maciekb has quit [Quit: Ping timeout (120 seconds)]
maciekb has joined #openwrt-devel
fakuivan has quit [Remote host closed the connection]
fakuivan has joined #openwrt-devel
Danct12 has quit [Quit: WeeChat 3.8]
Danct12 has joined #openwrt-devel
astro1 has joined #openwrt-devel
<astro1>
Thanks schmars for the help yesterday. The link you have given is for how the frontend calls. I wanted to understand how when the backend answers it. I have now got a better idea that individual packages register their keywords using UBUS_METHOD() and UBUS_METHOD_NOARG() in their own packages. I have made a search for this method and found a bunch of them.
<astro1>
Would anyone know where the packge for all the calls under system are located, i.e., where is the code that registers for the keyword in command like 'ubus call system info'?
<Znevna>
isn't ubi supposed to deal with bad blocks?:P
<mrkiko>
astro1: only a possibility, but I would look at procd
MaxSoniX has joined #openwrt-devel
<astro1>
@mrkiko: I have searched for any call to memory in package/system/procd/ and also in the system parent folder and didn't anything calling any get function for total/free memory and such over there.
<astro1>
mrkiko: Is there anywhere I could maybe search for this function that inquires memory?
<hgl>
In openwrt/gh-action-sdk, "make defconfig" is run after "feeds update -a", but before "feeds install", this seems to generate a .config containing no pkg config info. I wonder in this case, when you run "make package/foo/compile", what config is used for that package? like if it has subpkg, does CONFIG_PACKAGE_foo-mod-bar gets set to y?
<dwfreed>
defconfig should run again, so you'd end up with whatever the default is
<hgl>
dwfreed: that makes sense, thanks
<hgl>
Ansuel: I'm debugging your PR. The error probably came from code like "ifeq ($(CONFIG_PACKAGE_nginx-mod-naxsi),y)"
<hgl>
in sdk, their default value is m
<hgl>
Ansuel: I'd be happy to continue the work on the nginx PR, if you're fine with that
danitool has joined #openwrt-devel
Kufat_ has joined #openwrt-devel
Kufat has quit [Read error: Connection reset by peer]
<Mangix>
philipp64: I'm in Europe
bluew has quit [Ping timeout: 480 seconds]
astro1 has joined #openwrt-devel
<astro1>
mrkiko: Thank you so much.Exactly what I was trying to find. I just searched for the line you mentioned in my source repo and I am not able to find it in my source. Where would this procd submodule be in the complete openwrt source or do we get a binary of this during make?
<mrkiko>
astro1: well, if you already built openwrt you may search for system.c; otherwise, just clone the git repo and you'll get the code
<mrkiko>
astro1: I have it in build_dir/target-arm_cortex-a7+neon-vfpv4_musl_eabi/procd-default/procd-2023-01-16-190f13a7/system.c so you see there are some factors changing the path like the target you build for
tidalf has joined #openwrt-devel
astro1 has quit [Quit: Konversation terminated!]
tidalf_ has quit [Ping timeout: 480 seconds]
Danct12 is now known as Guest12324
Danct12 has joined #openwrt-devel
MAbeeTT4 has joined #openwrt-devel
Guest12324 has quit [Ping timeout: 480 seconds]
tidalf_ has joined #openwrt-devel
MAbeeTT3 has quit [Ping timeout: 480 seconds]
goliath has joined #openwrt-devel
tidalf has quit [Ping timeout: 480 seconds]
astro1 has joined #openwrt-devel
<astro1>
mrkiko: I don't have a procd-default in the directory you mentioned.My custom packages like hello-world package following the tutorial at https://openwrt.org/docs/guide-developer/helloworld/start got cloned over here but there is no other folder.
<astro1>
mrkiko: You mentioned that you have procd repo in build_dir/target-arm_cortex-a7+neon-vfpv4_musl_eabi/procd-default/procd-2023-01-16-190f13a7/system.c. For my source repository, in folder build_dir/target-i386_pentium4_musl/ I have a folder called helloworld-1.0 that created by following the tutorial to create one's own package at https://openwrt.org/docs/guide-developer/helloworld/start .Other than this helloworld, I only have an empty stamp folder
<astro1>
and a toolchain folder but no procd folder.
<mrkiko>
astro1: I don't know how to help; did your build complete successfully?
<mrkiko>
revolt: why not try out?
<revolt>
mirko_:
<revolt>
there is a problem coz no enought space
<revolt>
or mabe im doing something wrong,
<revolt>
mrkiko: ;)
<revolt>
mrkiko: mabe You could help ?
