<mangix_>
rsalvaterra: pretty sure there are other ones that openwrt does not carry
mangix_ has quit []
mangix has joined #openwrt-devel
<mangix>
anyway, finally got the ax router
<mangix>
goodbye ath10k
<Slimey>
hehe
pmelange has joined #openwrt-devel
pmelange has left #openwrt-devel [#openwrt-devel]
valku has quit [Quit: valku]
jlsalvador has quit [Quit: jlsalvador]
jlsalvador has joined #openwrt-devel
nitroshift has joined #openwrt-devel
<neggles>
I have shifted all of my openwrt build dirs etc. over to 4x250GB SATA SSDs in a striped zfs pool, and, definitely should've done that sooner
<neggles>
everything builds *so much faster* (probably half because the build_dir fits entirely in ARC)
danitool has joined #openwrt-devel
danitool has quit []
rmilecki has joined #openwrt-devel
nlowe has joined #openwrt-devel
Acinonyx has joined #openwrt-devel
<Pepes>
mangix: Which one? You don't need to buy a new router each time for new WiFi standards, when you can swap miniPCIe cards in Turris.
nlowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
nlowe has joined #openwrt-devel
danitool has joined #openwrt-devel
felix has quit [Remote host closed the connection]
felix has joined #openwrt-devel
goliath has joined #openwrt-devel
ecloud has quit [Ping timeout: 480 seconds]
ecloud has joined #openwrt-devel
cptn3m0 has joined #openwrt-devel
nlowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Acinonyx_ has joined #openwrt-devel
nlowe has joined #openwrt-devel
<f00b4r0>
is there a preferred way to preset root password with image builder?
Acinonyx has quit [Ping timeout: 480 seconds]
aleksander has joined #openwrt-devel
cptn3m0 has quit [Quit: Page closed]
<mangix>
Pepes: ax cards are not cheap
<rsalvaterra>
mangix: This is interesting… the generic kconfig disables CONFIG_DEVTMPFS_MOUNT by default. However, some targets enable it (layerscape/{armv7,armv8_64b}, mediatek/mt7622, oxnas, rockchip/armv8).
<rsalvaterra>
I wonder if these targets need it at all…
<mangix>
and actually in the context of the Omnia, I'd need those duplexer things. 2 antennas to one. I broke both of the originals
<mangix>
rsalvaterra: no
<mangix>
at least not in the context of openwrt
<Pepes>
mangix: Hmm, card itself is not expensive, but the shipping, VAT from the China could be, but there was some bulk order on Turris forum, which could save you some pennies. But in large quantity, it makes sense
<rsalvaterra>
In that case, according to the original commit message, the backport shouldn't be needed. We should just disable that kconfig.
<Pepes>
mangix: Regarding diplexer, I think I might be able to help you or at least figure out something for you
<stintel>
[ 16.883111] mtk_soc_eth 1b100000.ethernet eth0: Link is Up - 2.5Gbps/Full - flow control off
<stintel>
U6LR ^
<rsalvaterra>
stintel: \o/
<mangix>
stintel: nice
<stintel>
credit to dangole
<rsalvaterra>
Yep, I saw the patches. :)
<stintel>
but that thing is so annoying, sysupgrade takes ages due to large NOR flash
<mangix>
stintel: slow SPI speed maybe?
<rsalvaterra>
Heh. That's actually good, NOR is more durable than NAND. :P
<stintel>
[ 15.962776] jffs2_build_filesystem(): erasing all blocks after the end marker...
<stintel>
[ 145.023272] done.
<mangix>
Pepes: that would be great
<rsalvaterra>
stintel: Just two minutes. :P
<stintel>
that
<stintel>
that's additional to the sysupgrade time
<stintel>
I'm not seeing that symbol in openwrt.git and not in linux.git
<stintel>
someone is playing with 5.15 :)
<mangix>
stintel: I'm not actually
<mangix>
oh I see
<stintel>
well that symbol is not in openwrt.git so there is no patch adding it, and linux.git doesn't contain it
<mangix>
local patch
<stintel>
in the 5.10 branch
<mangix>
yeah it's a backport
<rsalvaterra>
mangix: I'm not going to backport the CONFIG_DEVTMPFS_MOUNT patch. Rather, I'm fixing procd in order to mount /dev noexec, since procd creates/populates /dev for us.
