<Ansuel_>
mangix: problem of china is that they like to steal stuff, doesn't put credis and makes money by them...
<Ansuel_>
they also fix stuff but never contribute
<slh>
probably functional access, less affected by the GFW
<Ansuel_>
example openwrt would be superiori and much ebtter if chinese community would contribute and help us
<mangix>
Ansuel_: that's not fair. There are a lot that do contribute
<Tusker>
Ansuel_: I don't think China is any worse than other countries really...
<Ansuel_>
Tusker: don't know if it's the same for other target but for qcom soc it's very evident... all the work from nss we worked for month on the same thing... they just pulled patch and added stuff to it... same thing about ipq807x
<fda>
from xposed: 50% of the apps are chinese, maybe good but for me useless as i dont understand what they are doing
<Ansuel_>
and at one point we had chinese guy coming on openwrt forum asking for help for source on their chinese repo o.O
<fda>
btw, i found my an droid apps only on russian websites...
<fda>
(warez)
<Ansuel_>
we were like WHAT? and they linked an acient lede repo with patches in it cleared from robi and mine signed off...
<Ansuel_>
preatty sad
<Ansuel_>
don't telling that it's all like this but it's very common :(
<mangix>
Ansuel_: you talking about immortalwrt or some other one?
<mangix>
those dudes absolutely submit patches
<Ansuel_>
mh nope
<Ansuel_>
again i can talk only about ipq806x/ipq807x so probably it's better for other target
<Ansuel_>
mediatek probably?
<mangix>
yeah mediatek employees develop the wifi driver and submit patches for others.
<Tusker>
Technicolor, make heavy use of OpenWRT in their products, do they actually contribute back ? (like they say on their https://www.technicolor.com/connect/homeware page)
<Tusker>
i think it's a common problem for all open source projects. people see patches and integrate it into their product, and forget that contributing back helps them
<fda>
immortal has kernel: bump to 4.9.285, 4.14.249, 4.19.209
<fda>
i do not like different forks
<fda>
its much useless work to merge around
<Ansuel_>
technicolor is a big f to openwrt entire project...
<Ansuel_>
but it's all brcm stuff so...
<dangole>
you can find technicolor firmware also on intel (ax. TI) puma SoC DOCSIS devices...
<Ansuel_>
anyway time to sleep only last thing mangix how do you see status of qca8k?
<Ansuel_>
do you think if we manage to merge all the patch upstream openwrt will consider adding support for it ?
<mangix>
Ansuel_: I haven't tested series 7 but it should be good. It seems barely anything is getting merged currently. qca8k backport PR needs updating.
<mangix>
and yes I do.
<Ansuel_>
nice! then we just need to wait andrew for the final words
Ansuel_ has quit [Quit: Probably my PC crashed or time to sleep.]
fda- has joined #openwrt-devel
<fda->
funny, this reboot was not ' jail: can't read from parent' message
<fda->
but i noticed on e8450 this message 2x and on avm7320 only 1x "netlink: 'iw': attribute type 302 has an invalid length."
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
swalker has quit [Remote host closed the connection]
swalker has joined #openwrt-devel
bluew has quit [Quit: Leaving]
dangole has quit [Remote host closed the connection]
hitech95 has quit [Remote host closed the connection]
valku has joined #openwrt-devel
victhor has quit [Ping timeout: 480 seconds]
valku has quit [Quit: valku]
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
floof58 has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
floof58 has joined #openwrt-devel
aiyion_ has joined #openwrt-devel
aiyion has quit [Ping timeout: 480 seconds]
zheng has joined #openwrt-devel
<Tusker>
weird, the t30-w has a 32kb spi nor flash that is empty... (mr25h256). why would they bother ?
zheng has quit [Remote host closed the connection]
<Tusker>
OK, now working on sysupgrade support... is there any "standard" way to write to an sdcard mmc device for sysupgrade purposes ?
nitroshift has joined #openwrt-devel
Tapper has joined #openwrt-devel
<mangix>
posted patches to github. hopefully they don't due
<mangix>
*die
<Tusker>
anyway, need to log off, back later
Tusker has quit [Quit: Time wasted on IRC: 2 days 8 hours 27 minutes 29 seconds]
dedeckeh has joined #openwrt-devel
rmilecki has joined #openwrt-devel
<rsalvaterra>
Oh, the new MIPS eBPF JIT has been backported!
