danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<neggles>
JiiPee: mention that, and once it's been accepted in master you can ask for it to be backported
ptudor has quit [Ping timeout: 480 seconds]
arjun has joined #openwrt-devel
ptudor has joined #openwrt-devel
arjun has quit [Remote host closed the connection]
arjun has joined #openwrt-devel
<arjun>
Where would the root.squashfs maximum allowed size be specified. My board has 256 MB NAND but when i create a build close to 50Mb packing the bin fails. WARNING: Image file .../build_dir/target-arm_cortex-a7+neon-vfpv4_musl_eabi/linux-ipq80xx_ipq80xx_32/root.squashfs is too big, size is 41418756 compare to max allowed 40036k
<csrf>
i built an initramfs image for a TL-WR840NV5. I tested the image & it works fine. Now, I need to flash my squashfs image
<csrf>
How do I go about loading/flashing the image via the uboot prompt?
cbeznea has joined #openwrt-devel
<Habbie>
csrf, i don't know, but can i ask what openwrt version? i have a stack of 841 that could enjoy some openwrt :)
<csrf>
Habbie, I built 19.07, but since it only has 4MB, it's a barebones openwrt image
<Habbie>
ack - i have a 841 with 18.06 + luci but i don't need luci
<csrf>
only needed wireguard, so i added only that to the base system
<Habbie>
anyway, things i think i know - do you maybe need a factory image instead? can you trigger tftp recovery? etc.
<csrf>
Habbie, i actually ordered 16MB chips off of aliexpress to do some upgrades
<Habbie>
ohh
ptudor_ has joined #openwrt-devel
<csrf>
the router is at remote location, and there's no one there right now to hold down the reset button on power-up for tftpboot
<csrf>
i'm in the serial console
<Habbie>
ah
<Habbie>
don't lock yourself out
<csrf>
i can load an image manually via tftpboot, but i need to know how to save it to flash
<neggles>
arjun: IMAGE_SIZE in target/linux/targetname/image/Makefile (or the subtarget makefile)
ptudor has quit [Ping timeout: 480 seconds]
<neggles>
csrf: boot into initramfs image, SCP sysupgrade image to /tmp/sysupgrade.bin, `sysupgrade -n -p -v /tmp/sysupgrade.bin`
<csrf>
neggles, should I be worried about overwriting the factory partition(s) by doing that?
<neggles>
csrf: not if you have configured sysupgrade and the device tree correctly, no
<csrf>
ok
<neggles>
on a nor based device with a monolithic squashfs image, it will just write it to whichever partition is flagged with the `compatible = "denx,umage"` or `openwrt,uimage`
<neggles>
or labeled 'firmware'
<csrf>
so i dont have to worry about memory addresses; the board-memory layout in the dts file for my build should take care of all of that?
ekathva has joined #openwrt-devel
rua has quit [Ping timeout: 480 seconds]
<neggles>
csrf: if the load address is correct in the uImage then it's fine
<neggles>
anyway
<neggles>
if you have serial console
<neggles>
you'll be able to see if it goes wrong, fix it, netboot initramfs again, etc
jlsalvador has quit [Read error: Connection reset by peer]
jlsalvador has joined #openwrt-devel
bluew has quit [Ping timeout: 480 seconds]
borek has joined #openwrt-devel
shibboleth has joined #openwrt-devel
cbeznea has joined #openwrt-devel
robimarko has joined #openwrt-devel
<f00b4r0>
is there an uci config to set (force) the DHCP client (udhcpc) lease time? the wiki doesn't list one
ptudor_ has quit [Ping timeout: 480 seconds]
cbeznea has quit [Quit: Leaving.]
cbeznea has joined #openwrt-devel
<karlp>
you mean, what it requests?
<f00b4r0>
no. I want it to ignore the server's lease time and use a custom (shorter) one, for $reasons
<karlp>
seems like an odd thing to waste configurability on, whe the server will decide anyway, so I'd suspect it not to be available.
<karlp>
wouldn't be very "u" style.
