danitool_ has quit [Remote host closed the connection]
<fpsusername[m]>
Redid the stuff. Download repo, rebase with main line, building now
Tapper has quit [Ping timeout: 480 seconds]
robimarko has quit [Quit: Leaving]
<fpsusername[m]>
seems to compile toolchain/gdb fine now :)
Tapper has joined #openwrt-devel
<fpsusername[m]>
YEEE, it compileeed
srslypascal is now known as Guest6464
srslypascal has joined #openwrt-devel
Guest6464 has quit [Ping timeout: 480 seconds]
Borromini has quit [Quit: leaving]
<aparcar[m]>
hauke: what's the status of wolfssl? I see now snapshots failing, too?
Tapper has quit [Ping timeout: 480 seconds]
bluew has joined #openwrt-devel
<hauke>
aparcar[m]: wolfssl is now build by the package feed again
<hauke>
you haev to wait till it is build
<aparcar[m]>
But on on snapshots right? We should backport it
dedeckeh has joined #openwrt-devel
<fpsusername[m]>
I'm not sure what I'm doing wrong, but for whatever reason, openwrt compiles witin 10 minutes, but I don't see things such as LuCi compile (and thus when I boot the system, I can't access the web interface)
<fpsusername[m]>
It's enabled in my config
<fpsusername[m]>
oh ffs, there's something stupid going on. I load a config, then save and start building. For whatever reason it then builds an almost blank image
<Habbie>
did you use scripts/install ?
<Harm_>
fpsusername[m]: if you want more range you can also lower the channel width, e.g. 40 or 20MHz
<fpsusername[m]>
Habbie: That's before running make menuconfig
<Habbie>
ack
schwicht has joined #openwrt-devel
<fpsusername[m]>
Harm_: Hm, didn't know that would increase range. I only know that a higher bandwidth increases crossovers, but if nobody else has a network in the range of mine, then nothing is happening
<fpsusername[m]>
But now the config does run well
<fpsusername[m]>
I saved the default config as .config instead of config.buildinfo. IYes I loaded it as config.buildinfo before. Not sure why it doesn't work well
<fpsusername[m]>
Now the question is, I'm not sure if I can reflash/update my extender without using uart. lan doesn't really want to connect. Trying things out on my test unit first now
<xdarklight>
hauke: I have yet another question about GSWIP (related to the recent discussion I started on the ML): how does the switch know when to forward a frame/packet to the CPU? is this only possibly through flooding or is there another way?
<fpsusername[m]>
Harm_: What extras did you enable?
<Harm_>
fpsusername[m]: not much
<Harm_>
fpsusername[m]: you can diff it with yours
<fpsusername[m]>
I'm going to try sqm, but other than that, just the basics
<Harm_>
fpsusername[m]: you want to use it as your main router/firewall?
<fpsusername[m]>
As an access point
<fpsusername[m]>
I did get it working already, then I tried updating it with the mainline commits (and rebuild an image)
<fpsusername[m]>
That went horribly wrong, so now I'm rebuilding
<fpsusername[m]>
69 new commits from openwrt. The kernel version seems bumped as well iirc
<Harm_>
I wonder how much SQM will help on just an access point
<fpsusername[m]>
Especially the fact that 2-3 devices will be connected at most that probably won't even drain the network
<fpsusername[m]>
Just playing around with it
<fpsusername[m]>
Not going for actual micromanagement stuff
<fpsusername[m]>
Then I'd first start looking at increasing range. Not sure if changing antennas would actually allow for more range if the power stays the same.
<fpsusername[m]>
* the same. (unless I use directional antennas)
<Harm_>
The currently installed antennas are just a simple board with a trace on it, I have no idea how much worse/better it is than having external antennas.
<hauke>
xdarklight: it is controlled with the GSWIP_PCE_PMAP1, GSWIP_PCE_PMAP2 and GSWIP_PCE_PMAP3 registers
<hauke>
we also put the CPU port into the bridge with the LAN ports
<hauke>
the CPU port is a normal port it can also learn its mac address
<fpsusername[m]>
Harm_: I've never done research on PCB antennas, but it's not just a simple trace. Antennas are finetuned for their frequencies and optimal radiation (in quite some directions).
<fpsusername[m]>
The way the experia extender has two of the same ones makes me think the field is donut shaped, meaning that you have almost everywhere the same range.
<fpsusername[m]>
I do know the longer the antenna trace, the more sensitive it will be. On my desktop I upgraded from 14 to 20cm antennas, didn't do much on the dBs though. I think only 1 or 2 dB better.
