minimalist has quit [Read error: Network is unreachable]
minimal has quit [Read error: Connection reset by peer]
minimalist has joined #openwrt-devel
Luke-Jr has quit [Ping timeout: 480 seconds]
minimalist has quit []
Luke-Jr has joined #openwrt-devel
Luke-Jr has quit [Ping timeout: 480 seconds]
Luke-Jr has joined #openwrt-devel
Luke-Jr has quit [Ping timeout: 480 seconds]
danitool has quit [Ping timeout: 480 seconds]
MaxSoniX has joined #openwrt-devel
MaxSoniX has quit [Quit: Konversation terminated!]
mrnuke has quit [Ping timeout: 480 seconds]
csrf has quit [Remote host closed the connection]
csrf has joined #openwrt-devel
goliath has joined #openwrt-devel
ptudor has quit [Ping timeout: 480 seconds]
ptudor has joined #openwrt-devel
zkrx has quit []
zkrx has joined #openwrt-devel
csrf has quit [Ping timeout: 480 seconds]
zkrx has quit []
goliath has quit [Quit: SIGSEGV]
zkrx has joined #openwrt-devel
EqUaTe has joined #openwrt-devel
ptudor has quit [Ping timeout: 480 seconds]
ptudor has joined #openwrt-devel
zorun has joined #openwrt-devel
danitool has joined #openwrt-devel
<oliv3r[m]>
Hi, I'm very naieve and unfamiliar with DSA. I have a Realtek based switch, and so far, it works, in that from the local ethernet port, I can ping the world on any LAN port (e.g. any lan port has an internet connected cable inserted). I would like however to turn my switch into a 'dumb unmanaged switch' if you will, e.g. disable dsa, and have it 'just work' all ports switched. Plugin device in port 1 and 12, they can communicate. This
<oliv3r[m]>
is currently not possible with my default DSA setup. For testing and development purposes, I'd just like to completly 'turn off'/'disable' these fancy DSA features. Not sure if it is fully possible however; as the switch chip might always need at least 1 vlan?
<jow>
putting all per-port netdevs into a bridge should achieve that
<oliv3r[m]>
ah, well right now, it doesn't :(
<jow>
maybe the implementation is bugged then
<oliv3r[m]>
unless 'dhcp' is somehow blocked of course :)
<oliv3r[m]>
i can try setting up a fixed IP on the second device, and see if ping works, but I'd expect dhcp also to work ...
<jow>
could you pastebin your current /etc/conf8g/network somewhere? Just to cross-check
danitool has quit [Remote host closed the connection]
danitool has joined #openwrt-devel
goliath has joined #openwrt-devel
<aska>
Hi, I have a EA8300 that does not boot. I have gain serial access and see the error "nandboot" not defined. If I run the command "run bootpart1" it will boot to the Linksys firmware as if everything is normal. I flashed the Linksys firmware again, but it does not automatically fix this.
<aska>
I am on ubuntu and "screen" cannot page up the log, is there a way I can page up under "screen"?
<jow>
aska: okay so that script changed the "bootcmd" variable to "run nandboot" but no "nandboot" is defined on your device
<jow>
aska: you said that "run bootpart1" successfully boots your firmware, so the solution would be
<jow>
aska: setenv bootcmd "run bootpart1"
<jow>
aska: the above command applies to the uboot shell
<jow>
aska: if you use fw_setenv from a booted firmware, it should be fw_setenv bootcmd "run bootpart1"
<aska>
the first command not work
<aska>
let me try the 2nd one after boot
<jow>
Mangix: do you want me to trick into recompiling cmake yet again? :)
<Mangix>
:)
<jow>
I honestly have no opinion on it, but if it reduces redundnancy then go for it?
<oliv3r[m]>
jow: no go :( wireshark see's DHCP requests being broadcasted, but nothing comes back on the wire
<Mangix>
just need to wait to get the tools/zstd changes merged to get rid of the rpath hack
<jow>
oliv3r[m]: hmm, that does look like an implementation bug though. The config I pasted should be equivalent to a dumb, do-nothing switch
<jow>
oliv3r[m]: this is one of those realtek switches, right?
