<hanetzer>
didn't work before. maybe I tried it with the cruft.
minimal has quit []
fda has joined #openwrt-devel
fda- has quit [Ping timeout: 480 seconds]
dangole has quit [Ping timeout: 480 seconds]
fda- has joined #openwrt-devel
Tusker has joined #openwrt-devel
fda has quit [Ping timeout: 480 seconds]
<hanetzer>
ok. so this is being far more dickish than I thought. maybe worthwhole to just spin up a vm lol
<victhor>
Tusker: is your image installed the same way as mine were? Flash factory image to first partition?
danitool has quit [Ping timeout: 480 seconds]
<Tusker>
it is setup to flash to .main, so overwrites the factory main, and keeps factory bkup
<Tusker>
i did all my testing via tftpboot and bootm, but the ruckus header is all setup inside the sysupgrade image
<Slimey>
you also like enterprise waps? :P
<Tusker>
yeah, can pick them up cheap, and often with warranty and support contracts still valid :)
<Slimey>
i have a bunch of adtran waps, working on 192x, 2030, and the big guy bsap-3040
<Tusker>
and it's a bit of an addiction... have a watchguard xtm330 under my desk once I get the ap3935i wifi working again (i broke it somewhere along the line)
<Slimey>
yes heh
<Slimey>
i just need more hours in the day
<Tusker>
definitely a great learning process
<Tusker>
first time using dyndbg yesterday for me
Grommish has joined #openwrt-devel
<victhor>
they sure are fun, maybe a bit too much sometimes...
<Slimey>
dyndbg ?
<victhor>
anyway, got it. I'll try it tomorrow, too late to mess around with hardware right now
<Tusker>
dynamic debug, pr_debug(...) statements from kernel modules
<Tusker>
OK... so ap3935i wifi problem solved by booting back into factory and then not loading the incorrect precal data... let's see if it survives power off and on
decke has joined #openwrt-devel
<Tusker>
yup, power off makes the power die...
goliath has quit [Quit: SIGSEGV]
danitool has joined #openwrt-devel
<rsalvaterra>
mangix: DSA on the C7 coming along nicely, heh? :)
_lore_ has quit [Read error: No route to host]
_lore_ has joined #openwrt-devel
<rsalvaterra>
mangix: Can't wait to test on my TL-WDR3600, should be trivial to port the changes.
<mangix>
rsalvaterra: 4300 has a different switch configuration
<mangix>
with that ag71xx patch I linked to earlier, it becomes 571mbps
<slh64>
tl-wr1043nd v1 should also work with realtek_smi, but I don't know if anyone has tried that so far (8/32)
<rsalvaterra>
Over 70 Mb/s difference…! That's… pretty wild.
<mangix>
that's iperf. routing should be faster.
<rsalvaterra>
I think the problem with the DSA migration is being an all-or-nothing thing. You can't migrate gradually, changing a .dtsi will affect all devices downstream of it.
<mangix>
sure but if the switch configuration is in a dtsi file, all devices downstream work the same
<mangix>
safe assumption to make
<mangix>
also note that mvebu since 19.07 has had both swconfig and DSA devices
<mangix>
*cough*omnia*cough*
<rsalvaterra>
mangix: I think DSA support for the Omnia goes all the way back to 18.06.
<mangix>
kernel 4.14 IIRC
<mangix>
yeah
<mangix>
18.06 was a dual kernel release
<rsalvaterra>
mangix: I know it's just proof-of-concept for now, but the changes to ath79/{config-5.10,Makefile} break all non-DSA devices. We'll probably have to live with both drivers for the time being. :(
<mangix>
sure. Someone has to figure out how to stop swconfig from hijacking things.
<mangix>
then it can coexist
<slh64>
there are quite some pitfalls beyond just the common dts changes, as we've seen on ipq806x with qca8k. both driver side (initial mdio issues with the early qca8k code) and all the custom firstboot setups (different ways to assign MAC address etc.)
<rsalvaterra>
By the way, does the LED configuration for the switch change?
