<vulpes2[m]>
hauke: the rockchip ATF still works fine but we are currently using external binaries compiled from upstream ATF, so I thought it'd be nice to migrate it to a source based build just like everything else. Since I've got both rk3328 and rk3399 boards, I will look into this later this week and see if I can migrate it over.
<vulpes2[m]>
a few days ago I found out the default baud is set to 115200 for rk3399 which is complete nonsense (most things on this platform use 1500000 by default), so we would need to do patch that too
<vulpes2[m]>
atf upstream isn't going to accept a change of the default baud rate because 115200 is apparently the default that some chromebooks use
dansan has quit [Quit: The C preprocessor is a pathway to many abilities some consider to be unnatural.]
<hauke>
vulpes2[m]: that explains why the "build" works fine
<vulpes2[m]>
lol
<hauke>
I think rockchip needs a different compiler for some parts of the ATF
<hauke>
at least that was the reson to use a pre build ATF on other platforms
<vulpes2[m]>
it needs one for cortex-m and one for cortex-a apparently
<vulpes2[m]>
guess I'll bump it to v2.8 first, the binary builds has been updated and I've already tested them on 3328 & 3399
dansan has joined #openwrt-devel
dansan has quit []
goliath has quit [Quit: SIGSEGV]
<f0g>
damex: thanks. Is there any document of this part?
minimal has quit [Quit: Leaving]
f0g has quit [Remote host closed the connection]
SlimeyX has joined #openwrt-devel
tSYS has quit [Quit: *squeak*]
tSYS has joined #openwrt-devel
danitool_ has quit [Remote host closed the connection]
Mangix has quit [Read error: Connection reset by peer]
SlimeyX has quit [Read error: Connection reset by peer]
SlimeyX has joined #openwrt-devel
<oliv3r[m]>
robimarko: found a few interesting bits; my (working) bootlog ommits the first few seconds, e.g. where we normally see gcc version and stuff. So none of the early stuff. Gonna dig into that now. Got zboot debug somewhat working, in that it prints some garbage chars (might be related of course :p) so first gonna turn garbage into characters :)
hhasert has joined #openwrt-devel
robimarko has joined #openwrt-devel
<robimarko>
oliv3r[m]: good to know there is progress
noltari has quit [Quit: Bye ~ Happy Hacking!]
noltari has joined #openwrt-devel
danitool has joined #openwrt-devel
Borromini has quit [Quit: Lost terminal]
danitool_ has joined #openwrt-devel
danitool has quit [Ping timeout: 480 seconds]
cbeznea has joined #openwrt-devel
<mrkiko>
aiyion: any progress with the Brume2?
<aiyion>
Currently at work, my progress is a burning pile of not properly apllying patches, of which I'm not sure, whether they are remotley complete ;)
<aiyion>
Did not compile yesterday, but building the toolchain alone costs me about two hours on my laptop. Maybe I'll get to that after work.
<oliv3r[m]>
Is there a clean way, to take a string from Kconfig (e.g. double quotes `"unsigned int"` and then drop those quotes so i can use it as `CONFIG_TYPE var;`
Borromini has joined #openwrt-devel
<mrkiko>
nbd: thanks!!
Borromini has quit [Quit: Lost terminal]
Borromini has joined #openwrt-devel
borek has joined #openwrt-devel
borek has quit []
borek has joined #openwrt-devel
borek has quit []
borek has joined #openwrt-devel
wvdakker has quit [Ping timeout: 480 seconds]
robimarko has quit [Remote host closed the connection]
<Backhand4630>
This situation is really unfortunate. And I guess the new bootloader of Mikrotik RouterOS v7 isn't changing that either. Too bad. Thank you for answering to my issue so fast.
<robimarko>
Pretty much as long as they are using YAFFS its not gonna be solved
<Backhand4630>
One last question: Would this issue also be present in a 19.07 build? (Because I am running a LEDE build since 2016 on it, and worry to be affected from this.)