<rsalvaterra>
I wonder if this would allow us to remove split-gso from cake…
<neggles>
f00b4r0: that should be how that works
<f00b4r0>
neggles: network device rename?
yrrips- has joined #openwrt-devel
<neggles>
f00b4r0: yeah
<neggles>
though what are you trying to achieve
<f00b4r0>
renaming eth0 to some other name
<neggles>
eth0 will always be there and always be eth0
<f00b4r0>
ah. bummer.
<neggles>
you can change the name of `lan` or `wan` or `some_vlan_number` on it
<neggles>
and you could technically rename it via some Shenanigans that would require a custom image
<f00b4r0>
it's easy to rename netdevs on "standard" linux, so I guessed it could be done here too.
<f00b4r0>
s/guessed/hoped/ :)
<neggles>
systemd udev does make it rather easy
<f00b4r0>
exactly
<neggles>
i am not sure what we use? but it is not that
<neggles>
because systemd has no place in a friggin' router
<f00b4r0>
sure not
<f00b4r0>
though you don't need systemd to do device renames
<f00b4r0>
anyway, thanks, now I know it's a dead end
<neggles>
give it six months and you'll need systemd to use udev...
<neggles>
why do you want to rename it?
<f00b4r0>
neggles: sure hope not ;P
<rsalvaterra>
neggles: I'd say that depends on the size of the router… not in an embedded system, for sure. :)
<neggles>
rsalvaterra: okay, no place in a router running openwrt :P
<rsalvaterra>
Fair enough. ;)
<neggles>
f00b4r0: oh no that's not doom and gloom, it's facts :(
<stintel>
SD card on my M300 has plenty of space for systemd ¯\_(ツ)_/¯
* stintel
hides
<neggles>
you could always create a bridge/lan device and add a single port to it
<f00b4r0>
for particular shenanigans. Making different hardware present an "homogenous" view to sysadmin
<f00b4r0>
neggles: i figured that too, but that seems rather overkill
<neggles>
stintel: just because you *can* doesn't mean you *should*
<rsalvaterra>
stintel: Just install Ubuntu on it, then. :P
* rsalvaterra
runs
<f00b4r0>
and I'm not sure the overhead is desirable either :P
<f00b4r0>
stintel: lol
<neggles>
I don't know that there would be much overhead on a single-port bridge
<stintel>
rsalvaterra: you're the worst, systemd I can live with, but ubuntu, really?
<neggles>
*distant snapd noises*
<stintel>
*puke*
* neggles
screams and brandishes a cross
<rsalvaterra>
I use Ubuntu (MATE). Works For Me™. :)
yrrips has quit [Ping timeout: 480 seconds]
<neggles>
I use ubuntu more than I use any other linux distro for all the usual reasons ("I can google things and someone else has probably already fixed it") - but my cloud-init has a script that obliterates every last remnant of snapd, apt-mark holds it, and installs a pin file setting it to -999
<rsalvaterra>
With Compiz. And wobbly windows. Because of course. xD
<neggles>
aaaaand tbh i've started just installing straight debian instead
<stintel>
oh dear
<stintel>
OpenWrt on embedded, Gentoo on anything else
<neggles>
^ not just gentoo, musl gentoo
<stintel>
because everything else is getting much like macOS: outdated soft/libs, and the OS decides how you do things
<f00b4r0>
devuan ftw :
<f00b4r0>
:)
<rsalvaterra>
I used Gentoo for about 7 years, or so… before Ubuntu. I like Debian, but the UI needs tons of polish (which Ubuntu does).
<stintel>
hardened musl gentoo on servers/hypervisors/VMs, glibc gentoo on workstation / laptop
<stintel>
and on my HiFive Unmatched because no musl support for RISC-V
<neggles>
if I had to use a linux desktop as my primary i'd probably end up on kde neon or something approximating that on arch
<neggles>
but my dayjob involves far too much windows
<stintel>
I'm about to "order" a Thinkpad T14s with corp managed Fedora to replace the piece of shit macbook/macOS combo
* rsalvaterra
's dayjob involves too much macOS.
