<pbb>
Not like on recent x86 macbooks where the SSD can not be accessed from outside Mac OS
<samueldr>
well, it's possibly a similar situation and solution
<samueldr>
no one getting paid to _work_ on that, so no progress
<pbb>
what is the solution on x86?
<pbb>
using an external boot drive?
<samueldr>
no clue
<samueldr>
that's a way you can do it though
<samueldr>
[20:14:30] <samueldr> well, it's possibly a similar situation and solution
<samueldr>
I meant that whatever method to access the SSD is probably going to be similar, hopefully
<pbb>
oh, yeah
<samueldr>
so maybe asahi linux work will help
monk has left #nixos-aarch64 ["Error from remote client"]
dstzd has joined #nixos-aarch64
<pbb>
I wonder how much emulation of aarch64 on x86_64 could be accelerated by optimizations
<pbb>
like, everyone talks about the other way, but this would be much more relevant for me
dstzd has quit [Client Quit]
<samueldr>
I seem to recall a thread about that lately, that it's way harder than it seems
dstzd has joined #nixos-aarch64
<samueldr>
too much for me to understand though
<pbb>
I mean nowadays aarch64 is not a risc architecture anymore. So there might be better translations for complex x86 instructions than are currently used. But then again both probably have many quirks.
<samueldr>
(and obviously don't have the link)
dstzd has quit [Client Quit]
<samueldr>
I think at that point it might be cheaper to just put an aarch64 cpu on a PCIe card
<samueldr>
which... you can get... within network accelerator cards
<pbb>
interesting approach :D
<samueldr>
well, there's that xeon phi accelerator which is basically a big bunch of atom CPUs on one PCIe card
<pbb>
I'd really like to change get a powerful aarch64 device and run NixOS on it, but ... well, there are no powerful aarch64 devices really
<pbb>
s/change //
<samueldr>
the honeycomb
<samueldr>
but yeah, I do too
<pbb>
> Honeycomb LX2 ?
<{^_^}>
error: syntax error, unexpected ')', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):463:1
<pbb>
oops
<samueldr>
yep
<samueldr>
at the very least it's supposedly workstation-grade
<samueldr>
too expensive for my hobbyist use
<samueldr>
(unless it was guaranteed to work well as a workstation)
<samueldr>
I don't want a "but [...]" machine
<samueldr>
like the pinebook pro is a "but wifi is not great" machine
<pbb>
only available from the manufacturer i guess?
<pbb>
pinebook pro is horrible tbh
<pbb>
the keyboard and trackpad are completely unusable for my standards
<pbb>
and it's slow af
<pbb>
even for the price, it's not a good deal
<samueldr>
meh, keyboard is good enough, but the trackpad could use an upgrade
<samueldr>
I say the pinebook pro keyboard is not _objectively_ bad, it's really in the territory of opinions
<pbb>
I want something thinkpad x230 grade
<samueldr>
yeah, you have really high standards here :)
<pbb>
nah, x230 is faster and cheaper and has better keyboard
<samueldr>
I meant comparable machines released new
<pbb>
oh, yeah m)
<samueldr>
and uh... that's not globally applicable
<samueldr>
an x230 here is about 3× as expensive a pinebook pro
<pbb>
huh?
<samueldr>
it's probably cheaper to buy it from germany and get it through customs
<pbb>
where are you located? :o
<samueldr>
Canada
<pbb>
iiinteresting
<samueldr>
like, you can probably get one for a comparable price from an individual sale
<samueldr>
but refurbs are way overpriced
<pbb>
they're getting more expensive here as well especially if you want a good (not broken) case
<pbb>
I bought a few for 100€ per piece in 2019
<samueldr>
I think there's just not many floating on the refurb market, they probably all ended up in another market (like in the US) and got mingled to their inventories
<samueldr>
pbb: few 100€ each or for the lot?
