<dwfreed>
jow: I'm not sure disabling kernel v6 would be possible, because that would still hit the issue with nft_reject_ipv6 not being built
<dwfreed>
I think that's the real underlying cause of #9580; openwrt CONFIG_IPV6=n results in kernel CONFIG_IPV6=n, and then nft_reject_ipv6 doesn't get built and the build fails
danitool has quit [Remote host closed the connection]
dangole has quit [Ping timeout: 480 seconds]
rua has quit [Quit: Leaving.]
valku has quit [Quit: valku]
minimal has quit [Quit: Leaving]
ptudor has quit [Quit: Strict-Transport-Security: max-age=48211200; preload]
ptudor has joined #openwrt-devel
ptudor_ has joined #openwrt-devel
ptudor has quit [Ping timeout: 480 seconds]
f0g has joined #openwrt-devel
<f0g>
Hi all. After reading the implementation of uci system, I have a question about the design. Why uci system write changes to /tmp/.uci/<package> instead of write to the /etc/config/<package> file directly? It looks like 'commit' is a useless step for generating config files for services.
<f0g>
To update a service, I need 'uci set <package>...' first, and then 'uci commit <package>', excuting '/etc/init.d/<package>'
<jayk>
its a symlink
<jayk>
perhaps yooooounix isnt for you, try luci!
<f0g>
jayk: do you talk to me?
<mrkiko>
I have an HLOS partition dump for ipq4019 (gl-b2200). I know there are multiple DTS files in the image, for different models. I don't remember how to extract them
<dhewg>
jow: do you have time to check the remaining iwinfo patches? While I recently added 3 more, there's not much left. It would be nice to have luci support 6g out of the box
<dhewg>
or just ack them and I'll poke someone else then, that'll be fine as well
<dhewg>
(got a yelling kid in the bg too ;)
mrkiko has quit [Remote host closed the connection]
cbeznea has quit [Quit: Leaving.]
robimarko has joined #openwrt-devel
rua has joined #openwrt-devel
<f00b4r0>
ynezz: hi, just curious: any hope to test buildbot changes soon?
Strykar has quit [Quit: /quit]
Strykar has joined #openwrt-devel
mrkiko has joined #openwrt-devel
goliath has joined #openwrt-devel
<mrkiko>
I'm back :D I would need some help
<mrkiko>
needing proper nandsim parameters to emulate a nand device which is 128 MiB in size, has a 128 KiB blocksize, a 2048 page size and 128 OOB size
<mrkiko>
for now googling wasn't so useful, because I guess I should find a way to calculate the proper IDs to pass to nandsim
<robimarko>
Its easier to just find the NAND model you dumped from
<robimarko>
And read the ID-s from the datasheet
dangole has joined #openwrt-devel
cbeznea has joined #openwrt-devel
rua has quit [Quit: Leaving.]
rua has joined #openwrt-devel
<mrkiko>
robimarko: mhm, from the datasheet would be a problem for me ... or at least sometimes it can be. Now I am looking if there is a way to let the kernel tell me or maybe just look in the source code
minimal has joined #openwrt-devel
<robimarko>
mrkiko: Well, that is gonna be way easiert than trying to figure out the ONFI spec
<mrkiko>
robimarko: sure. My problem is sometimes I have a hard time understanding how informations are laid in PDF files. Luckily most datasheet contain also ascii data and not only images, otherwise reading them with screen reader becomes harder
<robimarko>
mrkiko: Yeah, I get your issue
<robimarko>
As most datasheets have completely random layout
<robimarko>
If you know the NAND model, I can get the info from the datasheet for you
<robimarko>
Cause, 128 OOB is rare so I dont know any model on top of head
<mrkiko>
robimarko: I know it's gigadevice, GD5F1GQ1UC
<robimarko>
mrkiko: Oh, SPI-NAND
<mrkiko>
robimarko: yes
<robimarko>
Yeah, I dont think that is gonna work with nandsim
<mrkiko>
robimarko: what I am trying to do here is read the UBI file of the original firmware to understand better how they handle the LTE modem... and to do so what I was doing is follow the "classic" nandsim approach + modprobe ubi mtd= ... but I need to simulate e alfsh with the same characteristics of course
<robimarko>
mrkiko: There is no need to bother emulating NAND then
<robimarko>
Why not just extract the UBI images with ubireader?