<neggles>
my dayjob also involves too much macos because i am the only guy in the company (small company) that knows how anything mac works
<neggles>
so definitely do not have time to be screwing around with gentoo or even arch :P
<stintel>
and when I'm doing OpenWrt related stuff for day job I'm using my own gentoo workstation anyway
<rsalvaterra>
neggles: I only know how to use macOS when I open the terminal. The rest is a mystery. :P
<mangix>
rsalvaterra: you seen the compiz style gnome extensions?
<rsalvaterra>
mangix: Gnome died for me after 2.x. It's horribly bloated.
<f00b4r0>
holy cow. busybox shell is driving me nuts
<neggles>
yeah i should say, i use ubuntu on most of my linux boxen, but none of them have a gui
<f00b4r0>
now it errors on '[ "$var" ]' with $var unset
<neggles>
f00b4r0: ash is a fickle beast at times
<f00b4r0>
that's a polite way to say it's a piece of s***
<neggles>
hey be nice
<neggles>
it's very smol
<f00b4r0>
true
<neggles>
you would want `[ -z "$var" ]`
<stintel>
good idea to use path where lots of installers also drop their crap (ok, macOS doesn't need an installer, but unfortunately every other software company ignores that), then does not recommend to change the path
<rsalvaterra>
neggles: Even my parents use Ubuntu. There hasn't been a machine running Windows here since about 10 years ago. :)
<karlp>
which is what you _should_ be doing anyway, even in bash :)
<f00b4r0>
neggles: I initially used `[ -n "$var" ]` but that didn't work either
<stintel>
ugh, I'm gonna close this window for a while because ranting about OSes will kill my producitivty
<neggles>
hehe
<neggles>
fair
<rsalvaterra>
xD
<neggles>
lets go back to complaining about ash then
<neggles>
robimarko: ah so, if my ttyMSM0 is at 0x16340000, i'd pass `earlycon=msm_serial_dm,0x16340000` in cmdline?
<neggles>
I found the phys/virt addrs for ipq8064 easily enough, though I don't know if those change depending on how the uart/pinctrl is wired
<robimarko>
Yeah
<robimarko>
The phys/virt also need to match the UART adapter used
<robimarko>
So phys for it would be 0x16340000 as well
<neggles>
oh interesting
<robimarko>
You also need to pass the earlycon=ttyMSM0,115200
<robimarko>
Virt address needs to be calculated, I dont know whats it mapped on IPQ8064
<neggles>
ah
<robimarko>
On IPQ4019 it was just 0xf and then phys
<neggles>
I found a pair of addresses that are supposedly accurate for the ipq8064 dev boards, which seem to also use gsbi4 serial
<neggles>
most of these things do
<neggles>
so `ubi.mtd=rootfs loglevel=8 console=ttyMSM0,115200 earlycon=msm_serial_dm,0x16340000 earlycon=ttyMSM0,115200` ?
<robimarko>
GSBI4 is the most common for sure
<robimarko>
Yeah, that should work
<robimarko>
Nope
<robimarko>
Actually, the earlycon=ttyMSM0,115200 needs to be earlyprink=ttyMSM0,115200
<neggles>
ah
<neggles>
robimarko: Success!
<neggles>
...well sort of
<neggles>
ah wait yeah alright heh
<neggles>
it helps when you point the serial port at the right address in the dtb
<neggles>
earlycon worked then I lost it at regular console :P
<robimarko>
Yeah, it can break the regular console
<robimarko>
As it avoids the driver init basically
<neggles>
yeah, this is my fault though
<neggles>
12490000.serial: ttyMSM0 at MMIO 0x12490000
<robimarko>
Haha, so wrong stdout
<neggles>
ttyMSM0 should be at 0x16340000
<robimarko>
Or alias
<neggles>
h-uh, device tree alias is pointing to the right spot
<neggles>
oh
<neggles>
right, the bluetooth UART on gsbi2 is getting probed before the console one
<neggles>
that explains why the oem firmware uses the "qcom,msm-hsuart-v13" for the BT uart
<neggles>
driver gets probed later
<robimarko>
But that shouldnt matter at all
<robimarko>
Is the bootloader maybe passing the wront console=?