<oliv3r[m]>
yep; so I wouldn't be supprised :p
<jow>
might be that there's additional smartness that needs to be disabled outside of the network uci config
<jow>
not sure if there's extra ethtool knobs etc. for switch ic specific features
<oliv3r[m]>
(some context, I have 2 USB dongles on my host, 1 runs the dhcp server, connects to the switch on port1; eth0 (on the switch) is configued via dhcp, and pulls one from port1 (but can pull it from any port if i hook it up) ping etc on the switch works from any port. I can 'short' the 2.5g and the SFP ports (DAC cable) and I see duplicate ping requests going in out, so all that works fine. I plug the second USB eth into port8, but that
<oliv3r[m]>
port fails to get an ip
<oliv3r[m]>
Let me first boot the vendor firmware, because it should work there (the switch is the 'dumb'/unmanaged) variant by stock
<oliv3r[m]>
from 'u-boot' (after running rtk network on, which inits the hardware) itdhcp works immediatly :)
<oliv3r[m]>
so at least no hardware fault
<aska>
it now automatically boot. thanks jow!
<jow>
aska: yw
<aska>
I need to look into what command should be the proper one
<aska>
it now only boot to the first part even I issue command to boot to altkernel
<aska>
part 2 should be called altkernel in this router
<ldir>
Ideas for solving this: I want to extract the number from the end of a string that must end with either a single 4 or 6 - so foo4 or foo6 or foo_4 would be valid, foo44 or foo64 or foo1234 would not. The returned value must be either nothing, 4 or 6.
<ldir>
local family=$(echo "$set" | sed -nre 's#.*([46]$)#\1#p') is close-ish but not right
* ldir
has learned 'changing the escape char' and 'saving a string' in sed this morning
<jow>
... | sed -nrE 's#^.*[^0-9]([46])$#\1#p'
zorun has quit [Ping timeout: 480 seconds]
<jow>
this will rule out sole `4` or `6` (yields epty then) - but that probably makes sense
<jow>
upside or learning Perl as first language, comes with a lifetime regex mastering subscription
<ldir>
lol
<oliv3r[m]>
jow: how would one normally 'block' port - port isolation with DSA? or is that not a thing?
<jow>
oliv3r[m]: usually by not making the port part of a bridge
<jow>
or part of a different bridge
<oliv3r[m]>
that's hardware briding (via brctl through dsa) yeah?
<jow>
unless you mean per-port isolation
<oliv3r[m]>
i do mean per-port isolation :)
<jow>
as in, no isolated port can talk toanother isolated port but only to non-isolated (upstream) ones
<oliv3r[m]>
exactly
<jow>
sec
<jow>
add one or more sections like that to your /e/c/network:
<jow>
config device
<jow>
option name lan1
<jow>
option isolate 1
<jow>
repeat for each port you want to isolate
<oliv3r[m]>
ok; how does that boil down to the cli?
<aska>
Thanks again jow. it is working properly now. found the original bootcmd command in the toh wiki page
<jow>
let me figure out the corresponding cli command
<aska>
very happy now ^_^
<jow>
oliv3r[m]: not sure if there's an ip or bridge subcommand
<jow>
but netifd does this: echo 1 > /sys/class/net/$portname/brport/isolated
<oliv3r[m]>
right, fair nuff
<jow>
bridge link set dev $portname isolated on/off
<dwfreed>
was just going to say that
<jow>
"bridge" is usually not preinstalled though
<oliv3r[m]>
cool :) thank you!!
<[Pokey]>
Oh hey, jow, I hope you don't mind me catching you. Apparently you're the person to talk to with regards to adding a toggle switch to LuCI for a specific model of device which has a GPIO toggle to switch between an internal and external USB. Would this be correct?
