<nick[m]1234>
Client <-wifi-> AP <-ethernet vlan-> Core Router; Clients packets arrive at corerouter; corerouter responds; response can be seen on switch0.40 on AP, also on br-dhcp but never passed to wifi5-ff
<nick[m]1234>
22.03-rc1 works, but 22.03 not and also snapshot is broken
danitool has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<nick[m]1234>
we have to bisect the patch that breaks everything
<nbd>
nick[m]1234: that seems odd
<nbd>
yes, please let me know which patch is responsible and i will look into it
<nick[m]1234>
nbd: thanks I will ping you again when I know the patch
<rsalvaterra>
mrkiko: I've had the radio dying a couple of times (in about a year and a half, it's very rare). From the system point of view, everything was working perfectly, no errors in dmesg, nothing. But there simply was no signal being broadcast.
<rsalvaterra>
mrkiko: Oh, wait, on the RT3200? Never. This only happened on my Redmi AC2100.
<rsalvaterra>
However, I've only had the RT3200 for about a month. :)
dibdot has joined #openwrt-devel
dibdot_ has joined #openwrt-devel
dibdot has quit []
dibdot_ is now known as dibdot
goliath has quit [Quit: SIGSEGV]
dibdot has quit [Quit: dibdot]
f00b4r0 has joined #openwrt-devel
borek has joined #openwrt-devel
guidosarducci_ has quit [Remote host closed the connection]
guidosarducci has joined #openwrt-devel
Borromini has joined #openwrt-devel
dibdot has joined #openwrt-devel
dibdot has quit [Quit: dibdot]
bluew has quit [Quit: Leaving]
borek1 has joined #openwrt-devel
borek has quit [Read error: Connection reset by peer]
borek1 is now known as borek
borek has quit [Ping timeout: 480 seconds]
<mrkiko>
rsalvaterra: thanks! Well, I could connect to the 2.4GHZ radio, devices would get DHCP but no other traffic it seems. Experienced it after one day or so of uptime; 5 GHZ seems to work well
<stintel>
did anyone start bisecting that already ?
<rsalvaterra>
mrkiko: Right, that's a completely different issue. :)
rua has quit [Ping timeout: 480 seconds]
guidosarducci has quit [Remote host closed the connection]
guidosarducci has joined #openwrt-devel
rua has joined #openwrt-devel
indy has quit [Ping timeout: 480 seconds]
goliath has joined #openwrt-devel
indy has joined #openwrt-devel
noltari has quit [Read error: Connection reset by peer]
noltari has joined #openwrt-devel
<neggles>
matoro: just saw your octeon question from the other day
<neggles>
the best thing you can do with that board is *throw it in the trash where it F#$@ING BELONGS* (for context, I own more Octeons than I care to count, mostly because counting them would make me very depressed)
<neggles>
however, your issue sounds devicetree related - we would need your .dts file and a boot log dump to investigate any further
* Borromini
is gonna repurpose his ER Lite as testing device
<neggles>
Borromini: you should repurpose it as a bookend
<neggles>
or perhaps a paperweight
<neggles>
blanket warmer for the end of the bed where your feet go
<Borromini>
:P
<Borromini>
i locked myself out of my RB5009UG configuring fw4 and failsafe doesn't seem to work (yet?) on that stuff
<Borromini>
so i'm going to play on the erlite first now, at least it has easy serial in case anything goes wrong
<Borromini>
any new issues that popped up on octeon? you don't seem to be too happy with it...
rua has quit [Ping timeout: 480 seconds]
<neggles>
Borromini: the gigantic memory leak issue is fixed, it was a kernel bug that broke all of MIPS64 for 18 months
<neggles>
apart from that the only major issue is Being An Octeon
<Borromini>
:)
<Borromini>
i thought Marvell was slowly trying to upstream some of the support?
rua has joined #openwrt-devel
Borromini has quit [Ping timeout: 480 seconds]
minimal has joined #openwrt-devel
<rsalvaterra>
neggles: Being An Octeon™ sounds like a feature to me… ;)
<Pepes>
hurricos1: We tried it on OpenWrt 21.02 with kernel 5.4, but it uses kernel 5.10 for mac80211 backports and sorry for late reply.
<neggles>
because that's the hacked-to-the-skies octeon SDK kernel
<matoro>
nope. using latest git stable 5.18.0
<matoro>
just took the config from the SDK
<neggles>
interesting since that kernel config shouldn't be remotely compatible
<neggles>
it's got a whole bunch of symbols that don't match
<neggles>
and it's for 4.9 or 4.4 iirc
<matoro>
I used olddefconfig on it
<matoro>
but my problem now is with the octeon_mgmt driver for the ethernet
<neggles>
the openwrt tree kernel should work just fine, assuming your devicetree is correct
<neggles>
and it is probably not
<neggles>
which board do you have, specifically?
