<mangix>
rsalvaterra: shared/static is completely broken with cmake
<mangix>
this is why i always suggest meson
<rsalvaterra>
Still, I'm failing to reproduce the build issue. :/
<mangix>
you need to have libunbound and libidn2
<mangix>
built
<rsalvaterra>
Yes, I enabled both.
<mangix>
run make package/stubby/{clean,compile}
<rsalvaterra>
Heh. I deleted the build directories for libunbound, libidn2, getdns, stubby and libevent2. I even did make clean, removed the tmp dir and rebuilt.
<rsalvaterra>
mangix: Have you tried to repro?
rua has quit [Ping timeout: 480 seconds]
hitech95 has quit [Remote host closed the connection]
rua has joined #openwrt-devel
<mangix>
rsalvaterra: hahaha. different error
<mangix>
/home/mangix/devstuff/openwrt/staging_dir/target-mips64el_mips64r2_64_musl/usr/lib/libidn2.so: undefined reference to `libintl_gettext'
<rsalvaterra>
WAT
<mangix>
I build with FULL_NLS
<mangix>
after adding nls.mk to the stubby makefile, I get the same missing dependencies error
<rsalvaterra>
So, I added CMAKE_OPTIONS += -DENABLE_GETDNS_STATIC_LINK=OFF to the stubby makefile. Let's see…
minimal has quit []
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
goliath has joined #openwrt-devel
<rsalvaterra>
mangix: Nope. Definitely can't reproduce. Oh, week… I'll call it a day and hit the bed now.
<rsalvaterra>
*well
tty6 has joined #openwrt-devel
tty6 has quit [Quit: Page closed]
victhor has quit [Remote host closed the connection]
dangole_ has quit [Remote host closed the connection]
fda has joined #openwrt-devel
james_su has joined #openwrt-devel
fda- has quit [Ping timeout: 480 seconds]
james_su_ has joined #openwrt-devel
fda- has joined #openwrt-devel
<james_su_>
Hi All, happy new year.
fda has quit [Ping timeout: 480 seconds]
<james_su_>
I have my own AR9344-based device with OpenWrt 15.03 running stably in it. I am working to migrate it to 19.07. Now I am facing with an issue. I already made a 19.07 stable build for my device; however each time when I delete the staging_dir to let it rebuild a new image, I would get a kernel panic during this new image bootup period as below\
Luke-Jr has quit [Ping timeout: 480 seconds]
<james_su_>
[ 2.947195] NET: Registered protocol family 10 [ 2.957115] Segment Routing with IPv6 [ 2.960942] NET: Registered protocol family 17 [ 2.966109] 8021q: 802.1Q VLAN Support v1.8 [ 2.975569] VFS: Mounted root (squashfs filesystem) readonly on device 31:3. [ 2.983739] Freeing unused kernel memory: 208K [ 2.988292] This architecture does not have kernel memory protection. [ 3.621419] here here Kernel panic - not synci
<neggles>
james_su_: are you deleting build_dir as well?
<neggles>
need to delete build_dir + staging_dir + tmp all at once, but you also shouldn't need to do that in general
Luke-Jr has quit [Ping timeout: 480 seconds]
Luke-Jr has joined #openwrt-devel
aiyion has quit [Ping timeout: 480 seconds]
aiyion has joined #openwrt-devel
jlsalvador has quit [Quit: jlsalvador]
jlsalvador has joined #openwrt-devel
<Grommish>
james_su_: to scrub the system, I use: rm -rf bin/ build_dir/ staging_dir/ tmp/ .ccache/ (if you use ccache).. That'll start you as fresh as possible
<Grommish>
But, if you've hosed the system, you can't beat it (even at the expense of rebuilding the tools and toolchain)
valku has quit [Quit: valku]
fda has joined #openwrt-devel
ecloud has quit [Ping timeout: 480 seconds]
<james_su_>
Thanks Grommish and neggles, yes, I also tried to get a total fresh build, however still got kernel panic. Seems that only the buildroot in the initial backup works, once it is rebuilt totally, the kenel panic would appear.
fda- has quit [Ping timeout: 480 seconds]
ecloud has joined #openwrt-devel
nitroshift has joined #openwrt-devel
<PaulFertser>
james_su_: you should pastebin full output somewhere and give a link here. Do not paste to the channel.
<PaulFertser>
james_su_: also you should be probably targetting current master rather than old 19.07 branch.
<PaulFertser>
james_su_: also, build with KERNEL_KALLSYMS to have an easier to use backtrace.