<digitalcircuit>
Aside, someone emailed me earlier today remarking they ran into the same issue, too (same newer eMMC module, had to use TFTP to recover due to no serial console - after all this, I'll have to poke the GPL source and see if ZyXEL put any external toggle mechanism for the boot flag).
<digitalcircuit>
Brief note - NBG6817 router stuff is not overlooked, just had other things come up the past few days. I'll try to get to the "5.10 + MMC core driver debug log" build before too long, then get back to the CPU crash/reset issue.
danitool has joined #openwrt-devel
dangole has joined #openwrt-devel
decke has joined #openwrt-devel
<nbd>
rsalvaterra: are you using ebpf?
<rsalvaterra>
nbd: Not yet, but MIPS was probably the last big arch missing it, wasn't it?
<rsalvaterra>
Besides, seems like bpfilter is finally gaining momentum.
<nbd>
though i'm more interested in making sure that we won't need bpfilter for decent performance for most use cases
<nbd>
for that i'm writing an ebpf based qos classifier that can be used with cake
<nbd>
so that qos works well with sw flow offload
pmelange has joined #openwrt-devel
<rsalvaterra>
That's surprising. I believed SQM was fundamentally incompatible with flow offloading.
pmelange has left #openwrt-devel [#openwrt-devel]
<rsalvaterra>
Due to the NAT lookups, at least.
<nbd>
depends on the level of functionality needed, i guess
<rsalvaterra>
Yeah, I was thinking specifically of cake with the nat argument. :)
<nbd>
if you do qos based on matching ports, or destination addresses (ipset like functionality), but don't need explicit fairness between LAN hosts, there's no incompatibility with flow offloadiing
<nbd>
as long as you get rid of the need to do the matching in netfilter
<nbd>
i think even for the host isolation there are ways to make it work
<rsalvaterra>
Heh… right now I'm just back to the default fq_codel. My access got bumped from 200/20 to 500/100. I don't have powerful enough hardware to run cake with my previous configuration. :)
<stintel>
so who here is using firewall4 :)
<rsalvaterra>
stintel: You? :P
<rsalvaterra>
Oh, and dangole too, for sure. :)
<stintel>
well ... I've built an image with fw4 enabled for my U6LR but it's a dumb AP so not really useful
<rsalvaterra>
Right, you got the LR! How's it behaving?
* rsalvaterra
had a crazy week, not a lot of time to be around…
<stintel>
had some weirdness with HE160 but changed channel to auto and now that's working again
<stintel>
and sometimes the connection seems to freeze, no pointers whatsoever in either dmesg
<stintel>
but this could be intel ax200 too
<rsalvaterra>
Still no joy with the 2.5G PHY?
<stintel>
cannot test, the AP is PoE-PD only, so even getting a 2.5GbE USB adapter wouldn't help a thing
<stintel>
has to wait until 1/11
<mangix>
Who has 2.5gbit usb adapters?
<mangix>
I know aquantia has upstream support.
<rsalvaterra>
Speaking of 2.5G and PoE, kobi said that the XS1930-12HP I'm planning on getting will probably be supported in OpenWrt soon, so I couldn't be happier. :D
<rsalvaterra>
The price tag, though… ;_;
<stintel>
the netgear is half the price or so
<rsalvaterra>
stintel: Which one?
<stintel>
I was about to order it but then noticed "no stock" and no estimate
<stintel>
MS510TXM
<rsalvaterra>
Not all ports do 10G.
<stintel>
do you have >4 10GbE devices? :P
<rsalvaterra>
The Zyxel does 1G/2.5G/5G/10G in all ports.
<rsalvaterra>
No…
<rsalvaterra>
… not yet. :P
<stintel>
:)
<stintel>
I have 8 so even combination of 2x netgear would still use up all of my 10G copper ports :(
<rsalvaterra>
stintel: #firstworldproblems :P
<stintel>
so yeah if I want to replace my Huawei with something OpenWrt supported I'm gonna have to break the bank. again
<stintel>
but when OpenWrt supports LACP with members on different switches for true redundancy it's going to happen :)
dangole has quit [Remote host closed the connection]
nitroshift has quit [Quit: Gone that way --->]
minimal has joined #openwrt-devel
victhor has joined #openwrt-devel
dedeckeh has quit [Remote host closed the connection]
victhor has quit [Ping timeout: 480 seconds]
hitech95 has joined #openwrt-devel
<hitech95>
Hi guys does someone have an example/reference of a kernel option? I'm reworking the spi-gpio driver and I would love to add an option for non dts implementations. But I cannot find how / what to search for!
robimarko has joined #openwrt-devel
<robimarko>
Why do you need an option for platform aka non OF support?
