Ansuel has quit [Remote host closed the connection]
Ansuel has joined #openwrt-devel
Ansuel has quit [Quit: Probably my PC crashed or time to sleep.]
indy_ has joined #openwrt-devel
indy has quit [Ping timeout: 480 seconds]
Q__ has joined #openwrt-devel
nick[m]1234 has joined #openwrt-devel
<digitalcircuit>
Errp, I missed responding to Ansuel. Unfortunately, I don't have any source for the new MMC - whether that's ZyXEL's code (I've got an older edition of the GPL code) or the MMC hardware itself (other than buying a new NBG6817). I got Ansuel's comment on GitHub though, to try reverting a specific patch: https://github.com/openwrt/openwrt/pull/3954#issuecomment-938166442
<dwfreed>
I have a "brand new" nbg6817
<mangix>
digitalcircuit: he's probably asleep by now
MatrixTravelerbot[m] has joined #openwrt-devel
<dwfreed>
there's no manufacturing date on it, but I can dig through anything you want
<dwfreed>
(note I am on 19.07 still)
<digitalcircuit>
dwfreed: If you get a local console (SSH, serial, etc) and run this command: cd /sys/block/mmcblk0/device/ && tail -v cid date name manfid fwrev hwrev oemid rev
<digitalcircuit>
mangix: Noted! Tab key checking showed "Ansuel" as offline; I figured I could still share my reactions (and I'll follow-up with Ansuel next time it's possible).
<dwfreed>
cid is different, date is 2 months newer, everything else is the same, except no rev file
<dwfreed>
cid is 7001004d3632373034011f30906426e4
<dwfreed>
date is 02/2019
<dwfreed>
compared to yours, not slh
<dwfreed>
ie, name says M62704
<digitalcircuit>
dwfreed: Huh... Noted, thanks! That sounds like ZyXEL decided to stick with Kingston for the new MMC storage devices. Upshot, we probably don't need to worry about TOO many variants so far.
<dwfreed>
who knows how old this device actually is
<digitalcircuit>
Well, presumably not older than February 2019, at least in terms of being a completely assembled unit :)
<dwfreed>
right
<digitalcircuit>
(The 2018 dates for both of my 2019/2020-purchased NBG6817s suggest there's about 1-2 years of stock.)
<digitalcircuit>
(I have very little knowledge of manufacturing processes, this is just a guess)
<slh>
the ASRock G10 is still for sale by some regular/ big sellers (usually not resellers of old stock), although ASRock dropped all support in 2018 and switched of their streamboost servers in early 2019
<slh>
grrr, I hate that firefox/ chromium have switched off ftp support, probably as a result ZyXEL has apparently removed their ftp servers (which were still around 3-4 weeks ago)
<dwfreed>
yeah, it's really annoying
<slh>
what was so difficult about ftp://ftp2.zyxel.com/ARMOR_Z2_(NBG6817)/firmware/ - compared to that webinterface
<dwfreed>
slh: I still get a dir list if I curl, it just doesn't have nbg6817 on it
<slh>
dwfreed: that server was full with different devices (e.g. also ftp://ftp2.zyxel.com/GS1900-24E/firmware/ ftp://ftp2.zyxel.com/GS1900-24/firmware/ ftp://ftp2.zyxel.com/GS1900-8/firmware/ and hundreds more) just over 2 weeks ago, now the list is down to 3 devices (probably ones where the original firmware only checked ftp for upgrades)
goliath has quit [Quit: SIGSEGV]
<digitalcircuit>
Oof, that's frustrating. I had no idea there was an FTP server for the firmware.
