<Vavooon>
I'm trying to complete Zerotier support for Luci and not completely sure how to display ZT interface IP address and other standard details in Luci details. Is I send only 'proto_send_update' it shows bare minimum e.g. TX/RX packets. There's also `proto_add_ipv4_address` method which works in my case but I'm not sure whether I should use it because it reassigns IP addresses for the interface. Perhaps there's a way just to make netifd/luci/whatever
<Vavooon>
reading already assigned IP addresses from the interface and displaying them in LUCI?
<neggles>
disclaimer: i have rarely touched luci so may be telling you things you already know :P
<Vavooon>
Znevna: in Wireguard IP addresses seem to be stored in settings
<Znevna>
oh and LuCI reads it from there? x.x
<Znevna>
didn't check, sorry :p
<Vavooon>
neggles: Thanks, looks like I don't need to work with Luci directly (since it reads standard data from somewhere), but rather with netifd/ubus which is responsible for storing all the interface data
<neggles>
luci is calling into ubus i believe
<Vavooon>
It's not hard for me to read these IP settings from interface, I'm guessing that there should be some common logic which reads all the IP data directly from the interface so I don't do the same work twice
<neggles>
if you know the interface name, `ubus call network.interface.name status`
<neggles>
(unsure how that looks on the other end)
<neggles>
assuming you're working on making essentially `luci-mod-zerotier` then what `luci-mod-network` uses is probably the way to go?
<Vavooon>
Alright, thanks, will try to play with ubus and see what it returns
floof58 has quit [Remote host closed the connection]
floof58 has joined #openwrt-devel
Vavooon has quit [Remote host closed the connection]
indy has joined #openwrt-devel
Vavooon has joined #openwrt-devel
<mrkiko>
Out of curiosity - if I have a router + dumb AP setup, and two hosts communicate over wi-fi, will the traffic arrive to the router or not? Guess yes
<Znevna>
if they are on different subnets yes
<Znevna>
if not no
tlj_ has quit [Remote host closed the connection]
tlj_ has joined #openwrt-devel
floof58 has quit [Remote host closed the connection]
floof58 has joined #openwrt-devel
lmahmutov has joined #openwrt-devel
<lmahmutov>
hi2all
<lmahmutov>
can anyone help me) how to correctly debug hostapd with gdb? I rebuild it with config_debug=y but gdb cannot parse symbols
<lucenera>
Is there any way to compile OpenWrt without an overlay? The goal is to have an installation that is not overwritten with each upgrade.
<lucenera>
Have an ext4 or btrfs file system.
<oliv3r[m]>
If I have a MIPS34kc CPU, single core, with 2 VPE's, how do I setup the operating points? I can set it up all fine with a single core, and that works, but adding the operating points to the second VPE, things 'slowly grind to a halt'. I don't even think any frequency scaling is happening at this point yet (userland fixed freq). but regardless, the CPU works at the Clock, and both VPE's would always follow suite
wo0f has quit [Quit: No Ping reply in 180 seconds.]
<mrkiko>
Znevna: thanks a lot! So they work simply because the AP does broadcast ARP requests and so on?
<mrkiko>
lucenera: yeah - you would like openwrt to behave more like a classicaOS in x86_64 world, where you actually update the files - e.g.: you would like a non-image approach
<mrkiko>
lucenera: was wondering the same... but I guess you need to implement something to have that. It would be cool if sysupgrade could e.g.: interact with btrfs snapshots and so on. But I guess this is pretty out of scope for openwrt atm.
wo0f has joined #openwrt-devel
floof58 is now known as Guest3135
floof58 has joined #openwrt-devel
Guest3135 has quit [Ping timeout: 480 seconds]
lmahmutov has quit [Quit: Page closed]
<neggles>
mrkiko: the AP knows the MAC addresses that are attached to it - it functions the same as a switch does
<neggles>
if it receives a packet destined for a MAC that's in its client list, it sends it onwards (unless you enable client isolation)
<mrkiko>
neggles: thanks! Very nice.
<neggles>
if you have P2P and/or TDLS turned on, traffic doesn't necessarily even have to pass through the AP
<neggles>
so TDLS is Tunneled Direct Link Setup, 802.11z, I've no idea if openwrt supports it
wo0f has quit [Remote host closed the connection]
<neggles>
it lets two clients arrange to talk to each other directly, with the AP coordinating
<mrkiko>
neggles: and for p1p mode you need to do it yourself I gues
<neggles>
P2P is Wi-Fi Direct
<mrkiko>
neggles: and what happens when two clients on the same AP are on different subnets? I mean, how does the process go so that the router can intervene?
<neggles>
so if the destination IP is set to an IP outside the local subnet
<neggles>
the destination MAC will be the MAC of the default gateway
<mrkiko>
neggles: the client asks to gateway and so on
<neggles>
yup
<neggles>
AP only cares about MACs
<mrkiko>
neggles: ok, clear - thanks...
<neggles>
it does not care even slightly about layer-3
<mrkiko>
neggles: clear
<mrkiko>
neggles: from it's perspective, the main router is just another client with a MAC address
<neggles>
yep
<neggles>
from a logical perspective a wifi ap looks exactly like a switch, it's just one that's typically not permitted to have more than one downstream hop
<neggles>
(without 802.11s/WDS/whatever)
<neggles>
fwiw while TDLS and P2P/self-organized wifi direct are both Things That Exist
<neggles>
I have very, *very* rarely seen them used in the wild.
