<Lechu>
Got my hands on next Ruckus devices, this time AR7161 based. I got almost everything working, save for Ethernet port using RGMII and MV88E1116 phy. Regardless of PLL data, no data is ever received by the interface, though the TX works. PHY is detected properly. Any ideas what to check next?
<Lechu>
The same happens for ZF7351 and ZF7363.
<Lechu>
In statistics I see zero received packets, not even error frames.
<Lechu>
PHY statistics also show zeros in ethtool
<Lechu>
I get throughput when link speed is reduced to 100 or 10Mbps
<Lechu>
I managed to extract something from U-boot's logs based on different speeds and setting for GbE was different. Let's see if that helps.
<Lechu>
There is one more elephant in the room, RTL8363 switch in ZF7363.
minimal has quit [Quit: Leaving]
Ansuel has quit [Quit: Probably my PC decided to sleep or I decided to sleep.]
Tapper has quit [Ping timeout: 480 seconds]
<Lechu>
And nope, still no throughput at GbE.
danitool has quit [Remote host closed the connection]
isak has quit [Remote host closed the connection]
tSYS has quit [Quit: *squeak*]
tSYS has joined #openwrt-devel
SlimeyX has quit [Remote host closed the connection]
isak has joined #openwrt-devel
hanetzer has quit [Quit: WeeChat 3.7.1]
dangole has quit [Ping timeout: 480 seconds]
hanetzer has joined #openwrt-devel
valku has quit [Quit: valku]
aiyion has quit [Remote host closed the connection]
jayk has quit [Read error: Connection reset by peer]
jayk has joined #openwrt-devel
<stintel>
can we have a bot that automatically does s/OpenWRT/OpenWrt/g ?
<jayk>
people know what it means, why bother
<stintel>
because it's wrong
<Znevna>
some devs use the "wrong" casing anyway
<stintel>
then they should be corrected also :)
<Znevna>
¯\_(ツ)_/¯
<Znevna>
as long as they fix bugs I wouldn't care about the casing
<Znevna>
and probably for some WRT has more meaning
<stintel>
how would you feel if someone you work with keeps consistently spelling your name wrong ?
<jayk>
shouldnt let little things bother you
<Znevna>
but the spelling is right
<stintel>
look at it differently: if you can't be bothered writing it correctly why should we bother looking into your problem?
<Znevna>
why is it Wrt and not WRt ?
<Znevna>
or WRT? :P
mcbridematt has quit [Remote host closed the connection]
<jayk>
openwrt is easier to type
<jayk>
i prefer that!
<jayk>
..and the outcome is the same..does not effect operation, only user
lmore377 has quit [Remote host closed the connection]
<stintel>
I wasn't there when they decided on the name, I just know it's OpenWrt and not OpenWRT
<stintel>
and we care enough to make that the 2nd slide of presentations
<jayk>
perhaps the channel should be renamed then, too
<jayk>
#OpenWRT-Devel
<Znevna>
Wrt-devel
lmore377 has joined #openwrt-devel
<jayk>
*open***-*evel
<jayk>
how bout that
<jayk>
fill in the blanks!
<oliv3r[m]>
still strikes me as odd; Wrt vs WRT; WRT is an abbreviation, Wrt reads like CamelCase and doesn't make sense :p
<stintel>
I don't mind openwrt. but if you use caps, use them correctly
<oliv3r[m]>
'correctly' with big quotes, as 'what is correct' :p
<jayk>
ask alexa
<jayk>
i dont think she cares much about spelling
<jayk>
+syntax
<stintel>
there's are reason the logo is OpenWrt. there's a reason we write OpenWrt the website/forum/...
<jayk>
dns doesnt care
<jayk>
itll get there if you type oPeNWrT.oRG
<stintel>
wasn't talking about DNS
<jayk>
poor logic
<stintel>
right
<jayk>
dont change what aint broke
<stintel>
open openwrt.org or forum.openwrt.org, do a case sensitive search for OpenWRT
<stintel>
you'll find not matches
<Znevna>
since there's nowhere mentioned what "wrt" stands for and why only W is uppercase is hard for people to type it "correctly"
<jayk>
should rename tomato tomatoe then
dansan has quit [Ping timeout: 480 seconds]
borek has joined #openwrt-devel
<slh64>
doing automated string replacements has unintended consequences, as it is prone to 'fixing' things that must be 'wrong', e.g. in configuration files, code snippets or patches
<oliv3r[m]>
Wrt comes from the linksys WRT54g doesn' tit; where WRT stood for 'WirelessRouTer' if memory serves me well
<Znevna>
that's what google tells us
<oliv3r[m]>
hence, OpenWRT would stend for Open WirelessRouTer :); however now, it's just wrt without meaning, but camelcased :) the logic got lost. While 'lede' was a poor name, it was a very destinct name
<oliv3r[m]>
also, afaik OpenWrt won't be just rireless routers for long, generic networking gear solution is comming closer (or has come already?) and who knows, hopefully we can run it on (door)cams soon too :)
<robimarko>
I mean, its just a flexible distro, so you can run it on anything
<Znevna>
and back in 2004 the webpages had <title>OpenWRT</title>
<oliv3r[m]>
In that sense, you can argue 'wrt' no longer stands for anything, its just a (confusing) name with some history )
<Znevna>
so it was confusing even when that went online ;P
<oliv3r[m]>
time for some rebranding :)
<oliv3r[m]>
My bet, is some camelcase lover at some point figured 'it must be camelcase!'
