<mangix>
found out why DSA is not working. kernel 5.4. AGH
<Tusker>
stintel: sleep is for the weak :)
<stintel>
:P
<stintel>
well I have an automation in home-assistant that turns off lights, tv, amplifiers, ... at 3AM
<Tusker>
ah, now you can hear the m300 fan
<stintel>
:D
<stintel>
no no
<stintel>
but as internal dns is handled by dnsmasq still, and I have yet to HA that, internal hostnames don't resolve anymore, and my home-assistant tab in browser is broken so I can't turn on the lights again
<stintel>
there we go, main m300 is back
<stintel>
I was doing a setup with 2x OC200 and kea dhcp and was going to add probably powerdns auth for dns in full HA DHCP/DNS setup
<stintel>
but I nuked one of the oc200 while handling comments to upstream the NOR flash chip
<stintel>
so it's currently on one leg
<stintel>
and I kind of decided to abandon the oc200
<Tusker>
i have a static IP and advertising a /24 off my main router, it would be interesting to handle proper failover for it, does kea dhcp handle distribution of the leases ?
<stintel>
what do you mean by distribution of the leases?
<Tusker>
ie, the hosts that have obtained the lease, and then you fail over dhcp to another box, it won't have /tmp/dhcpd.leases for example, and DNS will fail until the client renews, right ?
<stintel>
leases are synced between the DHCP nodes
<stintel>
DNS ... I had not added that part so cannot comment on that
<stintel>
it would be a complex setup for a home network
<stintel>
but as I'm half HA already, it only makes sense to add that
<stintel>
well, to me at least ;)
<stintel>
already happy that external connectivity keeps working when main router is dead
<stintel>
so I can easily access the offsite backup ;)
<Tusker>
it would be interesting to be able to take out my main router for maintenance and not have to worry about other people in the house complaining
<stintel>
aye
<stintel>
well, I have no other people in the house
<stintel>
but I complain a lot myself :D
<Tusker>
:) if you don't complain, how do you know you are doing anything wrong?
<stintel>
but that's the main idea why I have dedicated dual router/firewall, multiple APs
<stintel>
and then came the idea to move DHCP/DNS/MQTT to yet other separate nodes
<stintel>
the OC200 looked nice, but it's a major pita to work with
<stintel>
getting TTL is insane
<stintel>
and apparently clamping the NOR is also impossible
<stintel>
I managed to get flashrom to detect the chip once in 200 attempts or so
<Tusker>
did you desolder it ? remove the reset lines ?
<stintel>
used different clamps, different readers, different jumper wires, ...
<stintel>
no
<Tusker>
there is probably a ground sink or something that is messing with your attempts
<stintel>
it works, 100% sure, I've done it once, with a clamp that I threw because low quality, and on 3.3v
<stintel>
which nuked the flash chip as it's 1.8v :P
<stintel>
but it worked twice
<stintel>
and got twice the same image (same checksum)
<stintel>
so I bought new clamp, 3M, turns out that's even bigger shite
<stintel>
I'd RPi but find a nice case compatible with PoE HAT ...
<stintel>
also not entirely open
<stintel>
so meh
<Tusker>
also, I find it's too popular, so not interesting enough
<stintel>
and then there's Ubiquiti with their damn habbit of making non-standard PoE-PD devices
<stintel>
maybe I'll just throw everything on the m300
<stintel>
instead of using separate devices for it
<stintel>
but first I want to get that qoriq target in a state that is acceptable to add to master
<Tusker>
yeah, good plan
<Tusker>
btw, m300 ethernet DTS ?
<Tusker>
is it marvell behind etsec ?
<stintel>
hmm, did I not write that ?
<stintel>
maybe in my matrix window which seems a bit bugged
<stintel>
17|01:46:25 < stintel> basically I started from t2081qds.dts
<Tusker>
OK, thanks
<stintel>
which on its own is a complete mess of includes
<Tusker>
ah, different to the xtm330 then
<stintel>
on the other hand makes https://git.openwrt.org/?p=openwrt/staging/stintel.git;a=blob;f=target/linux/qoriq/files/arch/powerpc/boot/dts/fsl/watchguard-firebox-m300.dts;h=871f5410f221ff2e502c235fc0ea635c292f2907;hb=refs/heads/qoriq relatively clean
<aparcar[m]>
mangix meson semm to work fine.
<Tusker>
it does use a marvell switch though, so pretty close
<aparcar[m]>
mangix do you know what the minimum required version is?