<robimarko>
Yes
<robimarko>
As long as the YAFFS is just statically generated without any knowledge of the underlying NAND-s bad blocks it can affect anybody
<Backhand4630>
yes, but was this also the case 2019?
<robimarko>
Yes
<Backhand4630>
that OpenWRT relied on YAFFS?
<Backhand4630>
ok
<Backhand4630>
thanks
<robimarko>
Its not OpenWrt that relies on YAFFS, but Mikrotik
<robimarko>
As RouterBoot will not boot anything else
<robimarko>
We would really like to get rid of YAFFS completely
<robimarko>
That is why kernel2minor tool look great
zer0def has joined #openwrt-devel
<Backhand4630>
My knowledge does not go deep enough to be good informed about this unfortunately.
<robimarko>
Basic issue boils down to YAFFS is not in the kernel, it newer will be
<robimarko>
And nobody wants to maintain that out of tree just for MTik devices
<Backhand4630>
But if I understand correctly: It does not matter if I run an old LEDE build, OpenWRT 19.07 or the newest code of OpenWRT. Everyone of those would have the issue, right?
<robimarko>
That is correct
<Backhand4630>
Okay thank you!
<mrkiko>
Backhand4630: because this all thing is out of openwrt control
<Borromini>
and MikroTik really only wants you to run their software on their hardware
<robimarko>
I had hope for MTik as they finally went UBI/UBIFS route on IPQ40xx and then they ressurected YAFFS2 on RB5009
<Znevna>
you wrote bad software wrong
vicencb has joined #openwrt-devel
<f00b4r0>
robimarko: in my totally unbiased opinion (not), their technical decisions are completely retarded. It also means that they have to support an out of tree port of YAFFS2 just for their own use, which sounds like an absurd waste of resources... *sigh*
<robimarko>
f00b4r0: This would be one of the cases where we fully agree
<f00b4r0>
those happen xD
<robimarko>
Cause they had to port that POS of FS to 5.6
<robimarko>
And FFS, why are the even using 5.6 kernel?
<robimarko>
They also seem to change the YAFFS2 format every now and then
<f00b4r0>
I'm tempted to think they have no idea what they're doing
<Znevna>
imagine them maintaining some butchered openvpn, dhcp and dns services
<robimarko>
Its even worse, seems like they are writing support for services themselves
<robimarko>
Hence, the nice number of CVE-s per year
<Znevna>
well, yeah.
<f00b4r0>
I think they special in building honeypots
<f00b4r0>
specialize*
<robimarko>
Well, with the number of people running outdated longterm builds which are outdated by definition anyway
<robimarko>
They are perfect for exploiting
<f00b4r0>
*nod*
<Znevna>
and with no firewall :P
<robimarko>
Znevna: They learned on that mistake and if its supposed to be a router they have a default config that drops everything
<Znevna>
on the soho devices
<robimarko>
But then you get the perfect advice on how to port forward to just disable the firewall
<Znevna>
on the entriprise targeted .. no
<Znevna>
enterprise*
<robimarko>
Cause, only experts are using those
<robimarko>
Or they wish
<Borromini>
they're still getting good press by STH :-/
<Borromini>
who seem to ignore the software side completely it seems
<robimarko>
Well, its your traditional user
<robimarko>
It does XYZ for X money
<robimarko>
You get some SW that works, so its good
<Znevna>
if you don't ask too much of the software, it's fine
<Znevna>
I have a few that I maintain
<Znevna>
but I guess they'll die with version 6.46.7 lol
<robimarko>
I am using MTik gear, cause its damn easy to configure
<robimarko>
Also, they dont really have competition for the price
<Znevna>
true
<Znevna>
you can do some nice things with the firewall
<Znevna>
and easy.
<robimarko>
Yeah, I know some basics about networking, just enough to get around
<robimarko>
So MTik is perfect
<robimarko>
But I really, really hate their behaviour towards OSS
<mrkiko>
ubnt seems also pretty hostile or am I mistaken? u-boot not printing to serial and so on ...
