<digitalcircuit>
NBG6817 ipq806x CPU L2 1.4 GHz cache bug update: thanks to Ansuel's feedback, I'm starting to investigate router power stability. Alongside trying modifying the CPU voltages in the DTSI file (as noted in their email it helped some of their routers), I've ordered a USB current meter to measure my USB 3.0 HDD's power draw, and I've also ordered a USB test load in order to recreate any power draw scenarios.
<digitalcircuit>
I *have* managed to get the router to reboot with stress-ng doing L2 cache stress tests with 1.4 GHz L2 enabled, but I haven't yet run the test with 1.4 GHz L2 disabled long enough to be confident it triggers the same issue (un?fortunately, stress-ng is very good at causing all sorts of new, seemingly-unrelated hard-reboot scenarios).
<slh>
thanks for being so persistent, it's just necessary to stay on top of these issues - especially as they're so difficult to debug
Rentong has joined #openwrt-devel
<digitalcircuit>
slh: Oh! Sure thing, and I appreciate knowing it's welcome. By now I figured I was annoying folks by dragging this up every so often :)
Rentong has quit [Ping timeout: 480 seconds]
<slh>
it may be annoying, but it's also very important
<slh>
any bug is annoying, especially ones that are hard to debug or reproduce
<digitalcircuit>
That makes sense to me, and thank you too, I'll keep this in mind
<nick[m]12>
Tusker: I currently have no R2 available. however, I wanted to buy me a device ;)
<Tusker>
what are you going to use the SPI for ?
<Tusker>
actually, I have the BPI-M2, not R2 ... ah well, can't test it
<nick[m]12>
(what are you going to use the SPI for ?): probably connection to microcontroller
<Tusker>
that will work out pretty expensive if you buy the board and wifi cards...
<rsalvaterra>
Quick question… do we have *any* ARM targets *without* an MMU?
<PaulFertser>
rsalvaterra: we as in OpenWrt or as in Linux users?
<rsalvaterra>
"We" as in OpenWrt, not the whole kernel, of course. :)
<rsalvaterra>
I know you can run the kernel in a dead badger… :P
<PaulFertser>
generic/config-5.10:CONFIG_MMU=y
Acinonyx has quit [Ping timeout: 480 seconds]
<rsalvaterra>
I'd git grep around, but I'm doing write/compile cycles at the moment…
<stintel>
open a new terminal tab? ;)
<rsalvaterra>
stintel: It's a remote machine, don't want to start a new ssh session. Yeah, yeah, I know, tmux… :P
<stintel>
:P
<rsalvaterra>
OMFG, I fixed it.
<rsalvaterra>
\o/
<stintel>
gratz!
* rsalvaterra
goes to do a V=s compile just to make sure…
<Tusker>
anyway, time to log off for the week, have a good weekend guys :)
<stintel>
cheers, you too!
Tusker has quit [Quit: Time wasted on IRC: 10 hours 19 minutes 16 seconds]
<stintel>
ooof
<stintel>
I'm glad I don't keep track of that time wasted on irc :D
_lore_ has joined #openwrt-devel
<Habbie>
'27 years and counting'
<rsalvaterra>
Binutils' 2.37 ld is way smarter when it comes to getting rid of non-explicitely maked as used sections… :P
<stintel>
Habbie: :D
<Habbie>
:)
<rsalvaterra>
*marked
<rsalvaterra>
And one of the sections it dropped (on arm32, at least) is… __ex_table. :P
<rsalvaterra>
Not cool.
Acinonyx has joined #openwrt-devel
<rsalvaterra>
Ok, apart from the usual stack size warnings, the kernel built just fine.
Acinonyx_ has joined #openwrt-devel
Acinonyx has quit [Ping timeout: 480 seconds]
<rsalvaterra>
I miss the times when it was safe to build with 4 kiB kernel stacks… now it's even impossible to select the stack size… :)
<rsalvaterra>
It's possible to dive in the code and hack THREAD_SHIFT, but it will break the system spectacularly with crazy complex filesystems, lvm and/or graphics devices. Probably only doable on embedded… wait a minute.
