devel
<rsalvaterra>
aparcar[m]: Looks like CI is happy now.
<aparcar[m]>
wonderful
<rsalvaterra>
It failed building the target, not it's past that point and building the packages. :)
<rsalvaterra>
*now it's
<aiyion>
Good morning, just wanted to thank everyone that helped to to work through the onion-omega device support. Seing it in master and 21.02 is really rewarding! russell-- karlp pinky_ PaulFertser Tusker blocktrron
<karlp>
you're very welcome :)
<russell-->
+1
goliath has joined #openwrt-devel
<blocktrron>
aiyion: great to hear :)
<PaulFertser>
aiyion: and it was even backported to 21.02? Congrats!
<aiyion>
Writing the backport was easier than I thought, after reading the wiki-page with the policies.
<PaulFertser>
And was it accepted that late in the release cycle?
<rsalvaterra>
PaulFertser: It's rather tightly contained, so it doesn't surprise me much. :) https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=5cc05358006d592d3712e6565c253762819fe010
<PaulFertser>
Lucky aiyion :)
<aiyion>
Indeed :)
<PaulFertser>
BTW, I think Onion is violating GPL knowingly.
<aiyion>
PaulFertser: How's that?
<PaulFertser>
aiyion: distributing a proprietary wifi driver that violates GPL.
<PaulFertser>
I tried to contact them several times but got no reply at all.
<PaulFertser>
aiyion: do you probably have any contacts to them?
<aiyion>
Contacting them is on my agenda afer my last exam;
<aiyion>
For now the only conatc I have is a community-matrix chat, that has not been very helpful, but frienbdly.
Tapper1 has quit [Ping timeout: 480 seconds]
<PaulFertser>
aiyion: I've just forwarded you my email to them. Please feel free to ask about it, I'd want an official answer.
danitool has quit [Ping timeout: 480 seconds]
<aiyion>
I intend to ask for a bunch of test devices, for my ath79 porting efforts; could suit them well, as it would improve the likelyhood of me having an eye on the target, and for them to have a more sustainable productcycle.
<aiyion>
Thanks, will do.
<PaulFertser>
aiyion: IMHO they MUST stop distributing that driver in any form unless they're pro-violation.
<PaulFertser>
And if they are they should be shamed rather than endorsed.
<aiyion>
indeed
<PaulFertser>
There's no way this module can be distributed outside of direct Mediatek contractors.
<stintel>
sigh, gs108t-v3 unreachable after adding that vlan config
<stintel>
and apparently the TTL pin layout on the wiki is also wrong
rua has joined #openwrt-devel
<stintel>
ok, the GS108Tv3 is just super flakey apparently
<stintel>
if I connect the RX pin on the device, it stops booting, it doesn't even blip the LEDs when I plugin an UTP that is plugged into a PoE injector
<stintel>
it boots fine with the original PSU
<stintel>
but that's not really the point of buying a PoE-PD switch
rua has quit [Quit: Leaving.]
minimal has joined #openwrt-devel
rua has joined #openwrt-devel
jbowen has joined #openwrt-devel
cp- has joined #openwrt-devel
goliath has quit [Quit: SIGSEGV]
<Slimey>
heh
rua has quit [Remote host closed the connection]
rua has joined #openwrt-devel
aleksander has quit [Quit: Leaving]
<SherlockDomes>
I'm a bit confused about signal-noise-ratio. Is the SNR different for the AP and client? I assuem it must be as the noise differs between the station and the AP.
rua1 has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
valku has joined #openwrt-devel
nitroshift has quit [Quit: Gone that way --->]
<PaulFertser>
SherlockDomes: yes, and also nl80211 doesn't have a way to expose the measured SNR value from the driver.
<PaulFertser>
SherlockDomes: with some drivers, e.g. ath10k you can get SNR by adding some value to reported RSSI as ath10k doesn't (can't?) really report RSSI and instead the driver fakes it.
<SherlockDomes>
Interesting PaulFertser
fda has quit [Ping timeout: 480 seconds]
<SherlockDomes>
Is that rate algorithm also unidirectional? Specfically does the AP have a rate-control algorithm for it's TX and the station have its own algorithm for its TX?
<PaulFertser>
SherlockDomes: I would assume so
<SherlockDomes>
Thanks PaulFertser
goliath has joined #openwrt-devel
<SherlockDomes>
Quick question, if you're trying to usptream a patch for a bug fix is it worth generating a bug report?