<Harm_>
fpsusername[m]: right. Maybe a hi-gain antenna (those 'flatten' the donut shape right?) or directional would improve the signal in your garden.
dedeckeh has quit [Remote host closed the connection]
<Harm_>
fpsusername[m]: re: flashing without uart. When OpenWRT is booted, you can reset the config by holding the reset button on the back for 5 seconds, release when the power LED starts flashing red. maybe that allows you to regain access
valku has joined #openwrt-devel
danitool has joined #openwrt-devel
mrkiko has quit [Quit: leaving]
Tapper has joined #openwrt-devel
gladiac is now known as Guest6473
gladiac has joined #openwrt-devel
mrkiko has joined #openwrt-devel
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<mrnuke_>
aparcar[m]: That's an example config. Currently, users are expected to write their own -- not ideal, true
<mrnuke_>
One concern is mapping between lan port and port_id. Not all devices are "monotonically increasing", and the only way to know for sure is for someone to plug in a PoE sink into each port, one by one
<aparcar[m]>
ethtool -p ethX
<mrnuke_>
Gives me "Cannot identify NIC: Not supported" on my engenius,2910p. I assume that means "number of ports" ?
<mrnuke_>
If so, you can't always assume the number of PoE ports is the same as the number of ports. You have devices with a mix of normal and PoE ports, and also have ports on SFP slots
<aparcar[m]>
this seems to be a solved problem, it's encoded in upstreams 02_network
<mrnuke_>
I'm going to look into that. I was under the impression it includes _all_ DSA ports, not just the PoE ones
<aparcar[m]>
if possible please fix your script accordingly so it works automatically
<mrnuke_>
02_network predates me. He's not my son! :p
<xdarklight>
hauke: thanks! according to the GSW140 datasheet the PCE_PCTRL_3 (per port) register has a LNDIS (learning disable) bit. I cannot find this in sources for the xRX200 GSWIP. is it missing in the sources or was this bit only added in later revisions of GSWIP?
<mrnuke_>
aparcar[m]: The (edited) 02_network script gives lan1->lan10, but on my specific switch lan9, and lan10 are SFP cages
<mrnuke_>
though I see a world where the ""poe' JSON node is generated correctly, and a different script converts that into an /etc/config/poe file automatically
philipp64 has quit [Quit: philipp64]
<fpsusername[m]>
<Harm_> "me neither. I'm not on latest..." <- Doesn't build on the latest. uboot doesn't build. Will probably take an hour again to run it with V=s. Not sure if you can run the make command without remaking already made modules/parts
<aparcar[m]>
mrnuke_: I created a patch to fix the SFP issue
<mrnuke_>
Here I was lollygagging and making a function named whaaaaaaaaaaaaaaat_do_you_want() connected to SIGINT, and you guys are trigger-ready on the merge button :p
<mrnuke_>
(That was a happy joke. I'd be glad to get this package upstreamed)
<aparcar[m]>
mrnuke_: what device is that?
<aparcar[m]>
looks like we need to add a mapping then to each poe device
<mrnuke_>
it's in the hostname :)
<hurricos>
aparcar: as is done with uboot-envtools, yes?
Tapper has joined #openwrt-devel
<aparcar[m]>
hurricos: i wouldn't know
<hurricos>
is the preferred still to build a shell script?
<hurricos>
(which has case statements for each of the board models?)
<aparcar[m]>
mrnuke_: could you add a PR for your engenius switch?
slh has joined #openwrt-devel
slh64 has joined #openwrt-devel
Misanthropos has joined #openwrt-devel
<aparcar[m]>
hurricos: you're okay to become the maintainer?
<hurricos>
aparcar: pong, yes.
<aparcar[m]>
so I'd go ahead an merge then?
<aparcar[m]>
mrnuke_: if you add your device please manually specify the order of lan/poe ports, that should automatically work
<hurricos>
10357 is fine.
<mrnuke_>
aparcar[m]: sure. sounds like a plan :)
<hurricos>
/etc/board.d/* is going to be converted to ucode one day anyway, it doesn't have to look pretty, just make sense ;)
<hurricos>
aparcar: but you're talking about 18763.
<aparcar[m]>
merged. Now we should come up with a proper way to automatically create /etc/config/poe based on board.json
<hurricos>
Sounds good.
<mrnuke_>
aparcar[m]: Awesome. Then we can remove the config/poe from the realtek-poe package. Makes sense!
<mrnuke_>
hurricos: speaking of "Why have this be a part of openwrt.git, and not the realtek-poe package". To me it makes sense that this be part of openwrt config scripts, since the info originates there.
<mrnuke_>
someone adds a device, they don't need to also add a PR for realtek-poe. It's all in one place
<aparcar[m]>
let's give the package a few days before moving it to openwrt.git
<aparcar[m]>
do you two want to work on the poe config script or should I?
<mrnuke_>
You want to move realtek-poe source code to openwrt.git?
<aparcar[m]>
no, just the makefile
<aparcar[m]>
this way it's possible to have it installed by default
<mrnuke_>
okay. Makes sense.
<mrnuke_>
aparcar[m]: regarding the config script, you can have that. I'm trying to fix an annoying bug in realtek-poe regarding error replies from the MCU
<hurricos>
mrnuke_: refresh the page. I edited my comment :P
<hurricos>
mrnuke_: I had not initially realized that this was an existing script
<hurricos>
Other consideration is that there can be more than one PoE implementation out there; this unifies them all.