<matoro>
it's a 68xx eval board
<neggles>
but *which*
<neggles>
there are a few :P
<neggles>
rsalvaterra: *hisses*
<matoro>
the reference board from cavium. CN6800-EVB-BASE-BOARD-Y22
<neggles>
cavium made more than one reference board :P but that answers it
<neggles>
rsalvaterra: my attitude towards octeons is the same as my attitude towards turbo subarus. they are terrible and bad and dumb and stupid and fragile and horrid and flaky and you can pry them out of my cold, dead hands
<neggles>
there's some cognitive dissonance here.
<matoro>
oh I see. it has board ID 39 in cvmx-helper-board
<neggles>
okay cool
<matoro>
I can't test if the driver in the openwrt kernel works or not because the octeon_ethernet driver has a panic line in it that it hits
<matoro>
which also happens the same in the upstream kernel. I have to disable that driver for it to go anywhere
<neggles>
well yeah, it panics because it's unable to set up the ethernet controller
<matoro>
I'm fine with just the one management ethernet
<matoro>
which is where the problem is now, in octeon_mgmt
<neggles>
so there's a thing you need to understand about octeons: a normal SoC has a CPU at the center, with a bunch of peripherals wrapped around it
<neggles>
an octeon has a gigantic pile of accelerators at the center, with the CPU glued on to the side
<neggles>
the CPU cannot do jack shit without the accelerators working properly, it can't even send bytes out a UART without help
<matoro>
oh I see
<neggles>
as a result
<neggles>
wait, missed a bit
<neggles>
a few of these accelerators are critical to the functioning of nearly everything on the SoC
<matoro>
so on the mgmt0 interface right now, the driver picks it up, link lights come up, it says carrier = 1, I assign an address and ping, now I'm sniffing on the other end and it's sending bytes out just fine, but it's not seeing any packets coming back
<neggles>
and because the chips are built to be very, very flexible, they can be configured in a nearly infinite number of ways
<matoro>
not even arp replies
<neggles>
which means when you get the config wrong
<neggles>
yep
<neggles>
that happens
<neggles>
or you get a kernel panic
<olmari>
Not exactly networking related, but how about certain berry and it's media centric roots? ;)
<neggles>
that's not quite the same, but it's similar yeah
<neggles>
difference being that the videocore is necessary for bootup, but it's not necessary for most normal CPU operations
<matoro>
I have tried probably 40 different configs in the past week, octeon_ethernet panic'd on all of them
<neggles>
you will need the out of tree patches that are in the openwrt kernel, *and*
<matoro>
ohh, I looked through the openwrt patches a while back and somehow missed 110-er200-ethernet_probe_order.patch
<neggles>
specifically, the pip: section and smi sections
<neggles>
the mainline DTSes are pretty much all f$%#ing wrong
<matoro>
well that would explain that, I tried appending the mainline dts and it did nothing besides make more error messages
<neggles>
yeah... this process is unpleasant
<neggles>
did it with the CN6640-SNIC10E
<neggles>
finally got that to be happy
<neggles>
took 2 weeks and a solid chunk of my sanity
<neggles>
we've not come across a cn68xx yet though, and they have some extra upgrades over the CN66xx
<neggles>
so may need yet another patch
<matoro>
there is one more issue, wonder if you saw this one yet - this thing should have 32 cores, but kernel hangs if I boot with numcores=32. it works with numcores=16 or anything below though
<matoro>
I'm less worried about this one than the networking thing though
<neggles>
can you pastebin a boot log including u-boot?
<neggles>
and your u-boot environment?
<matoro>
absolutely. let me try all of these things you just suggested and will get back to you
<rsalvaterra>
neggles: Stockholm syndrome, then… xD
<matoro>
thank you so incredibly much for the help
<neggles>
rsalvaterra: kinda. it's the same with subarus
<neggles>
the hardware is actually pretty good
<neggles>
but the software support and documentation is a steaming dumpster fire
<neggles>
(the analog in subaru town being the engine itself vs all the supporting components)
<matoro>
there's still some 6300, 6600s available, and apparently an 8100 now
<matoro>
neggles: the device tree in that sdk is identical to upstream, and unfortunately even with the openwrt patches + the device tree it still panics
HackOS_ has quit [Remote host closed the connection]
borek has joined #openwrt-devel
borek has quit [Ping timeout: 480 seconds]
<rsalvaterra>
nbd: I noticed qosify starts even if all interfaces are disabled. Is it possible to avoid it? It takes a bit of RAM and CPU time just for being there…
Tapper has quit [Ping timeout: 480 seconds]
<rsalvaterra>
Does anyone around here fancy (soft) bricking an MT7621 device? For science, of course. :P
<rsalvaterra>
For some reason, my Redmi AC2100 really dislikes Linux 5.15.42+. And I haven't soldered a serial header on it.
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
mangix has quit [Ping timeout: 480 seconds]
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
mangix has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
mangix has quit [Ping timeout: 480 seconds]
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<Lechu>
I'm finally trying to boot that bloody Ruckus ZF7372 from flash, and... nothing. Got to the stage when LZMA-compressed kernel unpacks successfully, entry point matches, and the device just stalls