<unnamed>

Public Key

npub19wj8tgf5p67elrkz0kgh8n28779aum6cwc838p2rqwaxh4hn4ycq28g8gs

Profile Code

nprofile1qqszhfr45y6qa0vl3mp8mytne4rl0z77dav8vrcns4ps8wnt6me6jvqprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmnt0cjehd

Publishing to

relay.momostr.pink

Last Notes

2026-08-27 23:54:27 CEST

"Sir, we believe there might be hackers living in this house"
https://asdf.donotsta.re/media/76822b46a34d98e6931c9ecd1c3a362981f561690354e0751082f658971ae228.jpg

2026-08-27 00:33:23 CEST

Now I understand why Germans like FRITZ!Box'es so much. Delightful hacks :akko_giggle:
https://asdf.donotsta.re/media/74c8c5e681ba5d785370d6659c1a0065b50ddd0166e1c4c27e731c76cdc2d06d.jpg

2026-07-27 22:00:27 CEST

- reply

`swapon /dev/sr0`

2026-07-26 05:13:56 CEST

- reply

It's a typical AliExpress special, it's really good... but somehow went up in price by 60% since I bought it :acat_huh:
https://aliexpress.com/item/1005005029695255.html?

2026-07-26 04:41:11 CEST

- reply

I wouldn't recommend it for larger soldering, tips are really tiny

2026-07-26 04:30:58 CEST

- reply

I do and carry it basically everywhere with me, works better than cheap "rework station" I got last year. I would say that unless you're doing microsoldering or something like busbars it's going to cover all your needs.
My work laptop's PSU (Framework GaN 180W) gives it extra "oomf" (aka can heat up and transfer the heat much quicker) but it works just as well with something like a 100W powerbank. Pretty sure you'll find pictures from FrOSCon of me soldering on the cafeteria table :akko_shrug:
For microsoldering I would recommend Miniware TS21, one such example of when it comes handy (wire in this picture is roughly the same thickness as human hair):
https://asdf.donotsta.re/media/7c9d85dbe3489d9e8f39d676ba17f9e96de652d7a0233a4e66c5a01be0dc85d6.jpg

2026-07-25 23:40:09 CEST

- reply

tela-purple/tela-purple-dark is my favourite

2026-07-24 13:39:47 CEST

"Look ma, we're running off of battery!"
ACPI on AMD and Framework-specific Chrome EC customizations are kicking my butt, but once I fix the suspend this will already be daily-drivable (with ~30% better battery life than stock firmware thanks to already having better power management than InsydeH2O) :ablobcatbongo:
Also after dropping AMD's GOP into IDA I suspect it silently fails because it can't store NVRAM variables due to SMMSTORE being unavailable (SmmStore region initialization fails and throws an exception, I freakin' hate EDK2).
Given that GOP is patching the VBIOS in some way, I suspect fixing SMMSTORE will fix GOP init, and that in turn will fix DP AltMode.
With that sorted the only remaining things TODO will be Thunderbolt 4 and controlling bifurfication on RootPort exposed as an expansion bay :blobcat3c:
https://asdf.donotsta.re/media/e485399782856ac6017af29ffcae8501bba0891dbf732326eb24f40070b1e330.jpg

2026-07-24 13:09:08 CEST

- reply

it probably wouldn't be able to use the full address space though, much cleaner solution is to create a SPI spider (solder the same exact chip on top of the current one with CS lines lifted, solder some wires to a 2-position switch and flash respective chips with different FW).
If you install the card in PC you set switch position to 1, if you install it in OF system you switch it to 2, voila

2026-07-23 23:29:38 CEST

- reply

nope, for Zen 2 you'd likely have to hack FSP from Chromebooks (someone actually did that on X370 and went quite far)

2026-07-22 11:20:43 CEST

- reply

yeeah, that's a sign that the battery is cooked

2026-07-20 22:38:09 CEST

- reply

well, I know from 4 separate sources that AMD measures employee performance in token usage... I just did not expect that was the case over a year ago
https://asdf.donotsta.re/media/d26aec053af2d62497ca698905278e774b10c57d82317f80580857bc1fc712ab.jpg

2026-07-20 22:23:19 CEST

- reply

ok, I'm pretty sure AMD's GOP driver is vibe-coded :neocat_floof_explode:

2026-07-18 23:31:10 CEST

Been doing some more sorting and clean-up today so I can finally have a bit more space in my (tiny) flat and it got me thinking:
I'm sitting on *a lot* of hardware which is at least partially broken (I would feel bad giving it away as it can possibly die at any moment) but it would be perfectly service'able as buildservers or systems people need temporarily (i.e testing something on older platforms).
However, I don't want to build those systems into cases due to space constrains and cost associated with it... but I could build a dedicated space for it under my electronics workbench.
Regular ATX mainboards are smaller than 19in racks (I guess they're roughly 12 inch long according to wikipedia as I don't speak 'Murican units) and some plywood I can get at a hardware store not only would do the job, it would support the workbench in case I put something heavier on top of it (after all, cheap IKEA benchtops are made pretty much out of cardboard).
Then I could 3D-print sleds for ATX/mATX mainboards (or even laptop mainboards) that would slide in perfectly, the only problem being "how do I cut perfectly straight sled guides without tools or space to do so".
In terms of power: ATX PSUs are super expensive these days... but server PSUs aren't, getting one that spits-out 12V even up to 1kW is pretty cheap (costs roughly 20 - 30EUR) and provides advanced monitoring over PMBus. The only downsides are noise they make (but who says you can't cut the case and attach bigger fans) and (usually) custom connectors that would need some RE-ing to yoink 12V busbar and i2c from them
Mounting them on an open-air side of that box for safety reasons (maybe covered with perforated aluminium cover) sounds like a perfect solution - very efficient PSUs that can provide a lot of power and advanced monitoring for pretty cheap, and ATX boards can get even cheaper PicoPSUs for 5 and 3.3V (basically DC-DC conversion and relay controlled by SuperIO) and be powered on/off on demand by GPIO on MCU.
Kinda curious if anyone built something like it in the past, or if there are any considerations that I missed?

2026-07-17 21:06:33 CEST

Why does Brno use Windows 98's ding.wav as public transport's announcement sound? :neocat_googly_woozy:

2026-07-16 03:48:33 CEST

- reply

absolutely, I'm always terrified when people suggest cross-flashing firmware from another motherboard (especially on forums) because you can absolutely release a magic smoke this way

2026-07-16 03:39:04 CEST

- reply

on dGPUs I would absolutely expect that, you need specific VBIOS for RAM present on the card, VRM etc

2026-07-16 03:31:12 CEST

Asked around and were told that AMD ATOMBIOS wasn't storing device-specific configuration and I found that sus because I couldn't get eDP initialized with generic one provided by 3rdparty/amd_blobs and Intel has VBT for this exact reason...
Turns out my hunch was right, we're gaming :acat_sunglasses:
https://www.youtube.com/watch?v=7XlH_1Z0slk
https://asdf.donotsta.re/media/0e8f7add03faddb63e9a935b05f750e83df2a5741e59fa7049a1eb315be6fbc4.jpg

2026-07-15 00:17:19 CEST

It has been 0 days since `acpi.poweroff_on_fatal=1` being new default trolled me :akko_giggle:

2026-07-14 08:58:05 CEST

How come that 'Muricans didn't port postmarketOS to Flock cameras?
They're free (you can just take them home) and horribly insecure (Android 8 running on SDM624) :blobcat3c:

2026-07-10 22:34:39 CEST

Apparently TigerLake supported AVX512, while RaptorLake doesn't... this will take a while :akko_tired:
https://asdf.donotsta.re/media/fb068a44cf67062e390760f4bede09403fd2defadc910d38289b6a7184ee4533.jpg

2026-07-07 23:07:23 CEST

- reply

rotation of hardware (including thonkpad parts and chromebooks) within our friend froup is kinda insane :akko_giggle:
I'm pretty sure @domi is currently posting using an Acer mechanical keyboard from ~1995 with French AZERTY layout... that came into my possession in 2012, monitor to my left side is an NEC from 2007 (1920x1200) that she left at my place. The X200 is question went to either @BluRaf or @Lili after the screen died circa 2020. Kinda wondering what happened to my cursed T420 (with coreboot, i3-3220M, keyboard from X230), don't remember who took that one

2026-07-06 11:19:49 CEST

*oven ding* it's done :ablobcatbongo:
Took me a while to finish this because of life-related reasons, but finally sat down and fixed (most) remaining issues during the weekend.
Pretty much everything* is working - including S3 suspend/resume, automatic fan control and so on (plus it draws less power than stock, I have no idea what AMI is doing). Feels kinda surreal to have a fully-functional computer :akko_giggle:
If you have an "MSI PRO B760-P WIFI DDR4" motherboard (or "MSI B760 GAMING PLUS WIFI DDR4", pretty sure they're the same just with different UISetup skin) you can now flash #coreboot on it.
DDR5 versions can be easily added if someone is willing to work/test it, just a matter of telling FSP you're using a different memory type.
Will document stuff (like making flashing cable that won't void warranty) later. Enjoy :blobcatsalute:
https://review.coreboot.org/c/coreboot/+/93127

