<stintel>
yeah the downside of putting instructions in commit message ... :)
<f00b4r0>
:)
* russell--
pops open the box to recover from the sdcard
<stintel>
no need
<f00b4r0>
the sdcard isn't the problem
<stintel>
run the u-boot commands in https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=c4b499bc03ab0e2eea643c46d1d781ab64e78931 + saveenv + reset
<russell-->
how do i get to a u-boot prompt? i'm looking at the u-boot menu.
<stintel>
+ if (memcmp(wg_pw_hash, wg_csum_output, 20) == 0) {
<dwfreed>
serial's probably hiding some characters
<f00b4r0>
sha1? that's breakable is it not? ;)
<stintel>
may be brute force-ble
<dwfreed>
stintel: so raw sha1? ooh
<stintel>
hmm, maybe not
<stintel>
20 characters ..
<dwfreed>
is that what wg_pw_len is set to?
<f00b4r0>
sounds like an awful lot
<stintel>
wg_pw_len = cli_readline("password> ");
<stintel>
:P
<dwfreed>
the 20 is the length of the hash
<f00b4r0>
^
<dwfreed>
not the password length
<dwfreed>
sha1 is 20 bytes (160 bits)
<f00b4r0>
i'm willing to bet it's 8 chars
<Namidairo>
or some variation of vendor name
<russell-->
i'm in
<russell-->
thanks!
<oliv3r[m]>
Hey guys, I'm a bit puzzled. So I have a SwitchChip, that has no embedded PHY, everything is remote via SerDes. I know I can query the PHY over SMI to get link status/speed information. However, the SoC also has a MAC and PHY prefix register space, that also gives link-status/speed information. I bet 1 of those 3 is the 'remote advertisement' (which is also something that SMI can tell us, autoneg). So how should I intepret all of this?
<stintel>
so the hex has should be e597301a1d89ff3f6d318dbf4dba0a5abc5ecbea if I'm not mistaken
<dwfreed>
yes
<robimarko>
oliv3r[m]: Is that RTL?
<oliv3r[m]>
robimarko: it is :
<robimarko>
Cause Marvell does something similar in their switches where you can have external PHY-s
<robimarko>
They have an internal MDIO bus on the switch that you connect the PHY to and then you tell the switch the mapping
<robimarko>
And its using C22 to read the link status
<stintel>
doesn't look like it's available in one of the online databases
<mrkiko>
neggles: thanks for the explanation
<oliv3r[m]>
robimarko: so you are saying they are constantly sideband reading this information over the serdes, and just locally reflect things?
<robimarko>
oliv3r[m]: That is the only thing I have seen so far
<oliv3r[m]>
ok but then still, why can both a MAC and a PHY have a 'speed status'? I thought this was only for the PHY. Unless the MAC just represents whatever parameters are on the other side of the link
<robimarko>
Why not?
<robimarko>
There is a link between the MAC and PHY
<dwfreed>
stintel: yeah, I checked a few different ones already, got nothing
<dwfreed>
and searching the hash itself turns up nothing
<oliv3r[m]>
but isn't that link some fast MII fixed speed?
<oliv3r[m]>
I know for example that 10 vs 100 vs 1000 uses the same 125Mhz, but just fills the data bits with 0's
<oliv3r[m]>
so then the MAC speed is not relevant/accurate?
<oliv3r[m]>
also, what's the point of having a MAC at 10Mbit and a PHY at 1000Mbit? you'd get no functional link
<robimarko>
Yes, but there are always MAC status bits
<robimarko>
How would you know the MAC status otherwise
<oliv3r[m]>
I understand there's status related to a MAC
<oliv3r[m]>
but speed, duplex, link state; these are all status bits of the PHY; the MAC is a level higher
<stintel>
generating some rainbow tables for sha1 too atm :P
<robimarko>
Ok, I am stupid
<robimarko>
Been figthing with NVMEM cell to read out a fuse for 2 hours
<robimarko>
I just couldnt get the expected value
<robimarko>
And then I realized that 0x249 + 0x1 aint 0x250
<robimarko>
Well, that was a nice start of the day
<dhewg>
ENEEDSMORECOFFEE
<stintel>
:D
<dhewg>
stintel: maybe try a dict attack with: for i in $(cat /usr/share/dict/words | egrep -i '^[abcdefszogli]{8}$' | tr '[:upper:]' '[:lower:]' | tr 'szogli' '520911'); do echo 0x${i}; done
<robimarko>
Yey, on 0x24a it works
Ansuel has joined #openwrt-devel
<Ansuel>
any idea if ddimension and André Valentin are the same person ?
