<jow>
zorun_: this should be probably applied to firewall*4* as well
mrkiko has joined #openwrt-devel
cbeznea has joined #openwrt-devel
danitool has joined #openwrt-devel
Tapper has quit [Ping timeout: 480 seconds]
robimarko has joined #openwrt-devel
robimarko has quit [Remote host closed the connection]
robimarko has joined #openwrt-devel
<Habbie>
the openwrt/rootfs image on dockerhub no longer runs
<Habbie>
(the :latest on e)
<stintel>
does it walk now?
<stintel>
SCNR
<Habbie>
well, only time will tell, but i'm not optimistic!
<Habbie>
i'm happy to debug more but first i'll just complain in here to see if anybody goes "oh wait i know why"
<Habbie>
openwrt/rootfs:x86_64-openwrt-22.03 also does not offer a working shell
<Habbie>
nor openwrt/rootfs:x86_64-22.03.0
<Habbie>
nor openwrt/rootfs:x86_64-openwrt-21.02 so now I'm wondering if it's some local problem
<Habbie>
ok, things do work on the other machine (a pi4), not on my Debian amd64 desktop, I'll do the debugging myself later
valku has joined #openwrt-devel
valku has quit [Quit: valku]
valku has joined #openwrt-devel
<russell-->
i'm seeing a weird dmesg during boot of a mediatek,linkit-smart-7688, four messages like this before reporting the spi flash partition table: "OF: Bad cell count for /palmbus@10000000/spi@b00/flash@0/partitions"
<robimarko>
I see it on all devices, apparently this explains it:https://lore.kernel.org/all/2fff1d95-c257-2756-ff63-5efe7b3c528c@gmail.com/
valku has quit []
<russell-->
the thread seems to end non-conclusively
<Habbie>
alpine switched luajit to the openresty fork some time ago, with great success
<Habbie>
would openwrt consider such a thing?
<Habbie>
(or where would i have that discussion? mailing list?)
<russell-->
rmilecki: did that issue (above) get resolved?
<robimarko>
russell--: Well, its one of those bindings issues
<robimarko>
Though to me it looks like a bug
<rmilecki>
russell--: i wasn't resolved, I did what I had time to (debugged & reported)
<rmilecki>
i won't find time to work on it anytime soon
<rmilecki>
it's harmless FWIW
<robimarko>
Its kind of annoying
<rmilecki>
i agree
<robimarko>
As size and adress cells are properly set under the partitions node
<rmilecki>
if anyone finds time to patch OF (DT), i'd really appreciate that
<robimarko>
For all of the subnodes that actually use reg properites for mapping
<russell-->
thanks!
bluew has quit [Ping timeout: 480 seconds]
Tapper has joined #openwrt-devel
GNUmoon has quit [Quit: Leaving]
GNUmoon has joined #openwrt-devel
GNUmoon has quit [Quit: Leaving]
<lynxis>
aparcar[m]: how do you got with rebuilding the release for reproducible builds? are your scripts somewhere to checkout?
<stintel>
select PACKAGE_firewall if PACKAGE_firewall4<PACKAGE_podman
<stintel>
depends on !(PACKAGE_firewall4<PACKAGE_podman && IPV6) || IPV6
<stintel>
I'm fooling around with Kconfiglib and this breaks it, happens in gnunet, podman, etherwake-nfqueue
<stintel>
trying to understand what this < means, and if it makes sense (to know if this requires fixing in OpenWrt* or Kconfiglib)
<stintel>
turned up a bunch of trailing spaces here and there, sent PRs to fix those
<stintel>
meh, also in luci-ssl, and this is something I can't really ./scripts/feeds uninstall
minimal has joined #openwrt-devel
<jow>
PACKAGE_foo<PACKAGE_bar means the condition is true if foo is set to =m or =n while bar is set to =y
<stintel>
If value of <symbol1> is respectively lower, greater, lower-or-equal, or greater-or-equal than value of <symbol2>, it returns ‘y’, otherwise ‘n’.
