<Ansuel>
mh what happen here? dd: failed to open '/builder/shared-workdir/build/build_dir/target-arm_cortex-a5+vfpv4_musl_eabi/linux-at91_sama5/root.ext4': No such file or directory
<hanetzer2>
tbf if you wanted to make a super-portable host tools it may be worthwhile to look into static linking with musl libc :P
* hanetzer2
is currently awaiting his noms to arrive
<hanetzer2>
reuben panini and salad.
<Ansuel>
related image?
<Ansuel>
aparcar Build tools 28s with one package that had to be compiled as was missing
<Ansuel>
(with that workaround)
<aparcar[m]>
Ansuel: did you test your commit? It looks like you'd get something like tools-yy at times
<Ansuel>
the build all tools is experimental and not used currently in the actions... now that I think about it tools-yy is problematic right?
<Ansuel>
(i tested it on local build but I stopped is as llvm tools tooks one hour to compile i think)
<Ansuel>
well guess i will have to use ifeq and else logic for that commit to have correct tools-y
<aparcar[m]>
yes filter it out
cbeznea has quit [Quit: Leaving.]
<Ansuel>
aparcar (also another idea to improve the kernel workflow would be test only the target that has the bumped kernel if it's that kind of changes)
<Ansuel>
example kernel 5.10 bump compile only 5.10 target
<Ansuel>
change to package/kernel? compile every target
<Ansuel>
wonder if dump-target-info.pl can be tweaked for such task (for example to dump only target with a specific kernel)
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
<Ansuel>
aparcar isn't this problematic ? tools-$(CONFIG_TARGET_apm821xx)$(CONFIG_TARGET_gemini) += genext2fs
<Ansuel>
oh wait no cause we can't have both apm and gemini
<aparcar[m]>
🙂
<Ansuel>
succeeded 29 seconds ago in 25m 39s
<Ansuel>
ipq806x/generic kernel workflow
<aparcar[m]>
seems slow
<aparcar[m]>
I'm sure 15 minutes of that are actually tools/
<Ansuel>
Package kmod-atlantic is missing dependencies for the following libraries:
<Ansuel>
macsec.ko
<Ansuel>
hwmon.ko
<Ansuel>
or should i fix it mhhh
<Borromin1>
Ansuel: dangole just pushed one
<Borromin1>
a fix, that is
<Ansuel>
thanks but now i wonder why this wasn't present on the github checks
floof58 has quit [Ping timeout: 480 seconds]
<Ansuel>
echo CONFIG_ALL_KMODS=y >> .config
<Ansuel>
make target/compile doesn't compile kmods ?
floof58 has joined #openwrt-devel
<Ansuel>
aparcar ping ?
<aparcar[m]>
Ansuel: Pong
<jow>
Ansuel: CONFIG_ALL_KMODS=y only has an effect on fresh seeconfigs
<jow>
Ansuel: if various kmods are already stored as "# CONFIG_PACKAGE_kmod-xxx is not set" .config they will stay this way
<jow>
Ansuel: so the complete approach would be: sed -i -e '/CONFIG_PACKAGE_kmod-.* is not set/d' .config && echo CONFIG_ALL_KMODS=y >> .config && make defconfig
<Borromin1>
I suppose that would require a device definition.
Borromin1 is now known as Borromini
dansan has joined #openwrt-devel
<Ansuel>
ERROR: denied: installation not allowed to Create organization package
<Ansuel>
time to google
<Ansuel>
mh aparcar should we make tools container public or private?
<Ansuel>
(to make it public i have to enable public package for the organization)
<hanetzer2>
huu, mans0n seems to find my python venv approach to be interesting :P
robimarko has quit [Quit: Leaving]
noahm has joined #openwrt-devel
bookworm_ has joined #openwrt-devel
wvdakker has joined #openwrt-devel
noltari has quit [Ping timeout: 480 seconds]
noltari has joined #openwrt-devel
<Mangix>
aparcar[m]: just updated to 22.03
noltari has quit [Ping timeout: 480 seconds]
bookworm_ has quit []
bookworm has joined #openwrt-devel
bookworm is now known as bookworm_
<aparcar[m]>
Mangix: how was it?