<PaulFertser>
james_su_: are you using normal 19.07 procd as init?
Guest5844 has quit [Remote host closed the connection]
neoraider has joined #openwrt-devel
danitool has joined #openwrt-devel
nlowe has joined #openwrt-devel
nlowe has quit []
nlowe has joined #openwrt-devel
nlowe has quit [Remote host closed the connection]
<james_su_>
I have tried current 19.07 master and 19.07.8
<james_su_>
I will try KERNEL_KALLSYMS to see if I could get more thanks
ecloud has quit [Ping timeout: 480 seconds]
nitroshift_ has joined #openwrt-devel
nitroshift has quit [Read error: Connection reset by peer]
ecloud has joined #openwrt-devel
f00b4r0 has joined #openwrt-devel
<james_su_>
@PaulFertser, I realise that CONFIG_KERNEL_KALLSYMS=y is already enabled in my build. Please refer to the link I mentioned above for the backtrace.
<aiyion>
I've been reading around 802.11ah (halow) in the past few days; and found that openwrt does apparently support it already in some way:
<aiyion>
Can any of you recommend wiki aticles or similar, that reflect the state of 802.11ah in openwrt?
<aiyion>
*articles
cptn3m0 has joined #openwrt-devel
<cptn3m0>
does the image builder support custom kernel config file like buildroot
<jow>
no
<jow>
it can only repack the rootfs from precompiled artifacts
<neggles>
ERROR: new format image overwritten - must RESET the board to recover
* neggles
sighs
rua has quit [Ping timeout: 480 seconds]
rmilecki has joined #openwrt-devel
rua has joined #openwrt-devel
cptn3m0 has quit [Remote host closed the connection]
nitroshift has joined #openwrt-devel
nitroshift has quit [Read error: Connection reset by peer]
nitroshift_ has quit [Read error: Connection reset by peer]
<PaulFertser>
james_su_: why are you not on OpenWrt master?
<PaulFertser>
james_su_: why are you building for ar71xx?
<PaulFertser>
james_su_: it looks like procd is getting killed, have you tried running with init=/bin/sh instead?
dangole_ has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
<neggles>
hurricos: if you thought the AP330 u-boot was bad, the sophos APX530 u-boot's gunzip() function only allocates an 8MiB buffer
<neggles>
it's ipq8064. with 512MB of NAND and 1GB of RAM.
<neggles>
*why*
<jow>
because its stupid code written by stupid people
<jow>
people who don't care about the quality of their work because they're badly paid and under time pressure
<stintel>
the main problem here is capitalism ;)
<neggles>
stintel: preach
rua has joined #openwrt-devel
<neggles>
they use uncompressed images for their own firmware, so it doesn't get used - I assume 8mb is just the default in their ancient u-boot build
<jow>
which brings me back to "don't care about the quality of their work"
<jow>
if they don't use it, why leave it enabled
<jow>
if you use it anyway, why unsuitable defaults
<f00b4r0>
a ploy to make us miserable? :)
<jow>
ignorance
mattytap has joined #openwrt-devel
<rsalvaterra>
stintel: Amen!
<f00b4r0>
speaking of ignorance, I'm trying to find doc on the new device vs interface configuration syntax in config/network but the wiki is somewhat lacking. Is it possible to assign arbitrary _device_ names to switch vlans? I.e. could I setup eth0.2 to appear as "vlan2" instead?
<jow>
yes
<f00b4r0>
ok. how? :)
<jow>
config device; option type 8021q; option ifname eth0; option vid 2; option name vlan2
<jow>
ifname = base device
<jow>
name = target name
<f00b4r0>
oh, that works with switch_vlan too? I was under the impression target name was not modifiable there. Thanks!
<jow>
yeah, switch_vlan just programs the switch IC it does not care how you name the linux netdev that grabs the tagged frames off the cpu port trunk
<f00b4r0>
i see, that makes sense now! Thanks again
victhor has joined #openwrt-devel
Brillie has joined #openwrt-devel
Brillie has quit [Remote host closed the connection]
nitroshift has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
nitroshift_ has joined #openwrt-devel
jlsalvador has quit [Quit: jlsalvador]
jlsalvador has joined #openwrt-devel
nitroshift has quit [Ping timeout: 480 seconds]
<james_su_>
Thanks PaulFertser, I will try init=/bin/sh and let you know the result.
<PaulFertser>
james_su_: why are you not porting to ath79?