<robimarko>
Cause I dont know what has higher priority, DT or cmdline
<neggles>
robimarko: bootloader cmdline is identical to dt one
<neggles>
and i have serial0 pointed at gsbi4_serial which is the console dev
<neggles>
oh wait
<neggles>
console=ttyMSM0,115200
<neggles>
and ttyMSM0 is gsbi2
<neggles>
do I need to pass console=ttyMSM1?
<robimarko>
Looks like it
<robimarko>
Looks like kernel uses cmdline if available
<neggles>
hmm... can i use a devicetree alias in console=?
<neggles>
i suspect no
<robimarko>
Dont know what has higher priority
nlowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
nlowe has joined #openwrt-devel
fda- has quit [Ping timeout: 480 seconds]
<neggles>
robimarko: okay so it's apparently just decided that whatever the first msm_serial device it probes is, that one's the console
<neggles>
which is problematic when it probes gsbi2_serial first, *then* gsbi4_serial
<neggles>
so i think for now I will just turn off the bluetooth uart
<robimarko>
Hmm, that is not supposed to happen really
<neggles>
yeah, something's not right
<neggles>
there's two different uart drivers in the OEM fw, gsbi2_serial is using the other one which comes up as ttyHS0 and gets probed about 0.25s after msm_serial
<hurricos>
... so, for what it's worth: I've moved to Fedora for both server and desktop use
<stintel>
nice
<hurricos>
a combination of bleeding-edge and stable has always sat well with me. I also support a community of non-technical users and am starting to play around with Fedora Kinoite
<hurricos>
it also helps that I need to know Red Hat for work :^)
<rsalvaterra>
hurricos: Well, we can start another one, say… vim vs emacs? :P
<hurricos>
and while RHEL8 has dropped support for LSI2008
* rsalvaterra
uses nano and doesn't care.
<hurricos>
Fedora 35 still supports it
<hurricos>
I use $EDITOR
<hurricos>
I'm an emacs guy except when I already know where the file I'm editing is
<Slimey>
yeah never trust those plastic junk anchors, adtran supplies the metal ones shown but bigger since those waps weight a ton and are basically a giant heat sink
<Slimey>
but i try to avoid walking under them just in case ;)
<rmilecki>
nbd: i've rootfs_data on bcm4908 almost fully working, but I'm dealing with one last issue
<rmilecki>
I use /lib/preinit/75_rootfs_prepare to call "ubimkvol /dev/ubi0 -n 20 -N rootfs_data" (if it doesn't exist yet)
<rmilecki>
the problem is: [ 43.563490] mount_root: failed to mount -t ubifs /dev/ubi0_20 /tmp/overlay: Invalid argument
<rmilecki>
it's cuased by missing /dev/ubi0_20 (it doesn't seem to get created automatically that early)
<rmilecki>
nbd: if I reboot, then /dev/ubi0_20 exists and mount_root works fine
jlsalvador has quit [Quit: jlsalvador]
<rmilecki>
I use /lib/preinit/75_rootfs_prepare as "rootfs_data" UBI volume needs to exist before "mount_root" kicks in
<rmilecki>
(validation can happen after "mount_root" mounts it)
<rmilecki>
is there a trick to create /dev/ubi0_20 that early manually? without procd which (as I believe) handles it later
jlsalvador has joined #openwrt-devel
<jow>
rmilecki: mknod?
<rmilecki>
ah, i was looking for "mkdev" command
<jow>
or try if you can use /etc/hotplug.d/hotplug-preinit.json
stintel[m] has left #openwrt-devel [#openwrt-devel]
<stintel>
LOL
<stintel>
I gave that leave command more than 24h ago
stintel[m] has joined #openwrt-devel
<jow>
unsuccessfully, it appears
<stintel>
no no, I joined again
<stintel>
I couldn't speak
<stintel>
and I still can't 😂
<rmilecki>
i've a working rootfs!!! =)
<rmilecki>
thanks jow!