<robimarko>
Yes
<Ansuel>
i was merging the nbg7815 pr and i just notice that the name is not connected to the user
<robimarko>
That should be the same eprson
<Ansuel>
ok then no problem
<robimarko>
The branch he is making a PR is from ddimension/openwrt so I guess its him
dangole_ has joined #openwrt-devel
gch9812138 has joined #openwrt-devel
bluew has joined #openwrt-devel
gch981213 has quit [Ping timeout: 480 seconds]
gch9812138 is now known as gch981213
Lynx- has quit [Ping timeout: 480 seconds]
<Znevna>
can we has mt76 bump
goliath has joined #openwrt-devel
minimal has joined #openwrt-devel
<robimarko>
Have you made a PR?
<Znevna>
nbd usualy bumps the version with a short changelog :P
<f00b4r0>
Znevna: the user you've been replying to on the forum seems to believe that ipv6 magically stops at the router; and seems to assume ipv6 "can work (for the internet?)" on an ipv4-only lan.
<f00b4r0>
in other words, they don't understand anything.
<Ansuel>
lol?
<Znevna>
I thought I just misunderstood what he's saying
<f00b4r0>
no you didn't.
gch981213 has quit [Quit: Ping timeout (120 seconds)]
<Ansuel>
well first time i will use attended sysupgrade... curious how will work
gch981213 has joined #openwrt-devel
<Znevna>
"ipv6 is enough for wan" i was like, ok, I won't ask.
<Ansuel>
(in short don't download host tool if already build and installed)
<mrkiko>
Ansuel: Znevna: minimal: here (Italy) Internet works well - even tough I'm stuck in FTTC. Using a FB7530 with pppoe bridge (from IPS) - I tried to connect from openwrt with vr11 VDSL and it works, but syncs at lower rates unfortunately
<Ansuel>
how much lower? anyway i think you will be covered to ftth in the next year
<Ansuel>
if you are lucky to be in a white zone
<mrkiko>
Ansuel: with FB7530 running as bridge I get 17.2 MB/s, when using it with openwrt 15.5 or 16 or something
<f00b4r0>
anyone needs italian-compatible SFP ONTs? I bought too many ;D
<Znevna>
coax doesn't fit in them :P
<Znevna>
tho aren't those pretty much programmable?
<Znevna>
*not just for italian ISPs)
<f00b4r0>
they are. But unfortunately for me, it seems French PONs are a tad more secure than their italian counterparts
<mrkiko>
I would like to try a VDSL SFP module but I guess I am going to have more troubles than benefits in using that
<mrkiko>
and well yeah - I let go as I hope to be out of the dsl world soon
<Znevna>
I'll stay with my current tiny bridge only thingy they gave me
<Znevna>
makes support calls way easier :P
<f00b4r0>
DSL is way more reliable than FTTH here tho
<Znevna>
ouch
<f00b4r0>
well apparently some genius thought it would be a good idea to allow "competition" for the last mile connection. Guess what, it didn't end well.
Ansuel has quit [Read error: Connection reset by peer]
<mrkiko>
Znevna: is it really bridge only? I know mine is not :D
<robimarko>
f00b4r0: its gotta be wpad
<robimarko>
Ansuel: The cpu variant is the default on target
Tapper has joined #openwrt-devel
<f00b4r0>
robimarko: makes sense.