<jow>
[Pokey]: yes
<[Pokey]>
How would I go about starting that process off?
<jow>
[Pokey]: it needs to be implemented as uci option in openwrt first, then we can cover it in LuCI
<jow>
[Pokey]: above sounds like something that should go into /etc/init.d/system
<jow>
or /etc/config/system rather, with a custom, target specific init script reading from it
<oliv3r[m]>
but netifd does this: echo 1 > /sys/class/net/$portname/brport/isolated; I don't even have that node, 'find' won't even find any 'isolated' :(
guerby has quit [Remote host closed the connection]
<dwfreed>
is the port in a bridge?
<[Pokey]>
Right. I'm guessing this isn't something that as an example from before on another device? I am a complete noob when it comes to OpenWRT contribution, I am just barely scraping by with help from others and heavy copy paste in order to add support for a couple of device variants
<oliv3r[m]>
brctl shows all ports as part of the bridge, bridge link show, shows lal ports
<jow>
oliv3r[m]: ls /sys/class/net/lan1/brport
<jow>
should list a bunch of bridge port specific resources
<jow>
[Pokey]: do you have an openwrt buildroot cloned?
<oliv3r[m]>
interesting; 'find' doesn't find this name ...
<oliv3r[m]>
find /sys/class -iname 'isolated' says 'no sir
<jow>
oliv3r[m]: it's symlinks to locations elsewhere
guerby has joined #openwrt-devel
<oliv3r[m]>
but usually find finds any filename?
<jow>
find might not be following them
<dwfreed>
busybox find may be acting differently than gnu find
<oliv3r[m]>
ah your right, I was searching /sys/class/net, my bad
<[Pokey]>
jow: I do, yes. I'm currently able to compile a target bin for my device
<oliv3r[m]>
ok so that works; thanks a lot; I have enough dsa now to try to debug this
<jow>
[Pokey]: a couple of targets/boards already ship custom "fixup" init scripts
<jow>
check: find target/linux/ -type f -path '*/init.d/*'
<[Pokey]>
Ah yes, I see
<jow>
you can find examples for matching your specific board in those files, e.g. target/linux/apm821xx/base-files/etc/init.d/hwmon_fancontrol
<jow>
once your script figured out that it runs on the (a) board that requires gpio toggling, it can use uci commands to read user settings and act upon them
guidosarducci has quit [Remote host closed the connection]
<[Pokey]>
Oh I see. Right
<jow>
you need to come up with a place and option name to store the desired state (external/internal)
<jow>
my first suggestion would be defining a new section in /etc/config/system
<[Pokey]>
So theres support to add custom stuff like this just its a framework rather than a targeted thing here
<jow>
config usb; option mode internal/external
<jow>
then query with uci -q system.@usb[0].mode
<[Pokey]>
sounds like a plan
guidosarducci has joined #openwrt-devel
<jow>
[Pokey]: this looks like a precedent, I guess there is other boards that require similar, hardwqare specific toggles
<jow>
first thing that comes to mind is poe power with is typically realized by gpio toggling as well
robimarko has joined #openwrt-devel
<[Pokey]>
jow: Yea, it is not a surprise that is so. So from my understanding (I'm wasting time with theory rather than practical here whilst I flash some firmware) I can add a config section and it will just immediately be picked up by uci, then that init script can be set up to just run the uci command to read the setting and do the appropriate action based on it's value? Then presumably LuCI can use uci commands internally to write that option
<[Pokey]>
too
<jow>
[Pokey]: that's the plan, yes
<[Pokey]>
Wonderful, much easier than playing wit DTSI files haha
<jow>
once uci understands that option (and something acts upon it), LuCI can be easily extended to expose the setting as widget
<jow>
need to figure out a suitable place
<[Pokey]>
I have an outstanding patch for adding device support. Should I roll this into the same commit later do you think?