<robimarko>
I dont see a reason for it to need to be turned on optionally
<hitech95>
robimarko: I would have just created an OF prop, but the driver is used also on acpi machine I suppose, so spi devs required an userspace controllable property. I have no idea how to do that.
decke has quit [Quit: Leaving.]
Dgrey has quit [Ping timeout: 480 seconds]
<robimarko>
Why does it need to be controllable?
<robimarko>
Cant it just be enabled by default
<hitech95>
robimarko: it is a threshold that changes the behaviour of the driver, right now the spi-gpio driver blast out the data but some chips need a slowdown. due to the overhead of the ndelay() function is impossible to have high throughput in some cases like spi flashes. so a threshold must be specified to know ehen to use the delay.
<robimarko>
Ok, then I really cant help. Sorry for the noise
<hitech95>
np :)
victhor has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
pmelange1 has joined #openwrt-devel
pmelange2 has joined #openwrt-devel
<Slimey>
:O
pmelange1 has quit [Ping timeout: 480 seconds]
robimarko has quit [Quit: Page closed]
indy_ has joined #openwrt-devel
indy has quit [Ping timeout: 480 seconds]
pmelange2 has left #openwrt-devel [#openwrt-devel]
indy_ is now known as indy
philipp64|work has quit [Quit: philipp64|work]
valku has joined #openwrt-devel
Ansuel has joined #openwrt-devel
<Ansuel>
hi
<Slimey>
hi
Dgrey has joined #openwrt-devel
<fda>
are 10 1G devices also okay?
Ansuel_ has joined #openwrt-devel
Ansuel has quit [Ping timeout: 480 seconds]
<mangix>
Active day today
<Slimey>
yes configuring new core switches in stackwise virtual on 2x C9500-12Q-A
<stintel>
cool
<fda>
when you want some activity: i still dont know why this message coorus during boot: netlink: 'iw': attribute type 302 has an invalid length.
<stintel>
this is since the mac80211 bump I think
<fda>
hm, it its a kernel module.
<fda>
i searched all mac80211* files for 'attribute type' but no hots
<fda>
./build_dir/target-mips_24kc_musl/linux-lantiq_xway/linux-5.4.152/lib/nlattr.c: pr_warn_ratelimited("netlink: '%s': attribute type %d has an invalid length.\n",
<fda>
./build_dir/target-mips_24kc_musl/linux-lantiq_xway/backports-5.10.68-1/compat/backport-5.2.c: pr_warn_ratelimited("netlink: '%s': attribute type %d has an invalid length.\n",
<fda>
./build_dir/toolchain-mips_24kc_gcc-11.2.0_musl/linux-5.4.152/lib/nlattr.c: pr_warn_ratelimited("netlink: '%s': attribute type %d has an invalid length.\n",
<robimarko>
No, its an error that shows up when starting an AP
<fda>
Ansuel_: its also on 2,4ghz only devices
<fda>
1x per radio
<fda>
robimarko: thx! but a bump of iw wold prevent this patch - and package/network/utils/iw/patches/001-nl80211_h_sync.patch could deleted completely
<Ansuel_>
fda: in theory considering they are in a separate branch that is merged after some time no it's not a problem replace patches
philipp64|work has joined #openwrt-devel
<Ansuel_>
robimarko: don't know in a private email he sent me that he wanted to wait for merge... some patch still had to be reviewed so don't know why they merged it right away... not that they were ready but andrew wanted to wait his final review
<robimarko>
Ansuel: Well, its not your problem that one of the maintainers merged them.
<robimarko>
I wouldnt worry about it
<Ansuel_>
now i just need to take a decision about reg field and another patch about mirror/fast ageing support
Ycarus has joined #openwrt-devel
<Ansuel_>
there is an rfc on openwrt but i still have to test the mirror feature
<Ansuel_>
and then start working on the led mess
<robimarko>
Fast ageing looks like something that should be easily pushable
<robimarko>
What do you mean by LED?