<mrkiko>
revolt: Huwei E3131 is NCM-based, right? Or do you have an hilink version? I don't know if an hilink version of the E3131 exist, but you never know :)
<astro1>
mrkiko: Yes. My build did complete sucessfully. My question is that it does the ubus call system info which we know is from procd repo but in my source folder, I don't see that procd has been cloned as source. I want to get to to know where this procd may have been cloned inside my main openwrt source repo.
<revolt>
mrkiko: say it,,,
<revolt>
mrkiko:
<revolt>
mrkiko: NCM-based,?
<revolt>
mrkiko: you want me to say i need reprogram my huawei ?
<mrkiko>
astro1: can't help you on that, I would use the find command
<astro1>
mrkiko: Sure. Thank you for all the help till now.
<mrkiko>
revolt: no no, absolutely. Just wanted to understand what flavor you're using... that said, I don't think I can help you much since I don't know the TP-Link device you specified well enough. I have only an idea of the E3131
<mrkiko>
astro1: yw
<revolt>
mrkiko: ok so probably i cant do much...
nixuser has quit [Ping timeout: 480 seconds]
<mrkiko>
revolt: what's the device's flash size?
djfe has joined #openwrt-devel
djfe has quit []
<revolt>
mrkiko: Free space: 39% (148.00 KB)
<revolt>
;(
<revolt>
probably i can remove some data but...
<Ansuel>
ngl what did you notice?
<Ansuel>
hgl*
<Ansuel>
oh maybe i know
<Ansuel>
let me test something
<hgl>
Ansuel: umm, are you there?
<Ansuel>
i think the problem is that CI build package with m
<hgl>
Ansuel: yes
<Ansuel>
and that conflict with the ifeq that checks for y
<hgl>
I'm also working on the PR 😂
<Ansuel>
so a simple ifneq should fix it
<Ansuel>
if the ci succeed then i will merge my version and then you can propose the macro changes you are pushing, OK for you ?
<hgl>
Ansuel: sure, hopefully there won't be much conflict
<Ansuel>
from the original pr i just changed the ifdef
<Ansuel>
ifeq*
valku has joined #openwrt-devel
<hgl>
Ansuel: I had a build where the ifeq was fixed, but then it failed with: ./configure: error: unsupported LuaJIT version; ngx_http_lua_module requires LuaJIT 2.x.
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<djfe>
Asus RT-AX54 - MT7621 device. The easy way to install right now is: flash initramfs using Asus Recovery Tool to flash and boot the device (I know hacky). If the user now calls sysupgrade -n from that rom, it tries to mount a ubi partition that hasn't been initialized, yet. https://forum.openwrt.org/t/openwrt-support-for-asus-ax1800hp/149891/40?u=djfe
<djfe>
Why is initramfs accessing/mounting mtd7 (ubi) on sysupgrade? Because it was booted from flash? Can we prevent this from happening?
<djfe>
(this is the same device Znevna mentioned this morning at 9o'clock CEST
<djfe>
The "old" way of installing is ssh into oem rom and flashing factory image with mtd-write.
nixuser has joined #openwrt-devel
<djfe>
Is initramfs broken for the device or would sysupgrade work just fine, when actually booted from ram?
<Znevna>
x.x
<Znevna>
I don't know what's happening there
<mrkiko>
djfe: if it's there a way to do it without switching to Windows, would be fine for me
astro1 has quit [Ping timeout: 480 seconds]
<djfe>
yes there is: via ssh and mtd-write and initramfs via uboot and serial
<djfe>
Ok I actually forgot that he initially flashed via ssh and it worked fine a couple weeks ago. When he tried booting two days ago, the device wouldn't come up. You can read the rest on the forum. The question for me still remains: if you sysupgrade and ask to forget config, will openwrt actually try to mount the user data partition, if so why?
<djfe>
or can anyone make sense of the verbose output of sysupgrade on the forum :)
astro1 has joined #openwrt-devel
<djfe>
We are mostly trying to figure out whether this is an overall issue that needs to be fixed or whether it's related to the users hardware somehow.
<Znevna>
obviously the user has a device with a bad block
<djfe>
shouldn't ubi work fine despite bad blocks?
astro6 has joined #openwrt-devel
tidalf has joined #openwrt-devel
astro6 has quit []
astro6 has joined #openwrt-devel
<djfe>
Maybe nmap still shifted the block underneath ubifs?
