<dhewg>
nbd: mesh parameters aren't set, like mesh_gate_announcements. there's code for it, but they won't apply correctly since apparently they can only be set while joining/joined?
<dhewg>
there's e.g. iface_start() in wdev.uc, but that code path apparently isn't hit at all when having wpad?
<dhewg>
weird, maybe my installed openwrt it too old, lemme try to update it all
ar7ch has quit [Quit: ar7ch]
ar7ch has joined #openwrt-devel
<dhewg>
nbd: ok, seems to work after a full update. thanks and sry for the noise!
<dhewg>
do you know if those values are guaranteed to be always active now? I'm a bit confused about the mesh11sd package with its claims and approach. With your patch that's completely redundant now, right?
<nbd>
the mesh config should stay the same as it was applied initially
<nbd>
i think mesh11sd does some more things like dynamically reconfiguring the portal settings and potentially other parameters
<nbd>
not sure what that's needed or useful for
rua has quit [Quit: Leaving.]
<nbd>
pushed the fixes to master
ar7ch has quit [Quit: ar7ch]
<dhewg>
I think that resolves #9370 then :)
<nbd>
backported to 23.05 as well now
ar7ch has joined #openwrt-devel
<robimarko>
linusw: Is there are reason why CONFIG_USB_FOTG210 is built as a module?
<robimarko>
And, why USB HOST support is disabled on Gemini?
<linusw>
robimarko: not all Gemini devices have physical USB ports
<linusw>
but USB HOST should be activated, probably a mistake if it isn't :(
<robimarko>
linusw: I get that, but its built as a module in the target config
<robimarko>
Not as a kmod package
<linusw>
robimarko: oh that can be deleted I guess
<linusw>
the kmod package should override it
<linusw>
robimarko: are you patching it or should I?
<robimarko>
linusw: It probably does and selects USB HOST support which then prompts for CONFIG_USB_FOTG210_HCD as in the last kernel config refresh it was dropped
<linusw>
robimarko: that looks very much like a mistake that I would do :D
<robimarko>
linusw: No worries, I will make a PR to drop the driver selection via kernel config
<robimarko>
Enable USB HOST support and fixup the module
<linusw>
Excellent thanks!
rua has joined #openwrt-devel
rua is now known as Guest6265
rua has joined #openwrt-devel
Guest6265 has quit []
<ar7ch>
PaulFertser: so it turns out that hostapd is indeed not used. LuCI "ProxyARP" checkbox corresponds to `uci set wireless.<radio-iface>.proxy_arp='1'`. `/etc/config/wireless` is not changed. Here is the demo snippet showing that proxy_arp attribute is activated on the bridge: https://gist.github.com/ar7ch/235723c40314701737b8a9402fcd8c6b
<ar7ch>
However, in this case only `proxyarp_wifi` attribute is activated on the bridge. For my use case, I also need `proxyarp` attribute to be set. So I would like to find the code handling changes in UCI configuration where those attributes are applied
<ar7ch>
As I understand, changes in UCI configuration are applied on `ubus call network reload`, so I’m looking for the code that handles this method…
<ar7ch>
Is there a way to tell which process registered given object on ubus? i.e. I need to know who handles `network.reload` method
<robimarko>
linusw: Is CONFIG_USB_FOTG210_HCD required as well?
<linusw>
robimarko: I have a patch set making use of it on the D-Link DNS-313
xback has quit [Ping timeout: 480 seconds]
<robimarko>
linusw: Then I will add it to the kmod
<robimarko>
As it was selected via kernel config but not in the kmod
<linusw>
Ah OK thanks
<robimarko>
Ah, not CONFIG_USB_FOTG210_HCD but CONFIG_USB_FOTG210_UCD
<robimarko>
Basically, it enables OTG support AFAIK
<linusw>
Actually that was what I meant
<robimarko>
Great, will enable it
<linusw>
HCD is host controller so you usually want that, UCD is the device controller, and on DNS-313 I have some hacks that turns its device port into a serial console.
<linusw>
I need to finish that patch set.
ar7ch has quit [Quit: ar7ch]
<hitech95>
Hi, who is the maintainer for mediatek/filogic? Me and friend have a PR open since July. I know we're all busy, and I don't want to come across as pushy. I would just like to understand if there is something I have to fix to get the PR approved.
<robimarko>
I see that config is being set properly and USB kmod is being built
rua has quit [Ping timeout: 480 seconds]
SnarlingFox has joined #openwrt-devel
<SnarlingFox>
Hi folks, I'd like to try and get a Zyxel WSQ50 ported to OpenWrt but I'm not quite sure where to begin. I've managed to get full access to both the operating system (seemingly based on an older OpenWrt release) and the u-boot boot loader.
<SnarlingFox>
I have a bit of Linux and some basic u-boot experience but I really don't know what I'm doing :D
<SnarlingFox>
Whatever Zyxel built this from, it seems derived from Attitude Adjustment 12.09-rc1 based on the outputs of /etc/openwrt_release
ar7ch has joined #openwrt-devel
Daaanct12 has quit [Quit: WeeChat 4.1.1]
rua has joined #openwrt-devel
<robimarko>
First question is what the SoC being used is
<SnarlingFox>
It appears to be ARMv7 based around the IQP4019 SoC.. Can I post pastebin links here?