<robimarko>
Cause if you are thinking about adding support for configuring them in the driver, then dont even try
<robimarko>
They are alergic to that
<robimarko>
There was a push for a PHY LED framework but it didnt get anywhere
<Ansuel_>
i want to try proposing an rfc for a generic solution for phy leds
<robimarko>
Ok, that is something that is needed
<robimarko>
There were patches last year for that
<Ansuel_>
the solution they proposed was too specific and not that generic
<robimarko>
I mean, PHY/switch LED-s are kind of specific
<Ansuel_>
so they really tried to make a generic solution for their specific device... and got obviouls rejected
<robimarko>
Ideally it should allow using the HW triggers to do the blinking etc
<Ansuel_>
my idea is to add 2 additional function to leds struct and create a dedicated trigger to make them configurable
<robimarko>
Cause all of the switches and PHY-s have HW LED controllers that can do everything on their own
<robimarko>
Its just that the configuration is non existant in a uniform way
<Ansuel_>
the 2 additional function will regulate the enabling of the hw mode and the configuration. the dedicated trigger is used to modify the blink option
<robimarko>
That sounds really good
<Ansuel_>
i think that should be generic enough... a driver can then declare how to configure a led to be hw controlled with his special regs access
<Ansuel_>
and they will just use standard binding declaration and cled api
<robimarko>
Ansuel_: That should solve 99% of the custom hackery that is going on with LED configuration in switches/PHY-s
f00b4r0 has quit [Quit: This computer has gone to sleep]
<Ansuel_>
yep i just need to find time to write actual code and implement it... and find the right way to rimplement the dedicated trigger... also andrew was positive about this kind of solution
f00b4r0 has joined #openwrt-devel
<robimarko>
Ansuel_: Yeah, Andrew was really advocating for something like that and refusing any more code that tried to add more bindings and hackery for LED-s.
<hitech95>
i sthere any tool to analyze the cal data to figure out how the antennas layout?
<fda>
the "netlink: 'iw': attribute type 302 has an invalid length" message is now gone afer adding the patch. should i test something?
<robimarko>
To me it seems like days are getting shorter and shorter all the time
<robimarko>
hitech95: Which radio model?
<hitech95>
Atheros, built in on AR9242
<mangix>
Ansuel: so what's the story with ar8216 again?
<hitech95>
I'm tring to figure out what the hell ubiquiti did. 2 antennas but 6 different signal paths on 2 different chips!
<robimarko>
hitech95: I think that docs for those older Atheros radios and the ART format leaked a long time ago, googling should get you the result
<robimarko>
But you will have to manually poke
<Ansuel_>
mangix: in short qca8k support only 83xx and 8216 have completely different regs so it doesn't work and require different function
<mangix>
Why add ar8216 support?
<robimarko>
Probably to be able to drop swconfig
<mangix>
ar8216 is the internal switch, right?
<robimarko>
Its actually external
<mangix>
Oh?
<robimarko>
But most of the Atheros MIPS CPU-s have a variant of it inisde
<robimarko>
But you can buy it as an external IC
<robimarko>
I mean there is an DSA driver for AR9331 and 9344
<mangix>
My understanding is that https://git.openwrt.org/?p=openwrt/staging/blocktrron.git;a=shortlog;h=refs/heads/dsa-ar9344 is a solution for that.
<robimarko>
Yeah, I was just writing that
<robimarko>
So no point in actually adding AR8216
<robimarko>
I dont know of any router that uses the external one
<Ansuel_>
owww didn't know about it
<Ansuel_>
also the question was do we have device with ar8216 ?
<mangix>
I just see ar8229 and ar8236 in compatible
<Ansuel_>
also does dsa supports some type of igmp snooping ?
<Ansuel_>
mangix: if you are referring to initval... that's not a good way... anyway could be that 8229 are used for 8216... let me check
<mangix>
No I'm reading driver source
<mangix>
compatible lines
<Ansuel_>
yes i'm also reading that...
<Ansuel_>
and you are rigth o.O
<mangix>
If you look at the link above, the ar9331 upstream driver gets used.
<mangix>
If you try to add this to qca8k, big fat rejection.
<Ansuel_>
it would be also stupid working on something already supporte
<Ansuel_>
anyway 8216 works only with compatible so no stupid initval like it's done with 8237
<Ansuel_>
we still have to fix the swconfig driver....
<mangix>
After all this qca8k stuff gets squared away, I'm testing ar9331. I had issues last time with phy-swap
<mangix>
Ansuel_: so I have this super amazing idea. Add an swconfig subtarget for ath79 for devices that cannot be migrated. Like the ones with realtek switches.
