goliath has quit [Quit: SIGSEGV]
dvn has joined #openwrt-devel
Daanct12 has joined #openwrt-devel
parazyd has quit [Ping timeout: 480 seconds]
Daanct12 has quit [Quit: WeeChat 4.2.1]
tSYS has quit [Quit: *squeak*]
tSYS has joined #openwrt-devel
Daanct12 has joined #openwrt-devel
minimal has quit [Quit: Leaving]
skynet2 has quit [Ping timeout: 480 seconds]
<KGB-2> https://tests.reproducible-builds.org/openwrt/openwrt_ath79.html has been updated. (98.7% images and 100.0% packages reproducible in our current test framework.)
gch981213 has joined #openwrt-devel
kwz has quit [Quit: Ping timeout (120 seconds)]
kwz has joined #openwrt-devel
Mangix has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
Misanthropos has quit [Ping timeout: 480 seconds]
Misanthropos has joined #openwrt-devel
Mangix has joined #openwrt-devel
<stintel> first look at EAP683 LR: no holes
<stintel> serial is gonna suck
<stintel> did spot 4 pads next to each other
robimarko has joined #openwrt-devel
<dwfreed> stintel: yeah, that is a very clean PCB
<dwfreed> not really even any vias
<dwfreed> except for the mounting holes
<stintel> why do I keep doing this to myself :D
<dwfreed> I let somebody else do all the adventuring
<dwfreed> So thanks for all the effort
<dwfreed> that firebox m300 works a treat
<stintel> ah happy to hear it :)
<stintel> I
<stintel> I'll be looking at some new stuff probably in the nearish future
<dwfreed> might see if I can get another one for here
<robimarko> That EAP looks quite neat
<stintel> got approval for the real estate loan, notary next week, so I'll be moving to a house soon
<stintel> big plans there :P
<dwfreed> nice!
<robimarko> congrats
<stintel> it's <2km from a datacenter, considering dark fiber and going nuts :P
<dwfreed> rofl
<stintel> robimarko: yeah I spotted it last week, there is a forum post asking about it, I looked up some of the components, looked like it should be easy to support, didn't consider serial console being such a PITA though :P
<robimarko> Have they connected TX or left it up to you to add a resistor?
<stintel> at this point no idea, I'll try some things when the cleaning lady is done with the office
<stintel> at least the enclosure is super easy to open, 6 screws, not hidden, no clips, it just straight opens after removing those 6 screws
<robimarko> That is a welcome break from glued together cases
rua has quit [Quit: Leaving.]
Misanthropos has quit [Ping timeout: 480 seconds]
<stintel> so ... https://www.linux-ipv6.be/OpenWrt/EAP683-LR/IMG_20240404_111225.jpg TP15 is GND, if the typical TP-Link layout, it should be TP13 RX TP14 TX TP15 GND TP16 3.3V
<stintel> then you have J255 with 2 big GND pads on left and right, then 4 smaller pads on top, they are connected to TP13 - TP16 in same order
<dwfreed> note that it's really hard to read the labels in that photo
<stintel> yeah can't do better sorry, tried one outside and it was also meh
<dwfreed> a closeup of that section of board would be sufficient
<stintel> right, why did I not think of that 😂
<dwfreed> in the FCC photos, that J255 is populated
<dwfreed> that screw is like right in the way, though
<stintel> there seem to be pads for 4 more antennas
<dwfreed> heh
<stintel> maybe they're working on a 3-band version
<stintel> so the J255 is actually a connector kind of thing
<stintel> I guess I should find one of those - the pads seem big enough for me to pull off soldering
<dwfreed> you probably just need to airgun it
<stintel> wouldn't that melt the plastic of the connector ?
<dwfreed> might :D
<stintel> first I need to figure out what type of connector it actually is anyway :P
<dwfreed> check the fcc photos
<stintel> I am, but I don't recognize it
<stintel> I'm not too familiar with all this stuff really
<dwfreed> pinout is bottom pin is TP13, then progresses upward the same
<dwfreed> with your closeup photo at 500% zoom, I can see the traces
<stintel> yeah I believe I mentioned that earlier ;)
<stintel> ugh I could try soldering wire to TP13 - TP15
<dwfreed> tacking a wire to the TP pads would be safer, yeah
<dwfreed> much bigger
Misanthropos has joined #openwrt-devel
<stintel> looks like I might need to bridge a few pads too
<stintel> T11 T12 T13
<dwfreed> probably just 11 and 12
<dwfreed> 13 should be the voltage ref
<stintel> yeah so TP13 seems connected to T11, TP14 to T12 and TP16 to T13
<stintel> ugh I should have ordered some of those SMD resistors
aiyion has joined #openwrt-devel
<stintel> I tried bridging some pads on an OC200 a while ago and ended up messing up the pads :P
<dwfreed> too much jpeg in the fcc photos for me to see what they did
<stintel> maybe I'll try finalizing ECS4100-12PH support first
<stintel> don't have SMD resistors anyway
<aparcar> robimarko: I think you can manually define the package suffix, so you can keep libdefalte on git base but make it use tar.gz, while the rest uses xz
<robimarko> aparcar: thing is that there is no point in fetching libdeflate from git then
<robimarko> Currently it was done so because XZ was built first
<f00b4r0> stintel: really hard to tell from the pic but the resistors seems to be populated, aren't they?