<PaulFertser>
SherlockDomes: not really
<Slimey>
anyone in the states abroad want to play with some free hardware, i just cant find the time to get to work on it
<Slimey>
or abroad*
<Slimey>
or would forums be better place to ask
decke has quit [Quit: Leaving.]
danitool has joined #openwrt-devel
rua1 has quit [Quit: Leaving.]
rua has joined #openwrt-devel
KGB-0 has quit [Quit: KGB-0]
Ycarus has joined #openwrt-devel
minimal has quit []
fda has joined #openwrt-devel
Tapper has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
fda- has joined #openwrt-devel
fda has quit [Read error: Connection reset by peer]
Tapper has quit [Ping timeout: 480 seconds]
<will[m]>
heya i can compile for x86_64 and ARM just fine but when i try for this MIPS target i get a weird glib2 / meson / ninja error that is wasting a ton of time every time i try something new. any ideas? i even torched and rebuilt my buildroot, same issue https://gist.github.com/zyphlar/a3cfdebe590e16bf1c7171abf65d5b56
<dwfreed>
what's in /media/my_username/my_path/openwrt/build_dir/target-x86_64_musl/glib-2.68.4/openwrt-build/meson-logs/meson-log.txt
<dwfreed>
will[m]: ^^^
KGB-0 has joined #openwrt-devel
niyawe has joined #openwrt-devel
<aparcar[m]>
jow do you have a comment on a JS linter?
danitool has joined #openwrt-devel
<will[m]>
dwfreed: a lot, 10k lines ending in the same ninja error. lemme upload
<dwfreed>
Sure, but there's probably some useful info before that error :P
<dwfreed>
you may need to just refresh your buildroot? ninja is now shipped in openwrt/openwrt (and patched to use make's jobserver)
<rsalvaterra>
Oh, lol. My Sid machine just spitted out this after calling which: "/usr/bin/which: this version of `which' is deprecated; use `command -v' in scripts instead."
<Grommish>
rsalvaterra: I've gotten to the point with Rust that I think ARmv7 is settled, v6 is still in the weeds, but someone said the ppc wasn't working or something. I need someone with build exp give me a hint on the feature-set to use for it
<rsalvaterra>
I guess it's now legit to send this upstream. :P
<dwfreed>
rsalvaterra: lol
shibboleth has joined #openwrt-devel
<Grommish>
and, does OpenWrt split out PowerPC properly I wonder
<slh>
deprecating such a trivial tool (<50 lines of code) from debianutils is ... a lot of unnecessary churn for no gain (gnu-which and FreeBSD which are about to be packaged instead)
<rsalvaterra>
Grommish: Not yet. As of now, there are only ppc(32) targets. That will change after merging stintel's qoriq target, which is ppc64.
<rsalvaterra>
slh: I beg to differ. Why using an external executable when a shell built-in is available?
<Grommish>
Any chance the build system will do it correctky and denote the arch as powerpc, powerpc64, powerpc64le?
<will[m]>
so many cases where someone's "improvement" makes them feel better about themselves but costs tens of thousands of programmers/admins multiple hours of time and frustration
<rsalvaterra>
will[m]: In this case, a trivial alias will avoid regressions.
<slh>
rsalvaterra: because which usage is ubiquituous and command -v behaves not identically (for shell internals)
<will[m]>
also me: `echo "command -v $1" >> /usr/bin/which`
<rsalvaterra>
That's horrible. :P
<will[m]>
same thing happened with me and php deciding to change how sizeof/count in v7 for no earthly reason
<Grommish>
stintel: I'll ask you instead :) Any chance when you push a powerpc64 target that the $(ARCH) will read powerpc64/powerpc64le? Will you be updating ./incude/targets.mk to denote the different powerpc archs?
<slh>
rsalvaterra: "command -v test" vs "which test" (test is a shell internal of bash and most other shells)
<will[m]>
i just wrote a function called myCount that did the needful and grepped my whole source code rather than evaluating the entire codebase for nulls
<rsalvaterra>
slh: The difference is that you'd never invoke 'which test' because you already know the output would be empty. ;)
<rsalvaterra>
I see no regressions there. :)
<dwfreed>
type -P test
<dwfreed>
the output of `which test` is not empty
<dwfreed>
$ which test
<dwfreed>
/usr/bin/test
<dwfreed>
test is part of coreutils
<dwfreed>
it is shadowed by the shell built-in
<rsalvaterra>
dwfreed: Right you are, I misinterpreted. Hm.