<karlp>
just set a static IP and use cron? :)
<f00b4r0>
but reading the udhcpc doc it seems the client doesn't implement that feature
<f00b4r0>
yeah, no.
<karlp>
you want big boy features, you need big boy tools, not "u" tools :)
<f00b4r0>
it should be about 5 lines of code to implement I suspect, but heh, if you say so :)
<karlp>
5 here, 5 there, before you know it's kernel 5.15 and it doesn't fit anywhere anymore...
cbeznea has quit [Quit: Leaving.]
cbeznea has joined #openwrt-devel
cbeznea has quit [Quit: Leaving.]
cbeznea has joined #openwrt-devel
cbeznea has quit [Quit: Leaving.]
srslypascal has quit [Remote host closed the connection]
srslypascal has joined #openwrt-devel
borek1 has joined #openwrt-devel
borek has quit [Ping timeout: 480 seconds]
borek1 is now known as borek
mattytap has joined #openwrt-devel
karlp has quit [Quit: WeeChat 3.1]
cbeznea has joined #openwrt-devel
srslypascal is now known as Guest211
srslypascal has joined #openwrt-devel
Guest211 has quit [Ping timeout: 480 seconds]
karlp has joined #openwrt-devel
ekathva has quit [Quit: Leaving]
<stintel>
blocktrron: any reason for keeping wnm_disassoc_imminent in hostapd ubus methods? I would just keep bss_transition_request and drop the old one, less code to maintain?
<stintel>
also did we ever try contributing hostapd ubus integration?
<stintel>
sure it's openwrt specific but with all those openwrt-based vendor SDKs out there I'd say there a billions of devices using it
<blocktrron>
stintel: No i did not try to upstream the ubus integration
<blocktrron>
regarding wnm_disassoc_imminent - we should be able to remove it now.
<blocktrron>
My intention was to keep it for 22.03 so we have one release which allows usage of both
<stintel>
gotcha
<blocktrron>
But i think DAWN is the only consumer now
<blocktrron>
nick[m]1234: ^
<stintel>
also usteer still
<stintel>
I can send a patch / PR for both if you'd like
<stintel>
not sure if a README.md in there is the best location, I followed the example from openvswitch kinda
<nick[m]1234>
go for it
<nick[m]1234>
sending a patch ;)
<stintel>
will do
<stintel>
any feedback on that README is also welcome :)
robimarko_ has joined #openwrt-devel
robimarko has quit [Ping timeout: 480 seconds]
<blocktrron>
stintel: nice
<blocktrron>
stintel: "also usteer still" <-- are you sure? I do not have the code with me ATM, but afair i've removed usage of this method, as otherwise usteer can not handle transition-rejections.
<blocktrron>
As wnm_disassoc_imminent always sets the DI bit, which starts a disassoc timer within hostapd
<stintel>
blocktrron: yes, there is 1 user left
<blocktrron>
stintel: oh, this user is dead code
<stintel>
ahh
<stintel>
so I can just remove usteer_ubus_notify_client_disassoc
<blocktrron>
<rant> also apple devices handle disassoc-imminent bit really poorly. If they reject the request and subsequently get disassoced, their WiFi part just ends up in a connected - no internet state until you reboot </rant>
<blocktrron>
Also, their link measurements make no sense if you parse them according to spec
<blocktrron>
stintel: in case you are using usteer, i recommend you to update to the latest version in packages
<stintel>
I am using it
<stintel>
and I noticed a recent update, one of my APs is already running it
Guest213 has quit [Ping timeout: 480 seconds]
<blocktrron>
Yeah, the roaming routine was fucked and always immediatly kicked clients
<stintel>
now for some reason my xps13 was kicked with reason 5 (load too high) and it's now on 2.4G :/
<stintel>
but I don't even have that load balancing enabled
<blocktrron>
Every kick has this reason code to it
<stintel>
oh
<stintel>
didn't look at it in detail yet, only noticed that today
<stintel>
as I was seeing RTS/CTS packets from Intel client and I'm like wtf
<stintel>
why is it on 2.4G at all even
<blocktrron>
Does your XPS / wpa_supplicant have CONFIG_WNM enabled?