<robimarko> And then it could be unpacked in order to be able to unpack gzipped archives
<stintel> f00b4r0: you mean the FCC photos? ack
<stintel> f00b4r0: T11 T12 T13 not populated, the FCC photo does at least seem to have something on T11 and T12
<f00b4r0> these aren't resistors. I'd expect diodes despite the T prefix
<f00b4r0> they're likely ESD devices
<f00b4r0> so most likely you can ignore them if they are wired the usual way (shunt to rails)
* f00b4r0 lookups FCC
<stintel> well I'm not getting anything if I connect RX to TP14
<stintel> so I suspect something is missing
<f00b4r0> T == Transil, most likely
<aparcar> robimarko: okay just wanted to let you know that you could use .tar.gz but I see your point
* f00b4r0 checks FCC pdf
<f00b4r0> can't see squat
<f00b4r0> stintel: if my assumption is correct, you're likely to find either GND or +VCC on either side of some of these pads, might want to check that before bridging anything :)
<f00b4r0> (some of these Txx pads)
<f00b4r0> the silkscreen symbol is definitely that of a diode
<stintel> T13 pad on the left side (thin line) is connected to GND
<stintel> and right side (thick line) is connected to TP16 which I suspect is VCC if it follows usual TP-Link layout
<stintel> so definitely don't bridge T13 :D
<f00b4r0> that ;)
<stintel> looks like I'm going to the local hacker lab
<aparcar> Mangix: nbd did you ever managed to build libxml2 on macOS?
<KGB-2> https://tests.reproducible-builds.org/openwrt/openwrt_mpc85xx.html has been updated. (100.0% images and 98.3% packages reproducible in our current test framework.)
<stintel> hmm so the pads on the sides of the thin line of T11 and T12 are also connected to GND
<stintel> ugh like I said, I'm not familiar enough with this stuff
<stintel> no idea what to do :P
<dwfreed> stintel: they may be pulled low by resistors elsewhere (or internal to the SOC; are you testing with power applied?
<stintel> no
<dwfreed> may be necessary to apply power to get good readings, though the RX pin will still look like it's connected to ground, and that's normal
<stintel> ugh
<stintel> damn hairy yaks
xes has quit [Ping timeout: 480 seconds]
<stintel> alright make clean it is
<stintel> sigh
<f00b4r0> stintel: this suggests T11 and T12 protect against reverse polarity
<f00b4r0> from what you described so far, i think you don't need to solder anything in these slots
<f00b4r0> (provided of course that you are very careful with what you connect to the protected signals)
<f00b4r0> reverse polarity or overvoltage if they are indeed meant to be Transil diodes
<f00b4r0> anyway, I'd leave them unpopulated
<stintel> ok so I'd just need to figure out which kind of socket fits on J255
<stintel> socket, connector, whatever
<f00b4r0> you'd need to determine the pitch between the pads
<f00b4r0> it looks similar to the SOIC-8 package in the upper part of your pic so I'm gonna guess 1.27mm
<stintel> that looks like it could be right, I don't have tools to measure it properly
<stintel> what would I be looking for, know there are 4+2 pins (2 on the side?)
<dwfreed> side bits serve as mounting support, that's all
goliath has joined #openwrt-devel
<stintel> yeah but looking at some JST stuff I don't see anything with such mounting support
<dwfreed> (they might be grounded, but the connector would not have any connection to that)
<stintel> hmm
<dwfreed> look at like 4 pin battery connectors
<dwfreed> like what's on an esp32, just 4 pin instead of 2
<nbd> aparcar: you mean the host variant?
<nbd> (of libxml2)
<aparcar> yes
<nbd> if i add a host build dep on libiconv-full, it seems to work
<nbd> yep, that's why the host build dep is needed
<aparcar> i installed libiconv via homebrew and added it to my envs, but can't seem to find it
<nbd> will add it
<aparcar> great thank you
<nbd> done
<aparcar> does work indeed
<aparcar> amazing RTT
Daanct12 has quit [Quit: WeeChat 4.2.1]
Daanct12 has joined #openwrt-devel
<stintel> anyone familiar with realtek target around? is there a reason non of the devices use FIT images (e.g. realtek u-boot doesn't support / enable it?) or is it an oversight?