<PaulFertser>
james_su_: is there anything special at all about the way you build OpenWrt? procd dying early is unusual. Probably you tweak some kernel config options from defaults?
nitroshift_ has quit [Quit: Gone that way --->]
yrrips- has joined #openwrt-devel
<james_su_>
@PaulFertser, so far my strategy is to migrate to AR71xx as before so that I could reuse those source code as before. Actually this is just a "add a new device support to OpenWrt" as normal. Yes, "procd dying early is unusual" which may me quite confused. But as you know, I did have one stable build in one VM to get it run up.
yrrips has quit [Ping timeout: 480 seconds]
<PaulFertser>
james_su_: do you have any kernel mods or anything? What's good about ar71xx compared to simply writing a DTS for your board?
<PaulFertser>
james_su_: I do not see any sense in adding to ar71xx, it's unmaintained since long. And it doesn't seem to provide any advantage anyway.
<stintel>
we are not accepting new ar71xx devices
<stintel>
in fact, ar71xx target does not exist in master anymore
nitroshift has joined #openwrt-devel
<james_su_>
@stintel, I know. thanks. This is why I didn't select ar79 at this moment and select 19.07 for ar71xx.
<PaulFertser>
james_su_: you never told why you stick to ar71xx.
<james_su_>
Hi PaulFertser, there is no special reason to stay in ar71xx. just because we used to use ar71xx in 15.03 and think 19.07 still support ar71xx.
<PaulFertser>
james_su_: frankly speaking, I'm afraid you do not plan to contribute to the OpenWrt community if you have reasoning like that.
<neggles>
james_su_: just to be clear, ath79 == ar71xx. they are the same target for the same devices, but ath79 uses modern device configuration (which is a lot easier!)
<stintel>
also, 19.07 is scheduled for EOL in just 2 months
<neggles>
if a device worked on ar71xx it can be made to work on ath79
<mattytap>
Talking of EOL 19.07, is it possible to port the Meraki MR18 to ath79?
<Habbie>
stintel, is that EOL anywhere on the website?
<stintel>
if a device wasn't ported, none of the people owning such devices cared enough to do so
<PaulFertser>
mattytap: if I had it I'd port it right away I think.
<mattytap>
Ok - well I've got two so I will have a crack at MR18 it once we have WatchGard M200 sorted.
<stintel>
<3
<mattytap>
I have three still on 19.07
<neggles>
ipq806x is doing my gd head in
<mattytap>
they are same price as MR16 (whih I also have on 21.xx) and much more plentyful as Cisco gave these away at conferences
<PaulFertser>
It's probably not that straightforward to make it work as it mentions some strange SGMII/SERDES quirk.
<PaulFertser>
But doable.
<PaulFertser>
Wished I got one :/
<neggles>
802.11n is laaaaaaaame though (I haven't recently spent far too many hours on the aerohive AP330 what are you talking about </s>)
<mattytap>
Question on dts and dtsi files. IMHO the WatchGuard T30-W should have a brand new dtsi to replace p1020RDB.dtsi. WatchGaurd say the T30 is p1020RDB compatable but use different switch & flash
<mattytap>
It looks way better as a stub, what is the policy?
<neggles>
mattytap: have a look at the aerohive ap330
<neggles>
there's pre- and post- dtsis
<neggles>
it's weird but that's how freescale/nxp roll
<neggles>
the AP330 is also based off the p1020rdb
<mattytap>
thanks neggles.
<neggles>
maybe i *should* pick up that T35-W...
<mattytap>
I've been testing PR4697 as I have been using my own T30 build for 6 months now and I'm wanting to write it 'properly'
<stintel>
avoid duplicate code when possible I'd say, so if you can reuse existing dtsi and only have to override a couple of things that's better I think
<mattytap>
there are two T35 on UK ebay at moment but no power supplies which make them a bit useless (52volt) unless you happen to have one
<mattytap>
hi stintel. there is no freescale config for this marvell switch off the mpcxx target so it needs writing, if this goes in a new dtsi then it can be used by multiple devices
<mattytap>
there are examples under arm
<stintel>
there was also no config for the marvell switch in the m300, I did not put it in a dtsi either
<stintel>
maybe it might make sense to move it out of the m300.dts and in a dtsi to reuse for the m200 (if they are compatible)
<mattytap>
good point, Ive already been referring to your m300.dts for guidance on formatting/standards etc :P. I will be to see if the M200 and T30 use the Marvell switch in the same way and if so move to dtsi (which Ive already written for myself)
<stintel>
by all means, if it can be reused, please extract it. and the m300 dts ... I just massaged it until it worked, it's not necessarily the best way, but I tried to avoid duplicating existing code and it works
<mattytap>
I need some help in interacting with the PR though. My son knows git so will ask him
<stintel>
I should probably consider sending that dts upstream, at the risk of being flamed for doing horrible things :P
<mattytap>
I agree, up until now Ive just fiddled with fdt and reversing the dtb to get something to work. Ive followed the marvell.txt off linux docs for the dtsi
<mattytap>
makes it easier to review though
nitroshift_ has joined #openwrt-devel
nitroshift has quit [Ping timeout: 480 seconds]
nitroshift_ has quit [Quit: Gone that way --->]
hitech95 has joined #openwrt-devel
ecloud has quit [Ping timeout: 480 seconds]
ecloud has joined #openwrt-devel
Borromini has joined #openwrt-devel
valku has joined #openwrt-devel
nlowe has joined #openwrt-devel
victhor has quit [Remote host closed the connection]
svlobanov has joined #openwrt-devel
<svlobanov>
hi. which OS is used for runners for packages CI? as I understood, it is "ubuntu-latest" but I would like to get exact version. and which arch (amd64?)