<mangix>
for the C7s I have, no
<mangix>
it might be bootloader dependent
<mangix>
no idea
<mangix>
I use breed, which has been amazing for testing qca8k
<Namidairo>
wasn't the web ui for that in chinese
<rsalvaterra>
My WDR3600 also has an alternate bootloader. The one from pepe2k, but with overclocking (running at 680 MHz instead of the factory-default 560).
<mangix>
Namidairo: not really an issue
<mangix>
han zi becomes easier after a couple of years
<rsalvaterra>
mangix: After a couple of years of what? :)
<mangix>
well, let's not go there :)
<rsalvaterra>
XD
<Namidairo>
iirc there was chrome extension or similar to translate it anyway
<rsalvaterra>
I had to use my phone and the Google app for translation, while flashing my Redmi AC2100. :P
<mangix>
1000mhz technically work, except it doesn't
<rsalvaterra>
mangix: Not even at CAS 5?
<Namidairo>
yeah you get weird clock drift at high overclocks on those qualcomm mips platforms don't you?
<slh64>
Namidairo: not really that useful (or sufficient), if you need to use it on a router (had lots of fun with that on my ax3600), probably even worse on the bootloader
<mangix>
It behaves as if clocked to that but in reality is clocked to NINT -1
<Namidairo>
it's funnier when you see a dozen people start cross-flashing breed images for other models
<mangix>
Namidairo: it's specific to 1ghz and above. Something strange happens at 1ghz
<Namidairo>
*points at rsalvaterra*
<mangix>
rsalvaterra: why would I switch to CAS 5?
<PaulFertser>
Is breed still proprietary?
<mangix>
PaulFertser: it's made by a mediatek employee who has contributed a lot to OpenWrt
<PaulFertser>
mangix: yes, and sent patches to OpenOCD for JTAGing MTK SoCs. But that doesn't answer my question :)
<rsalvaterra>
mangix: If the clocks are dependent (derived from a master clock), increasing the CPU clock will also increase the DDR clock. Increasing the CAS wait state could possible stabilise the RAM (unless, of course, it's the CPU which becomes unstable at 1 GHz).
<mangix>
the answer is that it's proprietary. I imagine it has trade secrets as part of the source
<PaulFertser>
mangix: thank you for answering, and my next question was just the one you answered too. OK.
<mangix>
rsalvaterra: uhh, no? did you check that picture?
<rsalvaterra>
mangix: Yeah, right after writing. :P
<PaulFertser>
Impressive amount of clock tree tweaking facilities, for a web interface.
<mangix>
PaulFertser: anyway, I find no issues with breed. I don't see Weijie as malicious
<mangix>
hmmm I thought the c7v2 had DDR1, not 2
<mangix>
oh wow, it's DDR2
<rsalvaterra>
Yay, my LGBT keyboard just arrived. :P
<stintel>
Ducky ? :P
<rsalvaterra>
FFS, why do I have to be stuck with RGB-lighting crap if I want a good mechanical keyboard…?
<mrkiko>
PaulFertser: yes, I would like very much for it to become OSS as well.
<ldir>
rsalvaterra: come on, come out of the closet ;-)
<rsalvaterra>
mangix: Don't get too happy about being DDR2… It's an embedded system, so it's probably 16-bit wide. :P
<rsalvaterra>
ldir: No!
* rsalvaterra
closes the closet door
* ldir
hangs some pretty coloured lights
<Namidairo>
the c7v2 was okay when it was released
<rsalvaterra>
ldir: I'm not falling for that one. :P
<Namidairo>
but I think it's less useful now because it's just a wave-1 card
<Namidairo>
makes it a pain to put it into a 802.11s mesh
* ldir
plays Gloria Gaynor's I will survive
<rsalvaterra>
ldir: Christ. I've sung that in a karaoke, together with a friend, at a birthday party. 20 years ago. There are no records of it. :P
* ldir
segues into Madonna's Vogue
<rsalvaterra>
The most troublesome part is, at the time I didn't drink at all. :P
<ldir>
rsalvaterra: oh so you might have actually been in tune as opposed to thinking you were in tune :-)
<rsalvaterra>
I sing in a choir, so I probably wasn't too out of tune. It's not a song for my voice, of course. :P
<mangix>
Namidairo: card can be replaced.