<robimarko>
Well, no idea why conflicts arent honored
Ansuel has joined #openwrt-devel
Ansuel_ has joined #openwrt-devel
Ansuel has quit [Read error: Connection reset by peer]
<robimarko>
ESS DTSI is included and I just compiled with the device included
<Znevna>
mrkiko: yes, I don't have better pictures, sorry :P and mind the cables. those are two models, with catv out and without https://s.go.ro/055mz2xn
Mathew has joined #openwrt-devel
mcbridematt has quit [Remote host closed the connection]
<Znevna>
the one with catv is a little lighthouse too >.>
<robimarko>
Ansuel: I can confirm that OPKG patchseries resolves the issue
<robimarko>
So, I replied on the mailing list to bring some attention to it
<f00b4r0>
👍
<Ansuel>
we will have to snipe jow to receive the correct feedback
dangole_ has quit [Remote host closed the connection]
<robimarko>
Yeah, as it seems the same issue is hitting libustream-wolfssl as well
Lynx- has joined #openwrt-devel
<mrkiko>
Znevna: thanks. I can't see the photo (being blind) but apreciate the fact you searched for it
<Znevna>
oh :*
<Znevna>
:( * sorry
<Znevna>
small devices made by some company called Luster
<Ansuel>
this thing almost worked but it does crash
<Ansuel>
as I should run it in a real linux installation and not wsl2...
<Ansuel>
actually now that i think about it... i should really try on a live disk
<dhewg>
doesn't github ci offer macos builders?
<robimarko>
Its slow as hell
<robimarko>
I tried using that to debug the CI build, its like 2+ hours per build
<Ansuel>
BIG BRAIN TIME
<Ansuel>
i will just packet mkimage depens tool
<Ansuel>
send them as artifacts
<Ansuel>
make the ci downloaded them and preinstalled tools
<Ansuel>
MH
<dhewg>
It's been a few years, but there was a rather simple kvm patch to run macos under kvm, but that obvious won't work for wsl
<Ansuel>
dhewg it works but it crash a lot and it's not stable at all...
<robimarko>
Ansuel: I think Scaleway finally has per 24 hour M1-s in stock
<robimarko>
And nope, they had them earlier today
<dhewg>
mkimage as in uboot fails to build on macos?
<Ansuel>
yes
<dhewg>
that's a rather small tool
<Ansuel>
FK THIS FK ALL i let me abuse github ci cache system
<Ansuel>
wanted to have a full env but not worth
<Ansuel>
openwrt-22.03-abuse-ci lovely branch name
Tapper has quit [Quit: Tapper]
<slh>
dhewg: current qemu/ kvm doesn't seem to need patching (at least Debian's packages work), last tried just under a year ago (really wasn't my kind of OS, and the emulation was rather fragile as well)
<dhewg>
ok, nice. I only tried it to check if builds succeed there too. No real interest in the OS
<dhewg>
some years ago there were even a metric ton of binutils/gcc patches so you could build a cross compiler
<Ansuel>
honestly we should really invest in container and say to user to use that
<slh>
for years I've been rather curious about osx, one day with it under kvm healed me from that ;)
<slh>
I was probably just holding it wrong
<dhewg>
heh, but now with m1/2 the picture changes again. and asahi fixes the os problem ;)
<slh>
sure x86_64 is shit, apart from everything else ;)
<Ansuel>
the instruction set is too bloated
<Borromini>
ok.