<aparcar> ynezz: I see your comments and will rework it, wasn't aware of this corner case
<stintel> looks like the ECS4100-12PH u-boot does not like the initramfs image, suspect it got too big, I'm reading something about 6MB in commit message of other realtek device
<robimarko> stintel: I dont think the U-Boot version RTK uses supports FIT images
<stintel> U-Boot 2011.12.(2.1.4.53590)1.0.2.1 (Dec 01 2020 - 02:16:11)
<stintel> mips-linux-uclibc-gcc (GCC) 3.4.4 mipssde-6.03.00-20051020
<stintel> GNU ld version 2.15.94 mipssde-6.03.00-20051020
<stintel> geez
<stintel> Linux OpenWrt 5.15.153 #0 SMP Thu Apr 4 12:11:38 2024 mips GNU/Linux
<stintel> looks like 6.5MB initramfs worked \o/ Data Size: 6825171 Bytes = 6.5 MB
rua has joined #openwrt-devel
skynet2 has joined #openwrt-devel
<ynezz> aparcar: there are some, yes, thats why I've asked Ansuel for more details in the first place to have complete picture
<ynezz> but this change looks like it shouldn't hurt, so LGTM
<aparcar> thanks, I'll merge it
Daanct12 has quit [Quit: WeeChat 4.2.1]
fakuivan has joined #openwrt-devel
minimal has joined #openwrt-devel
parazyd has joined #openwrt-devel
theJoker8814 has joined #openwrt-devel
<stintel> oooh hostapd memory leak fixes
<stintel> maybe I can postpone retiring the EAP615-Wall
<robimarko> \x: I am leaning more towards making the default configurable via KConfig
mcbridematt has quit [Read error: Connection reset by peer]
DLange is now known as sleepingclown
sleepingclown has quit []
DLange has joined #openwrt-devel
minimal has quit [Quit: Leaving]
<owrt-images-builds> Build [#146](https://buildbot.openwrt.org/images/#/builders/190/builds/146) of `master_at91/sam9x` completed successfully.
<mrkiko> robimarko: regarding the nbg7815: how are the two 5 gGHZ PHYs supposed to be used? How are they used in the stock firmware ? I ask because I did read this comment on the device topic - https://forum.openwrt.org/t/openwrt-support-for-armor-g5-nbg7815/98598/798 and, in case, wondering if some day athk11 might implements this. Are the two PHYs supposed to use the same frequency? Any pointer on the
<mrkiko> topic? Thanks a lot.
<robimarko> mrkiko: So it registers 2 5GHz WIPHY-s?
<robimarko> AFAIK that was the "8x8" in QSDK
mattsm has joined #openwrt-devel
<mrkiko> robimarko: yes, it registers 2 but I had some difficulties bringing up the second one
<mrkiko> but yes you end up with a 2.4 ghz phy and 2 5 ghz one
<mrkiko> robimarko: wondering if one day this might be supported or never... not holding my breath, just curiosity
<robimarko> mrkiko: Sadly, I dont know how its supposed to work
<robimarko> but I have a feeling that it doesnt really work as a standalone one
<mrkiko> robimarko: I imagined it could work on same frequency with a sort of synchronization or coexistence scheme or whatever. I was thinking at first glance they could use different channels but the second phy doesn't look like you can use it straight away
hanetzer has joined #openwrt-devel
philipp64 has left #openwrt-devel [#openwrt-devel]
<stintel> svanheule: ping
<svanheule> stintel: o/
<stintel> thanks for the review!
<stintel> the setenv command is a remnant, shouldn't be needed (afaict)
<svanheule> that's good! 4MB is really small...
<stintel> yeah
<stintel> I had trouble booting >7MB initramfs but that was because I didn't use what I wrote in the commit message *facepalm*
<stintel> tftpboot 0x8f000000 is fine for a larger image it seems
<svanheule> happens to the best of us :P
<svanheule> nice, then this device has some years of support ahead of it :)
<svanheule> that's to say... if we can manage to keep doing those kernel bumps on realtek
<stintel> changing the edgecore,reboot to gpio-restart causes crash during boot, adding open-source causes this during reboot: https://gist.github.com/stintel/02e7ac4cd3d1affd50117757a0b60ad7
<svanheule> if that's one of the external GPIOs, that might be due to a chip reset
<stintel> also that commit message is from 2 years ago, man I have no idea where I got that info (maybe from the TIP wiki)
<svanheule> people have mentioned the reset-at-boot behaviour before, but nobody really dug into it IIRC
<stintel> [ 0.505472] RTL839X Chip-Info: a0036290, version D
<robimarko> svanheule: any hope for 6.1 on realtek?
<svanheule> stintel: doesn't is also print the full rtl839.. part number somewhere in the boot log?