<stintel>
crazy enough, the decompiled DTS of the M200 contained vitesse-9953
<stintel>
which is complete and utter nonsense
<stintel>
but what do you expect from OEMs :)
<stintel>
I removed the heatsinks on the M200, it's the same marvell as on the m300
kenny has joined #openwrt-devel
danitool has quit [Read error: Connection reset by peer]
<slh>
who knows, perhaps they used that during development
<mangix>
wow so this makes no sense
<mangix>
build_dir contains only linux-5.10 but when flashing the image it says 5.4
<mangix>
uname that is
<Tusker>
usually when that happens the build has failed and didn't create the image
<stintel>
I'd be curious to see if using that globally fixes the chrony crash for me, and maybe even the crashes and ext4 corruption with CONFIG_HARDENED_USERCOPY
<stintel>
and maybe even the use of squashfs images
<stintel>
good thing I can build a full image in < 15 minutes
<stintel>
hmm, now it was routing inter-vlan on same physical interface at 22MB/s
<stintel>
before the reboot 10 times slower
<stintel>
that's not good
<aparcar[m]>
mangix: ping
<aparcar[m]>
your meson add commit message is empty
<aparcar[m]>
could you please provide a short pitch why it's great?
<stintel>
Tusker: unfortunately chrony still crashes here :(
<stintel>
ok, further debugging is for tomorrow / weekend
<Tusker>
sleep well
<stintel>
thanks
<stintel>
also looks like the kernel didn't pick up that extra flag
<stintel>
but that shouldn't matter for userspace
<stintel>
anyway
* stintel
&
goliath has quit [Quit: SIGSEGV]
fda has joined #openwrt-devel
fda- has quit [Ping timeout: 480 seconds]
<mangix>
aparcar[m]:ok
kenny has quit [Ping timeout: 480 seconds]
<aparcar[m]>
mangix: ping me and I'll make it happen
<mangix>
aparcar[m]: done
<mangix>
aparcar[m]: also, minimum required version for what?
<aparcar[m]>
it's fine, minimum is 3.6
<mangix>
yeah they want compatibility with Ubuntu AFAIK
<aparcar[m]>
ninja: fatal: chdir to '/home/user/src/openwrt/openwrt/build_dir/host/pkgconf-1.8.0/openwrt-build' - No such file or directory
<mangix>
btw that @!arc dependency is somewhat worrying. Although for development it's perfectly fine.
<aparcar[m]>
I'll ask upstream at some point
<mangix>
Upstream is Alpine Linux. I doubt they care about a non musl supported architecture. Speaking of which, Synopsis really needs to add musl support...
<aparcar[m]>
upstream does care, there is some integration happening
<mangix>
anyway, replacing opkg with apk sounds good
<mangix>
apk is extremely fast on my desktop
<mangix>
it and pacman are very close
<aparcar[m]>
how big is pacman?
<mangix>
157368 locally. it's probably glibc only
<mangix>
size is not really the goal anyway
<mangix>
alright. time to get rid of gettext
<aparcar[m]>
why?
<mangix>
it was only needed for uClibc, which is now gone.
<mangix>
The delta of ath79 swconfig to DSA is ~70
<mangix>
now to get these LEDs working
chadn has joined #openwrt-devel
rmilecki has joined #openwrt-devel
mangix has quit [Ping timeout: 480 seconds]
chadn has quit [Ping timeout: 480 seconds]
valku has quit [Quit: valku]
<Namidairo>
don't forget to bump the image version when you're done so you get *less* people complaining on sysupgrade
mangix has joined #openwrt-devel
<slh>
the former is the easy part... the later will happen either way (just count the number of complaints and questions about dsa on ramips and mvebu - or even the complaints about 'presumably dsa related issues' on ath79 in 21.02.0...)
<slh>
damned if you do, damned if you don't
chadn has joined #openwrt-devel
<slh>
(and I'm only counting complaints about a "strange message preventing the sysupgrade", not genuines questions of how to implement a(ny) slightly complicated configuration with dsa (this I do have some issues with myself))
<mangix>
slh: what?
<mangix>
i seem to have missed something
<slh>
mangix: 07:32:28 < Namidairo> don't forget to bump the image version when you're done so you get *less* people complaining on sysupgrade
<mangix>
what are these dsa issues on ath79?