<mrkiko>
well, because I fail to compile python-lzo at the moment
<robimarko>
What distro are you using?
<robimarko>
Almost all ship python-lzo already
<robimarko>
Or you can just use pio
<robimarko>
*pip
csharper2005 has quit [Ping timeout: 480 seconds]
<mrkiko>
Archlinux. You have a point. Using pip now, and all went well so far.
<robimarko>
You can also install the ubireader with pip
<robimarko>
That is how I have done it and its been a crazy good tool so far
<mrkiko>
robimarko: I knew about it, but in the past I had problems with it and resortd to the nandsim ... but giving it another try
<mrkiko>
at least this is what I remember but who knows
<mrkiko>
robimarko: however, if the kernel knows these IDs, it would be very useful to have them available somwehwere
robimarko has quit [Remote host closed the connection]
robimarko has joined #openwrt-devel
mrkiko has joined #openwrt-devel
xes has quit [Quit: bye..]
<oliv3r[m]>
So I finally figure dout, that we cannot use NO_EXCEPT_FILL with our current bootloader setup. However, it (seems for now) that this is the only option holding us back from using GENERIC_MIPS_KERNEL. Is it worth-while to investigate this deeper, and try to converge on the generic kernel, or 'screw it, just define a custom target, there's tons already'. I think moving this flag out of the generic kernel config, (or make it something
<oliv3r[m]>
fancyfull might be worth while ...) thoughts?
f0g has quit [Remote host closed the connection]
cbeznea1 has joined #openwrt-devel
<robimarko>
Anybody knows where I could rent MacOS VPS per hour?
<Habbie>
github actions has macos runners, depending on your needs, that might be something, or not
<robimarko>
That is what I am trying now but its painfully slow
<f00b4r0>
robimarko: depending on your needs a vm might be an option. IIRC macOS can be run in e.g. VirtualBox
jenders has joined #openwrt-devel
jenders has quit [Remote host closed the connection]
dangole has quit [Ping timeout: 480 seconds]
<robimarko>
f00b4r0: I basically just need a way to test macOS changes for OpenWrt in a faster way than waiting hour and half for GH to finish
<stintel>
get a mac mini :P
<jayk>
macos can barely run in virtualbox
<jayk>
better to use vmware workstation pro with DrDonk unlocker
<f00b4r0>
jayk: ok. I wouldn't know, I run native ;D
* stintel
just got an apple tv 4k 128GB to be able to upgrade his eve thread devices to the firmware with matter support
<stintel>
this requires said apple tv or homepod with thread support
<stintel>
I'm hoping to figure out how to trick the eve app into believing there is an apple thread border router in the network to do the upgrade without an actual apple thread border router
<stintel>
aka next rabbit hole
<f00b4r0>
stintel: at some point I start wondering what's the ratio of time saved thanks to your setup vs time spent getting things to work xD
* f00b4r0
hides
<stintel>
:P
<stintel>
this thread/matter stuff is just too new
<stintel>
I could just upgrade both my eve devices with the apple tv and be done with it but where's the fun in that
<f00b4r0>
bleeding edge got you ;)
<stintel>
as usual ;)
<jayk>
f00b4r0: thats the way to go!
<robimarko>
stintel: I strongly dislike Apple
<robimarko>
So no way I am buying anything, especially not to run OpenWrt CI
<stintel>
saying I strongly dislike apple would be an understatement ;)
<Habbie>
yet you keep giving them money
<stintel>
the ipad and the apple tv are for the greater good
<stintel>
however I'm still considering an m2 macbook air because everything just sucks
<Habbie>
your girlfriend?
<stintel>
well the apple tv will be a stop-gap solution until I introduce a "home" VLAN and SSID for her to connect to so she can access the other smart/multimedia/... stuff
<Habbie>
hehe
<stintel>
but no mostly because currently I need it to continue my experimenting with thread+matter (need to be able to upgrade my eve devices)
<robimarko>
We all spend decent money on experimenting, but I aint funding Apple
<robimarko>
Dont really have a "favorite" company
<hauke>
robimarko: you by QCA stuff ;-)
<jayk>
apple is probably heavily infiltrated like twtr
<jayk>
but i dont mind using an unlocker on linux or windows to use macos/xcode to unlock or turn on dev mode on their devices
<minimal>
jayk: infiltrated?
hanetzer1 has joined #openwrt-devel
<robimarko>
hauke: Yeah, cant avoid them for now
<robimarko>
But I have MTK stuff as well
<robimarko>
Heck even RTL
Ryncewynd has quit [Read error: Connection reset by peer]
hanetzer has quit [Remote host closed the connection]
tchebb_ has joined #openwrt-devel
hanetzer has joined #openwrt-devel
<mrkiko>
so, in general i'm kind of pragmatic. but the fact I would need a signed iboot to use my hardware makes me little bit scared. What if they change their minds at some point ? Ok, I can avoid updating, but it would make me feel little bit ... pressured.
