<neggles>
@PaulFertser: thanks! yeah, I found the script, but I'm not entirely clear on the syntax
<neggles>
is it just 'caldata_extract "<art partition name>" <start_address> <length>' ?
<neggles>
on radio disablement, it's actually the other way around; the factory image for these APs has a hardcoded cmdline with 'BOARD=ap55c' / 'BOARD=ap100c'
<neggles>
if you load the AP100C uImage into an AP55C, it turns into a 3x3 AP
<neggles>
same with the AP55, you just have to plug in a third u.fl to SMA pigtail to get the third 5GHz chain
<neggles>
the sophos GPL package for these is actually really good??? it includes (1) all of the source tarballs for opensource components, (2) all the patches and config files they used, (3) a list of every build command executed, (4) the resulting ipkg files from the build, (5) prebuilt kernel and full OS images from all stages of the build process, including ones with debug symbols still present
<neggles>
i'm pretty sure I could dump all of this into a copy of the 15.05 tree and have a fully functional build environment
hexa- has quit [Ping timeout: 480 seconds]
rua has quit [Quit: Leaving.]
hexa- has joined #openwrt-devel
hexa- has quit [Ping timeout: 480 seconds]
hexa- has joined #openwrt-devel
<neggles>
well i'll be damned
<neggles>
it booted
hexa- has quit [Ping timeout: 480 seconds]
hexa- has joined #openwrt-devel
<neggles>
@PaulFertser: Available Antennas: TX 0x3 RX 0x3
<neggles>
on both radios (this is an AP55C)
<neggles>
literally the only difference between the US$295-sticker AP55C and US$595-sticker AP100C is... a config string. Even on the factory firmware it reports 3x3 in 'iw list', but it's configured to only use two chains...
CN_SZTL has quit [Read error: Connection reset by peer]
Rentong has joined #openwrt-devel
hexa- has quit [Ping timeout: 480 seconds]
hexa- has joined #openwrt-devel
hexa- has quit [Ping timeout: 480 seconds]
Rentong has quit [Ping timeout: 480 seconds]
hexa- has joined #openwrt-devel
hexa- has quit [Ping timeout: 480 seconds]
hexa- has joined #openwrt-devel
hitech95 has joined #openwrt-devel
<PaulFertser>
neggles: hey, I see.
<PaulFertser>
neggles: that's quite an uncommon solution. And I assume it's not appropriate to try to use the third chain if nothing is connected to it.
<hitech95>
Hi, I´m working on a custom status page in LUCI, I´ve defined a modal using the s.addModalOptions = function(s)... the modal opens but it is always empty. Teh render function is called but no html code is attached to the dom section. Any idea?
<PaulFertser>
neggles: I do not see a clean solution to this basically purposefully crippled hardware problem other than documenting it in the wiki. Hm, another idea: like the MAC is patched in extracted ath10k "eeprom" you can also patch out the third chain and provide instructions on the wiki how to modify that variable if the end user decides to add the missing antenna.
<PaulFertser>
neggles: TX 0x3 RX 0x3 means only two antennas (0x1 | 0x2) are enabled, not three.
<PaulFertser>
neggles: please recheck, something's still fishy.
<neggles>
ah okay
<neggles>
I should clarify: there's two physical models, the AP55C/AP100C and AP55/AP100. 55C and 100C are physically identical; both boards have all three radios, all the antennas for them, etc; though it appears that the ART does indeed only contain two chains worth of calib data
<neggles>
sorry, two radios, three chains
<neggles>
the AP55/100 models have internal 2.4GHz antennas and external 5.8GHz, again both have all 3 chains present on-board and all 3 2.4GHz antennas, but the AP55 is missing one of the 5.8 externals.
<neggles>
So, this is fine, all working as you'd expect - sorry for the confusion! I've worked out where they keep the MACs, too, which is nice.
rua has joined #openwrt-devel
<neggles>
er, *though it appears that the ART does indeed only contain two chains worth of calib data for the 55/55C
<PaulFertser>
neggles: so no additional disabling needed? That's good in the sense it'll work properly out of the box and bad in that just adding a pigtail won't make the third chain work as it's missing the essential data, right?
<neggles>
appears so
<neggles>
I'm showing "3 streams: not supported"
<PaulFertser>
neggles: so do you have any other questions still not clarified?
Rentong has joined #openwrt-devel
f5- has joined #openwrt-devel
<neggles>
I _think_ I've got it mostly sorted
<neggles>
with 11-ath10k-caldata, if I need to extract board.bin, I do that under the 'ath10k/pre-cal-pci-0000:00:00.0.bin' section, yes?
rua has quit [Quit: Leaving.]