Mangix has quit [Read error: Connection reset by peer]
<djfe>
s/nmap/nmbm
Mangix has joined #openwrt-devel
astro6 has quit []
astro6 has joined #openwrt-devel
astro6 has quit []
tidalf_ has quit [Ping timeout: 480 seconds]
astro1 has quit [Ping timeout: 480 seconds]
<Ansuel>
well nice was right about the ci problem!
<hgl>
Ansuel: Do you want to base your new work on my refactor? It removed a lot of repetitive code (but not too much to force ourselves into a corner). I worry it might create huge conflict
<Ansuel>
i prefer to first merge mine and then you create a new pr for the refactor
<Ansuel>
i checked the artifacts from CI and package are correctly created
<hgl>
if you can merge without much further change, cool
<hgl>
strange, I wonder how I got the luajit2 error
<Ansuel>
i guess your conflict are related to rebase your current pr right?
<Ansuel>
or you are referring to package conflict?
KevinDB is now known as ldir
<hgl>
not sure I understand. I didn't fork your repo. I simply forked openwrt/packages, removed the net/nginx folder, and put yours there. So after you merge, I probably have to remove your folder and put in mine again.
<hgl>
If you're just fixing the ifeq, I don't think we will lose anything, I already fixed it in the refactor
<Ansuel>
yep i just changed that
<Ansuel>
stupid me for not catching that some days ago
<hgl>
nah, it's a pretty subtle bug
<Ansuel>
ok then i'm merging the pr
<hgl>
github has run out of runners, mine has been stuck for half an hour, lol
<Ansuel>
wiki should be updated to warn user of these new modules
<hgl>
let's finish the package first, and update the wiki in one go
<Ansuel>
hgl nothing stuck just waiting for a free slot runner are shared across all the org
<philipp64>
stintel: and you win the kewpie doll. apparently the "option disabled true" applied to both wan and wan6. WTAH...
<philipp64>
so how does one disable IPv6 but not IPv4? my ISP only provides v4...
<stintel>
philipp64: don't define wan6 ?
<philipp64>
Do I need "option proto none" or what?
<stintel>
but seriously change your ISP :)
<f00b4r0>
heh
<philipp64>
I'm in an underserved area, so limited choices...
<stintel>
OVH started charging monthly fee for IPv4 IPs ...
<stintel>
I'm gonna move some stuff to v6 only
<stintel>
not having IPv6 is unacceptable
<philipp64>
I've been thinking about Starlink or Unlimitedville.com ... but then I *also* will need an IPv4 statically assigned and a VPN tunnel to host my mail, web, ftp, and VoIP services... and then a DNS account... it's all such a pain.
<Ansuel>
ipv4 starvation finally started for real ?
<stintel>
Ansuel: it started many years ago but moron ISPs prefer cgnat over IPv6
<philipp64>
you'd never know it from talking to US ISP's...
<philipp64>
Yeah, Starlink uses cgnat
<philipp64>
apparently addressing *is* harder than rocket science.
<stintel>
well, they do also offer IPv6
<Ansuel>
stintel sure but as ISP use cgnat and still continue to buy ipv4 block nobody care to actually invest in switching to v6
<stintel>
but many ISPs ignored IPv6 entirely and just implemented cgnat
<stintel>
it's a total shitshow
<stintel>
I have IPv6 connectivity since at least 2002 ¯\_(ツ)_/¯
<philipp64>
stintel and schmars[m]: anyway, thank you both... mystery solved...
<Ansuel>
stintel my idea idea is that for some reason device still handle it bad...
<Ansuel>
and ISP can't configure them correctly
<stintel>
Ansuel: because it's not a priority
<stintel>
while it should be
<Ansuel>
also i read recently android not supporting dhcpv6
<Ansuel>
scenario for italy
<f00b4r0>
ipv6 is a shit show in many places, sadly
<Ansuel>
TIM dropped experimental ipv6 support (was a /64 dual stack and was served from a single BRAS in milan)
<f00b4r0>
and cgnat made sure it would be so
<Ansuel>
Fastweb offer ipv6-rd... ipv4 is cg but ipv6 needs to be manually enabled and requested to the isp
<stintel>
I heard android even handled DHCPv6 PD and assigns a dedicated prefix to mobile hotspot
<stintel>
if your mobile provider assigns you > /64
<philipp64>
proto=none
<Ansuel>
Every other ISP is ipv4 public bynamic
<stintel>
philipp64: you do not need config interface 'wan6' at all
<Ansuel>
mobile connection are all cg
<stintel>
but sure you can keep it around with proto none
<philipp64>
stintel: I know, I'm just trying to change the canned configuration with the fewest number of lines changed so that if there's a new config in the future that adds new options, it's easier to switch over...