<jow>
so the idea of `select PACKAGE_firewall if PACKAGE_firewall4<PACKAGE_podman` is to select firewall if podman is =y and firewall4 unset or =m
<jow>
it basically enforces the firewall||firewall4 alternatives
<stintel>
ok so it makes sense and is a bug in Kconfiglib
soxrok2212_ has quit [Read error: Connection reset by peer]
<stintel>
jow: thanks
soxrok2212 has joined #openwrt-devel
torv is now known as Guest248
torv has joined #openwrt-devel
soxrok2212 has quit [Read error: Connection reset by peer]
soxrok2212 has joined #openwrt-devel
Guest248 has quit [Ping timeout: 480 seconds]
<stintel>
is the process of how to test a pending PR in a package feed documented somewhere?
torv has quit [Remote host closed the connection]
<stintel>
there are several options and I don't feel like explaining all of that in detail
<zorun_>
jow: I will revert that commit, my intention was to apply it to firewall4 only
soxrok2212 has quit [Ping timeout: 480 seconds]
<stintel>
zorun_: yw
<jow>
zorun_: keep it, makes sense to keep the policies in sync
<zorun_>
jow: ok
<zorun_>
I'll just push it to firewall4 repo then, and it can be pulled in openwrt in the next bump
Tapper has joined #openwrt-devel
<stintel>
geez, managed to segfault python with Kconfiglib ... core dump ... 42460 calls
<mrkiko>
stintel: segault python with C code or python code? I guess the answer is C but ...
<stintel>
mrkiko: python
<mrkiko>
stintel: oh! So it makes sense to report it
<stintel>
apparently happens when passing a list as an argument to .format()
<stintel>
should probably try a few line reproducer
<stintel>
ugh
<stintel>
yak shaving day
Lynx- has joined #openwrt-devel
<mrkiko>
stintel: "{}".format(["this", "is", "a", "list"]) is not enough for me to segfault it ... congratulations in any case, you managed to find interesting issue
<Lynx->
auc command to switch to master?
<Lynx->
"auc -b master" is a no go
<stintel>
mrkiko: well it's probably more complicated than that
<mrkiko>
Lynx-: ah, so in the end in master it works!
<Lynx->
mrkiko how do you mean?
<Lynx->
Ah you mean with the modem reconnect issue? I imagine so. I haven't tested. I should really test and then pull rather than badger to get pulled and then test, right?
<mrkiko>
Lynx-: I am the person who repliedto you in the MM mailing list when you asked about autoconnecting issues, if I am not mistaken
<Lynx->
namely we see modemmanager gives signal: [3gpp] regular-deactivation, and state changed (connected -> registered), I lose internet connectivity and my interfaces are still up in OpenWrt
<Lynx->
and I have to manually restart, but I think this patch ties the loop and feeds the info to netifd
<stintel>
mrkiko: LMAO - TypeError: Kconfig._warn() takes from 2 to 4 positional arguments but 20338 were given
* stintel
gives up on Kconfiglib, if it's comparing 20338 symbols surely it's going to find a duplicate dependency
<stintel>
sigh
<mrkiko>
Lynx-: so, to solve this same issue I did wrote some smallish C apps that did exactly that - monitor bearers via glib-mm, and when a bearer was disconnected, reconnect it. But nowaydays it's not useful anymore I guess - at least in the scenario you want "standard" reconnection. But i should admit I don't know if you need more support for MM and therefore you should update MM as well or if the patch
<mrkiko>
you linked is enough (haven't looked at it admittely... ). I would try master and see the results.
<Lynx->
mrkiko understood, if I try master and it works might you be convinced to pull to 22.03?
<Lynx->
I've got auc up and running and ready to press fire
<Lynx->
ah yes, MM would need updating + this patch, that's true
<mrkiko>
Lynx-: I have no way to pull :D ...
<mrkiko>
Lynx-: out of curiosity, what device are you using ??
<Lynx->
Zyxel NR7101
<mrkiko>
stintel: sure, but it shouldn't be able to segfault the interpreter anyway.
<Lynx->
It seems fantastic, would highly recommend.
<Lynx->
albeit it's not exactly cheap at 700 GBP in the UK
<mrkiko>
Lynx-: what I was worried for that device is - is it easy to fecover or not if you brick it?
<mrkiko>
Lynx-: what troughput are you able to achieve more or less? And what provider are you using?
<Lynx->
ahh, I think it can be since the interface is exposed. Vodafone UK and since my cell tower is only band 20 at 10Mhz bandwidth I max out at 70 MBit/s download and 25 Mbit/s upload
<Lynx->
I am the guy behind the 'cake-autorate' project with the help of Sebastien Moeller
<Lynx->
btw with my RT3200 I used to chop and change from 22.03 to master without any issues - would you not expect the same with this device?
<mrkiko>
Lynx-: well, I was talking about recovery if the OS doesn't boot (from bootloader)
<Lynx->
seems like it would require serial console
<Lynx->
so maybe I just need to build ModemManager package myself?