<Namidairo>
are 3x3 mini-pcie cards easy to find
<mangix>
Can't wait for Turris people to release their ax card
<mangix>
Namidairo: yes
<Namidairo>
I went googling and found Broadcoms and more qca9880's
<Namidairo>
although I suppose you could just stick a 2x2 11ax card in there too
<mangix>
No real issue with that
Tapper has quit [Ping timeout: 480 seconds]
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
Tapper has joined #openwrt-devel
<karlp>
rsalvaterra: das keyboard?
<stintel>
that reminds me I should capture the "save" packet of the ducky software on the USB bus
pmelange has left #openwrt-devel [#openwrt-devel]
<stintel>
because I crashed my USB bus (cheap logic analyzers ftl) and after that the keyboard went back to lgbtqwtfbbq mode
<rsalvaterra>
stintel: I got the Logitech G512 Carbon.
<mangix>
All of you with your fancy keyboards
<Habbie>
stintel, sniffing USB on a VM is safer ;)
<rsalvaterra>
stintel: I was torn between blue and brown switches, but the I remembered I have neighbours, so I went with brown. :P
<karlp>
wife felt that brown was still far too loud here :)
<rsalvaterra>
Ouch… :/
f00b4r0 has quit [Read error: Connection reset by peer]
f00b4r0 has joined #openwrt-devel
<rsalvaterra>
karlp: I've been eyeing the Das Keyboard models, but they're overpriced and I've never seen them with PT layout. And the G512 has a much thinner bezel.
dangole has joined #openwrt-devel
<karlp>
rsalvaterra: I got custom keycaps from wasdkeyboards to get icelandic keys on mine.
<karlp>
but yeah, they're certainly not cheap.
<karlp>
I like hving the volume dial and the built in usb3 hub though.
<rsalvaterra>
Fair enough. :) I got the G512 for under 84 € including shipping, which wasn't too bad. Still the most expensive keyboard I ever got. :P
<karlp>
mine was a bday present to myself a couple of years ago.
<karlp>
only way to justify it's price.
<karlp>
but hey, no RGB :) met the stated specs ;)
victhor has joined #openwrt-devel
<stintel>
Habbie: yes that's how I did it
<stintel>
rsalvaterra: working from home since long before the pandemic, so blue!
<stintel>
I have 2 even :P
Tapper has quit [Ping timeout: 480 seconds]
<rsalvaterra>
stintel: Yeah, I know, redundancy… ;)
<stintel>
:D
<stintel>
just a second one for my secondary location
<stintel>
it's heavy enough to make hand luggage > 10kg
<rsalvaterra>
I think I'm going to bit the bullet and try to give my Archer C6 v2 the DSA treatment, later today…
<rsalvaterra>
I may have to nag someone for device tree guidance, though. :P
<victhor>
Tusker: good news! Your image is booting from flash
<Tusker>
victhor: good news :)
<Tusker>
hmm... is there a way to pull the caldata from the pci cards (ath10k) - if I power off and power on, it loses the caldata, and performance is really low
victhor has quit [Remote host closed the connection]
victhor has joined #openwrt-devel
<victhor>
Tusker: may I suggest changing the port configuration? I seem to remember in my later commits I pushed a configuration change that set all ports as LAN and enabled DHCP by default
<victhor>
as it stands, the PoE port is a WAN port in OpenWrt by default, which is rather inconvenient
<Tusker>
victhor: it can be configured after the fact, but to follow usual OpenWRT standards I've done it that way
<victhor>
I see. Some APs were set by default that way when I did the port, so I thought it was acceptable
<Tusker>
it might be acceptable :) I have a PoE switch in between the r500 and upstream
<Tusker>
so maybe I am biased
<victhor>
perhaps, my main switch is a PoE one
<victhor>
https isn't working, but you probably already noticed it
<Tusker>
https where? in luci ?