<pbb>
100€ each
<pbb>
but in okay condition
<pbb>
importing the pinebook pro ended up costing about 350€
<samueldr>
eek
<samueldr>
though, the pinebook pro I see it as "investing" in the pine64 mindshare so they (and others) can try and shake things up
<samueldr>
like the pinephone
<samueldr>
terrible phone on paper
<samueldr>
but they're doing things to shake things up
<pbb>
kinda true
<samueldr>
(well, maybe not terrible, but it can be made to look worse in comparisons)
<pbb>
If I get some aarch64 linux box it'll probably be the MNT Reform or its successor (if there will be one)
<pbb>
if you have a look at layer 4 ("Ebene 4"), you'll see why I need it
<samueldr>
I personally would prefer 105-keys type layout to be universal
<pbb>
oh yes please
<samueldr>
but personally, as long as every key is where it should be, according the standard, you are on your way to a passing mark, even if what's printed on them is wrong
<samueldr>
since at least it can be fixed in software
<samueldr>
remove or add keys where they shouldn't be? that's a paddlin'
<pbb>
I don't need things printed on keys
<samueldr>
same
<samueldr>
anyway most of the time it doesn't match
<samueldr>
I use the "french canada" layout, which is not being sold anywhere anymore
<samueldr>
since the CSA layout is the official layout for french speakers
<samueldr>
and otherwise QWERTY US
<samueldr>
(which here in reality it means choosing between 104 and 105 keys layouts)
<pbb>
ouch it really sucks that they don't sell the keyboard you need anymore. fortunately I use the 105 key and the 105 key is the common variant en DE.
<samueldr>
meh, they're only prints on keys :)
<pbb>
oh okay then I got that wrong
<samueldr>
and actually I think they're sold, but it's a crapshoot if a "french" keyboard in Canada is CSA or French CAnada
<samueldr>
the "french canada" layout is probably one of the most useful "default" layout for european and north american languages, since it treats most accents as dead keys
<samueldr>
there's only é which is pre-baked since it's used abundantly, but its dead accent is available as Alt+Gr+é
<samueldr>
for geman it'd only be missing ß
<samueldr>
german*
<pbb>
neo-layout has ß in the top row :D
<samueldr>
for those... uh icelandic?... languages it'd be missing ø
<Ke>
sphalerite: does is look like you will get the EFI port done quick?
<Ke>
I kind of don't have time for my system right now, so might as wll wait
cole-h_ has joined #nixos-aarch64
cole-h has quit [Ping timeout: 246 seconds]
cole-h_ is now known as cole-h
zupo has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<LinuxHackerman>
Emil Karlson: not sure, I have a lot of stuff on my plate but this is something I'm keen on
<LinuxHackerman>
I'll almost definitely put some time into it this weekend, can't say at this point how much actual progress will come of that though 😅
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-aarch64
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
dstzd has joined #nixos-aarch64
dstzd has quit [Client Quit]
<gchristensen>
DigitalKiwi: are you still interested in that PR?
<DigitalKiwi>
gchristensen: yes
<gchristensen>
cool
<gchristensen>
I'm looking forward to it :)
<DigitalKiwi>
it's trading hours so i got distracted :(
<gchristensen>
no worries
zupo has joined #nixos-aarch64
dstzd has joined #nixos-aarch64
Darkmatter66 has joined #nixos-aarch64
alpernebbi has quit [Quit: alpernebbi]
rajivr has quit [Quit: Connection closed for inactivity]
tmountain has joined #nixos-aarch64
tdeo has joined #nixos-aarch64
<tmountain>
Hi, I've been trying to get a NanoPC-T4 (RK3399) SBC running with Nix for a few days without success, and I'm wondering if anyone could be so kind as to help me out.
<sphalerite>
tmountain: what have you tried? How has it failed?
<sphalerite>
I'm running nixos on a nanopi-m4 which is basically the same AFAIK (just with more stuff broken out into ports), so at least I can tell you that there are good chances of getting it working :D
<sphalerite>
unfortunately I don't remember exactly how I did it
<tmountain>
sphalerite: I've tried a ton of stuff. I can boot the SD from Armbian, and I have a working u-boot, which I have used to boot both Armbian and Arch linux. I have been extracting the root file system from nixos-sd-image-20.09.2594.*-aarch64-linux.img and installing it to the appropriate partition. I'm modifying extlinux.conf with the following APPEND: APPEND systemConfig=/nix/store/jq8c49ddc4nkszja1qbaiwcdnawn8gh6-nix
<tmountain>
It gets as far as "Starting kernel" and the screen goes blank.
<sphalerite>
Do you have a serial console?
<tmountain>
I've experimented with different ttyS* settings in extlinux.conf.
<tmountain>
I do not have a serial console. I'm trying with an HDMI.
<tmountain>
I even tried copying the rockchip kernel from the armbian install. Same results.
<sphalerite>
yeah that'll be tricky :/
<sphalerite>
did you grab the initramfs from nixos as well?
