<mrnuke>
Am I the only one that's getting increasingly irritated with the inconsitent maintainer requirements of submitting kernel patches, and the idotic idiosyncraes of how scripts/get-maintainer.pl works when a patch series touches mutliple subystems?
skynet2 has quit [Ping timeout: 480 seconds]
SlimeyX has quit [Remote host closed the connection]
vincejv has joined #openwrt-devel
vincejv has quit [Remote host closed the connection]
<jakllsch>
mrnuke: heh
<jakllsch>
linux is a PITA to contribute to
vincejv has joined #openwrt-devel
vincejv has quit [Remote host closed the connection]
vincejv has joined #openwrt-devel
<mrnuke>
jakllsch: yeah. Why are the maintainers so angry?
mcbridematt has quit [Read error: Connection reset by peer]
xes has joined #openwrt-devel
mrnuke has quit [Ping timeout: 480 seconds]
xes has quit [Quit: bye..]
<mrkiko>
stintel: whats's the cable category needed for 10GB?
<mrkiko>
stintel: ethernet I mean
<mrkiko>
stintel: asking because I am trying to cable the home
robimarko has joined #openwrt-devel
<Mangix>
mrkiko: IIRC 7a
xes has joined #openwrt-devel
<Mangix>
even 6 works at short distances
<mrkiko>
Mangix: thanks a lot
<mrkiko>
Mangix: the electrician proposes cat7, doesn't tell if it's 7a or not
<mrkiko>
s/tell/say/
<dwfreed>
mrkiko: you only need 6a for 100 meter 10 GBASE-T
<dwfreed>
so 7 is plenty
<\x>
ive ran 10 GbE on cat5e ;) but ofcourse it was like < 10m
<dwfreed>
if you can get away with 55 meters, cat 6 also works
<dwfreed>
\x: if you've got 350 MHz Cat5e, you can do the 55+ meters that cat 6 can officially do
<\x>
yeah
<SwedeMike>
I've opted for regular cat6 since cable runs are less than 30 meters and it's easier to handle
<\x>
theres a lot of people that ran 10GbE on old lines without pulling new cables in and since theyre short enough, it ran fineeee
<SwedeMike>
yeah, standards have quite significant margins for bad cabling.
<SwedeMike>
so it often works anyway
<SwedeMike>
one thing to take into account is to have properly sized conductors so PoE works
<\x>
get some 24 awg ones
xback has joined #openwrt-devel
* f00b4r0
is running GigE on 1970s 32-pair telephone cable. Longest run 105m, no problem :)
<xback>
I would love to see an EYE diagram of that link :-)
<f00b4r0>
heh. "It Just Works™" :)
<f00b4r0>
i'm also running PoE over the same cabling since well, it's telephone cable (wider gauge than typical cat5)
<dhewg>
Ansuel: since you worked the ipq4019 6.6 bump, I'm getting this (apparently harmless?) warnings on a fritz7530: https://pastebin.com/raw/2q0XVb1E
<robimarko>
dhewg: Ahh, I had somebody report the same thing on ipq807x as well
<robimarko>
But I couldnt catch it in the act
<dhewg>
first and only boot on 6.6 yet, so no idea if its reproducable
<dhewg>
it kinda sounds like EDEFER and a sysfs register in the probe() callback?
<\x>
>inb4 swiotlb shennanigans
<robimarko>
dhewg: Well, I find it rather weird nobody else is hitting it
<dhewg>
maybe most are still on 5.15?
<robimarko>
Well, if its in the PCI generic code then they probably are using newer stuff
rua has quit [Quit: Leaving.]
asriel has quit [Quit: Don't drink the water. They put something in it to make you forget.]