<victhor>
I think the caldata for ath10k is copied by the driver from a mtd partition if configured to do so in the dts
<victhor>
yep, in luci
<Tusker>
victhor: yeah, for r500, it's good. the ap3935i doesn't have anything on the mtd
<victhor>
oh I see
<Tusker>
i don't use luci, so didn't test it. it is probably not related to the port
fda- has joined #openwrt-devel
<victhor>
yeah, it's just a package missing in the configuration or something like that
<victhor>
sysupgrade works on the r500 now, right?
<Tusker>
yeah, works on the current PR
fda has quit [Ping timeout: 480 seconds]
Strykar has quit [Quit: /quit]
<victhor>
excellent, great job
ecloud has quit [Ping timeout: 480 seconds]
<victhor>
so I shouldn't have to crack it open ever again, hopefully
<Tusker>
yeah, hope not :)
Strykar has joined #openwrt-devel
ecloud has joined #openwrt-devel
dedeckeh has joined #openwrt-devel
<Tusker>
PaulFertser: OK... so, it was the precal files after all... the ones from openwrt firmware are not "good enough" I powered off, and used the ones extracted from the filesystem on the factory firmware, and now I have -17dBm
<Tusker>
but, weirdly, the precal files work fine if the boards aren't powered down
<PaulFertser>
Tusker: oh, surprising, thanks for letting me know.
<Tusker>
not sure what to do to package it though... it's not like we have permission to just use random precal files from filesystems, can we ?
<PaulFertser>
Some ath10k experts needed to answer that
<Tusker>
ath10k_patch_mac doesn't work on the precal files though...
<Tusker>
(the new ones that have good signal)
<Tusker>
editing the precal files manually seem to work... odd :)
<Tusker>
anyway, have solved the problem with the device, will solve the project admin side another day :)
indy has quit [Read error: Connection reset by peer]
indy_ has joined #openwrt-devel
<Tusker>
victhor: let me know if you have any problems, I have a spare r500, not in current use, that I can use for testing
<Tusker>
ciao
Tusker has quit [Quit: Time wasted on IRC: 9 hours 59 minutes 27 seconds]
_lore_ has quit [Ping timeout: 480 seconds]
minimal has joined #openwrt-devel
nitroshift has quit [Quit: Gone that way --->]
chadneufeld has joined #openwrt-devel
_lore_ has joined #openwrt-devel
arifre has quit [Quit: Page closed]
arifre has joined #openwrt-devel
jbowen has quit [Ping timeout: 480 seconds]
neg2led has joined #openwrt-devel
neggles has quit [Read error: Connection reset by peer]
jbowen has joined #openwrt-devel
indy_ is now known as indy
Tapper has joined #openwrt-devel
floof58 has quit [Ping timeout: 480 seconds]
jbowen has quit [Quit: leaving]
floof58 has joined #openwrt-devel
<mangix>
rsalvaterra: thought you said you got it :P
decke has quit [Quit: Leaving.]
goliath has quit [Quit: SIGSEGV]
<stintel>
hauke: fyi, I'm going to apply the musl 1.2.2 series you sent and testrun it a bit
<Habbie>
ohhh
* stintel
nukes build_dir and staging_dir to avoid frustration ;)
<mangix>
stintel: expect lower memory usage.
<stintel>
probably won't notice that, don't have memory-constrained devices
<Habbie>
mangix, except for timestamps on 32 bit devices!
<luke-jr_>
anyone who can get me a wiki account?
<luke-jr_>
can't be bothered with joining a forum..
<PaulFertser>
luke-jr_: please send me desired user name and e-mail
goliath has joined #openwrt-devel
Tapper has quit [Ping timeout: 480 seconds]
danitool has joined #openwrt-devel
<rsalvaterra>
mangix: Got what? :)
chadn has joined #openwrt-devel
<rsalvaterra>
stintel: Do it from orbit. It's the only way to be sure. :P
chadneufeld has quit [Ping timeout: 480 seconds]
<stintel>
rsalvaterra: ?
<stintel>
ah, the nuking
<stintel>
:P
<rsalvaterra>
I have hauke's musl series in my tree already, but I've only build-tested.
<stintel>
flashing the dev m300 as we speak
* rsalvaterra
is waiting for the monster 5.10.65 release, tomorrow. 236 patches.