<PaulFertser>
neggles: it really depends, you see there's pci bus number mentioned in the filename.
<PaulFertser>
neggles: if you just enable &pcie0 and you see the card detected in dmesg then yes, that filename is what you would expect.
<neggles>
the original firmware extracts, cal-pci-0000:00:00.0.bin and cal-pci-0000:00:00.1.bin, along with board.bin
<neggles>
PaulFertser: see https://i.imgur.com/34R65wn.png - I believe I've got the ath9k / 2.4GHz covered with mtd-cal-data = <&art 0x1000>; in the DTS
<neggles>
since that came up just fine without extracting it
Rentong has quit [Ping timeout: 480 seconds]
<neggles>
and boot log shows it successfully reading the caldata from flash; the ath10k device is definitely at 0000:00:00.0, all I had to do to make it show up was turn on &pcie0
<neggles>
so, that should work fine - will run a build and see how it goes
dedeckeh has joined #openwrt-devel
Borromini has joined #openwrt-devel
<PaulFertser>
neggles: yeah
<neggles>
PaulFertser: looking at that OEM script again, isn't it just extracting the caldata for the 5G radio twice? board.bin being 0x5000-0xFFFF and caldata being 0x5000-0x5FFF?
<neggles>
This Seems Weird
<neggles>
the whole area from 0x5844 to the end is all FF bytes
<neggles>
but if I cut them out, it segfaults, so aaaaaaalrighty then!
<PaulFertser>
neggles: I suggest you follow what the existing ports for same wireless are doing, as vendor script might be tweaked to make out of tree driver happy.
<neggles>
PaulFertser: I did something wrong the first time, it works following the same process as the others now :) neat!
<PaulFertser>
neggles: good news! :)
<neggles>
thankyou very much for the help :)
<PaulFertser>
neggles: I didn't do much but hope some encouragement helped :)
<neggles>
hey, you did enough :P now I just need to sort out how much of the DTS i put in a shared file vs individual model files
<neggles>
and whether I should be basing off 21.02 or master branches...
<PaulFertser>
neggles: master
<PaulFertser>
neggles: dtsi should include as much as possible
<PaulFertser>
neggles: and if the identical DTS can be used on several boards (i.e. they are similar enough, identical from software point of view) probably good not to split it at all.
<neggles>
I need a slight change for the AP100/AP55, they have a USB port that the AP100C/AP55C doesn't have
<neggles>
otherwise they're all the same
<neggles>
OEM firmware only has a single mach-* file for the lot of them
<PaulFertser>
neggles: hm, that USB port, is it really hard to retrofit?
<neggles>
on the 55C/100C, it's an unpopulated footprint for a micro-USB connector which is also missing a few jellybean components
<neggles>
the controller's there, and if you turn it on nothing bad happens, it's just useless
<PaulFertser>
neggles: hm, if you get 5 V DC-DC module and micro USB-B connector and just short R327 and R329 won't it work?
<Namidairo>
that sounds like a terrible hack for an extra usb port
<neggles>
well, for a usb port
<neggles>
tbh it might
<neggles>
the OEM firmware doesn't use the USB port at all on either model; that beeper seems to have been hand-disabled at the factory, too
<neggles>
it might just be on a weird GPIO, poking at it with a multimeter atm
<Namidairo>
does it beep when you pull up the gpio
<Namidairo>
lol
<neggles>
OEM firmware doesn't specify which GPIO it's on >:(
<Namidairo>
I'd say try them all but that's usually how people end up factory resetting their devices accidentally
<neggles>
it's funny you say that because I did that
<neggles>
and it rebooted
<neggles>
there's also a reset button which appears to not actually be connected to the pin the OEM firmware says it's connected to, and is _completely inaccessible_ without disassembling the thing
<PaulFertser>
Namidairo: what's terrible about it?
<neggles>
I think the USB circuit there is for OTG switching between host/device mode
<PaulFertser>
neggles: doesn't look like that, to the left of the connector it's just missing components for the power supply, and to the top of it is the regular protection circuity.
<neggles>
oh that's interesting
<neggles>
the official use for that port is to add a 2nd LAN
<PaulFertser>
Namidairo: have you seen that USB hack where people were literally sanding off a bit of plastic packaging for the main SoC to get to the two QFN pads needed for USB host?
<neggles>
hey I'm sure if I bridge the 5V and either bridge or install the 67R resistors I'll get functional USB
<Namidairo>
don't breathe that in
<neggles>
so I guess there's no harm in turning it on
<PaulFertser>
neggles: if you have some self-powered device (e.g. a USB hub or printer) you can just try shorting those two resistors I'd say.