Rentong has joined #openwrt-devel
Tapper has quit [Ping timeout: 480 seconds]
Tapper has joined #openwrt-devel
<stintel>
in github, is it typically the reviewer that marks things as resolved, or the author?
<Habbie>
in just about every project where i've had to ponder that, i have resolved things as authors, and i have expected other authors to resolve their things
<Habbie>
*as author
Rentong has quit [Ping timeout: 480 seconds]
Ansuel has joined #openwrt-devel
Ansuel has quit []
gladiac has quit [Quit: k thx bye]
gladiac has joined #openwrt-devel
<russell-->
rsalvaterra: if you use screen instead of tmux, you get serial for free
<stintel>
you can do serial with tmux o_O
<russell-->
WUT? when did that get added?
<rsalvaterra>
Guys, please. I use tio for serial. :P
<rsalvaterra>
Previously I used picocom, but I found out tio is even smaller.
<stintel>
russell--: sorry, that was supposed to be a question, I misread what you wrote
<russell-->
oh, lol
<stintel>
let's blame it on Friday ;)
<russell-->
thank heavens, i was about to git clone tmux to look
<rsalvaterra>
xD
<russell-->
TIL about tio though, even though since i like screen it is useless to me ;-)
<rsalvaterra>
Well, but wasn't screen deprecated? On Fedora, at least?
<rsalvaterra>
I thought it was in very light maintenance mode.
* russell--
wears garlic to fend off fedora
<rsalvaterra>
Heh… I'm back to Ubuntu (MATE) after several months on Debian Sid…
<stintel>
Gentoo for lyfe!
<rsalvaterra>
Yeah, Gentoo still holds a special place in my heart, since it's the second distro I used… :)
* wb9688
uses Debian unstable
<rsalvaterra>
I think I need to try Arch one of these days…
minimal has joined #openwrt-devel
Tapper has quit [Remote host closed the connection]
Tapper has joined #openwrt-devel
\x has quit [Ping timeout: 480 seconds]
<dangole>
nick[m]12, Tusker: I got the R2 on my desk atm. wifi/bt are not so trivial, it's a mess of a driver (looks a bit like old Ralink proprietary driver codebase), ie. not even using cfg80211, weird userspace tools for firmware loading, ....
Tapper has quit [Ping timeout: 480 seconds]
<nick[m]12>
dangole: thanks!
rsalvaterra_ has joined #openwrt-devel
<nick[m]12>
dangole: is spi working?
<dangole>
nick[m]12: hm, I have to attach something to SPI to try, I guess...?
rsalvaterra has quit [Ping timeout: 480 seconds]
rsalvaterra_ has quit []
rsalvaterra has joined #openwrt-devel
<nick[m]12>
dangole: shouldn't there be some entry under /dev/ ?
Acinonyx has joined #openwrt-devel
Rentong has joined #openwrt-devel
Acinonyx_ has quit [Ping timeout: 480 seconds]
<rsalvaterra>
dangole: You probably know this better, do we support any ARM no-MMU device in OpenWrt?
<PaulFertser>
btw, it seems the only nommu musl target is (was?) SH-2.
<rsalvaterra>
PaulFertser: That settles it, thanks! :)
Rentong has quit [Ping timeout: 480 seconds]
<PaulFertser>
Apparently there was ubicom32 nommu target. Ten years ago.
<PaulFertser>
And it was the only one, introduced in 2009, removed in 2012.
<PaulFertser>
git log -G 'FEATURES.*nommu'
<rsalvaterra>
PaulFertser: And it's not ARM, it's some other proprietary architecture, if the datasheet is to be believed.
<rsalvaterra>
So, if you guys want to play with my previous binutils 2.37 patch and have arm systems, you definitely need this too. :P
ecloud has quit [Ping timeout: 480 seconds]
ecloud has joined #openwrt-devel
<fda>
dangole: IDLE: with min300mhz there is ~1 transition/? . with min125mhz ~30 transitions/? some LOAD: 25-30 transitions https://ibb.co/B3w6tC9
\x has joined #openwrt-devel
floof58_ has quit [Ping timeout: 480 seconds]
rejoicetreat has joined #openwrt-devel
<dangole>
rsalvaterra: no, i think we never officially supported non-MMU systems. blackfin is the only interesting embedded platform (for VoIP) without MMU I could think about that would have been relevant in the past...