<philipp64>
deleting an entire block isn't "low touch".
<f00b4r0>
tbh i've always found it weird to require separate interfaces for ipv4 and ipv6 in OpenWrt
<f00b4r0>
I sort of understand why it's done that way, but it's counter intuitive imho
<Ansuel>
well for ""standard"" configuration wan6 is normally a virtual interface
<Ansuel>
standard i mean a simple dhcpv6-pd
<f00b4r0>
yes. It still shows up as a separate interface in Luci
<Ansuel>
i have to ask is there a single ISP that offer ipv6 only and ipv4-over-ipv6 with cg ?
cbeznea has quit [Read error: Connection reset by peer]
<f00b4r0>
so I'm trying to use the SDK (for the first time - n00b alert), and the wiki instructions suggest it includes the toolchain, but it doesn't seem to be the case?
<Ansuel>
it is and should be already setup to use them
<tmn505>
it's in staging_dir/toolchain-whatever.
cbeznea has joined #openwrt-devel
<f00b4r0>
hmm I must have had a case of fat finger, the folder is missing
<Borromini>
f00b4r0: there's also 'dirclean' but unsure if that's less invasive (should be)
<f00b4r0>
Borromini: make -r dirclean: build failed. Please re-run make with -j1 V=s
<f00b4r0>
;P
<f00b4r0>
apparently it's trying to execute git clean in non-git repo folders
<f00b4r0>
no big deal, I'll just keep in mind not to clean :)
<Borromini>
:P
<Borromini>
sorry, been a while since I tried dirclean.
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
<f00b4r0>
np, thanks anyway :)
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
<tmn505>
f00b4r0: to clean something better use make package/foo/clean
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
<f00b4r0>
tmn505: sure but that's not ideal when you want to start from a clean slate and have a lot of packages built
MaxSoniX has quit [Remote host closed the connection]
MaxSoniX has joined #openwrt-devel
<philipp64>
anyone here stateside and using ddns with a non-static IP address?
Borromini has quit [Ping timeout: 480 seconds]
danitool has joined #openwrt-devel
<philipp64>
all... is there a guide to getting a stack trace decoding a segfault on OpenWrt? I created a VM that I'm testing a new image on, and syslog-ng is now segfaulting somewhere in Openssl but I can't tell where because of ASLR...
<Ansuel>
an idea might be disable aslr
cbeznea has quit [Quit: Leaving.]
<philipp64>
Ansuel: still need debug symbols... sigh.
gladiac has joined #openwrt-devel
<stintel>
Thu Apr 27 21:33:04 2023 daemon.err hostapd: nl80211: kernel reports: integer out of range for type '1'
<stintel>
ah netlink
<stintel>
the biggest horror to debug ever
Borromini has joined #openwrt-devel
MaxSoniX has quit [Quit: Konversation terminated!]
<philipp64>
stintel: it wouldn't be nearly as bad if it weren't a constantly moving target...
<Borromini>
robimarko: I think I fixed the breakage with the RB5009UG. Seems due to the 5.18 at803x fiber patches being modified to apply to 5.15. I returned them to their 5.18 state (so they apply cleanly on top of the QCA8081 backport). Would you mind taking a look if this is sane? It sure boots again at least :P
<tmn505>
f00b4r0: so seeing that You saw some git command when cleanig, I checked why it does that. So toplevel makefile in SDK has prepare target, introduced in 4685e4f27310 and changed in 49fda5e4d9d5. The prepare target stoped working at some point, because on SDK creation, prepare.mk is no longer copied to include directory.
c0xc__ has quit [Remote host closed the connection]
c0xc__ has joined #openwrt-devel
cbeznea has joined #openwrt-devel
<philipp64>
huh... I turned on NO_STRIP, turned on DEBUG... reinstalled libopenssl3*.ipk onto the target... I still can't get a stack trace with gdb locally.
gladiac has quit [Quit: k thx bye]
<philipp64>
given this line:
<philipp64>
[ 9602.814486] syslog-ng[7434]: segfault at 180 ip 00007f429ea0ed43 sp 00007fffa4f43818 error 6 in libssl.so.3[7f429ea03000+45000]
<philipp64>
how do I figure out the offset into the library?
tidalf has quit [Remote host closed the connection]
tidalf has joined #openwrt-devel
<robimarko>
Borromini: Great, nice to hear that it worked out
<Ansuel>
Borromini sorry if i caused any regression while backporting patch
<Ansuel>
the pr was stalled so i had to manually do things to merge it
<Borromini>
Ansuel: no worries, support is out of tree for that device, so...