<SnarlingFox>
I've got a pastebin log of the full boot up sequence dumped from UART
ar7ch has quit [Quit: ar7ch]
<robimarko>
If its IPQ4019 then it should be easy as the SoC is well supported
<robimarko>
And there are plenty of examples
<SnarlingFox>
Just neatening up the soldering to the UART, one sec
<linusw>
robimarko: looks good to me, approved.
<robimarko>
linusw: Thanks
ar7ch has joined #openwrt-devel
<PaulFertser>
ar7ch: hey
<PaulFertser>
ar7ch: "uci set wireless..." call certainly changes /etc/config/wireless after commit.
<PaulFertser>
ar7ch: and then it should be handled by https://git.openwrt.org/?p=openwrt/openwrt.git;a=blob;f=package/network/services/hostapd/files/hostapd.sh;h=271c1f7becd27064f3e244298e6058abb5f06177;hb=HEAD#l1152
<PaulFertser>
So it goes into hostapd.conf.
<PaulFertser>
You can see that hostapd.sh script handles all the other options in /etc/config/wireless too.
<SnarlingFox>
Huh, @robimarko I tidied up the soldering, reconnected a USB TTL adaptor and thought I'd try and tftpboot from another image that had the same SoC (zyxel_nbg6617-initramfs-uImage.itb).. it only went and booted up fine :)
<robimarko>
SnarlingFox: Please dont boot random images just because its the same SoC
<stintel>
lol glinet really doesn't want their customers contacting them. mail is rejected due to SPF failure (which is incorrect), form on website says captcha failed while it shows the green V
<SnarlingFox>
Oh I admit it's not ideal, but when you said "there's lots of examples" you _kinda_ led me there ;) but at least we now know it's "capable" of running a vanilla OpenWrt build, even if the wireless interfaces aren't showing (yet) nor the UI loading. It's a starting point.
<stintel>
sigh
<SnarlingFox>
I could do with some advice on what to do / how to do this _properly_
<SnarlingFox>
(I've already taken a full `dd` backup of mmcblk0 / mmcblk0boot0 / mmcblk0boot1 to USB)
aiyion has quit [Remote host closed the connection]
<SnarlingFox>
Ignore the no webui build - was just the firewall
aiyion has joined #openwrt-devel
<SnarlingFox>
Managed to get into the luci UI :)
<PaulFertser>
SnarlingFox: usually when adding support you look at a reasonably recent commit that added support for similar board and then work from there, starting with preparing DTS, sometimes using hints from vendor boot log etc.
<PaulFertser>
SnarlingFox: booting initramfs images is good idea at that stage if it works.
<robimarko>
Vendor DTS is also sometimes usefull, though usually also contains a bunch of nonsense
<SnarlingFox>
That's pretty much what I did Paul, it's booted from initramfs over tftp at the moment
<SnarlingFox>
Can SSH in (as well as having UART access) and the luci webui works too
<SnarlingFox>
There's no wireless adaptors but I'm not worried about that yet
<SnarlingFox>
I don't know how to dump the device tree though, or whether it's even now relevant / accurate booted from another targets initramfs
<SnarlingFox>
My hope is to get this thing fully supported because the hardware is still pretty good but Zyxel's software is atrocious
<SnarlingFox>
(And their support even worse "We'll consider allowing wireless channel changes in another product")
<PaulFertser>
SnarlingFox: do you have access to vendor's command line via UART probably? Then you could dump via /sys/firmware
<SnarlingFox>
I've got full access to the vendors u-boot and Linux environment, their root password was amazingly complex! "1234" XD
<robimarko>
I usually static compile dtc, or if I have access to the stock FW upgrade or dump just get it from there
<SnarlingFox>
Lemme reboot into their firmware, one sec
<PaulFertser>
robimarko: do vendors never alter DT in bootloader before passing to the kernel?
<robimarko>
PaulFertser: Depends on the vendor
<robimarko>
Most dont bother to modify U-Boot at all
<robimarko>
So, it will only add the QCA specific tags and update MAC-s
<PaulFertser>
I see, so extracting from update is as good as dumping via /sys/firmware then
<SnarlingFox>
I don't think we'd need to tamper with u-boot on the device, there's even a firmware update from USB option built in, if you boot up with specially crafted files on a USB stick it just flashes the mmcblk0 partition
<SnarlingFox>
I don't believe installing OpenWrt in future would even require UART access
Mangix has quit [Ping timeout: 480 seconds]
<SnarlingFox>
That said, something has changed, I can no longer log in with root / 1234 - just booted into failsafe mode, ran `passwd` and now booting back up again to see if I can get back in. Worst case I just wipe it and start again.
minimal has joined #openwrt-devel
<SnarlingFox>
Right, there is nothing in /sys/firmware/ when booted up from the factory image
<SnarlingFox>
There was chatter from the boot loader about loading a dtb though
<robimarko>
It always loads the DTB, its bootloaders responsibility to pass it to the kernel
<SnarlingFox>
Is it required to make an official OpenWrt image?
rsalvaterra has quit []
rsalvaterra has joined #openwrt-devel
<mrkiko>
robimarko: are there situations where booting an initramfs image can be dangerous?