<neggles>
I am quite well versed in the art of "it's not spec compliant, but it works" when it comes to USB
<PaulFertser>
neggles: HS USB shouldn't have 67R in series I think.
<neggles>
27R? it's some resistance or another
<PaulFertser>
22 I guess
<neggles>
ah yep
<neggles>
I stopped bothering with them once I realized every device I've routed USB traces for has been FS only
<PaulFertser>
0R should work too
<neggles>
and you can run FS over coathangers
<Namidairo>
not really useful for putting a network interface on
<Namidairo>
unless you can stand 10mbit in this era
<neggles>
you can get away with all kinds of things even with HS USB, which this has
<PaulFertser>
I guess it's HS capable actually.
<neggles>
as long as you don't try to hang a 5M cable off it
<PaulFertser>
And HS is how raspberry pi (<4) offers Ethernet :)
<Namidairo>
with the right components obviously
<neggles>
ehhh the resistors are a good idea but with typical 1-2 meter USB cable lengths it's fiiiiine
<Namidairo>
they now do the sane thing and hang everything off PCI-E\
<Namidairo>
...1x
<PaulFertser>
Other vendors have Ethernet MACs integrated though ;)
<neggles>
PCIe 2.0 x1
<neggles>
but the MAC is on the SoC
<neggles>
pi4 has dedicated ethernet MAC + PCIe 2.0 x1 for the VIA USB3.0 chip
<neggles>
unfortunately the IP they used for the PCIe controller is positively archaic and not quiiiite spec-compliant
<Namidairo>
there are some interesting people who disconnect the usb controller and replace it with a raid controller or other weird things
<neggles>
hello yes hi it's me
<neggles>
we use the CM4 now
<PaulFertser>
Yes, with a PCB that does pass through for the PCIe lanes.
<Namidairo>
that would make too much sense
<neggles>
look it's just really fun to plug an RTX 2080 into a Pi CM4IO board okay
<neggles>
it doesn't work, but it's fun
<PaulFertser>
Doesn't work due to PCIe BAR windows sizes?
<neggles>
no drivers, and the Pi PCIe controller does not support 64-bit transfers
<neggles>
32-bit transfer size only
<PaulFertser>
Heh
<neggles>
this also breaks the Google Coral accelerator module
<Namidairo>
no ML acceleration for you
<neggles>
you can use it in USB2.0 mode
<neggles>
Gumstix make a CM4-to-CM3 adapter with the Coral accelerator on it; their workaround was to put a PCIe to single-port USB3.0 chip on there as well
<neggles>
dual-port 10G NIC works, though
<neggles>
you can only push about 3.6-4Gbps through it, but, details
<neggles>
hmm, either this reset button needs another GPIO to pull it up, or it's not connected either
<neggles>
aaaand setting GPIO16 to an output will reset it
<neggles>
that's. interesting. OEM firmware says GPIO16 is BTN_WPS
<neggles>
oh, no, i'm wrong, it's gpio11, watchdog timer
<hitech95>
Hi, I´m working on a custom status page in LUCI, I´ve defined a modal using the s.addModalOptions = function(s)... the modal opens but it is always empty. The render/addModalOptions function is called but no html code is attached to the dom section. Any idea?
philipp64|work has quit [Quit: philipp64|work]
dedeckeh has quit [Quit: Page closed]
goliath has joined #openwrt-devel
rejoicetreat has joined #openwrt-devel
hitech95_ has joined #openwrt-devel
hitech95 has quit [Read error: Connection reset by peer]
Rentong has joined #openwrt-devel
rua has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
aleksander has joined #openwrt-devel
rua has quit [Quit: Leaving.]
rua has joined #openwrt-devel
jlsalvador has quit [Quit: jlsalvador]
jlsalvador has joined #openwrt-devel
Rentong has joined #openwrt-devel
madwoota has quit [Read error: Connection reset by peer]
madwoota has joined #openwrt-devel
Rentong has quit [Remote host closed the connection]
Rentong has joined #openwrt-devel
jlsalvador has quit [Quit: jlsalvador]
rua has quit [Remote host closed the connection]
rua has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
aleksander has quit [Quit: Leaving]
Rentong has joined #openwrt-devel
Rentong has quit [Remote host closed the connection]
Rentong has joined #openwrt-devel
rua1 has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
goliath has quit [Quit: SIGSEGV]
rua1 has quit [Quit: Leaving.]