<robimarko>
What is the point of rebranding now that everybody knows the name OpenWrt
<Znevna>
germans might argue that the R should be uppercase
<Znevna>
since .. Router
<f00b4r0>
more importantly, how do you pronounce the "Wrt" part? ;)
<oliv3r[m]>
'everybody' is a bit strong :p
<oliv3r[m]>
but I think the XBMC -> kodi transition went quite nicely in the end
<jow>
it was rebranded?
<f00b4r0>
sort of
<robimarko>
GH actions are pilling up
<Znevna>
from wiki: The ending of Xbox support by the original project was also the reason that it was renamed "XBMC" from the old "Xbox Media Center" name, and why it later was renamed "Kodi".
<jow>
I wish some PRs would be mergeable from the get-go
<robimarko>
Good one
<Ansuel>
ynezz i had a similar idea but i would make it for specific test and keep the current way
<Ansuel>
something like "I need this pr to be tested with each subtarget" and we add a tag and the test is triggered
<Ansuel>
(instead of creating TODROP commit to manually trigger tests)
<robimarko>
Ansuel: Since you are online, does IPQ806x have the VDD_APCC and VDD_MX logic
<robimarko>
For normal and turbo states?
<Ansuel>
mhhh how to check that?
<robimarko>
I mean, does it change the voltage source for memory controller etc once in turbo state?
<Ansuel>
mhhh no or yes? they are linked to cache voltage i thing and cache voltage is changed to max value when cpu are at max freq
<robimarko>
Ok, so its completely different then
<robimarko>
Cause, it seems that they moved to APM on newer SoC-s
<robimarko>
And here it just switches between 2 voltage "sources" depending on if a voltage threshold on CPU has been crossed
<robimarko>
Basically turbo mode for memory controller and peripherals once you are in high OPP-s
<robimarko>
Investigating all of that now that upstream CPR is being worked on again to get rid of the downstream implementations
<ynezz>
jow: thats what I like on that bors bot approach, you simply add `bors r+` and rest is handled automagically
<ynezz>
s/add/add comment/
<robimarko>
But isnt currently stuff getting manually picked to git.openwrt.org and then pushed which then syncs the github mirror?
<ynezz>
indeed, we would need to adjust for that
<Ansuel>
ynezz bidirectional sync is a nightmare from what i read online
<ynezz>
I didn't meant bidirectional sync
<ynezz>
we should finaly decide where the project should live, move there and stay there until it makes sense
<robimarko>
That was a bit of an hot topic
<Ansuel>
eh fun times since last meeting we were still questioning if github was a good idea
<ynezz>
yes, even mentioned on the last meeting, some other hosting
<ynezz>
exactly, so lets make a vote and move on?
<robimarko>
I mean, whats the competition?
<Znevna>
oh noes and I was just getting familiar with github
<robimarko>
Codeberg, Gitea and Gitlab?
<ynezz>
codeberg IIRC
<Ansuel>
codeberg i think? but honestly github ci are too good compared to other platform
<f00b4r0>
and then there's the matter of platform popularity (i.e. how much do we want to have a wider audience and make contribution easy)
<Ansuel>
ynezz we should first fix the other problem with the vote rules itself no idea if we made some progress on that
<robimarko>
Lets be honest, Codeberg is a complete unknown for 99% of the people
<f00b4r0>
^
<f00b4r0>
i only opened a gitlab account to be able to contribute to buildbot ;P
* f00b4r0
hides
<Ansuel>
but honestly moving to github and enabling merging pr would result in the repo getting bloated by empty commit with Merge #3238 from blah/bhu that I really hate...