<mrkiko>
SnarlingFox: whatever you do, remember to full backup the flash, all of it
<mrkiko>
SnarlingFox: especially calibration data, but backing it up fully is always good
goliath has quit [Quit: SIGSEGV]
<mrkiko>
SnarlingFox: the devicetree describes how the device is made to the kernel; so yes, usually it is needed. You are being very lucky here to get networking working and so on, but a proper DTS (S = source) is needed to describe things like where to find MAC addresses in flash; compiling a DTS results in the DTB.
<SnarlingFox>
mrkiko: I believe I've done that properly, I took a dd of mmcblk0 and its boot0 and boot1 partitions
xback has joined #openwrt-devel
ar7ch has quit [Quit: ar7ch]
Mangix has joined #openwrt-devel
SnarlingFox has quit [Remote host closed the connection]
<ar7ch>
by the reasons I described earlier, hostapd patched with `740-snoop_iface.patch` is not able to enable on its own proxy arp for the bridge
<PaulFertser>
ar7ch: no, it's generated for hostapd by the shell script
<PaulFertser>
ar7ch: it's not an "additional custom option" from uci config point of view.
dangole has joined #openwrt-devel
robimarko has quit [Remote host closed the connection]
minimal has quit [Quit: Leaving]
ar7ch has quit [Quit: ar7ch]
dangole has quit [Ping timeout: 480 seconds]
xback has quit [Remote host closed the connection]
xback has joined #openwrt-devel
dangole has joined #openwrt-devel
<hitech95>
dangole, ping
Danct12 has left #openwrt-devel [A-lined: This user has been AViVA-lined!]
Danct12 has joined #openwrt-devel
robimarko has joined #openwrt-devel
xback has quit [Remote host closed the connection]
xback has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
dangole has joined #openwrt-devel
dangole has quit [Ping timeout: 480 seconds]
<ldir>
Here's a 'fun' one - for $reasons I wish to have the reflection rules put in by a port forward WITHOUT having the port forward actually work from the outside world. Any ideas?
<dwfreed>
just make a NAT for the lan zone instead
ar7ch has joined #openwrt-devel
<ldir>
it must reflect based on the external internet facing IP address though.
<dwfreed>
what are you actually trying to do
<ldir>
I have an internal server which ddns's to a specific domain and internet IP address. I do NOT wish to allow internet access to this server from the internet, BUT for certificate reasons I do want to use that domain when access internally. The domain resolves to the external internet address.
rz has quit [Remote host closed the connection]
<ldir>
fw4 puts in 2 reflection rules which do what I want and it also puts in (what everyone else wants but I don't) the rule to do the WAN > LAN forwarding.
ar7ch has quit [Quit: ar7ch]
<dwfreed>
why not just override the name in dnsmasq to point to the local IP?
<dwfreed>
and then you don't have to worry about NAT at all
<ldir>
hmmmm. Worth a try!
<ldir>
my current workaround is to restrict the source IP & source port to something unlikely to appear on the WAN interface
ar7ch has joined #openwrt-devel
dangole has joined #openwrt-devel
goliath has joined #openwrt-devel
goliath has quit []
xback has quit [Remote host closed the connection]
SnarlingFox has quit [Read error: Connection reset by peer]
SnarlingFox has joined #openwrt-devel
dgcampea has quit [Remote host closed the connection]
xback has quit [Remote host closed the connection]
xback has joined #openwrt-devel
<ar7ch>
PaulFertser: I put some tracing and debug logs to hostapd. you’re right, the options are adjusted in `/var/run/hostapd-<phy>.conf` according to changes in uci (even without `uci commit`). However, as I described before, hostapd can’t set bridge parameters to 1 because of patched condition (also, seems like the radio interface is not added to the bridge during the initialization of hostapd, since the brport directory does not exist, and a write would
<ar7ch>
fail anyway).
<ar7ch>
jow: thank you for pointing out. so the actual attributes change is carried out in netifd:https://git.openwrt.org/?p=project/netifd.git;a=blob;f=system-linux.c;h=9097f6a56a1f2cb233de1ab74950d1653db01a0b;hb=HEAD#l549
<ar7ch>
and https://git.openwrt.org/?p=project/netifd.git;a=blob;f=system-linux.c;h=9097f6a56a1f2cb233de1ab74950d1653db01a0b;hb=HEAD#l943 . it also shows that only proxyarp_wifi attribute is assigned to the bridge port
ar7ch has quit [Quit: ar7ch]
ar7ch has joined #openwrt-devel
ar7ch has quit [Remote host closed the connection]
ar7ch has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
dangole has joined #openwrt-devel
SnarlingFox has quit [Remote host closed the connection]
hitech95 has quit [Read error: Connection reset by peer]