<jow>
later since it is kind of optional
<[Pokey]>
Okay. I'll start a new branch to tinker in then
<jow>
optional as in, without this you would need to do manual gpio toggling on the cli
<jow>
but the device would still work
<[Pokey]>
Yea, makes sense
<[Pokey]>
# uci -q show system.@usb[0].mode
<[Pokey]>
system.cfg03b3cf.mode='external'
<[Pokey]>
alrighty, a bit messier than I anticipated that output being but it works I guess
<oliv3r[m]>
jow: might be related to my USB (apple) adapter; which is only 100Mbit; could be the switch is not setting up 100mbit links properly (though it says it is a 100mbit link). Luckily I had 2 USB 3.0 gigabit adapters; so at least I know what direction to debug; stupid thing is; its setup properly in u-boot; so tftpboot works fine; anyway, plenty to debug :)
<robimarko>
\x: Was it you or somebody else with the cheap used Linksys from Amazon?
cbeznea has joined #openwrt-devel
<\x>
robimarko: I bought one yeah
<robimarko>
\x: Great, mine arrived yesterday
<\x>
i think mrnuke also bought one
<robimarko>
Its pretty much new
<\x>
ill be interested on how viable will the linksys dual firmware be like
<\x>
like, if i flash from it, which one will it replace, fw1 or fw2?
<\x>
and if I sysupgrade from openwrt, will it replace which one?
<robimarko>
\x: We are quite a long time from that
<\x>
well yeah in mainline
<\x>
The chinese already has qsdk working with 22.03 userspace
<\x>
I still havent tried it, something seems missing from the recipe
<robimarko>
\x: I couldnt care less about those
<\x>
ahaha well yeah
<\x>
but sure, im looking forward getting real openwrt support on it
<\x>
robimarko: would also be interested on a coremark run, just so I have some idea on cpu perf
<robimarko>
Though the core should be pretty much the same as IPQ8074
<robimarko>
So it could do 2.2, but I am not sure if the PMIC is capable of that current though
<Habbie>
CoreMark 1.0 : 7019.783025
<\x>
yeah
<Habbie>
wrt3200acm with openwrt 21.02
<\x>
Habbie, hella fast
<\x>
OC'd ipq4019 is 2600's coremarks
<\x>
1.1GHz MT7921A is 2750ish iirc, ill retry later
<\x>
MT7621A*
<\x>
robimarko: so it seems pretty linear not hitting any bottlenecks, so 1.2GHz (ipq6000) should be around 4000 coremarks
<robimarko>
\x: Yeah, it should be linear
<rmilecki>
ddns-scripts should get rewritten from scratch
<rmilecki>
change my mind
<rmilecki>
its code is so messy, absolutely painful to follow & understand
<rmilecki>
eval-s everywhere
<rmilecki>
variables named without any sense
Tapper has joined #openwrt-devel
<rmilecki>
verbose mode *disabling* actual ddns functionality (working like dry mode)
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
snh has quit [Read error: Connection reset by peer]
noltari_ has joined #openwrt-devel
noltari has quit [Ping timeout: 480 seconds]
noltari_ has quit [Ping timeout: 480 seconds]
noltari has joined #openwrt-devel
snh has joined #openwrt-devel
MaxSoniX has joined #openwrt-devel
snh has quit [Remote host closed the connection]
snh has joined #openwrt-devel
srslypascal has joined #openwrt-devel
<[Pokey]>
Hey \x I did also get the firmware from the manufacturer to confirm we haven't missed a trick. The USBs are DPDT controlled on that firmware too
noltari has quit [Read error: Connection reset by peer]
noltari has joined #openwrt-devel
srslypascal has quit [Ping timeout: 480 seconds]
srslypascal has joined #openwrt-devel
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
SlimeyX has quit [Read error: Connection reset by peer]
noltari_ has joined #openwrt-devel
aska has quit [Quit: Leaving...]
noltari has quit [Ping timeout: 480 seconds]
SlimeyX has joined #openwrt-devel
valku has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
dangole has joined #openwrt-devel
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
minimal has joined #openwrt-devel
<\x>
ah yeah
mrnuke has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
dangole has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
<Ibarakasen>
Hello, I found a potentially serious bug in UCI system: If the storage is full, uci commit wipes the config file.