<mrkiko>
Lynx-: congrats for the project, seems cool
ecraven has left #openwrt-devel [Bye]
<Lynx->
:s maybe I'll just keep badgering and hope that works
<mrkiko>
Lynx-: serial? Didn't that device have a recovery image or so? Mhm, but yes - seems a pretty cool device, especially if you are in an areawith 5G
<Lynx->
mrkiko maybe there is an OEM recovery image in kernel2
<Lynx->
but I think serial needed to switch to that
<stintel>
but it didn't work because of some missing backports :)
<stintel>
not that big a deal, currently building with kmod-ipt-core dep
<stintel>
it's for research / testing purposes anyway and the used device has enough flash
<stintel>
it's just tickling my ocd :P
Lynx-- has joined #openwrt-devel
Lynx- is now known as Guest278
Guest278 has quit [Read error: Connection reset by peer]
Lynx-- is now known as Lynx-
Lynx-- has joined #openwrt-devel
Lynx- is now known as Guest281
Lynx-- is now known as Lynx-
Guest281 has quit [Read error: Connection reset by peer]
shibboleth has joined #openwrt-devel
Lynx- has quit [Quit: Going offline, see ya! (www.adiirc.com)]
<hauke>
stintel: If you think it is ok, I can backport all of the linked patches
<hauke>
would probably simplfy something for me at work
<stintel>
hauke: it's not really adding new features, ok new kmod-sched-drr package, but it was already in the kernel ... I'd say do it
<hauke>
stintel: ok
<vulpes2[m]>
robimarko: re: earlycon: I can certainly try that, do you happen to know what tty device should I use on an mt7621 device?
<vulpes2[m]>
Annoyingly this vendor has disabled uart in u-boot and there's no way to turn it back on, since it doesn't have a separate env partition and the env compiled into the binary. Mainline u-boot doesn't seem to boot or give any indication of life either.
<stintel>
sounds like mikrotik
<vulpes2[m]>
Nah it's Arcadyan
<vulpes2[m]>
They seem to have done a fairly okay job at physical security but the software is trash as usual. Enabling telnet was super easy.
<Habbie>
there's a few known webui exploits for non-updated arcadyans
<Habbie>
how did you enable telnet on this one?
<vulpes2[m]>
It's not exactly an "exploit" since it only works when you have a valid session
<Habbie>
some vendors do consider those exploits too
<Habbie>
(and when installing openwrt, they do come in handy)
<vulpes2[m]>
The typical arcadyan stuff tend to have an `apply_abstract.cgi`
<vulpes2[m]>
Turns out it sends the `action` parameter as dbus calls…
<Habbie>
ohh
<Habbie>
different from the old one i looked at, i think
<vulpes2[m]>
and there was a dbus snippet on the filesystem that restarts telnet
<Habbie>
which is a different exploit that also works via apply_abstract.cgi
rua has quit [Ping timeout: 480 seconds]
<vulpes2[m]>
That writeup was the one I looked at but it's not applicable to my device, I didn't have anything that would write to the global settings (they deleted most of the juicy asp htm files on this fw)
<Habbie>
right
<Habbie>
but good source of avenues :)
<vulpes2[m]>
yup
<Habbie>
you should document it somewhere
<vulpes2[m]>
There will be a write-up someday after I get openwrt running on this thing :)
<Habbie>
:)
rua has joined #openwrt-devel
<Habbie>
i have collected two more arcadyan devices (although they are identical to the one in my post, they might have newer firmware) so i might be interested
<vulpes2[m]>
Experimenting with it was pretty annoying as Firefox clears out the web requests in the inspector if the page refreshes
<Habbie>
indeed
<vulpes2[m]>
I feel like the refresh is intentional
<Habbie>
i recall reading that you can disable that
<vulpes2[m]>
I just gave up and used mitmproxy instead
<vulpes2[m]>
worked great
<Habbie>
:)
<vulpes2[m]>
I still have some level of suspicion that it's a u-boot level problem, not a kernel problem
<vulpes2[m]>
because the device was bootlooping
<vulpes2[m]>
(indicated by the LED state)
<vulpes2[m]>
if there was a KP it would likely have just sat there and never rebooted
bluew has joined #openwrt-devel
rua has quit [Remote host closed the connection]
<Slimey>
anyone ever extract EnGenius firmware
rua has joined #openwrt-devel
<vulpes2[m]>
sometimes you kind of have to pull the flash :/
<vulpes2[m]>
there is no way to obtain a valid fw update blob in my case so I had to pull the flash and dump it
shibboleth has quit [Quit: shibboleth]
csrf has joined #openwrt-devel
<robimarko>
vulpes2[m]: sorry for the late reply, no idea what is the serial port
<vulpes2[m]>
I checked the other mt7621 targets and it appears to be ttyS0