<slh>
none (that I know of), as I understood it the hint was more generic (about setting the compat_version for dsa)
<Namidairo>
I think they're making fun of the fact that only one device in ath79 was ported to DSA but people are chalking up issues to it anyway
<mangix>
there are no ath79 devices using DSA
<Namidairo>
release notes say only the TL-WR941ND is
<slh>
only the tl-wr941nd v2/ v3 - but that one has done so for years already, without anyone noticing
<slh>
courtesy of its mv88e6060 switch
<mangix>
I don't see where it's using DSA. CONFIG_DSA is not enabled for the kernel
<mangix>
ah I see it now. It's is the tiny target
<mangix>
*in
<slh>
yep
<slh>
a fringe oddity
<mangix>
anyway, generic has no DSA. That's only WIP at this point
<slh>
(even ar71xx in 17.01.x was already using DSA for the tl-wr941nd v2/ v3 with mv88e6060)
<mangix>
that's somewhat strange
<mangix>
wonder why the swconfig driver was used for so long
<slh>
basically no other devices are using MV88E6060/ MV88E6063 switches
<slh>
ancient 100 MBit/s stuff, a decade out of fashion
<mangix>
sure but there's the modern marvell stuff
<slh>
no interesting devices using DSA --> no luci integration for DSA --> no impetus for DSA --> rinse and repeat
<mangix>
anyway, the most successful transition is probably the ramips one
<mangix>
DSA ramips is way better
<mangix>
no random driver crashes. HWNAT works as well.
<slh>
what I'm still partially fighting with, is when to use bridge vlan filtering (do I include wan into this underlying bridge) and all the stuff emerging from that
dedeckeh has joined #openwrt-devel
chadn has quit [Ping timeout: 480 seconds]
nitroshift has joined #openwrt-devel
<nitroshift>
morning guys
<mrkiko>
nitroshift: morning!
<nitroshift>
mangix, ping
<mangix>
nitroshift: pong
<nitroshift>
mangix, following yesterday's tests, all ports that have bogus mac address also don't have connectivity (as in device is unreachable when cable is plugged in them)
gladiac has quit [Read error: No route to host]
gladiac has joined #openwrt-devel
<mangix>
nitroshift: what's funny is I witnessed the same thing. I think it's an issue with that dualport DSA patch
<nitroshift>
also, there is only one mac address printed on the device label
<mangix>
I recommend abandoning dualport DSA for now
<nitroshift>
i was thinking exactly the same
<nitroshift>
let's see it upstreamed in the kernel tree and we'll take it from there
gladiac has quit [Read error: No route to host]
gladiac has joined #openwrt-devel
danitool has joined #openwrt-devel
<rsalvaterra>
Morns!
<stintel>
o/
<rsalvaterra>
stintel: Mate, do you even sleep? :)
<stintel>
I do
<stintel>
just got a phone call, the monitor I ordered last night after close of shop, will be delivered today \o/
<stintel>
finally bit the bullet and replacing 2x24 1920x1200 with 49" 5120x1440
danitool has quit [Read error: Connection reset by peer]
<rsalvaterra>
mangix: Holy hell, over 130 Mb/s faster just going from ar71xx to ath79? Is this the exact same configuration? No way this can be just the switch.
<mangix>
rsalvaterra: the comparison is ar71xx and ath79+DSA
<mangix>
swconfig is ~70 slower than DSA
danitool has joined #openwrt-devel
<mangix>
the ar71xx build is unmodified 19.07
<rsalvaterra>
mangix: Ah, right, I see it now.
<mangix>
the DSA build has...special sauce on it
<rsalvaterra>
stintel: Those are… atrociously wide. I like watching movies too, but for work, my dream is to have a *square* monitor. :P
<rsalvaterra>
… that's just depressing, considering what we have in target/linux/ath79/dts.
<mangix>
yep. there's lots to upstream
<rsalvaterra>
There shouldn't be much science regarding the upstreaming of device trees. They just sit there waiting to be used. But they *need* to be used to be upstreamed.
<mangix>
i mean do keep in mind openwrt dts files have stuff that is not immediately suitable for upstream
<rsalvaterra>
mangix: I know, and missing DSA is the obvious one.
<rmilecki>
noltari: can you check if bmips still works?
<mangix>
rsalvaterra: hmm how would I test the USB phy driver?
<rsalvaterra>
mangix: You'd need to change the device tree of the target device, I guess.
<mangix>
sure but what is there to test?
<rsalvaterra>
Uh… if USB is working? :)
<rsalvaterra>
Unless blogic_ has other ideas, it's his patch, after all. :)
<noltari>
rmilecki: sure, I will test it as soon as I can, but I’m working right now, so I can’t tell you when…
justas has joined #openwrt-devel
<justas>
Hello. If I have bridge wired and wireless interfaces and CONFIG_BRIDGE_NETFILTER enabled in kernel, does netfilter see traffic that passes through bridge?