<Ibarakasen>
I would expect it to try making an atomic change where it first writes the new files and then swaps the new and old file..
<Ibarakasen>
Tested on OpenWRT 21.02, not yet tested on 22.03.
<f00b4r0>
known "feature"
<Ibarakasen>
Is that an expected behaviour?
<f00b4r0>
not really. It's just that there hasn't been a proper fix implemented (yet)
<Ibarakasen>
Right, so not even 22.03 has it fixed yet.
<f00b4r0>
no
<Ibarakasen>
Okay, thank you for the information.
Ansuel has joined #openwrt-devel
gladiac has quit [Quit: k thx bye]
<Ibarakasen>
Are there complications that make the proper fix hard(er) to implement?
<Ibarakasen>
I looked through the code a bit, it seems that there is an atomic swap, but there seem to be no checks that the export to temporary file was successful.
<mrnuke>
Slimey: Check images in the comment. Params are 115200n8
<oliv3r[m]>
Anybody here have any experience with DaC (Direct Attach Copper)? I'm hacking on the realtek target, and I have 1 mikrotik DAC cables; 1m and 3m. My switch has 2 SFP ports, so i'm shortting those.. If I plug in the 1m in both ports, I get a link, can ping etc, so the links are working fine. If I do the same with the 3m cable, it doesn't work (get error spam, but that's for me to fix I suppose). The i2c data is very very little different
<oliv3r[m]>
between the two; so I assume the exchange is inband as part of the serdes link. The original author that got the SFP ports working described the same issue, and left it at 'hit or miss which ones will work'. As the author has since left, I'm trying to figure out what could be going on, but I have no idea where to start really. He mentioned something about calibration data, but that data would have to come from the module? Where could I
<[Pokey]>
Is it generally safe to keep writing firmwares to SPI flash as you test or does it risk wearing it out?
<mrnuke>
[Pokey]: How long will it take you to write 100000 images ?
<[Pokey]>
mrnuke: :P fair
<mrnuke>
[Pokey]: That being said, I generally prefer to run initramfs/tftpboot for testing because it's faster
<[Pokey]>
mrnuke: Fair. I just have a convenient Web UI
<mrnuke>
"convenient" ?
<[Pokey]>
mrnuke: More convenient than running a TFTP server and using the serial console yes
* mrnuke
laughs
<[Pokey]>
¯\_(ツ)_/¯
<Mangix>
Ansuel: have fun updating :)
<Ansuel>
mh?
<Mangix>
I've noticed that check seems to not be sensitive enough
MaxSoniX has quit [Quit: Konversation terminated!]
<mrnuke>
[Pokey]: I recommend you give the tftp method a try. Setup is annoying, but once you can 'Ctrl-R tftp put' and 'run bootcmd' it's waaay faster :)
<[Pokey]>
mrnuke: Sounds like I am missing context because I am not certain what you mean
<[Pokey]>
I assume you mean tweaking UBoot to start it easier
<Ansuel>
Mangix rip I have to merge it
<mrnuke>
[Pokey]: Yeah. you set up bootcmd to boot off of tftp, so uboot automatically does a tftp boot each time :D
<[Pokey]>
mrnuke: Ah. Yea I don't want it to do that because I test in bursts and sometimes I wanna be on flash for extended periods
<[Pokey]>
Good point for general concerns of testiong many images though
<mrnuke>
[Pokey]: And if you set your tftp server such that you can push files to it, it's trivial to push freshly built images: "tftp localhost -c put bin/... myimage.img"
<Ansuel>
me wondering why the router doesn't print anything on the serial
<Ansuel>
and doesn't notice the power cable was not plugged in
<Mangix>
reminds me of that one joke "Who will win? Experienced PC builder or Power switch on PSU"
<mrnuke>
Ansuel: I wish MR7350 were that simple :)
<robimarko>
mrnuke: MR7530 arrived yesterday, its brand new
<robimarko>
Will have to wait a day or two to open it up though
<robimarko>
I am digging through a bunch of broken IPQ6018 stuff anyway, good news is that CPU scaling PLL fixes were picked
bluew has joined #openwrt-devel
<Slimey>
mrnuke thanks
<Slimey>
thats what i used hmm i have to take it back to the lab and try again ;')
<Slimey>
have to pop in some pogo pins
<mrnuke>
robimarko: You have four screws on the bottom. Two are under the label. It's really annoying
<robimarko>
mrnuke: So the usual stuff
<mrnuke>
robimarko: I tried my eap610-outdoor branch. No serial output at all. It just reboots and mover very slow after reboot.