<robimarko>
I am scared to even look at the new "ax" devices from MikroTik
<Borromini>
how so?
<Ansuel>
robi anyway don't know if you notice pm... sorry for wasting time with the broken rcg patch
<Ansuel>
i assume this new version should work...
<robimarko>
They are all off the shelf IPQ stuff, but no doubt they managed to do things in a way so crazy to avoid 3rd party stuff running
<Borromini>
lovely.
<Ansuel>
also i wonder if i should duplicate the rcg2 struct instead of the union
<robimarko>
Ansuel: Yeah, saw the email and building now
<robimarko>
Borromini: Not spending my money again on MTik gear
<Borromini>
I'm personally not looking any further than MediaTek stuff anymore for wireless.
<Borromini>
robimarko: fully understand.
<Ansuel>
robimarko oh god... with new IPQ stuff you can hack stuff in your custom way in some many way
<Ansuel>
since the soc is new so you have tons of security feature for the more advanced stuff
<robimarko>
Ansuel: Their security method is RouterBoot with disabled UART plus probably some new variation of YAFFS plus bothed up DTB
<Ansuel>
plus all kind of fuse blown
<robimarko>
Nah, I doubt they are using secure boot as that adds cost during manufacturing
<Borromini>
robimarko: I suppose you're still using MTik for Sartura? Or do you have a say in what gets bought
<robimarko>
ARM64 with RouterBoot so you cant pass custom DTB-s and no UART with YAFFS is enough
<robimarko>
Borromini: Nah, we are not using it
<Ansuel>
btw with the eu act about security level on software product... i expect a tragic history for devices supporting custom fw
<Borromini>
robimarko: ok :)
<robimarko>
We got it as it was nice HW that we are familiar with, but then its so annoying to use 3rd party stuff on it
rmilecki has quit [Remote host closed the connection]
rmilecki has joined #openwrt-devel
rmilecki has quit [Quit: Konversation terminated!]
robimarko has quit [Quit: Leaving]
<Lechu>
I just took ZTE MF28D on my workbench and got everything working... almost. And have a WTF moment with Ethernet switch.
<Lechu>
It works, but only if I interrupt boot sequence and let U-boot initialize Ethernet interface, and then let it continue simply by calling 'boot'. Otherwise I get stalls from ag71xx, and swconfig cannot configure AR9344 built-in switch
<Lechu>
I presume the configuration is incomplete, somehow, but have no idea how it's possible, frankly. Regardless if I try current master or 22.03.4. My tree is here: https://github.com/Leo-PL/openwrt/tree/zte-mf28d
<Lechu>
Now question: any simpler ideas to diagnose this, other than comparing outputs of devmem from boot with and without Ethernet init by u-boot?
<Lechu>
That's old device, but has a nice perk: the VoIP driver from Dragino2 can be ported to it and used with Asterisk, it has plenty of space on NAND flash.
<Ansuel>
yes it's probably the configuration those switch where in a transition period so the reg conf changed a lot
<Ansuel>
i assume uboot have a tool to dump switch regs
<PaulFertser>
Lechu: what does the switch driver say? No communication over MDIO at all?
<Ansuel>
i would compare them
<Lechu>
It seems so, "Failed to connect to switch"
<Ansuel>
maybe it's helf reset ?
<Lechu>
Ansuel: the thing is, that simply interrupting boot causes the U-boot to reconfigure the interface
<Lechu>
so I'd have to build with CONFIG_KERNEL_DEVMEM and capture from Linux
<Ansuel>
you will never know the amount of shit they put in uboot
<Lechu>
maybe disabling ag71xx.
<Ansuel>
could be that some init are skip if the autoboot continue
<Lechu>
so it doesn't interfere
<PaulFertser>
It's unlikely they added power control for the switch but connecting a reset line to GPIO sounds plausible.
goliath has joined #openwrt-devel
<Lechu>
that would be funny, given it's a built-in switch
<PaulFertser>
Ah, silly me, sorry.
<Lechu>
i.e. the AR8216 inside AR9344
<Lechu>
but no-one can rule that out. I managed to crash the board while scanning GPIOs for LEDs and modem reset
<Lechu>
I'll post the faulty bootlog in a short while