<rsalvaterra>
dangole: Ohhh, blackfin is currently dead an burried… :/
<rsalvaterra>
So, I have about 90 % confidence my patch is fine. :)
floof58_ has joined #openwrt-devel
Rentong has joined #openwrt-devel
decke has quit [Quit: Leaving.]
dorf has quit [Remote host closed the connection]
dorf has joined #openwrt-devel
danitool has quit [Ping timeout: 480 seconds]
danitool has joined #openwrt-devel
dorf has quit [Remote host closed the connection]
jordbaer has joined #openwrt-devel
<jordbaer>
hi. my company will manufacture a ethernet-over-coax device with wifi in a wall socket. my task is to port openwrt to the device. it's a mediatek/ralink mt7628an chipset. is there some kind of README so I know how to create a new .dts and integrate it into the openwrt build system?
<rsalvaterra>
dangole: I see 24 targets/subtargets with kconfigs named "config-default" and it worries me… when adding a testing kernel, we don't know if the symbols can be shared. Why not just name all the kconfigs config-<kernel major>.<kernel minor>?
<rsalvaterra>
Sure, we might have duplicate kconfigs, but seems safer to me…
<dangole>
rsalvaterra: some of these config-default are not very big and really just select the target and apart from that do not diverge much from target/linux/generic/config-<kver>
<dangole>
rsalvaterra: if there are a lot of entries you will have to duplicate it into version specific files in the moment we want to offer testing kernel in addition to stable kernel for that target/subtarget
<hurricos>
I see you are playing with QLE3242/NC523SFP's, sintel :)
<hurricos>
stintel*
<stintel>
hmm, am I?
<stintel>
ah no, that was just a thing I ran into due to building with CONFIG_ALL_KMODS=y
Borromini has quit [Quit: leaving]
Tapper has quit [Remote host closed the connection]
Tapper has joined #openwrt-devel
<hurricos>
stintel: oh ....
<hurricos>
well in any case, thank you. I know now about those fantastic / cheap cards.
<stintel>
hurricos: hah, you're welcome ;)
<stintel>
hurricos: btw, not sure if you got my PM, but 1 of the cards you shipped is in Belgium now, in svanheule's hands :)
<hurricos>
stintel: I did not! Great to hear. I've been focused on working crap wages and recently found out a less stressful + more senior position my boss wanted me for 'never existed' ...
<hurricos>
I've wanted to grab your work, re-test and submit to linux-mips / mainline, but haven't found the time through all thta.
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
<hurricos>
stintel: was there any further work done to get those hw queues working?
<hurricos>
stintel: maybe I should learn kubernetes and move to Europe? :^)
<stintel>
hurricos: no I haven't done anything on this anymore
<stintel>
I've had a crazy 6 weeks
<hurricos>
stintel: I don't blame you, those cards suck eggs.
<hurricos>
stintel: Oh?
<stintel>
well I had planned to go hometown for the first time since the start of the pandemic, then all of a sudden a new job finds me, had some overlap between old and new job, and lots of stuff to take in for the onboarding etc
<hurricos>
stintel: argh, that sucks ...
<hurricos>
I have a similar situation. My grandmother was born 1931-09-13 ....... and she wants to see me
<stintel>
well .. I'm happy with the new job, but it's been a crazy time, I'm back home since Monday
<hurricos>
but I'm an American with two passports
<stintel>
so now things are settling down
<hurricos>
and the other one is British :^)
<hurricos>
so now I get to figure out how to get the paperwork to get to Europe. 2x dose Pfizer but I have no idea whether I need spare COVID-19 tests or what???
<hurricos>
stintel: that's good, though. At least things are settled.
Rentong has joined #openwrt-devel
Rentong has quit [Ping timeout: 480 seconds]
jordbaer has quit [Quit: jordbaer]
<fda>
stintel: ALL_KMODS? Interesting! I need all the time to flash if i miss some kernel modules. Are in this case all installed loaded automatically at startup?
<stintel>
fda: no, CONFIG_ALL_KMODS simply builds all kmod packages