<mrnuke>
robimarko: I presume it's the borked CPU scaling
<robimarko>
mrnuke: Whats the CPU model?
<robimarko>
CPU scaling thats upstream is meant to be used solely on IPQ6010/18
<mrnuke>
the /proc/cpuinfo says "Qualcomm Technologies, Inc IPQ6018"
<Borromini>
i thought i read there were some throughput regressions on 5.15 vs 5.10 but i'm unsure if that was specific to mvebu or another target? Anyone recall something similar?
<jow>
you could try reverting the zstd static thing first and await the next build
<jow>
if it still fails revert more
<jow>
before you do anything you probably should try a completely clean build locally first
<jow>
maybe the zdzd static commit only worked for the testers because they still had a dynamic copy of zstd.so lingering around in theeir staging_dir/host
<Ansuel>
buildbot compiled correctly with that commit tho but some user reported compilation error before the last commit got applied... ahhh i will revert both and just debug
<jow>
congratulations to your first buildbot breakign commit btw :)
<Ansuel>
this is a big one and sorry :( I also need to understand why workflow didn't failed....
<Ansuel>
Package 'cmake' is not installed, so not removed
<Mangix>
probably available in staging_dir
<Mangix>
wipe it and make tools/firmware-utils
<Mangix>
good news
<Mangix>
I reproduced the zstd failure
<Ansuel>
nice
<Ansuel>
meanwhile trying to repro the firmware-utils
<Mangix>
if it helps, I ran my run with make -j 31
<Mangix>
so there was no chance for cmake to be built
<Ansuel>
aand why libressl is replaced?
<Mangix>
cmake depends on it
<Ansuel>
right!
<Mangix>
and after my pr cmake will depend on zlib too
<rmilecki>
[Pokey]: if you compiled firmware once with file without +x and then changed ONLY +x, the OpenWrt's buiildroot will not rebuild package/rootfs properly
<rmilecki>
[Pokey]: modify something in that file and then rebuild firmware (make V=s)
<Ansuel>
news about the zstd ?
<[Pokey]>
rmilecki: okay, l try that tomorrow, thank you
<rmilecki>
[Pokey]: yw
GNUmoon has quit [Quit: Leaving]
GNUmoon has joined #openwrt-devel
GNUmoon has quit [Remote host closed the connection]
<Mangix>
Ansuel: trying to fix. not sure why AC_SEARCH_LIBS(ZSTD_compress, zstd) is failing...
GNUmoon has joined #openwrt-devel
<Mangix>
g++ -o conftest -g -I/home/mangix/devstuff/openwrt/staging_dir/host/include -static-libstdc++ -static-libgcc -L/home/mangix/devstuff/openwrt/staging_dir/host/lib conftest.cpp <-- where is lzstd ???
<Ansuel>
Mangix real question is why it does work on gcc12
<Ansuel>
did they refactor the scan lib logic?
GNUmoon has quit [Remote host closed the connection]
<Ansuel>
Support for Realtek's clock controllers (COMMON_CLK_REALTEK) [N/y/?] (NEW)