<Ansuel_>
not a bad idea
<Ansuel_>
but what device can't be migrated?
<robimarko>
Which switches are those?
<mangix>
rtl8366 and 7
<mangix>
They have many variants and only 1 is supported upstream
<mangix>
Ansuel_ it's maybe 6 or 7 of them.
<robimarko>
It hink that only rtl8366rb is upstream
<mangix>
Right
genuser1 has joined #openwrt-devel
<genuser1>
Hello?
<robimarko>
mangix: I know of b and s versions, any more?
fda- has joined #openwrt-devel
<genuser1>
Is anyone here familiar with the netgrear wac510?
fda has quit [Ping timeout: 480 seconds]
<robimarko>
genuser1: I worked on it a long time ago
<mangix>
robimarko: 6rb, 6s, and 67 are used looks like
<genuser1>
Is there any new development on it? I am having some minor issues with it. I have flashed the snapshot to it and while it boots and takes opkg updates (ssh/luci) the base upgrade files refuse to upgrade due to a "read only" system (according to the error)
<genuser1>
I attempted to compile the 21.02 rc4 code for it (wac510) but I have failed miserably
<Ansuel_>
genuser1 probably rootfs is full
<mangix>
Other oh there's a 7b as well
<genuser1>
Ansuel I was attempting to revert to stock firmware to hopefully "reset" the device to stock but i can't seem to get anything to tftp to it
<robimarko>
Have you tried connecting UART?
lynxis has quit [Remote host closed the connection]
<genuser1>
robinmarko I don't think I have the equipment for uart connection. I build drones and I have a usb stlink v2 programmer as well as another usb uart type board but I wasnt sure if those would work
lynxis has joined #openwrt-devel
<genuser1>
robinmarko besides the openwrt page for wac510, is there any other resources to help me try to the uart programming? I will just attempt to use what usb-uart programmers I have on hand
<genuser1>
they both have 3.3v and 5v pins so maybe they might work?
<robimarko>
You just need a 3.3V USB to UART
<robimarko>
Make sure not to connect the Vcc at all
<robimarko>
There is a header inside
<genuser1>
robinmarko I will attempt the uart method. if I can get the stock firmware restored I will try again to install openwrt. I would like to see if there is a way to get 21.02 rc4 for the wac510.
felix_ has joined #openwrt-devel
<hauke>
robimarko: fda-: yes the linked patch fixes the netlink error shown by iw when starting an AP
<hauke>
this is a problem in master
felix has quit [Ping timeout: 480 seconds]
Tapper has joined #openwrt-devel
<fda->
@hauke you patch is yet in the master!
<fda->
(upstream master)
pmelange has joined #openwrt-devel
f00b4r0 has quit [Read error: Connection reset by peer]
f00b4r0 has joined #openwrt-devel
Luke-Jr has quit [Ping timeout: 480 seconds]
pmelange has left #openwrt-devel [#openwrt-devel]
<genuser1>
robinmarko I have successfully connected to my wac510 via uart. Any idea on what firmware file I can use to restore the oem? I have tried the .img and the .tar files and it is reporting wrong image format for bootm command
<robimarko>
Just load OpenWrt initramfs
<robimarko>
And from it you can flash whatever you want
<genuser1>
Interesting.
danitool has joined #openwrt-devel
Luke-Jr has joined #openwrt-devel
<genuser1>
Ok, I got the initramfs itb booted to ram, how do i restore the oem firmware from this point? I am still learning openwrt and this device so apologies
<robimarko>
Just use nandwrite to write the stock firmware image
<robimarko>
Though, if you want to use OpenWrt there is no need to flash stock back
<robimarko>
Just flash the OpenWrt factory UBI using ubiformat
<genuser1>
I still want to use openwrt. I am just doing the exercise of restoring oem to document it for future use
dangole has joined #openwrt-devel
<robimarko>
Ok, I dont remember what was the stock format
<robimarko>
But you should be able to just flash it
<robimarko>
Ok, I looked at the stock FW
<robimarko>
Its just a UBI image
<robimarko>
So you can simply use ubiformat to flash it
<robimarko>
As far as I remember it uses dual FW so make sure to flash it to both as it also uses some weird bootloader logic
<robimarko>
Usually if the one of them doesnt boot it will just revert to the second one and that one will flash stock FW to the broken partition
<genuser1>
Thanks robinmarko. You say the stock fw is ubiformat. Maybe I am being tripped up...when I extracted the fw it has a .img extension. Is that correct?