rua1 has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
minimal has quit []
nlowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<f00b4r0>
Slimey: these anchors are definitely not suitable for drywall. Anyone knowing their left from their right hand should know that ;)
<Slimey>
lol
pmelange has joined #openwrt-devel
Borromini has joined #openwrt-devel
mattytap has joined #openwrt-devel
mattytap_ has quit [Ping timeout: 480 seconds]
mrkiko has joined #openwrt-devel
<mrkiko>
Hello all! Does anyone know if the MVEBU GL-MV1000 has kernel size limits? rsalvaterra ?
srslypascal is now known as Guest10411
srslypascal has joined #openwrt-devel
<rsalvaterra>
No idea… :/
bluew has joined #openwrt-devel
Guest10411 has quit [Ping timeout: 480 seconds]
<rsalvaterra>
WTF, cake has diffserv*8*? o_O
<hurricos>
yeah :D
<rsalvaterra>
I was casually reading the code and stumbled on the diffserv8 matrix. :P
zatwai has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
zatwai has joined #openwrt-devel
<Habbie>
(pretty cool really)
swegener has quit [Remote host closed the connection]
swegener has joined #openwrt-devel
<dwfreed>
would be interesting to see the patches for that functionality
<Habbie>
yep
<dwfreed>
presumably it operates on some kind of RSSI triangulation based on individual antenna reception
<Habbie>
could they get away with keeping this closed?
<Habbie>
i suspect they could
<dwfreed>
what's hostapd's license
<slh>
the first question would probably be if that's actually doing anything functional - or if it's more for show to its gamer audience
<Habbie>
slh, i did get the impression antenna orientation can matter
<Habbie>
slh, which doesn't mean this is useful, of course
<dwfreed>
slh: right, if the antenna is just a vertical and all that is moving is the plastic, it's probably useless
<slh>
dwfreed: BSD
<hauke>
it only has 4 antennas so more low end
<dwfreed>
slh: oof
<hauke>
or it has multiple other internal antennas in addition
<dwfreed>
then yeah, probably won't get this in a GPL dump
<hauke>
doing beamforming is probably less expensive compared to haveing motors
<hauke>
<ou can adjust the beam with beamforming much faster
<hauke>
but it does not look so cool
<Habbie>
ack
<dwfreed>
Habbie: antanna orientation matters insofar as radiation pattern and and polarization are optimal for your environment
<rsalvaterra>
It looks cool, but I can't imagine it not being noisy and distracting.
<slh>
and propne to breaking down the line
<Habbie>
rsalvaterra, you mean entertaining and cool
<Habbie>
needs more RGB though
<rsalvaterra>
Habbie: You already have lolcat. :P
<Habbie>
:)
<dwfreed>
Habbie: north of 10m frequencies, radiators are generally vertical wires, which results in an omnidirectional donut-shaped radiation pattern; beamforming basically uses principles of constructive and destructive interference, and the ability to adjust the phase of the produced signal to convert the omnidirectional radiation into a more directional "beam" which can have greater strength than the
<dwfreed>
original radiations on their own
<mangix>
rsalvaterra: nvim
<Habbie>
dwfreed, ack
<Habbie>
dwfreed, i knew like 15% of that :)
<Habbie>
but, wouldn't these motors move those vertical wires?
<dwfreed>
it seems like they're just rotating the plastic side to side
<Habbie>
i got the impression it had two elbows
<dwfreed>
if it's angling up and down, yeah, that can help
<Pepes>
philipp64: Yep, I noticed them too sometimes while updating git package. It is weird, but I noticed that only on CI, so it is not a big issue for me, though.
<philipp64>
It would be nice if CI/CD didn't have FP/FN's, but... I've yet to encounter one that didn't. So I should ignore them?
pmelange1 has quit [Ping timeout: 480 seconds]
<philipp64>
Who owns CI/CD these days?
rmilecki has quit [Ping timeout: 480 seconds]
dangole has quit [Remote host closed the connection]
Tapper has joined #openwrt-devel
dangole has joined #openwrt-devel
pmelange has left #openwrt-devel [#openwrt-devel]
Borromini has quit [Quit: Lost terminal]
rsalvaterra has quit [Quit: rsalvaterra]
Tapper has quit [Read error: No route to host]
rsalvaterra has joined #openwrt-devel
Tapper has joined #openwrt-devel
<neggles>
ahh okay, the HS UART is 4-pin, so i do actually need to use that