<Borromini>
just wanted to get perspective on your statement :)
<slh>
there just isn't anything better either. (and the M1/ M2 might be cool, but it's not really a competition either, not without a real software ecosystem and PC/ workstation like hardware)
<Borromini>
so you like RISC instead of CISC :)
<Znevna>
z80 is the shiznit
<Ansuel>
m1/m2 is full of compromise but it was a big example of a real arm cpu on laptop and how much is better suited for mobile use
<slh>
sparc was cool... up to the mid 90s
<Ansuel>
it's curious that riscv is getting some traction for google
minimal has quit [Quit: Leaving]
<dhewg>
if by compromise you mean the soldered/non-exchangable parts, while that is less flexible, it's how it archives its performance. which is seriously impressive for an arm soc
<Ansuel>
the bus is very limited
<Ansuel>
and m2 didn't improved it
<Ansuel>
i remember some test connecting a 4k external monitor + 10g port result in the 10g port being limited in speed
<slh>
dhewg: I can't really take a system seriously that doesn't allow replacing (and adding tons of-) block storage and RAM. maybe fine for a netbook, but not for anything I'd put under my desk (my current system has one small 250 GB SSD and 5 big HDDs - and I'm really fighting with myself to mount the old 3ware 9650SE to extend that by another 4 drives, if the (mainline-)linux drivers were just a tiny bit
<slh>
better)
<slh>
sure, real Apple user would use USB3 or a NAS instead, but network file transfers are just no good for the kind of file sizes (up to 400-700 GB) I'm often dealing with
<slh>
qsoldering the CPU onto the board, o.k. - the RAM, not really - the block storage, no chance in hell
<Ansuel>
same
<slh>
I've seen too many disks die - and they're chronically too small
<Ansuel>
dell started to solder the wifi card...
<dhewg>
to each his own, but don't underestimate the performance gain because the ram *is* soldered
<Ansuel>
had to buy the precision lineup for this reason as i didn't want that trash intel killer card
<Borromini>
Ansuel: in ultrabooks?
<slh>
not nice, but that wouldn't bother me too much (as long as I have the choice prior to the purchase)
<Ansuel>
Borromini yes new xps15 have wifi card soldered
<Borromini>
ouch.
<Borromini>
my 9350 came with Broadcom crap, I swapped it for an Intel AC and now Intel AX
<Borromini>
that's a major buzzkill.
<Ansuel>
yep and i hate it considering wifi card are cheap... an i'm stuck with ax201 instead of ax211
<Borromini>
oh but at least you got intel :)
<slh>
Borromini: from a performance point of view there isn't much of a difference between ax200 and ax210 on 6 GHz, it's roughly the same (but lower range). however if you're in a very crowded environment with lots of competition on the 5 GHz band, it can pay of
<Ansuel>
wait 210 supports wifi 6e tho
<slh>
yes, in practice performance is about 750-800 MBit/s for 5 or 6 GHz though, not much of a difference to be felt
<slh>
and that#s already same-room, no obstructions
<f00b4r0>
slh: I'm curious how you plan to saturate a thunderbolt 4 link with mass storage tho
<slh>
and mt7621au is actually currently slower (by half) on 6 GHz, than it is on 5 GHz (ax210 is roughly the same) - at least on linux (need to test yesterday's new firmware update for mt7621), with the windows drivers it's roughly the same there as well
<f00b4r0>
internal storage had a point up to a point ;)
<slh>
f00b4r0: I don't plan to, as long as there's SATA ;)
<Ansuel>
and the thing is just stuck....
<slh>
really, the extra costs and a second box to place next to the computer doesn't make sense to me
<f00b4r0>
was reacting to your argument about not being able to replace/add "tons of block storage"
<f00b4r0>
well, we're talking about a computer that's as big as few CDs stacked together.
<f00b4r0>
if it soothes your mind to have everything in one large box, just get an ATX case and put everything in there ;)
<slh>
f00b4r0: afaik, apple solders the SSD onto the board (at least for their macbooks, but don't they also do that for the desktop Macs now?), when it dies -and it will- the whole system is trash; not gonna play ball with that
<f00b4r0>
slh: i used to think that way
<f00b4r0>
I have a 8yo ssd that has yet to die.
<f00b4r0>
system ssd that is
<f00b4r0>
the system it's in became obsolete long before the ssd showed any sign of wear.
<f00b4r0>
so I tend to think that argument has no merit anymore.