<robimarko>
Havent used UBNT in a while
<robimarko>
But they went kind of hostile after the UAP AC series
<mrkiko>
I can be wrong of course
<robimarko>
Its all encrypted, tools missing etc
<Backhand4630>
Regarding Mikrotik: I am correct in the assumption that if I flashed the OpenWRT and it did work. I am fine for using the device as long as not upgrading to another version?
<robimarko>
Well, not really
<robimarko>
As NAND can develop a bad block anywhere
<robimarko>
And if that happens on the kernel part you have a brick
minimal has joined #openwrt-devel
<Backhand4630>
damn. ok... wow. now i feel lucky using this device for years with LEDE
<robimarko>
It doesnt seem to happen often as its just SLC ONFI NAND, but it has happened
<aparcar[m]>
does busybox shell aka ash supports some kind of restricted shell? Like having less access to things
<mrkiko>
btw, would like to use a different host as wireguard gateway rather than my main router and have all my host using it automatically. Is pushing the router via dnsmasq / dhcp the way to go or is there another way?
<Backhand4630>
you can forward your wireguard port to the new host, that is what i am doing
<Znevna>
and set a route on the main router
<Znevna>
but then that main router will also route your internal traffic to the wg host
Acinonyx_ has quit [Quit: Peer reset by connector]
<f00b4r0>
^
Acinonyx has joined #openwrt-devel
<f00b4r0>
this seems quite counterproductive. Running wg on the main router is the more "logical" way to do this.
<Znevna>
unless you set the route on your hosts, if not tha many
<Znevna>
depends how much horsepower the main router has :P if you want to offload the encryption/decryption to some other device it's fine
<f00b4r0>
wg isn't very taxing
<Znevna>
it is if you want more than 500Mbps
<f00b4r0>
(although there was apparently some bump between 19.07 and 21.02, as I witnessed myself)
<f00b4r0>
Znevna: if you want more than 500Mbps you probably already have a pretty beefy router anyway :)
<Znevna>
not my main one
<Znevna>
:P
<Znevna>
mt7621 is main, wireguard sits on rb5009
<Borromini>
:)
* f00b4r0
runs wg on as low as qca9531 ;)
<Znevna>
i don't wanna grow old watchign file transfers
<f00b4r0>
i have a mt7621 router running wg on a gigabit link, I should check how much it pulls over the wg link
<f00b4r0>
of course the difficulty is finding a fast enough peer
<f00b4r0>
(at least for me)
<Backhand4630>
what is the best (=cheap and fast) way to build openWRT. my linuxVM on my old 2015 macbook takes forever and since i changed profession i lack computing power in my home :(
<robimarko>
Quad core Ampere VM from Oracle
<robimarko>
Its free for now
<robimarko>
I am using that for Github Actions as GH VM-s are so slow
<Backhand4630>
cool, i will give it a shot
<f00b4r0>
patience is cheap, too xD
<f00b4r0>
Backhand4630: use ImageBuilder if you can. That's fast.
<Backhand4630>
f00b4r0: time also costs :') ... ImageBuilder is just a dockercontainer with the buildtools pre-setup. why would building with it be faster?
<f00b4r0>
no it's not
<f00b4r0>
building an image with ImageBuilder takes mere seconds.
<Borromini>
Backhand4630: it's all precompiled already with the ImageBuilder.
<Borromini>
what you're linking to is a docker container containing the imagebuilder, as the description states
<Borromini>
the imagebuilder can be grabbed from the OpenWrt download pages.
<Borromini>
i'm not sure if you can build images for disabled devices with it though, if you'd like to build for your MikroTik
SlimeyX has quit [Read error: Connection reset by peer]
<f00b4r0>
ah right, that you can't do.