<hurricos>
Is that an index? Is it the cfg\d{5} you see in uci show?
<hurricos>
Similarly, git-blame on /lib/functions.sh is useless. Nobody gives a concrete example.
hitech95 has quit [Remote host closed the connection]
<hurricos>
Oh no.
<hurricos>
it's Jo's commit message which misled me
<hurricos>
4a6795409 suggests using 'config get'
<hurricos>
s/'config get'/'config_load'/
<hurricos>
s/'config_load'/'config load'/, sorry.
<mattytap>
hurricos: I should have asked about the MR18 12 months ago! A mystery solved. Anyway I an confirm the MR16 build is rock solid - thanks
<hurricos>
config load sets no shell variables, config_load does
<hurricos>
mattytap: It's a nice device to be sure, just a little underpowered these days.
<hurricos>
OK, and there it is. the ID is e.g. cfg016cb7. This lets you do all sorts of magic with config_load it looks like
<hurricos>
so you can't collide between config_loads on accident
<mattytap>
I will let you know when I get round to porting. I've been bidding on MR33 and MR36 recently but as I had an MR18 a couple of years ago I used the JTAG method to get it working again and this was my way into OenWrt.
<hurricos>
yeah, needing to use a JTAG is not something I'm a fan of
<hurricos>
but it's populated and documented, so it's not that bad.
svlobanov has quit []
<mattytap>
hurricos: when you said tihs is done at 16:34UTC to what are you relying to?
<hurricos>
RE: AP330? I finished a build and it got synced to the web endpoint I linked.
<PaulFertser>
mattytap: that some MR18 support is available, but not upstream yet.
<f00b4r0>
can someone tell me when I try to create a named section with uci like this: https://pastebin.com/ZGKttZXx it does what I want but *also* add a spurious unnamed and empty "test" section as well?
Luke-Jr has quit [Ping timeout: 480 seconds]
<f00b4r0>
oh i see. I don't need to "add" a named section
dedeckeh has joined #openwrt-devel
linusw_ has joined #openwrt-devel
Luke-Jr has joined #openwrt-devel
svlobanov has joined #openwrt-devel
<svlobanov>
is it possible to trigger package CI to start building package without a PR?
<Habbie>
hi svlobanov
<Habbie>
you can PR against your own fork
<Habbie>
i wondered about why it was so tied to PRs too
<Pepes>
svlobanov: Yes, you need to fork the repository, edit the GitHub actions file and thats it. No need to open PR against your own fork.
<Habbie>
but the tools need to figure out the openwrt version and they do that by looking at the base branch
<Habbie>
so if you want to do it without a PR anywhere, you need to do a bit more work
<Pepes>
Habbie: Well, it is not so hard to specify which OpenWrt version it should use
<Habbie>
correct
<Habbie>
but i find that not modifying those files makes it easier to do the actual PR later
<Habbie>
either approach certainly works
<svlobanov>
Pepes: what exactly I need to change? sorry for this stupid question
<Habbie>
svlobanov, take the easy road - PR against your own repo :)
<svlobanov>
I tries to build package/feeds/packages/node , but it fails on debian11 and ubuntu20.04, but openwrt CI was able to build it 2-3 weeks ago
<Pepes>
it is forked from packages repository
<Habbie>
ah, you have a version matrix
<Habbie>
that's clever
<svlobanov>
do I need one more github account to send a PR to my fork of packages?