<stintel>
good question. Makefile enables it if CONFIG_MBO=y and CONFIG_MBO=y is enabled in .config
<stintel>
so I think so, but how to confirm ?
<blocktrron>
ubus call usteer connected_clients
<blocktrron>
bss-transition-management has to be set to true
<stintel>
thanks, that's useful. it is enabled
<blocktrron>
Okay, in this case kicks should (with the current version) only happen in case the lowermost RSSI threshold (min_snr) is violated
xback has quit [Remote host closed the connection]
<stintel>
I haven't set min_snr at all
<blocktrron>
Hmm, okay, then I'm puzzled why this happened
<stintel>
I have only option roam_scan_snr -70
<stintel>
the comment in the default config is a bit confusing though: Minimum signal-to-noise ratio or signal level (dBm)
<stintel>
what is it, snr or signal level, or both ?
csrf has joined #openwrt-devel
<blocktrron>
stintel: both
<blocktrron>
< 0: RSSI ; >0 SNR
<stintel>
ok
minimal has joined #openwrt-devel
<mattytap>
stintel: had a look at the box stuff you sent me. Will take a month of Sundays to read but the answer to our Rx problem is in there somewhere.
<mattytap>
A coupe of things standout immediately in the patches: queue counts; extra mem pages; level of detail and comments is what we need
robimarko_ is now known as robimarko
<mattytap>
Meanwhile I'm still running the m300 sysupgrade test. Took a while for me to compile on archlinux (I know!) glib2 etc; and in the end I've cloned your staging qoriq-5.15
<mattytap>
too early to give any feedback yet.
<stintel>
I think I added mmc dev 0; somewhere in the bootcmd
<stintel>
and that solved it
<mattytap>
OK. Good to know. useful though as I ended up reading up on the hardening techniques you are using.
<stintel>
hardening techniques?
<mattytap>
BPF, eBPF, stackprotector
<stintel>
ah :)
<mattytap>
BPF would not compile for me initially and when I saw 'Berkley' I went - aghh - never knowingly done this before
<mattytap>
All good in the end after a distclear
<stintel>
that reminds me I should spend some time pushing my BPF related stuff to master
<mattytap>
Also, had to resort to your staging in order for 5.15
<stintel>
but now that winter is gone, I'm having trouble staying inside after work :P
<stintel>
and with covid travel restrictions being completely canceled ... even having trouble to stay in the same country :P
romany has quit [Quit: Ping timeout (120 seconds)]
<mattytap>
good - nobody said on their deathbed 'I wish I'd spent more time on BPF'
romany has joined #openwrt-devel
<mattytap>
I've got a s***load of stuff to do like tax-returns and stuff for my dad so will come back to the m200 in a couple of weeks
<jow>
hmm
rua has quit [Ping timeout: 480 seconds]
<jow>
nft flow offloading headaches; if soft flow offload is in use we can use { eth0, eth1, ..., wlan0, wlan1 } in the flowtable
<jow>
enabling hw offloading fails though with the same list as it includes non-offload capable wifi devices
<jow>
so that means either hw offload without soft offloading for wifi, or soft offloading for all
<jow>
or can we somehow cascade flowtables?
<jow>
maybe the softflow case is just silently ignoring the wifi ifaces while the hardflow one does not
rua has joined #openwrt-devel
<nbd>
i think the api is a mess. i will ask pablo if we can simply add an 'auto' mode which will add the required devices with graceful fallback from hw to sw offload, like on xtables
<nbd>
i don't think flowtables can be cascaded
<nbd>
and i don't see a sane way to configure it for a mix of sw and hw offload
srslypascal is now known as Guest217
srslypascal has joined #openwrt-devel
mattytap has quit [Quit: Leaving]
<jow>
yeah, seems so
Guest217 has quit [Ping timeout: 480 seconds]
<jow>
ther is "flags offload" already, maybe just add another "fallback" flag
<jow>
nayone with a pppoelink can share his "devstatus pppoe-wan" with me?