<mrkiko>
neggles: and that's why I guess lot of people go their way out to isolate iOT devices or devices using lot of broadcast traffic
<neggles>
it's actually worse than that
<neggles>
for broadcast
<neggles>
because broadcast traffic *has* to be sent at the lowest possible modulation rate, to ensure every client that *could* be connected will receive it
<mrkiko>
neggles: ok, so airtime usage becomes inefficient and so on
<neggles>
very. if your min MCS is set to MCS0, 6mbps of broadcast traffic can consume 95% of your airtime
<karlp>
I tried to use wifi direct p2p shit for what it's advertisd for, and while the radios might support it, the software virtually never does, total fail. :(
<neggles>
there are 4 places you'll find functional wifi direct
<karlp>
I mean, printers were obvious, and they failed at it too.
<neggles>
printers (with an asterisk, since printers are often nonfunctional independently of wifi direct), Chromecasts, AppleTVs, and... what was the other one
<neggles>
oh sorry miracast
<neggles>
note how three of those are basically the same thing :P
<karlp>
it needed both the devices to do the right thing sw wise and the other side though, and rarely did that ever line up :|
<karlp>
ap+sta mode is super shit too, but at least it was understood shit.
<neggles>
hence chromecast and appletv where the software on both ends is controlled by the same people
<neggles>
and miracast where it's pretty much "intel said do this and microsoft baked it in so i guess it's a standard now"
minimal has joined #openwrt-devel
Knogle has joined #openwrt-devel
<Knogle>
Hey friends, how are you doing
<Knogle>
I'm trying to get 19.07 built for my TP-Link EAP120. Unfortunately i had to adjust the partition sizes in /tools/firmware-utils/tplink-safeloader.c and /target/linux/ar71xx/image/generic-tplink.mk, because they were to tight in order to fit the kernel.
<Knogle>
Now a sysupgrade and factory bin are being created finally, but they are only around 3.5M of size. The flash's size is 16M. I'm connected using an external flasher. What is the appropriate way to get this thing working? I guess i somehow need the new mtd partition table on the device as well, but don't know how i can get something running without the full 16M file, and only a part of it
<mrkiko>
Knogle: it's not a problem for the factory and the sysupgrade images to be smaller than flash size, they are not meant to be a bit-by-bit image of the flash in general, not in all cases at least
<Knogle>
Ah alright. So whatever is running currently on the device might have a different partition table. So i think i have to swap out the partition table, and the rest should be part of the sysupgrade stuff. Is "mtd" maybe the way to go here?
<Knogle>
Currently there is a 18.06 instance running
<mrkiko>
Knogle: well, if you have a flash programmer and you already backed up the whole flash, you can experiment with it; given you shouldn't move calibration data around, I guess I would try to enlarge kernel partition and move rootfs ahead, if that's possible and this is not one of the devices where rootfs is "behind" kernel. But hope more knowledgeable devs will help you
madwoota- has joined #openwrt-devel
<Knogle>
Alright thanks a lot :) The layout is more or less like this. spi0.0:128k(u-boot)ro,64k(partition-table)ro,64k(product-info)ro,2560k(kernel),13312k(rootfs),192k(config)ro,64k(ART)ro,15872k@0x40000(firmware)
valku has joined #openwrt-devel
<Knogle>
Removed 1024k for the rootfs, and added 1024k for the kernel
<mrkiko>
Knogle: that's what I would have tried - doesn't mean it's correct tough :D
<mrkiko>
Knogle: and how much RAM?
<Knogle>
It has 64M of RAM
<Knogle>
It's not the best thing though, but i have received 31 of them. They were meant to be scrapped
<mrkiko>
Well, it maybe worth it to port it to ath79 target if you feel like it
<Knogle>
And i think they are at least capable of running something more recent then 18.06, even without LuCi
<Knogle>
But i don't get the kernel stripped down to less than "1567k"
madwoota has quit [Ping timeout: 480 seconds]
madwoota- is now known as madwoota
<mrkiko>
Knogle: does the bootloader refuse to boot larger kernels?
<Knogle>
Well i could try. I have almost 0 experience with this stuff, but i love to try things out
<Knogle>
To be honest i don't know. But the tool which builds the sysupgrade and factory bin files in tools/firmware-utils/tplink-safeloader had limits set for the partition sizes. So i had to adjust these limits as well by 1024k, in order to get a sysupgrade and factory bin file thrown out
<mrkiko>
Knogle: so you might need to try out yourself to see if enlarging the kernel e.g.: by 2MB or more works or not. Or maybe the git history has some useful informations. But you're in a good position to try out things
<Knogle>
Alright, right now i'm doing my first try using the sysupgrade image heh
<Knogle>
Well it didn't like it haha, it's stuck in the tp link bootloader. But maybe someone else has an idea on how to proceed in this case here
<mrkiko>
may you paste the log to a pastebin?
<mrkiko>
Knogle: if you are able to determine empirically the problem is the kernel size, you may consider using one of the "loaders" present in openwrt ... but I wouldn't be able to suggest you what would be the right choice. Still, if you hang on here you may have some chances of getting more informations. Also, the mailing list can bea great place to ask
<Knogle>
Yeah, as i see, the TP Link safeloader loads uboot, and uboot loads the kernel
<Knogle>
Seems like something with the partition table is broke, i'll send you a pastebin
<Knogle>
Btw. is it possible to use "mtd" on a flash dump, or is there something useful i can get out of the flash dump?
<Knogle>
That's what i get on the serial console. Allocated memory for elf segment ok: addr: 0x80060000, size 0x189621
<Mangix>
nick[m]12: uhhhh parts of that patch look like they should be migrated to target/linux/generic/hack-5.10/300-MIPS-r4k_cache-use-more-efficient-cache-blast.patch