<tmountain>
Yes.
<tmountain>
Mounted the Linux partition from the nix img and moved it onto the root of the /dev/mmcblk2p1 partition.
<tmountain>
Mounted via loopback.
<sphalerite>
right
<tmountain>
and copied the fs with tar...
<sphalerite>
ah, you probably need to label the filesystem as NIXOS_SD
<tdeo>
hi, i've flashed mobile-nixos on my oneplus3t and it got to the part where it says switching to stage 2, but now the screen is just blank but the device seems to be on. is anything supposed to show on the screen after that? or do i have to configure it more
<tmountain>
sphalerite: have not tried that. can you advise how to do so?
<sphalerite>
tmountain: tune2fs -L NIXOS_SD /dev/mmcblk2p1 or whatever the root partition is
<tmountain>
sphalerite: thank you, I will give that a try.
<sphalerite>
tmountain: debugging this without serial is going to be very tricky though, if that doesn't help. I'd strongly recommend getting hold of one :)
<tmountain>
sphalerite: agreed 100%. was wishing I had one today.
<tmountain>
it was the damn filesystem label. I've been smashing into a wall with this for two days. thank you SO MUCH.
<tmountain>
incidentally, the whole thing boots without even modifying extlinux.conf, so it works out of the box.
<sphalerite>
yay!
<samueldr>
tdeo: that's because it booted successfully to a tty, on a system where the tty cannot be drawn
<samueldr>
tdeo: the vendor kernels from qualcomm *generally* can't use the tty
<tdeo>
yeah, i figured out i had to build examples/hello instead which works
<samueldr>
yeah, that's a better thing to use to confirm things work as expected
<samueldr>
and nice to know things work as expected
<tdeo>
i'm now starting to attempt to get sway to cross compile
<samueldr>
good luck if you're going with cross-compilation; anything in stage-2 is pretty much pure Nixpkgs/NixOS
<samueldr>
same caveats apply :)
<samueldr>
tdeo: quick survey: are you aware that if you were to modify the partition table of the device's internal storage you're at high risk to make the device unbootable?
<samueldr>
(in the oneplus3 case, it's not bricked, but in most qualcomm-based devices it'd be bricked for real)
<tdeo>
yeah
<samueldr>
I wonder if there should be more indications about that somehow in the docs
<Ke>
don't most of these devices have some hardware flashing mode
<Ke>
special software maybe required
<Ke>
like odin whatnot, but for qualcomm
<samueldr>
exactly
<samueldr>
but it needs the proprietary cryptographically signed "programmer"
<samueldr>
which is often unique to a model
<samueldr>
or if not a model, an SoC model × OEM
<samueldr>
which officially should never leave their control
<samueldr>
since it means it's trivial to dump the data from that device
<samueldr>
or change things in bad ways
<samueldr>
but then there are OEMs like oneplus who ships the programmers for their devices in flashing tools to unbrick their devices
<samueldr>
which is both good and spooky
<samueldr>
recently their programmers are limited to only allow verified images to be installed to the device
<samueldr>
but their previous programmers can do anything
<Ke>
would it matter, since the SoC would not boot?
<Ke>
without signed bootloader
<samueldr>
yes, because you may be able to force OEM unlock a device without erasing the userdata partition
<samueldr>
then, install a boot.img which is not signed
<samueldr>
it may show the scary boot prompt
<samueldr>
but on older devices it was much less scary
<samueldr>
and that, it's when you're thinking without bugs!
<samueldr>
if you end up findings bugs like on LG devices
<samueldr>
not necessarily the exact way it's been exploited
<samueldr>
but many, if not most, devices don't sign the logo partition
<samueldr>
the logo partition is used to host the different images the early bootloader can throw to the framebuffer
<samueldr>
if there was a similar bug on oneplus devices where you can just dump anything on the internal storage, you might even be able to work around oem unlock without actually unlocking it
<samueldr>
and boot unverified programs
<samueldr>
it's the swiss cheese analogy in security
<samueldr>
and it's been used (probably wrongly) as an end-user security measure AND a supply-chain security measure
ib07 has quit [Quit: No Ping reply in 180 seconds.]
ib07 has joined #nixos-aarch64
<Ke>
I guess
cole-h has joined #nixos-aarch64
<samueldr>
forcing an oem unlock through programmers is a known thing that is doable
<samueldr>
as long as the unlock bit is not handled in the rpmb via trustzone