<Habbie>
svlobanov, no
pmelange has quit [Read error: Connection reset by peer]
<Pepes>
Habbie: Thanks! :-) It's too pity that it is no longer used.
<svlobanov>
ok I will try to send a PR to myself
pmelange has joined #openwrt-devel
<Habbie>
Pepes, do you ever go to the CZNIC offices?
<Pepes>
Habbie: Yes. Why are you asking? :-D
<Habbie>
because i sent a box of candy there two weeks ago
<Habbie>
and you should make sure you get some
<Habbie>
:)
<Pepes>
Oh, we have those in kitchen! There was some coffee candy, etc. :D
<Habbie>
yes!
<Habbie>
good to hear
<Pepes>
I wondered who put it there:D
<Habbie>
:D
<Habbie>
did you like them?
digitalcircuit has quit [Quit: Signing off from Quassel - see ya!]
digitalcircuit has joined #openwrt-devel
<Pepes>
I think we are getting on offtopic here, let's hope it does not mind. Yeah, coffee was nice, but what I dont like is jelly candies, but that's my personal opinion. I tried I guess 3 candies today and that was it. I can make picture that it arrived safely, but I am not going tomorrow to the office.
minimal has quit []
<svlobanov>
Habbie: do I need to setup my own runner or I can use github runner?
<Habbie>
svlobanov, github, may need to go to the Actions tab and enable it once
<svlobanov>
it was activated, but actions don't work
<neggles>
hurricos: nope, their production kernels are uncompressed and almost exactly 16MiB
<neggles>
the firmware runs entirely out of RAM (save for one config cache file) in the OEM setup
<Slimey>
neggles what model is that
<neggles>
Slimey: sophos APX530, ipq8064
<Slimey>
ah
<neggles>
it’s got 512MiB of NAND and 1GB of RAM and it barely uses any of it
<Slimey>
:P
<neggles>
nand is all one big ubi partition with four vols
<Slimey>
sounds like some of the bsap-xxxx stuff i have lol
nlowe has joined #openwrt-devel
<neggles>
two 64MB FIT images with one kernel+initrd and 12 or so DTBs in them, a config cache partition, and a download cache partition for some reason
<Slimey>
yup sounds familiar heh
nlowe has quit []
<neggles>
i have OpenWRT booting now, I think - but I’ve goofed the device tree so I am not getting a console yet 😝
<neggles>
the SPI NAND it does initial boot from has _twelve_ partitions
<neggles>
in 4MiB
<Habbie>
wow
<neggles>
at least three of them aren’t even used
<hurricos>
hah 12 partitions
<hurricos>
have you seen chromeos?
<dwfreed>
chromeos no longer uses 20+ partitions
<neggles>
on the plus side the GPL tarball they provide is 100% complete and buildable, they include the private key for their “development mode” signing cert, and you can disable image sig verification with `setenv verify no`
<neggles>
hurricos: yes, yes I have
<dwfreed>
chromeos now uses 12 partitions
<neggles>
hahahaha
<neggles>
on x86_64?
<neggles>
i haven’t checked how many are in this ARM64 chromebook I have yet
<dwfreed>
yeah
<dwfreed>
it used to be 27, though
<neggles>
oof
<dwfreed>
had to dig out the post from Neverware where they talked about moving from 27 to 12
<neggles>
on an 8mib chip?
<dwfreed>
no, for ChromeOS
<neggles>
oh dear
<neggles>
that’s
<neggles>
why
<dwfreed>
redundancy, to an extent
<dwfreed>
and far too much splitting of data into different partitions
nlowe has joined #openwrt-devel
nlowe has quit []
<neggles>
dwfreed: ah ok, so they took a few too many leaves out of android’s playbook
<dwfreed>
basically
<dwfreed>
Or more likely the other way around
<dwfreed>
iirc, Android did not have redundancy until long after ChromeOS did
<neggles>
Android had a silly number of partitions from the get-go, even without redundancy
<neggles>
ok that’s actually kind of neat, they do A/B kernel/rootfs/recoveryFS but never actually update A
<neggles>
so it can’t get broken
<neggles>
and two of the partitions are just “maybe we will want kernel+rootfs C later”
nlowe has joined #openwrt-devel
<dwfreed>
neggles: it should be alternating which gets updated
<dangole>
sintel: I think I have fixed 2500Base-T on the UniFi 6 LR :) Try my staging tree, unifi6lr-2500Base-T branch: https://git.openwrt.org/?p=openwrt/staging/dangole.git;a=shortlog;h=refs/heads/unifi6lr-2500baseT