<f00b4r0>
well, removing a tag is rewriting git history, which is generally frowned upon. But since it wasn't formally released/announced, there is nothing to "remove"
<f00b4r0>
as stintel just pointed out before I could finish my sentence :D
<f00b4r0>
oh, indeed. These should probably be removed
mzvd has quit [Read error: Connection reset by peer]
<Ansuel>
mhhh what happent to rc4 ?
mzvd has joined #openwrt-devel
<stintel>
firebox m300 for eur89 on ebay, germany
<Habbie>
stintel, what's interesting about the m300?
<stintel>
4c/8t 1.5GHz 4GB ECC 8GbE SD storage
<stintel>
iirc
<Habbie>
yes :)
<Habbie>
oh 4GB is very decent, yes
<stintel>
1U RM and not fanless - might not be for everyone :)
<Habbie>
oh, not fanless is a pity
<rsalvaterra>
Also, the idle power consumption due to CPU bugginess…
<stintel>
but afaict it's faster than apu2
<Habbie>
my plans for the apu2 are somewhat overstretching the apu2, but i really don't want fans :)
<Habbie>
but i agree, for 89 EUR that's a very nice box
<nlowe>
The APU2 works well at gigabit assuming you don't need SQM - but at gigabit speeds, you probably then don't need it
<stintel>
that's a myth
<stintel>
most connections will benefit from sqm regardless of bandwidth
<stintel>
we've tested different connection types with different bandwidths in different continents and enabling sqm always improves latency, by quite a big factor even
<stintel>
so please stop spreading that misinformation
<rsalvaterra>
True. If you have the raw CPU power needed, SQM is *always* beneficial.
mzvd has quit [Remote host closed the connection]
mzvd has joined #openwrt-devel
<stintel>
sqm improves latency on a LAN with 10GbE connection even
<rsalvaterra>
Unfortunately, none of my devices are able to cope with my wire speed, at the moment.
<stintel>
m300 does >900Mbps with qosify enabled on WAN
<stintel>
good enough for me for now :)
<rsalvaterra>
Hm… I wonder if the RT3200 can do 500 Mb/s with qosify…
<rsalvaterra>
… it's the only one I haven't tested.
<rsalvaterra>
The Omnia craps out at about 250 Mb/s.
<stintel>
you know what to do ;)
<rsalvaterra>
stintel: Need to reconfigure it from scratch, I soft-bricked it with my 5.15.42 bump testing… been to lazy to do it, for the last week. :P
<stintel>
sysupgrade -b / -r ? :)
<rsalvaterra>
#yolo :P
<rsalvaterra>
"Aww, it's fine, pretty sure it can't break." <- famous last words
<stintel>
:P
<stintel>
does anyone know how to make iw scan show vendor specific IEs ?
mzvd has quit [Read error: Connection reset by peer]
<stintel>
ah -b
<stintel>
or not
mzvd has joined #openwrt-devel
<f00b4r0>
hmm that m300 is tempting
<f00b4r0>
rsalvaterra: is the idle power draw so bad?
schwicht has quit [Read error: Connection reset by peer]
frwol has joined #openwrt-devel
user95 has joined #openwrt-devel
robimarko_ has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
<rsalvaterra>
f00b4r0: It's much worse than it should be, but I don't have concrete numbers.
<f00b4r0>
rsalvaterra: ok, thanks still
<f00b4r0>
3 native interfaces and 2 links to the switch, on paper that looks very nice still
c0sm1cSlug has joined #openwrt-devel
mzvd has joined #openwrt-devel
robimarko has quit [Ping timeout: 480 seconds]
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
<f00b4r0>
oh well, I bought it :)
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
<Ansuel>
f00b4r0 the worl is not happy for your idle consumption
<f00b4r0>
well on the upside: if it can be fixed I'll look into it :)
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
<stintel>
f00b4r0: it does cake at near 1Gbps for me yes
mzvd has quit [Read error: Connection reset by peer]
<stintel>
f00b4r0: but CPU frequency scaling is hard disabled due to erratum
<stintel>
irclogs/2021/12/OFTC/#openwrt-devel.log:21|21:54:39< stintel> slh: the M300 with original PSU settled down to 23.2W idle - this is without cpufreq as there's an erratum disabling it on the e6500
<Ansuel>
flawed cpuidle no cpufreq WTH
mzvd has joined #openwrt-devel
<f00b4r0>
stintel: I see. 23W is not abominable but clearly not great.