<slh>
it was a treasure trove with all current- and past firmware/ driver/ manual/ MIB/ etc releases, for all of their devices
<karlp>
jow: I was (am) struggling to actualyl implement what you suggested for me last night, but while trying to simplify it, I ran into pretty unexpected behaviour with commented out code? https://paste.jvnv.net/view/aZBTJ
<karlp>
even deleting things, instead of commenting them out, I've still not got the right wording to call these: https://paste.jvnv.net/view/TqbZW I'm trolling through other luci apps for how to use this, but obviously missing osmething basic
<mrkiko>
aparcar[m]: rsalvaterra: that said, UML has the advantage to allow testing on machines without hw acceleration. but I ackowledge the maintenance cost is a little bit high
<mrkiko>
or, better, doesn't justify the achivement
<stintel>
nbd: is there a way to get state or event info about ustp? just lost part of my network due to STP flapping
<nbd>
not yet
<stintel>
alright, I guess I'll edit the init script and add -v
<stintel>
hmm, that doesn't seem to work either
<nbd>
just call the init script with stop and run it from the command line
<nbd>
so you have stderr
<stintel>
ok, I must be incredibly stupid, LOG_LEVEL_MAX 100 yet -v3 or higher claims invalid
<stintel>
also it seems to require -s when running on command line
<nbd>
stintel: i think it may need -DDEBUG in cflags
<rsalvaterra>
aparcar[m]: Oh, my…! There were still ath25 users a year ago: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=081e944be6684bf5c4ee7c9fa2683d568d5be288
<rsalvaterra>
I was about to send a pull request dropping it, but now I have to reconsider.
<xback>
blocktrron: ping
<stintel>
so strange that it's just kicking vlan30 offline
<stintel>
that's not even configured on the device, yet I plug it back in and starts flapping again
<rsalvaterra>
Ansuel: Most ath25 devices have 16 MiB of RAM. Only a couple of them have 32 MiB.
<Ansuel>
rip ath25 i assume ancient device
<rsalvaterra>
Probably old enough for university.
dangole has joined #openwrt-devel
<rsalvaterra>
Well, not yet, but nearly… The AR2315 is from 2005.
<Ansuel>
only 2005 ? wow
<rsalvaterra>
There's no way in hell a 16/4 device with a 180 MHz CPU will run OpenWrt 22.x. It may barely boot to init=/bin/sh, if the stars are aligned.
fieryeagle954[m] has joined #openwrt-devel
<Ansuel>
180MHz wat?
<rsalvaterra>
Hey, it was the state of the art, at the time.
<rsalvaterra>
PaulFertser: Great! And what does it do after booting? :)
<PaulFertser>
rsalvaterra: continues to work as a managed switch with dropbear to control it.
<rsalvaterra>
So, it's basically a 10/100 managed switch.
<rsalvaterra>
Uhh… fair enough, I guess.
<Ansuel>
who needs a 10/\\
<Ansuel>
100 managed switch in 2021 with wifi6 at 3 gbps ?
<rsalvaterra>
Ansuel: Not our business. If it works and people use it, support shouldn't be removed.
<Ansuel>
you are right!
<dangole>
AR2xxx were the first SoC we ever had FOSS wifi drivers for (ath5k) and esp. Fonera can be considered an important device in Openwrt history ;) and with 8/32 the Fonera2 should at least boot.
<PaulFertser>
rsalvaterra: my point is that apparently 16 MiB is still enough for basic OpenWrt userspace to work, not just /bin/sh
<PaulFertser>
rsalvaterra: btw, WPA3 works too but OOMs soon.
<dangole>
and AR2xxx boards are cool imho also because they run RedBoot (ie. eCos) as bootloader, ie. you have console over IP from early on, you never really need for serial console.
<rsalvaterra>
dangole: Thanks for the authoritative comment, I'm definitely not removing ath25. :)
<rsalvaterra>
Speaking of which…
* rsalvaterra
remembered he has a FON2200 somewhere…
* dangole
used FON2200 back in 2007 for my final thesis in university, ported Ruby 1.9 to OpenWrt and made it run mesh-based distributed transaction system, with local display (Noritake VFD) and keypad connected for POS application...