<Backhand4630>
okay. but thank you for mentioning :)
wvdakker has joined #openwrt-devel
robje has left #openwrt-devel [👋 #openwrt-devel]
floof58 is now known as Guest145
floof58 has joined #openwrt-devel
<oliv3r[m]>
so, I can't get zboot to output anything, garbage or not. Further more, using earlycon via CMDLINE will actually break the serial console and print out gibberish. What is more weird, is that first I see the linux version stuff being rpinted, printk: bootsconole enabled etc (due to early_printk_seutp) but thena bit later, earlycon gets parsed and applied, and from thereon, a bit of diareah followed by nothingness
Guest145 has quit [Ping timeout: 480 seconds]
<oliv3r[m]>
does the openwrt buildsystem do anything to ignore the compressed kernel? E.g. while Linux is built with compression enabled, openwrt will actually use the uncompressed kernel anyway? I see signs of this as u-boot states (as per target.mk) LZMA compressed, whereas the kernel is configured to be XZ compressed. Or are we double-compressing things for extra protection :)
<oliv3r[m]>
true, but afaik in kernel/uboot land they are treated differently?
<oliv3r[m]>
atleast in terms of naming :)
<Borromini>
no idea, sorry.
<oliv3r[m]>
only one way to find out :)
<oliv3r[m]>
ah, but then I'd have to find out what that one way is :D i guess disabling stuff in target.mk
<oliv3r[m]>
though right now, you can't even set 'kernel_uncompressed=y' without some hackery; doesn't mean openwrt can't use the uncompressed kernel of course
SlimeyX_ has joined #openwrt-devel
<vicencb>
Hi, i am trying to port the OpenWRT linux kernel on an AR9341-based switch. It has no wireless nor USB, just 5 ethernet ports. Currently i have linux booted with an initrd and busybox. The serial port and the gpio LEDs are working. The issue is with the network, i can't get link.
SlimeyX has quit [Ping timeout: 480 seconds]
SlimeyX_ is now known as SlimeyX
<vicencb>
dmesg reports these messages related to eth0:
<vicencb>
[ 2.609833] eth0: link up (1000Mbps/Full duplex)
<vicencb>
[ 1.372085] ag71xx 1a000000.eth: connected to PHY at fixed-0:00 [uid=00000000, driver=Generic PHY]
SlimeyX has quit [Read error: Connection reset by peer]
Backhand4630_ has joined #openwrt-devel
<damex>
robimarko: i really like what mikrotik do to some of their hardware leaving room for improvement and not butchering their devices. as in - cases are universal for some models with predrilled holes for antennas, boards with ms156 to manually install antenna when needed. crowded designs that is made for use and abuse compared to other vendors, decent description of internal topology and specs and etc. most important is that people are not screwe
<damex>
on tiny devices with antennas and that formfactor kept the same around 4x4 (NUC-LIKE)
Backhand4630 has quit [Ping timeout: 480 seconds]
<damex>
i am using hap ac2, hex-s and others
<damex>
would be nice to be able to get hap ax2 :)
<f00b4r0>
damex: ipq-6010. pretty much not happening I guess
<damex>
f00b4r0: yeah, they could choose something like mt7981/mt7986 or something from that filogic lineup but nope ;)
madwoota has quit [Ping timeout: 480 seconds]
madwoota has joined #openwrt-devel
<damex>
so that's why i am getting that (supported by openwrt) banana pi r3 with MT7986 to test. hope it goes really well for homelab use. size is bigger than expected 100.5mm x 148mm while i would prefer to not exceed 4'x4' or something like 111mm x 115mm
<Borromini>
do they sell cases as well?
<Borromini>
their specs say '5 GbE network port' but it looks like they mean 5x 1 GbE network port' >_>
<damex>
Borromini: yeah, they sell cases/antennas/powersupply/tiny heatsinks too. current case they sell is made of metal (not sure which one, maybe just painted steel)
<robimarko>
Though it looks like they dont ship all over EU like for most products
SamantazFox has joined #openwrt-devel
<danitool_>
This item cannot be shipped to your selected delivery location.
SlimeyX has quit [Read error: Connection reset by peer]
valku has joined #openwrt-devel
<aiyion>
Good evening everyone, say I got access to an m300, do I need to pay special attention while building the image?
SlimeyX has joined #openwrt-devel
<aiyion>
Mhm. That did not work.
vicencb has quit [Quit: Leaving.]