<mrkiko>
that said, I would avoid funding Apple if at all possible for their behaviour. due to how well they implement accessibility features tough, I end up using their devices (mostly phone)
tchebb has quit [Ping timeout: 480 seconds]
GNUmoon has quit [Read error: Connection reset by peer]
GNUmoon has joined #openwrt-devel
* f00b4r0
is a happy user of Stuff That Just Works™ and thinks there isn't a better alternative, would have made the switch otherwise
<Znevna>
what stuff just works
<f00b4r0>
the fruity kind.
<f00b4r0>
stintel: in other news, I had the immense surprise of receiving today what is probably my first ever DMARC detailed failure report (ruf) following my last email to the m-l, which I sent from the wrong address.
<f00b4r0>
i think it's due to the list breaking my domain's signature (as expected), so the report is relatively useless, as you'd expect
tchebb has joined #openwrt-devel
<slh>
robimarko: you can also run MacOS under qemu/kvm, never looked any deeper into it, but it works
<jayk>
its a pain in the ass to get macos to work in qemu, vbox, or vmware
<slh>
jayk: I played with it for around a day, yes, there were some quirks about orchestrating qemu properly, but it works (I never looked deeper into giving it a correct/ accepted Apple ID though and lost interest very quickly, not my kind of OS)
<jayk>
i found less support for qemu
<jayk>
virtualbox was slow af
<jayk>
vmware works but you got to add a lot of tweaks
<jayk>
but i was able to sign into the apple store, plug in my iphone, and run xcode with vmware workstation pro
<robimarko>
jayk: you know where you can rent a VPS by hour?
<jayk>
not really
<jayk>
free trial maybe
<zorun>
most VPS providers bill by the hour
<robimarko>
Only found one, and its gonna take multiple days for it to be active
<zorun>
"some users have positive experience with WSL and macOS, but those are not officially supported."
<jayk>
i wouldnt ever go the macos route to do that
cbeznea1 has quit [Quit: Leaving.]
<robimarko>
zorun: I personally dont need it to work, but we have CI test for it and I want to fix that
<robimarko>
No idea why, other than its annoying me
minimal has quit [Quit: Leaving]
<f00b4r0>
robimarko: if you had a suitable PC build to spare, I could send you an image that would boot into OSX :)
* f00b4r0
runs a few hackintoshes
<f00b4r0>
robimarko: also, glad to see we found another point to agree on (re coreutils ;)
<stintel>
r17753-163df797f6
<stintel>
damn that's old
<f00b4r0>
old? Seems newer than 21.02, or I'm not understanding these tags
<stintel>
we're >20000 agtm
<stintel>
iirc
<f00b4r0>
OpenWrt 21.02.5, r16688-fa9a932fdb
<f00b4r0>
that's my router here
<stintel>
ancient
<f00b4r0>
heh
<f00b4r0>
I keep forgetting. "Bleeding edge" xD
<stintel>
;)
<Mangix>
zorun: nbd uses macOS therefore it must be supported.
<stintel>
f00b4r0: I think I reverted the dmarc stuff for my @linu-ipv6.be domain
<f00b4r0>
stintel: sounds smart. I had forgotten I still had a ruf setting on this particular domain
<f00b4r0>
that really is the first and only ruf report I ever received. I didn't even think those existed. Now I know they're real :)
<stintel>
gonna be hooking up the apple tv and my ipad to a beryl and run some tcpdump writing pcap to sdcard hoping I can capture everything I need :)
<robimarko>
f00b4r0: I aint that invested to repurpose HW for hackintosh
<robimarko>
Yeah, requiring OpenSSL just for base64 is crazy
<f00b4r0>
robimarko: ok; the offer is there. In fact I might be able to set you up on one of my machines if your requirements aren't too exotic. But that'll have to wait, they're all powered off atm.