<slh>
Xiaomi BE7000 isn't that far away from being reasonable, ~140 EUR delivered/ incl. 19% VAT (if only Xiaomi wouldn't be that hostile), ipq9554+qcn5025+qcn6224 (the 2.4 GHz radio probably makes MLO impossible)
<\x>
well mlo can also be 2.4G + 5G or 5G1 + 5G2
<\x>
just a heads up that china does not have 6GHz for wifi, so mlo is like that
<slh>
well, the be7000 only has one 2.4-GHz-only ax/ ath11k radio and one 5-GHz-only be/ ath12k radio
<\x>
ah yeah
<slh>
sensible (MLO is far away from being fully there in mainline linux and hostapd), but not great for development purposes
<\x>
current meme are cheap ipq50xx / mt7981 but they mostly ship with 256MB so yeah fw mode 2 and run nothing extra
<slh>
ipq807x/ ipq60xx are good for what they are, ipq50xx, hmmm, why (yes, budget pricing, I know, but, from a technical point of view)...
<robimarko>
I dont want to repeat using Xiaomi for early development as its a major paint
<stintel>
cool, there is apparently also bootargs-append
<stintel>
let's see if that's enough
xback has joined #openwrt-devel
<stintel>
if a device has PCIe, but there are no devices attached to it, should we just leave it disabled in the DTS, or follow OEM, which enables PCIe?
gladiac has joined #openwrt-devel
<stintel>
personally I'd leave it disabled?
gladiac is now known as gladiac1
<xback>
stintel: is it wired to something?
<stintel>
the OEM bootlog suggests PCIe bus is there, maybe I wrongly concluded no devices are attached to it, but something fails probe -110
<stintel>
if anyone knows how to fix the wiki formatting issues with the bootlog on that page, please explain it to me, also I'm trying to add the OpenWrt bootlog but it doesn't even appear in the preview
<KanjiMonster>
stintel: the upstream pcie driver returns -110 / -ETIMEDOUT if link status is and stays down, which matches the output in the boot log, so likely nothing connected
<stintel>
yeah so the thing is, I tried enabling it and I get unable to get clocks, so I'm not sure
<stintel>
ah bugger. no traffic when the link is 2500Mbps
<stintel>
argh, now there is no traffic on 1000Mbps either
<stintel>
must have been tftpboot that flipped something
<stintel>
the factory dts sets fixed-link 2500/fd/pause on the mac
<mrkiko>
Wondering if there are differencies in quality/assembly of the devices or is pure trademark or whatever. For example - the nbg7815 is costly compared to other devices. Does this have a reasonin your opinion?
<torv>
stintel: Done.
<mrkiko>
In other terms, I am under the impression you can not evaluate if the price is worth the device without keeping in consideration otherfactors, I feel like the equation is missing some terms. But this is more an impression, well willing to change my mind
<stintel>
torv: thanks, appreciated! looking good
<mrkiko>
BTW, I know because they told me, on the nbg7815 tgere are actually 12 antennas. But I guess we are not using all of them in OpenWrt
<torv>
stintel: Haha, thank you!
<stintel>
now I need to figure out this phy issue still
minimal has joined #openwrt-devel
kirdesde has quit [Ping timeout: 480 seconds]
<stintel>
alright, think I got it sorted
<stintel>
have traffic from powercycle directly into OpenWrt on 1000 and 2500 Mbps
<stintel>
and 100 also seems to work
<stintel>
appears my switch doesn't support 10 anymore :P
<xback>
Was it a pll issue?
<stintel>
xback: OEM DTS has reset-gpios for the phy 10 ACTIVE_HIGH
<stintel>
but if I do that mdio-tools does not detect a phy at all
<stintel>
changed that to ACTIVE_LOW and removed the fixed-link
<xback>
Thanks for the details!
robimarko has joined #openwrt-devel
<stintel>
alright, great progress \o/
<stintel>
xback: ah, and you're welcome, of course
<stintel>
man I'm really hoping TP-Link comes out with a similar device with a 6GHz radio
<stintel>
there's 4 unpopulated antenna pads on the PCB
<stintel>
might be wishful thinking though
mrnuke has joined #openwrt-devel
<stintel>
interesting, OEM bootlog shows: [ 0.212287] pci 0000:00:00.0: [14c3:7986] type 00 class 0x000280
<stintel>
that suggests something is actually connected to the PCIe bus
<robimarko>
Isnt type 0 the endpoint?
<stintel>
right - the 14c3:7986 also hints at MT7986 which is the SoC
<stintel>
now if I can figure out how to get the CC2652 that should be on the board according to the SPI photos to work with OpenThread ... that would make this AP even more awesome