2026-07-06 11:06:36 CEST

- reply

that sounds waaay too expensive, I paid ~200EUR for my glasses (frames *and* lenses)

2026-07-06 01:06:25 CEST

- reply

same, I'm also a filesystem "boomer". XFS has zero advantages if you're not running using it for databases, BTRFS is just unstable, ZFS is out-of-tree and so on...
EXT4 is still fastest/most reliable :akko_shrug:
Also, is it a Smasnug drive by any chance?

2026-07-05 04:13:35 CEST

Well, not getting blasted with completely white screen at night is definitely an advantage :akko_giggle:
https://asdf.donotsta.re/media/2eb1597cd4f28d4c04224c94a4a3c52dabd8c8901bbc041cda6989404e4671b4.jpg

2026-07-03 20:12:09 CEST

PSA for folks considering buying an Anycubic Kobra S1: Extruder module is prone to having filament getting stuck in gears if you get a clog.
I couldn't fix it with any tricks my coworkers and friends had suggested, decided to open the extruder (since it wouldn't take a new filament, so I knew it was stuck in there). Looks like either cuts on gears aren't deep enough, or motor doesn't provide enough torque to push the clogged filament trough.
I was really angry when others said they had to replace the whole module (especially since I went trough *one* spool of filament on this printer and it already broke), but it looks fixable. Just very annoying to do so.
https://asdf.donotsta.re/media/786431b5bdc4cb8971a97bdb56df6eb98466cb0a3c56c56d5902abe342ad1abf.jpg

2026-06-24 12:31:11 CEST

Yeah, sure, upload your ID and picture of your face to yet another random website, what could possibly go wrong...
This should be illegal :akko_angry:
https://asdf.donotsta.re/media/996489bfb979fd152a9787bfa20947b37c35e9e3614cd8cafc8e01b19d2299bc.jpg

2026-06-19 23:55:22 CEST

- reply

I expected it to happen around 2028, I guess we're just 2 years early

2026-06-19 21:37:50 CEST

Where the heck is the tra- oh, I guess it won't be coming :akko_giggle:
*(Today was an absolute disaster, massive delays etc, likely caused by a heatwave. I almost passed out after walking to catch a tram for 10 minutes and I drink plenty of water)*
https://asdf.donotsta.re/media/b1adcc278e25dfbec5c3e311f002dd1a3741d4e4ecfd942e069336e957f26775.jpg

2026-06-17 23:00:11 CEST

- reply

backlight keyboard is a must these days tbh, can't use a laptop without one

2026-06-13 23:38:04 CEST

- reply

~~meanwhile my linkedin picture features me wearing Lynx R1 and holding glasgow connected to UART/JTAG port, it's the only publicly-accessible picture of me (not counting CCC talks)~~ :akko_giggle:

2026-06-12 23:53:27 CEST

In 2014 friends tried to get me into Ingress but it seemed nagging and sus, so I didn't.
Then in 2016 they tried to get me into Pokemon Go and it also scratched that part of my brain that screamed "I bet they're using this data for *something*" and I refused to play.
2026: News outlets report on "scandal" that Pokemon Go players trained AI for free. Everyone reacts with *surprised pikachu face*, I go "yep, freakin' told ya".
:akko_shrug:

2026-06-12 02:56:43 CEST

Got to the point of mapping clock sources and as a final test, I needed a PCI-E x16 card that only takes a single slot (to populate EVERY slot on the board)...
Going trough my stash of PCI-E cards, I found only one - first GPU I ever bought (Radeon HD4850).
It blows my mind that I just chucked a GPU from 2008 into a brand-new system, booted mainline Linux and card just... works, while Nvidia GTX1000 series that's only a decade old is now effectively unsupported :smugcat:
https://asdf.donotsta.re/media/9b8758b406f66aca180a2f466c1a87cb05a587bb02a22d3b11b759280ab5cd62.jpg