<rsalvaterra>
Maybe I should try to drag it to 5.10 (kicking and screaming, for sure)…
<karlp>
rsalvaterra: hard to know for sure honestly, because it's all just one target in openwrt land :|
<rsalvaterra>
dangole: Sentimental value too, I see. :)
<dangole>
rsalvaterra: I no longer got any board around though
<karlp>
we had ath25 with 8/32 iirc, but we binned all the old stock about 18 months ago iirc.,
<karlp>
not all redboots had console over ip though,
<dangole>
stintel, Ansuel: that UniFi 6 LR got Marvell Aquantia AQR112W connected to MT7622 which potentially could do 2.5G, but lacks driver/firmware (and I lack 2.5G link partner for testing)
GuruPrasathGovindarajan[m] has joined #openwrt-devel
dedeckeh has quit [Remote host closed the connection]
<stintel>
dangole: it's crazy hard to get for some reason + the lack of rj45 console port combined with opening case = breaking it is putting me off a bit :(
<stintel>
dangole: but if I can help with testing that 2.5G part (I have 2.5/5/10G link partner), I'll order one for that
pavlix has joined #openwrt-devel
robin_ has quit [Ping timeout: 480 seconds]
<stintel>
dangole: are you dangowrt?
Ansuel has quit [Read error: Connection reset by peer]
<nbd>
you should not call the bridge-stp script directly
<nbd>
when ustpd starts, it calls stp_init on netifd
<nbd>
and netifd notifies it of active bridges
<nbd>
and restarts stp bridges to give ustpd a chance to handle them
<nbd>
stintel: what kind of device are you using?
<stintel>
nbd: Unifi Switch Flex and Edge-Core ECS4100-12PH
<stintel>
nbd: so is there a way to enable it on a bridge, that would not stick after a reboot ?
<nbd>
i think the realtek driver stuff is still missing the part that sets up trapping BPDU packets to the CPU
<nbd>
and that messes up STP
<stintel>
I see, the Unifi is ramips though
<stintel>
but that might explain maybe why plugging the ECS4100 with STP enabled into my network made part of it go down
clayface_ has joined #openwrt-devel
clayface has quit [Read error: No route to host]
<nbd>
no idea if ramips implements bpdu trapping properly
<stintel>
nbd: on what did you develop/test ?
<nbd>
mediatek arm
<nbd>
and a hacked up tree for the realtek stuff
<stintel>
actually, I've enabled it on the unifi and it went briefly unreachable
<dangole>
stintel
<stintel>
but is reachable now (sorry multitasking)
<dangole>
stintel: i saw that, to me that whole cgroups-mount package is an obsolete hack which should be removed
<stintel>
dangole: :D
<dangole>
and yes, it does conflict with procd-ujail and also breaks the (rudimentary) cgroups functionality in procd itself (populating cgroups tree with processes)
<stintel>
I see
<stintel>
so it should conflict with procd and as a result be removed because procd is non-optional
<stintel>
I will add that to the PR, close it, and start something new, to nuke cgroupv1 support out of docker, then remove cgroupfs-mount
dansan has quit [Quit: The C preprocessor is a pathway to many abilities some consider to be unnatural.]
dansan has joined #openwrt-devel
dansan has quit []
dansan has joined #openwrt-devel
dansan has quit []
dansan has joined #openwrt-devel
<JiiPee>
building 21.02 on fedora 35 fails because detection of python and which, could someone who knows git stuff better do pr of master changes for python detection and which detection
<karlp>
hauke had some workaround for that iirc.
<stintel>
nbd: ok so on the unifi switch flex, if I enable STP, I see a few STP packets originating from the unifi, but stops after a few. behaviour confirmed on the other end
<stintel>
I guess it's not quite ready yet, and we should probably consider adding a dependency on only tested targets
<rmilecki>
JiiPee: is that the latest tree? python 3.10 support was added a week ago or so
<jow>
karlp: I guess changing { %s } into { %s\n } would fix it (didn't consider the possibility of a trailing // comment without newline)
gladiac has quit [Quit: k thx bye]
gladiac has joined #openwrt-devel
<dangole>
nbd: any plans to bump mt76 any time soon? forum users are complaining about that NULL-pointer dereference causing Oops every now and then which is already fixed in mt76 tree...