Rentong has quit []
Rentong has joined #openwrt-devel
rua has joined #openwrt-devel
goliath has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
hitech95__ has joined #openwrt-devel
hitech95_ has quit [Read error: Connection reset by peer]
greearb__ has quit [Read error: Connection reset by peer]
greearb_ has joined #openwrt-devel
KGB-0 has quit [Ping timeout: 480 seconds]
danitool has joined #openwrt-devel
greearb__ has joined #openwrt-devel
greearb_ has quit [Ping timeout: 480 seconds]
greearb__ has quit [Read error: Connection reset by peer]
greearb_ has joined #openwrt-devel
greearb_ has quit [Read error: Connection reset by peer]
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
notupus has joined #openwrt-devel
<notupus>
does anyone have any idea why the dsa is blocked for the lantiq target?
<PaulFertser>
notupus: can you please rephrase the question?
valku has joined #openwrt-devel
notupus has quit [Remote host closed the connection]
notupus has joined #openwrt-devel
<notupus>
DSA for the lantiq target was merged into master on jun 26 and xdarklight asked for it to be included in 21.xx however afaik no action to merge it has been taken. Also for some reason it disconnected me, sorry.
notupus has quit [Remote host closed the connection]
notupus has joined #openwrt-devel
<xdarklight>
notupus: to be precise: Martin Schiller asked for inclusion in 21.x - I'm happy that it's in master (as that's what I use)
<notupus>
xdarklight: Yeah sorry, didn't remember your full name, I also do custom builds from master most of the time but I wondered if someone had reported some more issues with it
notupus has quit [Remote host closed the connection]
notupus has joined #openwrt-devel
notupus has quit []
<Habbie>
fpsusername[m], so did you try my exploit on any updated devices? :)
<fpsusername[m]>
Didn't have the time to try it
<Habbie>
ack
lmore377 has quit [Quit: No Ping reply in 180 seconds.]
lmore377 has quit [Quit: No Ping reply in 180 seconds.]
lmore377 has joined #openwrt-devel
philipp64|work has quit [Quit: philipp64|work]
philipp64|work has joined #openwrt-devel
<grid>
hurricos: i just use /etc/hosts for dnsmasq, it does not care what domains or ips are in there
<grid>
that is, it will happily serve up anything you put there. for ptr records, it will pick the very first one for an IP if there are multiple lines, or multiple entries on a line
<grid>
10.0.0.10 foo bar\n10.0.0.10 baz, it will make 10.0.0.10 resolve to foo
<grid>
killall -HUP dnsmasq gets it to reload /etc/hosts without restarting
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
Borromini has quit [Quit: Lost terminal]
<hurricos>
grid: I neded up using umdns to figure out what I needed :D
<hurricos>
umdns lets you collect a *really* good view of what the broadcast domain thinks you should use to find e.g. _ipp._tcp
<hurricos>
(an IPP-compatible printer via TCP)
<hurricos>
PTR records are largely supposed to be dynamicallygenerated by e.g. avahi-daemon
<hurricos>
question: I'm porting Archer 'AC750' c2 v5. It's identical to the c2 v1 except the switch is swapped from realtek,rtl8367b to qca,qca8337
<hurricos>
Problem: I see no 'regular' boards using QCA8337 o_e
<hurricos>
never mind! I should have checked `git log --grep=QCA8337`
<grid>
ah mdns
mrkiko has quit [Ping timeout: 480 seconds]
<mangix>
hurricos: that switch is being converted to DSA for ipq806x platform
<hurricos>
mangix: I saw, it's already done for whatever MX65 is, too
<mangix>
I tried playing with it on ath79 but got nowhere. I might need to use the upstream ethernet driver
<hurricos>
I bet mediatek is going to be a 'really fun' DSA port.
<hurricos>
:\
<hurricos>
Do you happen to know how to find tplink hardware revisions, etc. for a given board?
<mangix>
huh?
<hurricos>
err, the switch, when attached to ramips.
<mangix>
mediatek should already be DSA
<hurricos>
oh.
<mangix>
same with ramips, mt7621 specifically
<hurricos>
oddly enough the swconfig binary is still in the archer c2 v1
<hurricos>
though it's a realtek switch
<hurricos>
I hope I'm not just being a fool RE: swconfig being there btw
<mangix>
oh that needs to be converted
<mangix>
there's an upstream driver for that switch
<hurricos>
gotcha. Yeah it's mt7620a
<hurricos>
ahh
<hurricos>
ok
<hurricos>
mangix (or anyone else): Do you happen to know how to find tplink hardware revisions, etc. for a given board?
<mangix>
RTL8367RB spexcifically
<mangix>
I do not. maybe grep through /dev/mtd1
<hurricos>
yeah, not exactly sure why they went from realtek + mediatek to QCA for both to qca + mediatek