<Ansuel>
mhhh W under load tho ?
<stintel>
well someone responded ipq807x does similar figures idle
<f00b4r0>
fair point
<f00b4r0>
i have an espressobin headed my way, I'm curious how that'll do, we'll see
<stintel>
dunno under load, can't measure either, not at home and don't have per-outlet measurement in my rack
<f00b4r0>
well under load shouldn't be very much higher if cpufreq doesn't work
<f00b4r0>
the PSU is rated 100W too.
<stintel>
I replaced the PSUs in mine and that lowered the idle draw by a few W iirc
<f00b4r0>
stintel: are there things needing love still or did you complete the support already :)
<stintel>
DSA setup .. only single CPU port is enabled to the switch
<stintel>
couldn't figure that out - did get GPL code since but haven't had time to look at it yet
<f00b4r0>
i see. well now I have some incentive to help :)
<stintel>
great :)
<Ansuel>
stintel btw there is a multicpu patch in rfc stage to finally add correct multicpu support instead of my bad hack
<Ansuel>
but they are lots of patch to backport
<stintel>
I tried getting the other CPU port to work standalone also and never managed to get it working
<Ansuel>
the idea is to create a LAG across the cpu port and use that
<Ansuel>
99% the switch should support it
<stintel>
then again I'm a noob at all this so I might have done something wrong
<stintel>
Ansuel: cool, thanks for the info, didn't know that
<stintel>
f00b4r0: ah you bought it? :P
<f00b4r0>
stintel: i did :}
<stintel>
cool :)
<f00b4r0>
was too cheap to pass
<stintel>
I do get sometimes this still: [3523012.539647] fsl_dpaa_mac ffe4e0000.ethernet eth1: Err FD status = 0x00000020
<stintel>
there's something about that in either of the patch series I sent to the ML
<stintel>
wanted to port some userspace utility that controls the FMan (iirc) but that requires the SDK driver - not compatible with upstream FMan driver
<stintel>
might just be some weird traffic coming from my ISP's Huawei ONT
borek1 has joined #openwrt-devel
borek has quit [Read error: Connection reset by peer]
borek1 is now known as borek
<mrnuke>
Ansuel: I had the misfortune of looking over the qca-ssdk yesterday. I wasn't trying to hurt myself, Was just trcking down a bug in the phy driver.
borek has quit [Ping timeout: 480 seconds]
ekathva has quit [Quit: Leaving]
valku has joined #openwrt-devel
<robimarko_>
mrnuke: Dont tell me that I did not warn you about the level of code
<robimarko_>
SSDK is just a nightmare
<Ansuel>
mknuke: Check the fal code :D Free nightmare once you notice the amount of abstraction with the define
borek has joined #openwrt-devel
<stintel>
just looking at serial console of a device running qsdk-based firmware is enough to make me nauseous
<stintel>
not sure I even want to look at their code :P
<Ansuel>
they are very chatty
<robimarko_>
stintel: You for sure dont want to look at SSDK
<robimarko_>
Newer stuff like NSS-DP is decent
<robimarko_>
But the older stuff is just insane
<Ansuel>
btw robimarko_ i should waste some time on ath11k and reworking some logic to make it more conservative... like there are tons of thing that are allocated doing nothing
<Ansuel>
but my fear is making things worse...
<robimarko_>
Dont poke the bear
<robimarko_>
Something will break for sure
robimarko_ has quit [Quit: Leaving]
minimal has joined #openwrt-devel
<mrnuke>
Ansuel: no. five levels of indirection to get to your ops function, which then dereferences a pointer hidden behind another five levels of indirection
<mrnuke>
Only reason I'm looking there again is to delete code. BTW, I found an upstream PHY driver in 5.18 :D
<mrnuke>
No bug there
srslypascal has quit [Remote host closed the connection]
srslypascal has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
benjamin_ has joined #openwrt-devel
<Ansuel>
mrnuke can you link that?