<dwfreed>
notably, type -P is not POSIX, though
<slh>
rsalvaterra: the regression is needless churn in 40 years of a grown eco system, just to save 50 lines of shell code
<slh>
50 lines of code that 'needed' around 5 changes in the last 15 years (and most of those are more vanity changes than functional modifications)
<will[m]>
listen we'll just make Linux one big binary that runs out of sysctl with incomprehensible arguments and unparseable output, and then everyone will think "gosh, if only this was Microsoft(R) PowerShell(TM)"
<slh>
in other words, the effort needed to get the deprecation implemented already needed more efforts in which itself, than another 15-20 years would have required (before even looking at changing the world and nuking which usage everywhere)
<dwfreed>
will[m]: you mean systemctl
<will[m]>
i'm too old to remember the difference (i'm in my 30s)
feckert has joined #openwrt-devel
<dwfreed>
sysctl is for manipulating kernel tunables
<dwfreed>
like whether your IPv6 stack does SLAAC
<will[m]>
i know, i just never have the brainspace for it
<shibboleth>
he prolly meant systemctl
<will[m]>
init.d scripts por vida
<rsalvaterra>
will[m]: You mean, "para a vida". ;)
<will[m]>
listen i don't talk good, or computer good
Tapper has joined #openwrt-devel
<rsalvaterra>
xD
<will[m]>
i hit things with hammers until they work and then duct tape them together
<rsalvaterra>
Wait, did you just describe the entire computing field? :P
<will[m]>
exactly
* ldir
needs a sandwich
noltari has quit [Quit: Bye ~ Happy Hacking!]
noltari has joined #openwrt-devel
<will[m]>
new idea, replace `find` with `command -f` but alter the output so it returns "Not Found" for every single directory a match isn't found in
<will[m]>
dwfreed: I edited meson.mk but that seems to exist for meson's compilation not its running, so I'm unsure where that gets called, seems like it might actually be outside of the make package/glib2/compile routine
<will[m]>
also pulled latest from git and cleared/updated a bunch of stuff
floof58 has joined #openwrt-devel
<will[m]>
ok i edited meson.py in the staging_dir and printed the PATH:
<will[m]>
it's true that i don't see much ninja stuff hanging around the filesystem
<will[m]>
./staging_dir/hostpkg/lib/meson/mesonbuild/backend/ninjabackend.py exists, which is a subfolder from ./staging_dir/hostpkg/lib/meson/meson.py but hostpkg/lib isn't in that PATH, unsure if it should be?
<dwfreed>
there isn't a file named 'ninja' in that PATH? then it sounds like your buildroot is not building it
<dwfreed>
maybe try a fresh buildroot
<will[m]>
weee
<will[m]>
i pretty much tried that, what's your process for making it fresh?
<dwfreed>
burn it to the ground and start over
<dwfreed>
rm -rf
<dwfreed>
(save any local changes/branches to patches outside of it, first)
<Grommish>
Ninja moved to the core
<Grommish>
Make sure your ./script/feeds/update -i -f and then install -f -a
adschm has joined #openwrt-devel
adschm has quit [Read error: Connection reset by peer]
<will[m]>
zzzZZZzzzz unfortunately "just redo your buildroot" always seems to involve a few hours of compilation
fda has quit [Ping timeout: 480 seconds]
<karlp>
yeah, a few people live in the world of "my customized build server builds in 15min, what's the problem" but not everyone gets that luxury
valku has quit [Quit: valku]
<slh>
i single build run, with just the stuff enabled that actually goes into the image, works within quite reasonable time - shouldn't be much more than and hour on half-decent hardware from the last decade. building 'everything' or for multiple targets is another topic though
<karlp>
sure, but an hour is a _LONG TIME_
<karlp>
especially when it's "maybe try this"
<slh>
sure it is, if you need to do it more often/ regularly.
<will[m]>
i didn't need to until apparently someone was like "what if we redid meson/ninja?" :P is there a command for "just build base/packages/libraries" so i don't have to wait for make -j5 to bug out
<will[m]>
i tried building glib2 direct but it complained that ld-musl doesn't exist