<robimarko>
Its just an extension, it can be whatever
<robimarko>
I looked at it with binwalk, its just UBI
fda has joined #openwrt-devel
fda- has quit [Ping timeout: 480 seconds]
<genuser1>
robinmarko thanks. Will continue this process of restoring oem just to document. I did want to add a note though. When i first tried to flash the snapshot I couldnt flash it through the webui of the oem firmware until I downgraded the firmware to v5 something.
<genuser1>
I got my wac510 within the past month so it has v9 oem on it
<genuser1>
had*
<fda>
hauke: before without you patch i notice som "invalid HT mode" errors in syslog (i have not saved them) but they are gone now also
<robimarko>
genuser: No idea, they probably changed the validation. I have the board somewhere but havent used it in a long time
<mangix>
I don't think anyone maintains 17.01 anymore.
<mangix>
that's kernel 4.4 I believe
<mangix>
yeah...imortal guys do 18.06
<aparcar[m]>
stintela
Tapper has quit [Quit: Tapper]
dangole has quit [Remote host closed the connection]
<Ansuel_>
aparcar[m] think you misspelled stintel
<aparcar[m]>
hiccup of my chat client sorry
<aparcar[m]>
anyone using firewall4?
<aparcar[m]>
masquerade for ipv4 doesn't work for me with the default settings
<Ansuel_>
another question about firewall4 jow: any plan about supporting it ? is it hard?
<Ansuel_>
(i mean in luci)
<hitech95>
guys is there a way to get from cli the list of services and its status? like running not running autostart etc?
<dwfreed>
for i in /etc/init.d/*; do $i status; done
<aparcar[m]>
hitech95: you can just type service
<aparcar[m]>
it will print a full list
<hitech95>
it is a really old version of openwrt, something around 2010.
<hitech95>
it does not have the status argument
<Ansuel_>
2010 O_O
<hitech95>
yea... proprietary SW based on openwrt in GPL violation...
Ansuel has joined #openwrt-devel
<hitech95>
the antenna_gain parameter reduce the gain of the antenna, right?
<genuser1>
robinmarko learned some interesting stuff in the oem firmware restore exercise. Right now my wac510 has the owrt snapshot on the first boot option (?) and oem fw v5 on the second boot option (?). When connected with uart I can switch between owrt and oem so I have a fallback between the two. Kinda cool. After doing the sequence of booting owrt to ram, then ubiformat of owrt and rebooting into owrt, installing luci with intern
<genuser1>
then using luci to upload the nand sysupgrade file (bin), I cannot upgrade the base files from the gui. Still gives the read only error. Any ideas?
<genuser1>
There is a root password set as well
gladiac is now known as Guest3035
gladiac has joined #openwrt-devel
Ansuel_ has quit [Ping timeout: 480 seconds]
Guest3035 has quit [Ping timeout: 480 seconds]
shibboleth has joined #openwrt-devel
dansan has quit [Remote host closed the connection]
Ansuel_ has joined #openwrt-devel
Ansuel has quit [Ping timeout: 480 seconds]
jlsalvador has quit [Quit: jlsalvador]
jlsalvador has joined #openwrt-devel
rmilecki has quit [Read error: Connection reset by peer]
rmilecki has joined #openwrt-devel
<mangix>
Habbie: ping
<Habbie>
hi
<mangix>
can you test the latest qca8k patch?
<mangix>
I bet it's going to fail because of mac6-exchange
<Habbie>
i can! you mentioned one of the 3 PRs i combined wasn't useful, so I'd love a clear definition of what to build
<mangix>
Habbie: the same PRs. the qca8k backport one was updated.
<mangix>
no patching needed anymore
<Habbie>
ok, so i take fresh 4562+4528+4622? and rm /etc/config/system and another one?
<Habbie>
i've been fully relying on just 'make' doing the right thing after checking out branches, merging them, etc.
<Habbie>
i know this is not valid in some projects
<Habbie>
but it should be here, right?
<Habbie>
Applying /home/peter/projects/powerdns/openwrt/openwrt/target/linux/generic/hack-5.10/780-2-Documentation-devicetree-net-dsa-qca8k-document-conf.patch using plaintext:
<Habbie>
can't find file to patch at input line 20
<mangix>
hrm. do you have something extra in your tree?
<mangix>
it's compiling fine here.