2026-06-11 21:39:54 CEST

- reply

well, pmOS now has a systemd option :neocat_blep:
That being said, while musl is generally a saner implementation it comes with it's own... quirks let's say. I were insane enough to roll some glibc libraries in /usr/local/usr-glibc and patchelf software I needed to run that would otherwise only run on something like Debian (i.e: geekbench) but that's a hacky workaround

2026-06-11 21:25:36 CEST

- reply

depends, more-or-less:
- Gentoo on the desktop
- Chimera on x86_64 Chromebook
- postmarketOS on phone/arm64 Chromebooks
- Alpine on servers and VMs
- Fedora on DUTs (quick and easy to install, grab packages I need etc)
All development I do is done within Docker container, same with gayming.
Some time ago people noticed that a beginner developer publishing some community VR project used a binary dependency from github repo owned by an otherwise empty account and building this dependency from source gave completely different results, so it was safe to assume people who used that project got infected.
I didn't, but after that I just decided to contain everything in the container as well (as a bonus, I can run this stack on a musl-based system like Chimera) :blobcatwink:
I need to add "--volume '/dev:/dev'" to startup script (otherwise data from controller IMUs cannot be read), but that's about it:
https://gist.github.com/ellyq/db86e914be584fab5f6c3547f4ddbf29
+ quick demo: https://www.youtube.com/watch?v=N26EgJWgx3g

2026-06-11 21:14:15 CEST

- reply

for what it's worth, I knew that AUR could be used to distribute malware when I last used Arch... in 2014 (yes, this was a problem 12 years ago!).
I know people will hate me for saying this, but AUR was never safe. It was one of two reasons why I ditched Arch (another reason was stability... or rather lack thereof, although I've heard it improved over the past decade)

2026-06-10 19:44:06 CEST

- reply

yuuuup

2026-06-09 22:50:46 CEST

- reply

I usually just order hardware from mindfactory, alternate or amazon.de (ew) because Germany has much lower tax rates (when I lived in France and wanted to buy a GPU I would've paid close to 600EUR if I bought it locally, same card on amazon.de cost me roughly 520EUR with shipping to Strasbourg)

2026-06-09 22:40:46 CEST

- reply

Oh, yeah, Gigaschrott is very incompetent (every single piece of hardware I ever touched from them had design/software flaws and died sooner or later). It would indeed be a good candidate, but support for modern AMD platforms is very much WIP (we'll be lucky if it'll be usable this year).
Try adding `acpi_os_name="Windows 2021"` to your kernel parameters, pretty sure this will help (it makes Linux cosplay as Windows, firmware will think you're booting Windows 11 and will use the same ACPI tables as it does when you're booting Windows): https://donotsta.re/notice/B2qAwVHAilptYyfgrA

2026-06-09 16:01:19 CEST

- reply

I love how this spiraled out-of-control with beowulf cluster made out of 6 desktops on the floor by the end of GPN :akko_giggle:
Our table also seemed cursed, I smashed my laptop on the ground on day 1 and on day 3 someone tripped sending their laptop flying in the same spot

2026-06-09 15:25:49 CEST

- reply

and there were no PSUs/HDDs installed, so the server was a bit awkward to carry but not heavy at all (definitely less heavy than my desktop)

2026-06-07 12:44:08 CEST

#carryshitolimpics w/ @9lore
https://asdf.donotsta.re/media/23e5590548ed4b5020cc7ea97607f1e33d4339324d497fdd73647da99cd852f0.jpg

2026-06-06 22:21:16 CEST

- reply

yes

2026-06-06 19:31:03 CEST

- reply

time to eat and hack on pmOS :blobcat3c:
https://asdf.donotsta.re/media/97e9d65f989862fc6271a938c74804bd49d7fa880224a3d4e968e06a48233a83.jpg

2026-06-05 17:56:15 CEST

- reply

yeeep, hinge is making crunchy sounds (inspected it, it's similar mechanism that smasnug used in first-gen Z Fold) so some gears likely became misaligned or snapped altogether.
It's fine as long as I don't use it in the tablet mode (kinda sucks to not be able to do it) and are careful while closing it, but fixing it is impossible. Hinges are fused with display assembly and insides of them are filled with epoxy so there's no way to remove the mechanism