<aiyion>
a bit optimistic to set SECCOMP I suppose. This looks better.
<aiyion>
I assume, building will take a while. There are two sets of instructions, how to flash/upgrade the device.
<aiyion>
Does one need to follow a combination of both to freshly flash the device, not that the firmware has FIT uImage support?
<aiyion>
*now that
<aiyion>
slh or stintel maybe?
<slh>
aiyion: I don't have a watchguard M300, I strongly considered buying one (still do), but didn't - went the easy way out --> x86_64
<aiyion>
darn, thanks
<slh>
it's cheap, it's fast - but for me personally, it's a bit too exotic for now and a tad too noisy and too much of an electricity hog (my Atom j1900 four-port fw-7543b needs 11 watts, an M300 needs 21-22 watts at least)
<aiyion>
Yeah, I am just building a Gluon Image for it. That thing might just be the powerful device to run the local refugee shelter.
<slh>
yeah, it's a very interesting device nevertheless
<aiyion>
f00b4r0: you don't own the device either, do you?
<aiyion>
Saw your name on the wikipage
<robimarko>
He does have the M300 AFAIk
<aiyion>
Maybe he knows how to flash the device properly.
<slh>
aiyion: I remember that there has been a change in the procedure, towards FIT images, should be described in the git history
<aiyion>
it is. My knowledge is just not profound enough to understand, whether part of the "new" instructions replaces part of the old, whether the "new" isntructions are just relevant for a sysupgrade from a non dts image of if the new just follow the old, before I saveenv and reset.
<aiyion>
Maybe we can improve on the instructions for a combined workflow, including a usb stick, so that one does not need to open the device at all for installation?
<aiyion>
f00b4r0: does the second instructionset even need saveenv and reset?
bluew has joined #openwrt-devel
<f00b4r0>
aiyion: yes it does it's how uboot works
<aiyion>
f00b4r0: you're back just in time ^^
<f00b4r0>
i didn't try USB stick install so I can't say about this, I find removing 4 screws easy enough. Plus I'd recommend replacing the SD card with a fresh one anyway: there's no telling what kind of abuse the original went through.
<aiyion>
compilation was successful, I just dd'ed the factory imag onto a USB-Stick
<aiyion>
true.
<aiyion>
Gotta test this at least once though, gluon supports only hardware that you can flash without screwdrivers and soldering irons -.-'
<aiyion>
But yeah I will definitely replace it, thanks.
<f00b4r0>
as much as I understand the soldering iron argument, I don't understand the screwdriver one.
<aiyion>
I'd say I start with the Instructions for USB which lead to the system being on the sd card, but the system not booting.
<f00b4r0>
heck, you're never going to install to an x86 box if you can't install RAM in there :P
<aiyion>
f00b4r0: I gave up discussing policies a long time ago.
<aiyion>
lol :D
<aiyion>
*which will lead
<f00b4r0>
then again i have no idea what gluon is tbh
<aiyion>
afterwords I go on with the uboot modification like noted in the second commit.
<aiyion>
gluon is a project that helps building a maintainable Freifunk firmware based on OpenWrt.
<aiyion>
Alright.
<f00b4r0>
sounds like too many levels of indirection for me ;)
xes_ has joined #openwrt-devel
<aiyion>
f00b4r0: Any idea how to enter u-boot in the first place, with an intact system?
<f00b4r0>
no. I don't remember if there's an input timeout configured, I suspect there isn't
<f00b4r0>
which is why you need to let it fail first
<f00b4r0>
stintel would know more
<dwfreed>
I was just about to boot my device, I can find out
<dwfreed>
my usb serial adapter arrived today
<f00b4r0>
perfect timing :)
<aiyion>
Sounds good.
<aiyion>
dwrfreed: Do you have an image?
<dwfreed>
not yet, but I can build one; right now my m300 is completely stock
xes has quit [Ping timeout: 480 seconds]
<f00b4r0>
anyway, nice to see we've got one more user to the party. Soon enough we'll need both hands to count ourselves xD