<jow>
basically I'm interested in the value of "devtype"
<jow>
also "ls -l /sys/class/net/pppoe-wan"
srslypascal has quit [Remote host closed the connection]
<jow>
grift: yep, it did - thanks. I just wanted to know if for pppoe devices there a `lower_*` symlink to resolve the underlying ethernet device
<jow>
but unfortunately there isn't
<grift>
i see
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
srslypascal has joined #openwrt-devel
ekathva has joined #openwrt-devel
<hurricos>
Grommish: Sort of. I'm looking at a 64M device run up from 6M to 20M slab in 12h. It might be how much traffic it's moving
<hurricos>
neggles: OK, OK, so it was inconsitently tracked by the kernel anyways
<hurricos>
Thanks, so this is probably isn't a memleak but just kernel needing way more space
<hurricos>
thankfully we now have Prometheus on the network
<hurricos>
now the person that runs it feels very validated that she wasn't just seeing things. She's trying to learn Lua to figure out how some of the web stuff shows what it does so she can record it :^)
cbeznea has quit [Quit: Leaving.]
<hurricos>
thanks for letting me use your brains folks
<Grische>
Is the source for that conversion online where I can have a look? Or who would/could I talk to about this?
pedropt has joined #openwrt-devel
<pedropt>
hi everyone
<pedropt>
need a little help on my dir 645 from dlink , suddenly without changing any config files the router stopped to deliver dhcp over the lan and can not get any dhcp over the wa,
<pedropt>
wan
<pedropt>
already reset it but stills the same , no changes at all
<pedropt>
i am accessing it over serial console , any tips to check if it is an hardware of software issue
<rsalvaterra>
pedropt: Check your logs. Judging from your ISP, you have IPv4 only. Check for any udhcpc related messages (regarding the WAN side).
<pedropt>
by the way , my openwrt version is 10.06.1
<pedropt>
18.06.1
<rsalvaterra>
pedropt: Any reason not to update to the latest version?
<rsalvaterra>
18.x is getting long in the tooth.
<pedropt>
no reason at all but i can`t
<pedropt>
can not get luci , dhcp over lan not working
ekathva has quit [Quit: Leaving]
<pedropt>
already restarted network and only Br-lan have an ip
<rsalvaterra>
Sure, I didn't mean at this moment. However, that router has USB, so a pen drive with the new image is a possibility. That's another issue, though.
<pedropt>
question , usb pen must be fat or fat 32 ?
<rsalvaterra>
Uh… ext2 or ext4, most likely.
<rsalvaterra>
Nevertheless, is you DHCP server running? It's probably dnsmasq.
<rsalvaterra>
*is your
<pedropt>
gonna try format it to fat (i am on windows right now) drop the upgrade there and try to mount it using serial to see if this works
<pedropt>
udhcpc: started, v1.28.3
<pedropt>
udhcpc: sending discover
<pedropt>
udhcpc: no lease, failing
<rsalvaterra>
That's the client.
<pedropt>
how do it restart the server ?
<rsalvaterra>
Anyway, your LAN bridge seems correctly set up. Try setting a static IP in the 192.168.1.0/24 range on your computer and accessing the router.
<pedropt>
however if it is the client , then client should have get an ip over wan
<pedropt>
i already did that before , setting a static ip on pc
<rsalvaterra>
Odd… have you tried the failsafe mode?
<pedropt>
not yet
<pedropt>
going to do it now
<pedropt>
sending a reboot on serial
<pedropt>
got many options , however before going to there witch is tftp upgrades and i need to disconnect from here to set a static ip here , i will try the usb pen to see if i get it mounted
<pedropt>
can not get any sdx devices on /dev
<pedropt>
and lsusb command is not even installed by default , probably to get usb working i probably need to install some packages witch i can not get now
<pedropt>
and dmesg dont even get any registration in log of any device input on usb port
<pedropt>
wait
<pedropt>
[ 5.779172] usbcore: registered new interface driver usbfs
<pedropt>
[ 5.790261] usbcore: registered new interface driver hub
<pedropt>
[ 5.801004] usbcore: registered new device driver usb
<pedropt>
[ 5.854154] phy phy-usbphy.0: remote usb device wakeup disabled