<Mangix>
worked fine
bookworm_ has quit []
bookworm has joined #openwrt-devel
<aparcar[m]>
💪
<aparcar[m]>
well spread the word
bookworm is now known as bookworm_
<Borromini>
Mangix: aren't you on master or is that just a few testing devices?
bookworm_ has quit []
bookworm has joined #openwrt-devel
bookworm is now known as bookworm_
<Mangix>
Borromini: not on my main router
bookworm_ is now known as bookworm
<Mangix>
RT3200 is well supported on 22.03
<Mangix>
dango backported a bunch of stuff before the release
Ansuel has quit [Read error: Connection reset by peer]
<Borromini>
:)
<Borromini>
how is 802.11ax for you?
fblaese has joined #openwrt-devel
<Mangix>
excellent
<Borromini>
=)
<Borromini>
signal strength seems a mixed bag with walls in between? how is your experience with that on MT79xx?
<Mangix>
better than anything I've used.
<Borromini>
lots of people complaining on the forum it seems. I have an open plan on the floor my MediaTek ax AP is on, so i can't really check.
<Borromini>
ok, good
<Mangix>
keep in mind I'm coming from ath10k.
Ansuel has joined #openwrt-devel
<Borromini>
was ath10k so bad for you?
csrf has joined #openwrt-devel
<Borromini>
(i am not really keeping tabs on ath10k anymore)
<Mangix>
it was slower and higher latency than mt7915. there's also the crashing every 5-6 days.
<Mangix>
otherwise no real issues
<Borromini>
ok
<slh>
ath11k is quite fine over the range
<Mangix>
isn't that still in development?
<slh>
yes, doesn't mean that it wouldn't work
<slh>
functionally, my xiaomi ax3600 has been fine for about a year now
<slh>
and the ipq807x is close to being PR material now
<Borromini>
:)
<Ansuel>
very fun github... it's incredible how everything related to container is poorly documented as soon as it's something that diverge from personal use... all the documentation assume things works like magic GOD DAMN...
<slh>
admittedly, only using it as AP - I've kind of defected to using OpenWrt/ x86_64 as router
<Ansuel>
slh you will change mind with the electrical bill ahahha
<slh>
Ansuel: my router uses less power power than my ipq8074a AP ;)
noltari has joined #openwrt-devel
<Borromini>
slh: what's your router on then? OPNSense or the likes?
<Ansuel>
if only qcom started using correct cpu instead of recycled defective cpu repurposed for socs
<slh>
Borromini: I don't like opnsense/ pfsense, OpenWrt/ master works just fine for x86
<Borromini>
slh: nevermind i read your statement wrong
<Ansuel>
slh i expect ipq50xx will consume nothing and be as powerful as ipq8074 just because it's a bit newer
<Borromini>
Ansuel: you mean stop recycling mobile SoCs?
<Ansuel>
yep...
<Mangix>
pretty sure everyone does that
<Ansuel>
it's sad that on router we are years behind and it's stupid honestly... considering now we are starting to handle 10gbps and wifi 7
<slh>
Ansuel: the ax3600 is already quite good at just under 6 watts, so ipq807x isn't necessarily bad
<Mangix>
Ansuel: behind what?
<Ansuel>
Mangix the rest of the world about arm cpu
<Ansuel>
m1 router FTW ahahahha
<Mangix>
that...wouldn't even make sense
<Mangix>
m1 has accelerators for graphics and whatnot
<slh>
that j1900 is fine with routing 1 GBit/s, sqm/ cake is a bit more challenging to that poor Atom CPU, hitting a wall around 830 MBit/s - but my contract 'only' gives me 400/200 MBit/s anyways, so it's doing fine - and 32g/4g is just nice ;)
<Mangix>
maybe you could buy a mac mini and use with a thunderbolt switch. does such a thing exist?
<Mangix>
oh it does. LOL.
<slh>
...even if I only use way under 50 MB SSD storage and under 90 MB RAM.
<Ansuel>
i wonder what would be the w consumption of ipq807x with the nss cores used
<slh>
I wouldn't expect much of a difference
<Ansuel>
same power but 3 gbps on wifi
<Ansuel>
not bad
<Mangix>
I should probably connect a smart switch to my router to measure the power usage
<Mangix>
*smartplug
<slh>
measuring (semi-)idle power is among the first things I tend to do before putting a device into 'production' use
<slh>
Borromini: btw. do you happen to have any advice for 2.5GBASE-T/ 5GBASE-T/ 10GBASE-T ethernet cards?