<justas>
looking at /proc/net/nf_conntrack doesn't seem like it, it's empty
<mangix>
rsalvaterra: bleh. sounds like upstream driver needs patching
<rmilecki>
noltari: sure, no rush
<rsalvaterra>
rmilecki: Something tell me that commit will conflict with my kernel bumps… :P
<rsalvaterra>
*tells
<rmilecki>
rsalvaterra: some b53 commits have "Fixes:" so they should end up backported
<rsalvaterra>
rmilecki: Sure, but we still have to deal with them until they're backported. :) Let me see how it goes here…
<rsalvaterra>
mangix: My branching is getting out of hand. For my WDR3600, I have *six* extra branches on top of master. :P
<Slimey>
interesting read, have to go over it more slowly
<stintel>
wonder if it's coincidence that it differs only 1 char from NSA
* stintel
hides
danitool has joined #openwrt-devel
<Slimey>
heh
<Slimey>
stintel hows those mpc targets
<stintel>
well I'
<stintel>
I've one M300 running as main router for >1m already
<stintel>
but there are some things I'd like to sort. my initial series to openwrt-devel should contain most if not all info about that
<rsalvaterra>
stintel: Are you sure that tinfoil in your hat is thick enough? :)
<stintel>
what hat?
danitool has quit []
<rsalvaterra>
DSA, NSA, conspiracy theories… tinfoil hat for protection. :P
<rsalvaterra>
Ok, fresh new build for the WDR3600, now with mangix's ag71xx cache line optimisations.
<stintel>
hey I'm vaxxed, should prove enough ;)
<rsalvaterra>
stintel: Proves you're not a twat, yes. :)
<stintel>
but I like twats!
<rsalvaterra>
A gay friend of mine called me a twat, once. I replied that we are what we eat. :)
<stintel>
😂
* stintel
keeps the word in his head to himself
<rsalvaterra>
(No offense intended or taken on either side, of course. ;))
<olmari>
Vax, 5g and NSA.. how bout 5g actually having modes on SA and NSA
<stintel>
see what I mean? ;)
<olmari>
Stand Alone and its counterpart Not Stand Alone, or so they keep telling us ;D
<stintel>
it's what they want us to believe!
<rsalvaterra>
The best vax is DEC VAX.
* rsalvaterra
runs
<stintel>
:D
<rsalvaterra>
VAX was actually supported in Linux, at a certain point.
<stintel>
there was this video on fb, where some guy supposedly ssh'd into a person after he had his sputnik vaccine
<Habbie>
lol
<Habbie>
as if your ssh client has GOST
<stintel>
I replicated said video about pfizer in ~1h, using the girl who posted the other video her name, and then insinuating something like "body status: horny"
<stintel>
man I had a good laugh
<rsalvaterra>
(As was the 31-bit IBM S/370, but nobody should want to go there.)
<stintel>
I wonder if I can find the original post
<stintel>
/home/stijn/projects/vaccinbs
<stintel>
found the code I used to replicate the video :P
<olmari>
Magnets also stick the spot, but general O'Neil had best... Mmmmagnets
<stintel>
someone was showing ssh to foo.sputnik.bar or w/e then running script that gave similar output claiming he was logging into a vaccinated person
<stintel>
oh the interwebz are fun
<mangix>
rsalvaterra: ar8129 i think. Dunno.
<rsalvaterra>
Flashing…
<rsalvaterra>
… and it's up! <3
<mangix>
So how are those speeds?
<rsalvaterra>
mangix: 100 Mb/s.
<mangix>
Ah
<rsalvaterra>
I don't even have to measure. My laptop only has Fast Ethernet. :P
<mangix>
Strange
<Slimey>
:(
<rsalvaterra>
mangix: There was a time manufacturers too too much of the piss.
<rsalvaterra>
*took too much
<mangix>
I have a knockoff gigabit usb adapter. It’s 100mbps
<rsalvaterra>
I'll have to test with another machine, but right now I need to translate the old network configuration to DSA.
<rsalvaterra>
Slimey: What am I even watching…? o_O
<Slimey>
lol im not sure but its catchy
<rsalvaterra>
Oh, wow…
<rsalvaterra>
I just noticed… when there are no enslaved devices in a bridge, the bridge itself isn't created at all, even if specified in /etc/config/network. Is this intended behaviour?
<mangix>
rsalvaterra: for that pahole patch, please mentiion that the first part is already upstream. Also if you could, Add your signed off and send to openwrt-devel
Forst has joined #openwrt-devel
ServerStatsDiscoverertraveler4 has joined #openwrt-devel