<stintel> [ 0.000000] RTL839X model is 83936806 ?
<svanheule> stintel: that's the one! so it also says 8393 there
<stintel> yep. on both units (I only opened and removed heatsink on one but both say 8393
<svanheule> robimarko: there's more downstream driver code than combined developer energy to keep things moving :(
<stintel> and someone ragequitting didn't help I guess :(
<svanheule> stintel: 8392 is the less powerfull chip, so the "lowest common denominator" in a way. But you have actual HW with 8393, so I have a *slight* preference toward using 93 in the filename etc.
<svanheule> stintel: certainly not :-/
<stintel> alright, I have ~2 weeks before I'd like to start using the 2 units I have so I should have some time to sort some things out :)
<svanheule> bmork has looked into 6.1 most recently, but he ran into an(other) vague issue with the VLAN dehavior
<svanheule> *behavior
<stintel> was the dedicated IRC channel still active ?
<robimarko> Well, considering the giant pile of wonderfull magic code it would be weird if bumping kernels was easy
<svanheule> the IRC channel is still present, but inactive
<stintel> ok so not much point in joining I guess
<svanheule> there's 10 users silently hanging out there :P
<stintel> also I don't remember if these units came with OpenWiFi from factory or with some other thing, in any case I don't have access to "vendor interface"
<svanheule> robimarko: I would love to strip some of the "extras" like L3 offloading to make thing more maintainable. Problem is that *also* takes time and energy
<svanheule> stintel: vendor FW is usually based on the SDK and only works with initramfs images (and strips the squashfs trailer of a sysupgrade image)
<stintel> if I plug in an SFP and try again -> reboot
<stintel> Verifying Checksum ... Bad Data CRC
<stintel> ERROR: can't get kernel image!
<stintel> argh
<stintel> so I *do* need that setenv thing
<stintel> the 4MiB might be an arbitary chosen value because the default wasn't big enough
<stintel> I guess I could experiment with bloating the kernel and see where it says no
<mrkiko> I am a little bt confused - when we talk about devices like the fritz!box 5590 that have a fiber modem - does it mean you can use them without an ONT ? Or am I off?
<svanheule> stintel: bloating an initramfs image might be easier
<stintel> right, both use bootm so I guess it has the same limitations
<stintel> bootm seemed to work fine with a ~7.2MB image
<stintel> funny how they fake the 8392 in u-boot :P
<stintel> so I commented out the gpio-restart part in the DTS, and the thing still reboots fine, does that mean I don't need the gpio-restart at all? it's not really clear to me tbh
<svanheule> There are multiple reboot handlers available
<svanheule> By default it will use the built-in watchdog
<svanheule> But that's not really reliable on rtl838x
robimarko has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
<svanheule> So usually a GPIO pin is wired to the hard reset line
<svanheule> If this goes via the external RTL8231, then I think there might be a glitch when the chip resets at driver probe
<svanheule> The driver could be updated to just leave the config as-is, if that doesn't cause driver/HW state sync issues
<svanheule> And they were probably just too lazy to update the logging from u-boot :P
<stintel> right, so this happens because the gpio-restart is connected to the rtl8231-gpio - if we can believe the DTS from TIP, that is
<stintel> target/linux/realtek/rtl838x/config-5.15:CONFIG_EXTRA_FIRMWARE="rtl838x_phy/rtl838x_8214fc.fw rtl838x_phy/rtl838x_8218b.fw rtl838x_phy/rtl838x_8380.fw"
<stintel> hah
<stintel> this is not set in the rtl839x config
<stintel> looks like there's no code for loading phy firmware on anything but rtl838x, iiuc
<Mangix> aparcar: I'd need ssh access to some macOS install to check failures
<Mangix> I don't own apple hardware
<Mangix> well, besides iphone watch and airpods
<aparcar> Mangix: let me think forabout it
<aparcar> *for a minute
<Mangix> I pretty much gave up with jailbreaking my iphone
<aparcar> renting a macmini vm costs 175$/m... let me see if I get sponsoring for it
<Mangix> cool, https://github.com/openwrt/openwrt/pull/14971 marked as ready. time to pull
<dwfreed> github has macos runners
<zorun> you have access to https://portal.cfarm.net/machines/list/ IIRC
<stintel> hmmm wait this thing has two rtl8231 chips
<Mangix> zorun: I do. I have no idea if I can build OpenWrt with them.
<Mangix> oh there's one remaining
<Mangix> aaaaand nope
hanetzer has quit [Ping timeout: 480 seconds]
mcbridematt has joined #openwrt-devel
<owrt-images-builds> Build [#146](https://buildbot.openwrt.org/images/#/builders/138/builds/146) of `master_bcm27xx/bcm2708` failed.
minimal has joined #openwrt-devel