<Borromini>
slh: my experience is very limited unfortunately. I have an RTL8125B in my new desktop, but it's just hooked up to a plain gigabit router still. I'm prepping a server for my brother with an X550-AT2 though, and that seems good
<Borromini>
but not cheap to come by
<slh>
yeah, that's the problem - RTL8125B is reasonably priced, but I don't really want to invest into 2.5GBASE-T (in terms of multiple cards/ switches)
<Borromini>
doesn't Aquantia offer 10GBASE-T?
<slh>
yes, for around 100 EUR a pop (e.g. from qnap)
<slh>
not really what I'm looking for in terms of prices <-- so I'll porbably have to wait a bit more for prices to come down
<Borromini>
understandably.
<slh>
especially as a single card is kind of useless, I'd need at least two (better three) to retrofit my main devices - and from there it ramps up quickly
<Borromini>
yep gets pricey.
<Borromini>
you're not concerned with increased power consumption?
<Borromini>
fiber seems to perform better on that front compared to copper
<Borromini>
been eyeing a second-hand X520-DA2 for my home server but it *is* complete overkill
<Borromini>
for me at least =)
<slh>
I am concerned about power consumption, but 1 GBit/s has been rather limiting for my needs for over half a decade now - and I don't really see fibre for short distance links so far
<slh>
e.g. the DAC cables would be 30-50 EUR a pop as well, which kind of offsets the cheaper fibre based cards
<slh>
and all the prosumer networking stuff (routers, APs) seems to be copper based anyways
<Borromini>
yeah you'd have to go pretty pro to go all fiber
<slh>
it's either ~40 EUR for an SFP+ DAC cable or twice that for two SFP+ fibre transceivers + the fibre itself. in terms of cable length, DAC would do for me. but my ipq8074a AP would already have copper based 2.5/5/10 GBit/s ports
<Borromini>
:)
<slh>
so... I don't really see the point of fibre (<7m distance for the relevant systems), in terms of prices, nor interoperability with 'affordable' APs
<Ansuel>
also more power consumption
<slh>
yeah, that#s the downside of copper there
<SlimeyX>
those qualcomm ref boards have sfp+ cages ;)
<slh>
yeah, but few of the consumer devices do (but they tend to have at least one port with 2.5GBASE-T, 5GBASE-T or 10GBASE-T)
<slh>
the reference boards are unobtainium, four-figure range and not really usable @home, without case/ antennas
<slh>
great for experimenting/ development, if you'd have the budget and connections to get them, but not to actually use them
<Borromini>
there's some newer MTK boards with SFP+ and multi-gig I've seen :^)
<SlimeyX>
im trying to get adtran to give a 6040 wap thats based off this design but missing the sfp+ and friends
<slh>
also a problem with the BPi-r3 board, as you can add over ~50 EUR on top for pigtails and antennas (and still have to cobble together some kind of mechanically sound case)
<slh>
(and you'd probably get crap antennas for those +50 EUR as well, as you can't really shop around/ test stuff to get good ones)
<Borromini>
personally i'm not interested in the wireless on those devices, and it looks like AX (AC even) will be plenty for most usage here at our place, so...
<Borromini>
backbone on multi-gig would be nice though.
<Borromini>
i'm out, goodnight
Borromini has quit [Quit: leaving]
KGB-2 has joined #openwrt-devel
<hauke>
I am surprised that there are no pro cunsumer devices with 10G SFP+ ports
<hauke>
SFP+ Port are pretty cheap 10G PHYs are expensive
<slh>
yeah, it's surprising, but I guess vendors are afraid of SFP cmpatibility issues and many consumers probably not having a clue what to do with SFP cages
goliath has joined #openwrt-devel
<slh>
(to be honest, it is a bit of a leap to get your feet wet, if you're new to SFP stuff)
noltari_ has joined #openwrt-devel
noltari has quit [Ping timeout: 480 seconds]
MaxSoniX has quit [Quit: Konversation terminated!]