<robimarko>
^ Second that
<robimarko>
I hate merge commits
<f00b4r0>
same
<ynezz>
you can enable fast-forward only mode IIRC
<f00b4r0>
correct
<robimarko>
Whats CI in Codeberg like?
<ynezz>
it's drone.io IIRC
<Ansuel>
and that won't solve anything since people will have to rebase each time or maintainers will have... Wonder if the merge bot can be instructed to rebase and merge to git.... we can consider giving him a ssh key and pass it with github secrets
<robimarko>
Github does have an option to rebase and merge PR-s
<robimarko>
ynezz: We use drone.io at work
<robimarko>
But havent personally had to work with it
<Ansuel>
right the rebase and merge way
<Znevna>
I don't see any "empty" commit from my PR and it got merged fine (I think?) how was that ok?
<robimarko>
There are no merge commits currently as its manually picked and pushed to the branch
<Ansuel>
Znevna i created a script... github flag pr as merged if the commits are merged in the base branch with fast-forward
<Znevna>
ok x.x
<Ansuel>
ynezz btw i would check if triggering tests with label is not too hard to implement... It would be very useful. Ideally the tag is removed as soon as the test is triggered.
<Ansuel>
easy change that should massively speedup tests
<robimarko>
Oh yeah
<Ansuel>
what I'm scared is the changed file detection... but at worst we will have the tag method to the rescue
<dhewg>
no idea if that's possible, but every now and then I force push but just change the commit messages, maybe it can skip ci if there're no incremental file changes?
<dhewg>
just happened on #11728 if that helps
<Ansuel>
force push brake the entire system... github internally lose track of the last event so you can't track what changed before and after so it's a big corner case hard to handle
<robimarko>
Ansuel: They have a concurrency system now
<Znevna>
that commit with "todrop ipq4019" scared me rofl
borek has quit [Ping timeout: 480 seconds]
Ansuel has joined #openwrt-devel
dansan has joined #openwrt-devel
<jayk>
/win 2
Lynx- has joined #openwrt-devel
<jayk>
oliv3r[m]: lolol camelcase lover...it should just stay as it is to avoid confusion and unintended breakages, not a good idea to change something that may have unintened consequences and isnt exactly broke but more of a personal preference..
<Lynx->
Is use of 'flock' in bash frowned upon in favour of lock-free solutions?
<jayk>
i thought flock() could do both, maybe not in bash tho
Lynx- has quit [Quit: Going offline, see ya! (www.adiirc.com)]
Ryncewynd has quit [Quit: Leaving]
<jayk>
lowest price i see is 75€
<robimarko>
Ansuel: It aint gonna get cheaper
<robimarko>
I assume its Amazon.de
<Ansuel>
yep
<Ansuel>
problem is that i will have to use drop shipping so + 20 of shipping
<Ansuel>
cause they only ship to de
<robimarko>
Yeah, I know
<robimarko>
I got mine off eBay via dropshipping while ago
<robimarko>
No idea why they refuse to ship outside of DE
<Ansuel>
ok so the way is find something on ebay
<robimarko>
It cost me 70 EUR for an used one
<robimarko>
So its not gonna be cheaper
<Ansuel>
my problem is the keyword
<slh>
there were a couple of dynalink offers on ebay until last summer, since then there haven't been any (at least not in the 2-figure EUR range); the 79 EUR amazon.de offer is the best you can find
<robimarko>
Doubt its gonna be cheaper than 69 EUR plus MailboxDE
<slh>
it wasn't any cheaper on black friday or cyber week, or singles' day either
<Ansuel>
true i have the extention the price didn't change
<Ansuel>
eh to ship 13,40
<Ansuel>
83,40
<Ansuel>
cheaper than the xiaomi one and less problem to install openwrt
<Ansuel>
also usb 3.0 + 2.5g port
<slh>
yeah, in retrospect I wish I'd have waited as well
<Ansuel>
WHY NOBODY PUT A FKING POWER BUTTON WTH
<Ansuel>
but still i think the mvp is the qnap 301w
<Ansuel>
but it's pricey
<slh>
I've been searching hard for an excuse to buy the dynalink as well, but with the xiaomi ax3600 at hand, I don't really find one...
<robimarko>
2.5G and USB?
<robimarko>
1G of RAM as well
<Ansuel>
ohhh
<Ansuel>
1g is even more important
<slh>
sure, if I didn't already own the ax3600, this would be a no brainer
<robimarko>
For the price Dynalink doesnt really have any competition
<robimarko>
But I really like my Qnap for deving
<Ansuel>
my parents are moving to a new home this march so that will probably be the house router
<Ansuel>
but i'm tempted by the qnap... but it's overkill