<slh>
f00b4r0: I just had a Verbatim Vi550 S3 256GB SSD die within 10 months of medium desktop type usage. sure, Apple will put a better one in there, but still - it happens
<Ansuel>
slh how much stuff you write wow
<f00b4r0>
key point: "apple will put a better one in there"
<f00b4r0>
Ansuel: that's what 250cpm looks like ;)
<slh>
and yes, my 10 year old 128 GB Samsung 830pro and 8 year old 128 GB Samsung 840pro are still going strong - despite lots of abuse
<slh>
but they will die nevertheless
<f00b4r0>
slh: and when they do, the system they're in will be long gone too
<f00b4r0>
so again, the argument is rather moot.
<f00b4r0>
a 10yo laptop is a dinosaur with very limited use.
<f00b4r0>
if you think otherwise I'm happy to set one of my old MBPs up for openwrt's use ;)
<slh>
foxtrot: not for me, sorry - there are still ~2005-2007/ 2008 vintage systems that want to talk to you right next to me
<slh>
grr, sorry for the wrong ping
<f00b4r0>
slh: desktops i presume.
<slh>
midi-tower, yes
<f00b4r0>
that's the point.
<f00b4r0>
we're not talking about anything like that here, and you're comparing apples to oranges I think.
<f00b4r0>
yeah, I have such a device, an HP thingy with a flip touch screen
<f00b4r0>
it's cute, I use it as a VT100 terminal.
<dhewg>
lol
<f00b4r0>
and thats only because it draws less power than a proper CRT VT100 ;P
<dhewg>
or look at it from another pov: with those expectations you'll never get off x86
* Borromini
is still happy with his 2016 XPS 13
<dhewg>
and a good 4*3.5" usb3.2 casing won't saturate the link for spinning rust
<slh>
apart from the battery, it's not quite mint condition, but still in good shape
<Ansuel>
xps13
<Ansuel>
smol
<Borromini>
?
valku has quit [Quit: valku]
<f00b4r0>
dhewg: preaching a convert.
<f00b4r0>
I'm on my way out of x86 for anything with a display nowadays
<Borromini>
Ansuel: is that slang for 'small' or what does it mean?
<f00b4r0>
Borromini: it is
<Borromini>
ok :)
<Borromini>
never heard it
<f00b4r0>
often spelled "smôl" ;)
<Borromini>
o_O
<f00b4r0>
gives it that Ikea vibe
<Borromini>
lol.
<Borromini>
still on the fence about a bigger screen size when it needs replacing. Portability (and battery life) are important even though I use it at home most of the time
<Borromini>
i'm out gents, goodnight
<f00b4r0>
well the fact that I can go an entire weekend on a single battery charge is definitely a very high selling point for my M1 MBP
<Ansuel>
Borromini yes
<Ansuel>
smol = small
<f00b4r0>
in fact I've got a personal record running an entire week over a charge.
<Ansuel>
gn
<f00b4r0>
which is insane, coming from x86
<Borromini>
i'm happy with a day ;)
<Borromini>
well i don't like macOS... And i'm not up to speed with the porting progress Avahi made
<robimarko>
f00b4r0: I am gonna give credit to Apple for the SoC
<robimarko>
But, for me its not usable, I dont want vendor lock-in
<robimarko>
And soldered SSD is a no-go for me
<robimarko>
Kudos to Asahi and Apple not preventing instaling a 3rd party SW
<robimarko>
But its always gonna be catch-up to make it usable and I aint gonna be fighting with macOS till that happens
<slh>
just out of curiosity, is there anything special to take care of to use openssl with cpu crypto (instead of wolfssl) on ipq807x? any special package (beyond the normal openssl/ libopenssl) or configuration?
<robimarko>
Nope
<slh>
thanks :)
<robimarko>
ARMASM should be enabled in OpenSSL by default
<robimarko>
As OpenSSL is smart enough to try the crypto extensions instructions and see if they throw and exception
<slh>
great, that makes my setup easier (as I'm using openssl for everything anyways and don't need/ want wolfssl in my builds)
<robimarko>
I hope to see mbedTLS finally get crypto extension support
<robimarko>
As until that happens its not really an option for me