<Habbie>
i shouldn't, but i just ran dirclean and am trying again
<Habbie>
then that's the one i meant by 'one was not useful'
<Habbie>
building again
<mangix>
yeah tp link decided to use actual GPIOs
<Habbie>
this i recall :)
Acinonyx_ has quit [Ping timeout: 480 seconds]
<genuser1>
I am trying to compile openwrt for the netgear wac510 using the 21.0.2.0-rc4 branch. I pulled the config.buildinfo file from the appropriate location and then ran make menuconfig but I don't see the wac510 option. I see netgear ex61x but no wac510.
philipp64|work has quit [Quit: philipp64|work]
<slh>
genuser1: it's usually easier (and much faster to build) to select just the device(s) you actually care about, that also makes debugging a little more obvious
<genuser1>
slh I agree. I was trying to select the netgear wac510 specifically to build (not including other ipq40xx devices) but in the menuconfig it is not showing up as a device option against 21.0.2.0-rc4.
<slh>
why "21.0.2.0-rc4" and not either 21.02.0, openwrt-21.02 or master?
<genuser1>
i tried 21.0.2.0 and master but it still is not showing netgear wac510 as an option so I figured maybe try against rc4. wac510 is available as a snapshot on 21.0.2.0 right now but I was hoping to compile it with luci already baked in.
<slh>
(I'd always recommend either of the last two, if you build yourself there's little reason not to go with HEAD of the individual branches)
<genuser1>
slh thanks. I have those lines in my current .config file but when I run make menuconfig it still does not show up. Still working on it though.
<Habbie>
mangix, still building
<Habbie>
mangix, my dirclean has a cost, clearly
<slh>
is came up just fine for me, after selecting ipq40xx as target architecture (admittedly, using make xconfig)
<Habbie>
mangix, given the time of day here it might be 24 hours before you get real answers from me
<slh>
genuser1: the view over the config is the same as menuconfig, just using qt5 to visualize it
<slh>
which is a tad more convenient in places
<slh>
but that's 98% personal preference, there results are the same either way
<slh>
your linked reference is correct and good to follow (but there are always nicer ways here and there)
<genuser1>
Ok I just switched to the master branch, re-pulled the config.buildinfo file and re-ran make menuconfig and now wac510 is showing. I am assuming that once I select wac510 exclusively that all other defaults should be good to go especially if wanting luci? I did the scripts/feeds install luci command too.
<slh>
./scripts/diffconfig.sh should tell you (make sure that luci is enabled in there as well)
<slh>
and all the other packages you generally tend to need/ install later
<mangix>
Habbie: OK
rmilecki has quit [Ping timeout: 480 seconds]
<Habbie>
mangix, still building, will report back later, have a good day :)
<Ansuel_>
mangix: anyway i think we are near actual working of multi-cpu port
<Ansuel_>
at least for qca8k
<Ansuel_>
hope e9hack won't write the scary message of "NOTHING WORK YOUR PATCH IS SHIT ?!?!?!?"
<genuser1>
slh thanks for your help so far. Once this kernel_menuconfig command completes I will attempt a build and I can start testing. :)
<slh>
would be great, as multi-cpu support is really necessary on ipq806x, ~600 MBit/s (guessed, my contract 'only' gives me ~430/220 MBit/s) vs ~230 MBit/s
<genuser1>
Question for anyone else willing to chime in. I have openwrt installed in a container on my proxmox server and it works generally with no issues. Its a cosmetic thing but is there a way to change the model string instead of it saying "Must be completed by OEM"?
<Ansuel_>
slh strange how multicpu patch makes that much of a difference o.o
<slh>
Ansuel_: it's surprising me as well, but the delta is really that massive, with one core pegged at 100% during the test
<will[m]>
is there something in 21.02.0 that loads appropriate kernel modules / network drivers that aren't covered by AutoLoad ?
<will[m]>
the patch to AutoLoad the igb network driver wasn't included in 21.02.0 so i'm trying to figure out how the driver would ever get loaded otherwise
<Ansuel_>
slh: using that config should enable bridge filtering?