Gaspare has joined #openwrt-devel
ephemer0l has quit [Ping timeout: 480 seconds]
benjamin__ has quit [Ping timeout: 480 seconds]
benjamin_ has quit [Remote host closed the connection]
<mrnuke>
I know what you're going to say. That's not how we do backports in openwrt. It was late at night, and I was trying to go to sleep. I ain't gonna marry that commit :D
mzvd has quit [Read error: Connection reset by peer]
borek has quit [Remote host closed the connection]
mzvd has quit [Read error: Connection reset by peer]
mzvd has joined #openwrt-devel
<nlowe>
When are the default builds for all supported targets going to change to the 5.15 kernel instead of 5.10?
<nlowe>
Seems odd to me, for example, that X86-64 is being built with a 5.10 kernel default at the moment
<nlowe>
In the master branch
<stintel>
this is per target
<stintel>
once enough testing has been done for target X, someone can decide to switch it to kernel X by default
<stintel>
it doesn't make sense to do this for all targets at the same time, it's not even possible because not all targets have support for 5.15 already
frwol_ has joined #openwrt-devel
mzvd has quit [Ping timeout: 480 seconds]
<stintel>
if you feel a target should change to 5.15 by default (e.g. you have tested it for a while, feel free to send a patch that does so)
frwol has quit [Ping timeout: 480 seconds]
mzvd has joined #openwrt-devel
<stintel>
s/,/)/;s/)$//
<Ansuel>
is that a regex ?
<stintel>
yes, to fix some errors in my message before it :P
<stintel>
it might need some \ but you should get the idea
<Ansuel>
anyway i have some patch from upstream to backport... they are related to devfreq and they are needed for ipq806x... should i place them in generic or in ipq807x ?
<Ansuel>
ipq806x*
f5- has joined #openwrt-devel
f5- has quit [Remote host closed the connection]
f5- has joined #openwrt-devel
ephemer0l has joined #openwrt-devel
f5 has quit [Ping timeout: 480 seconds]
frwol_ has quit [Quit: leaving]
guidosarducci_ has joined #openwrt-devel
guidosarducci has quit []
<mrnuke>
Ansuel: I just ran unifdef on qca-ssdk yesterday. 7000 lines of code removed.
shibboleth has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
srslypascal has quit [Ping timeout: 480 seconds]
srslypascal has joined #openwrt-devel
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
schwicht_ has quit [Read error: Connection reset by peer]
schwicht has joined #openwrt-devel
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
nlowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Ansuel>
well nice... just discovered dwmac driver apply wrong configuration values for their internal regs
<russell-->
i'm trying to restore an nvram partition on an old netgear wgt634u, which requires writing flash from linux (cfe is braindamaged), how can i make the right bits writable?
nlowe has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<PaulFertser>
russell--: kmod-mtd-rw provides a module which can be loaded with insmod mtd-rw.ko i_want_a_brick=1 and it makes all partitions writable.
shibboleth has quit [Quit: shibboleth]
<f00b4r0>
nice.
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
<Ansuel>
i_want_a_brick LOVE IT
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
<russell-->
PaulFertser: fixed, thank you!
<Ansuel>
Mangix: ping
nlowe has joined #openwrt-devel
mzvd has joined #openwrt-devel
mzvd has quit [Read error: Connection reset by peer]
<Ansuel>
still don't know if it's new gcc12 warn or false-positive
<Ansuel>
or gcc12 detection was flawed and all this thing only showed now
<mangix>
yep
<mangix>
I just saw that error in uqmi
<mangix>
error: pointer 'ptr' may be used after 'realloc' <-- wtf is this?
<Ansuel>
well i think i will have to tag jow since this looks to be a bug with the compilation process
<Ansuel>
if a Makefile have some package non-shared and some shared, stage2 will rebuild them and any package dependent on that will be based on the new one built by stage2 (and i described it in the worst way...)
<mangix>
a lot of this stuff should probably be converted to C++ to avoid all this gcc breakage
nlowe has quit [Read error: Connection reset by peer]