<will[m]>
is there any reason why my coworker building from master@1ed9fc663e550a47fe221f512 and me building off v21.02.0, using the same .config, would have different outcomes with regards to /etc/modules-boot.d/35-igb existing ? cuz he does and i don't
<will[m]>
i would say, looks like igb still doesn't load at boot, except this commit was a month before my coworker's commit, and also mine is later and works worse, and also the behavior is opposite of what this commit intends. sooo idk
<aparcar[m]>
jow: actually jaguar wasn't updated in 8 years and so the npm package. I'd just use it as is rather than extending the action. In case of any unforeseen updates of jaguar I'll take care of it personally
<mirko>
dangole: re rust and wayland: is realy so much stuff running on wayland depending on libsrvg? if so, why? because also written in rust? why not linking against libsvg?
Ansuel has quit [Ping timeout: 480 seconds]
Ansuel_ has quit []
<dangole>
mirko: my target is EFL and that needs librsvg... maybe KDE or Gnome / Phosh can work without it.
Ansuel has joined #openwrt-devel
<dangole>
mirko: and while weston works nicely on x86 (in qemu with virgl) on mt7623 (mali450 with lima) kmscube works great, but weston doesn't, so more testing on different targets/gpus is needed i guess
<mirko>
dangole: a side project is still having kodi fully running on openwrt with oss only, for which i target allwinner h6 with panfrost (for mali) and cedrus for hw video decoding
<mirko>
dangole: by efl so you don't mean the enlightenment libs do you?
<dangole>
mirko: i do :)
<mrkiko>
dangole: hi!!
<mirko>
since when do they depend on libs written in rust?
<mrkiko>
dangole: nice to see you
<mirko>
back then i had the whole EFL suite (including e itself) running on openwrt - which was mainly to have the GUI phone stuck running on the openmoko gta02
<mirko>
that was part of the deprecated xorg feed
<mirko>
s/stuck/stack
<dangole>
mrkiko: hi there
<dangole>
mirko: interesting... i bumped into EFL and decided to go for it because of Tizen making use of it.
<mirko>
back then the API was incredibly unstable and it was real pita keeping track
<stintel>
if a realtek switch is not forwarding packets from one port to another, but bridge vlan shows both ports configured for that vlan, what should I be looking at? where there some caveats with realtek still?
<jow>
aparcar[m]: no, afair the git.openwrt.org -> github.com mirroring uses an api access token generated by me
<jow>
aparcar[m]: so github.com credits random things to me
<jow>
in this particular case it apparently detected an external push as ff merge or something
<jow>
actually not api key but ssh key pair
<jow>
whose public key is added under my account
<aparcar[m]>
jow: I can just no longer use ff merges and only cherry picking
<aparcar[m]>
Or we use a OpenWrt-bot account
<Habbie>
i migrated 1500 tickets from Trac to GitHub for PowerDNS
<Habbie>
i did that with my own account
<Habbie>
people still ask me questions about tickets I filed ;)
rmilecki has quit [Ping timeout: 480 seconds]
<jow>
aparcar[m]: I guess a dedicated bot account is the way to go
<jow>
aparcar[m]: something like "openwrt-mirror" or similar
mrkiko has quit [Remote host closed the connection]
pmelange has left #openwrt-devel [#openwrt-devel]
<aparcar[m]>
I’ll make it happen I guess
<aparcar[m]>
Habbie: how did you do it? Got a script
<Habbie>
aparcar[m], migrate from trac to github?
<aparcar[m]>
Yea I want to scrape bugs.OpenWrt at some point