<Ansuel_>
in that case i'm testing all with this meme config
<slh>
the problem is that you'd first have to remove your existing bridging, before creating a new bridge-vlan config - possible (as described in https://github.com/openwrt/luci/pull/4307), but it's easier to start with a rough basic bridge-vlan config and extend/ modify from there
<slh>
according to my understanding, no - it just defines the switch
<Ansuel_>
oh ok it's to have all under switch
<slh>
the bridge-vlan interfaces below split up the switch
<Ansuel_>
well i just defined 2 switch ahhaha
<slh>
yep
<Ansuel_>
should be same thing
<Ansuel_>
never understood why they used this bad comunication... the axi bus is limited on this soc?
shibboleth has quit [Quit: shibboleth]
<mangix>
Sounds like some good crack
<hitech95>
Ansuel_: dumb question if I patch a Tim modem with the extended gui/luci can I roll back with a factory reset? I would like to test MAP-T on those routers with the unloked interface.
<Ansuel_>
we have wifi 5 ghz that have a max datarate of 1 gbps and more wifi 2.4 that has also a very big datarate with hack vht
<slh>
I can just guess that the interrupt load isn't handled as well as it could be
<Ansuel_>
hitech95 yep nothing is permanent.
<Ansuel_>
slh i think it's really datarate contrain...
<hitech95>
Ansuel_: Perfect! I will find some modems to run compatibility tests!
<Ansuel_>
they conencted a switch that can to 5 gbps or datarate with a 2gbps connection
<Ansuel_>
and we are running it with 1gbps
<Ansuel_>
so the shitty perf would be cpu load + datarate contrain of passing traffic from wan port to lan port
<slh>
hmm, yeah - but still, my 400/200 MBit/s line shouldn't challenge it that much (even with a single CPU port)
<Ansuel_>
400 from wan port 400 to your lan port
<Ansuel_>
800 mbps
<Ansuel_>
add some overhead from both port and load... and you are at 1gbps
<slh>
yes, but I only get ~260 MBit/s with a single CPU port
<Ansuel_>
but anyway we can hope someday multicpu port will get merged
<slh>
:)
<Ansuel_>
i think i will just make a pr on openwrt to merge my hack patch :D
<Ansuel_>
if we manage to make it stable
<slh>
have you updated the PRs (or anything I should test on ipq806x/ DSA)?
<slh>
I've kind of lost track a bit with the flurry of today's mails/ mailed notifications from the github PRs ;)
<slh>
ah, your latest mails describes it a bit
goliath has quit [Quit: SIGSEGV]
<Ansuel_>
yes but it's a mess aahahaha
<Ansuel_>
slh if you want to test i can put some effort in refreshing the commits
<slh>
yeah... maybe not tonight anymore ;)
<slh>
take your time, if you'd like a test, I'll be around - but I'll also be around tomorrow, or after your vacation
<Ansuel_>
fact is that i'm lazy
<slh>
we all are ;)
<Ansuel_>
i'm trying to implement and fix stuff and at the same time propose them upstream
<Ansuel_>
so i see refreshing the commit a waste of time
<Ansuel_>
but the real reason is that multi-cpu port is currently a mess and i need to polish them a bit
<slh>
it usually takes me ~45 for a build, per target - unless I'm debugging an issue, I always build for ath79, ath79-tiny, ipq40xx, ipq806x, ipq807x, lantiq/xrx200 and realtek, an hour each, looong time ;)
<Ansuel_>
owww i still have dsa-cpu-ports-multi in my linux repo
<Ansuel_>
that change everything !
<slh>
(and the PR wrestling betweek the interdependent DSA PRs doesn't make that too much fun at the moment ;)
<Ansuel_>
slh we should really start using vps and offload all this stuff
<slh>
genuser1: your build (master/ HEAD) should succeed, 7,1M bin/targets/ipq40xx/generic/openwrt-ipq40xx-generic-netgear_wac510-squashfs-nand-sysupgrade.bin --> http://paste.debian.net/1215606/
dansan has quit [Ping timeout: 480 seconds]
<Ansuel_>
anyway the only thing i need to verify is how vlan is handled when changing cpu port
<genuser1>
slh cool. I got time tonight so I don't mind waiting it out. I have other activities to do elsewheres. If I can at least get it compiled I would be content. Then I might try again against the rc4 branch for fun :)
robimarko has quit [Quit: Page closed]
<mangix>
Ansuel_: verifying that mac6-exchange is also not broken is a good idea
<mangix>
I also need to test what led-open-drain does on my unit
<Ansuel_>
mangix mhhh i think using port6 as primary cpu port is fine
<mangix>
we'll find out tomorrow
<Ansuel_>
ok now i'm scared
<mangix>
?
<Slimey>
heh this cobranet stuff is neat, works over layer 2 so no addressing is required