<robimarko>
stintel: What MTK device with AX and 128MB of RAM you have
<stintel>
eap615-wall
<robimarko>
Oh, good old MT7621
<Mangix>
Old and reliable apparently
<f00b4r0>
i have a yuncore AX820 and yes, it seems to do AX "just fine" (at least according to client)
<robimarko>
f00br40: Dont bother, I will give it couple more tries in GH actions
<stintel>
the apple tv remote with usb-c charging port is cute
<f00b4r0>
robimarko: re base64, I was tempted to suggest packaging a bespoke decoder. Should be a few hundred lines tops. Heck, I bet ChatGPT can write one ;D
<robimarko>
f00br40: I was thinking of suggesting the same
<robimarko>
But since its quite literaly just base64 and no post processing its kind of overkill
<f00b4r0>
oh, that's even easier
<f00b4r0>
I wonder if it could simply be done in shell then
<Mangix>
Apparently base32 is a thing
<slh>
well, I was suggesting openssl just because it's already in the main repo, so no additional maintenance at all
<Habbie>
and base36, and base36hex.. terrible things
<Mangix>
LOL
<Habbie>
or was it base32hex
<slh>
I did use it that way myself for some other tasks (encoding the certificates for strongswan) in the past, just because it was there
<robimarko>
slh: Yeah, but forces you to include openssl just for base64 sake
<Mangix>
base32. Apparently all the 2 factor auth URIs use base32
<slh>
sure, it's not nice - but not an issue with 4 GB eMMC either
<robimarko>
Agreed, but its really an overkill considering that even busybox has base64 applet
<robimarko>
But its not enabled by default
<f00b4r0>
robimarko: actually, I'm pretty sure it can be done with awk
<f00b4r0>
robimarko: we're going to get back to busybox applets that are or aren't enabled by default. Remember the telnet can of worms? ;P
<f00b4r0>
Habbie: yeah, i'd have been very surprised if it didn't work
<Habbie>
f00b4r0, ack - i had no idea what to expect :)
<f00b4r0>
Mangix: nice, but rather bigger. Also probably much slower :)
<Mangix>
Yep
<stintel>
guess it's good timing to test 5.15 on ramips with the beryl
<stintel>
I forgot to do it on my U6Lite in Belgium
<stintel>
and don't wanna risk it now that I'm back home
<f00b4r0>
Habbie: in fact that code has apparently been checked on busybox awk: "# We cannot output a NUL character using BusyBox AWK. See:"
<Habbie>
oh neat
<f00b4r0>
I'd say let's use that. I'll reply on m-l
<robimarko>
stintel: Back to slavland (couldnt resist)
<robimarko>
f00b4r0: agreed
csharper2005 has quit [Ping timeout: 480 seconds]
<stintel>
robimarko: да!
<Habbie>
aaaaaaaaaaa
<robimarko>
stintel: luckily I know some cyrillic
<Habbie>
i learned some more cyrillic today
<Habbie>
because stintel said this in another channel earlier :D
<robimarko>
I find it easy to read
<robimarko>
But writing is a whole new level
<Habbie>
i've learned that if you just know the alphabet, you can pronounce many things, and many words make sense just because you know other languages
<robimarko>
Yeah, especially if languages are similar
<Habbie>
i also learned that knowing a bit of greek alphabet is a good start :)
<Habbie>
as for what stintel just said, i wasn't aware that first letter looked like that - once i looked that up, i understood :D
<robimarko>
For me its great that whole region (With exception of Hungary) speaks rather similar language
<f00b4r0>
Habbie: you should try some водка then :)
<Habbie>
f00b4r0, i managed to guess what that was!
<f00b4r0>
same letter ;)
<Habbie>
yes
<robimarko>
Universal sign in the slav world
<Habbie>
and spanish is weird with b and v
<f00b4r0>
robimarko: lol
<robimarko>
f00b4r0: Though rakija is more popular in my parts
<robimarko>
Probably schnaps in german/english
<f00b4r0>
:)
<f00b4r0>
meanwhile, I should try to catch some "ЗЗЗ"s, and call it a night
<f00b4r0>
(surprising how some Cyrillic characters look like numbers)
<robimarko>
Same here, surprisingly long day for weekend