2026-06-03 03:14:51 CEST

Bundenspolizei dillema:
Is it a #postmarketos developer, or terrorist with a bomb? :akko_giggle:
- Wires sticking out of the phone
- Wall of text scrolling by after pressing the power button
- Phone making weird noises
*(In case anyone would want to play with this device at #gpn24, please **do not squish the right side of the case**, while it should be safe, I cannot guarantee you wouldn't puncture the battery if you would go "full ham" on it)*
https://asdf.donotsta.re/media/c6da6f0772da7f871ebd36518dcea79eca9303ce490569c0ef241e6184cc11e5.jpg
https://asdf.donotsta.re/media/6bb817b655726ea8edf6cef399f028b2610df22675394aa69772b3249df0ee28.jpg

2026-06-02 22:06:34 CEST

- reply

It's the same price as Glasgow though:
- https://1bitsquared.de/products/tigard
- https://1bitsquared.de/products/glasgow-c-aluminium-case

2026-05-29 12:24:57 CEST

- reply

I don't see why not, but temperature becomes a problem. PLA deforms at 65*C, PETG at 85*C so I would have to buy an ABS because that starts to deform at 105*C.
@npub1hz9…vgra has been advising me and infodumping about 3D-printing (which I really appreciate, I'm a noob) so I'm sure she'd have a better idea how to do it.
Still, gonna try to rummage through my basement and see if I can find it but my place is a total mess right now as I'm still in the middle of renovations

2026-05-27 21:46:46 CEST

@npub1wcd…n7s2 #postmarketos can have contactless payments, as a treat :blobcat3c:
https://asdf.donotsta.re/media/394a37949d3227916156972f681802aa778d5c2e8553a02cd137d603a5879927.mp4
#note1t9t…x8ju

2026-05-26 16:30:08 CEST

Does anyone know how to match thermopads? Of course GPU vendors never provide this information...
Back in 2024 I bought the fastest, most expensive card I ever had (~500EUR), it started crashing last summer and memory was idling at 70*C. It got better during the winter, but recently (before my motherboard died) it got worse.
Never had to replace thermopads before (even on 10 year-old cards they usually were fine), so I'd appreciate any tips.
I'm currently assembling a new workstation, thanks to friends ( @BluRaf, @TadeusTaD, @Remiberry, @npub10hk…28x2 ) I was able to afford an LGA1700 DDR4 motherboard (there's no way I'm buying DDR5 in current market conditions even if it costs a fair bit of performance) and i9-14900K which will also be ~50% faster than Erying with ES i9-11980HK that... disintegrated.
Over the past 2 weeks I tested all components from my (dead) workstation, with following results:
- Heatsink: Bent in shipping (thanks LaPoste), will use a publicly-accessible vice in OBI to strengthen it. It should be able to handle this CPU.
- Seasonic 620W: Works perfectly, dusted it off and swapped a new fan in
- 32GB of DDR4 I bought back in 2023: Works without XMP enabled, but instability at 3200MHz might be caused by either bad board design or heatwave.
- SSDs/HDDs: Perfectly fine, I got all of my data uncorrupted.
- GPU (Sapphire RX7800XT Pulse): Overheats even at idle (bad fan profile), fan control doesn't work (even though it was supposed to). Opened it up to find completely dry thermal paste and thermopads crumbling if touched.
Measuring current pads (obviously squished) with calipers gave me following results:
- Backplate -> PCB (above VRAM): 3mm
- VRAM -> heatsink: 1.8mm
- VRM: 2mm
Should I just order the same thickness, or do they have to be slightly thicker (since heatsink should compress them)?
Thanks! :neocat_heart:
https://asdf.donotsta.re/media/1b1f7675a32bb3a5cc8e87d9c03d5c02e6a8e1d4ba4ab65445ba0ae219fe87cf.jpg

2026-05-24 18:42:25 CEST

I have to say, this new MacBook Neo is really pretty :akko_giggle:
https://asdf.donotsta.re/media/05798fe6e7e552c3f957b3cf14ed1a4008474010ee831fe0eb6988136fd06d38.jpg

2026-05-19 14:21:51 CEST

I wonder if local police station has a bomb defusal squad :neocat_nervous:
https://asdf.donotsta.re/media/0859f6af35e1347e0612996eb6d7f57d3c9aafd2d15bad900d2b0c880b0c7527.jpg

2026-05-11 05:38:18 CEST

Welp, after spending 3 days and nights debugging my workstation I can say with 100% certainty that all issues I've been having recently are caused by silicon degradation.
First it was PCIe, then random (sporadic) lockups every ~month or so, then issues with USB controller, then iGPU and then lockups became so frequent that it became unusable (2 - 3 per day). AMDGPU issues might even be related to this.
Cleaned the system and put it together, but I expect it to crash while I will be sleeping... so I guess it's time to look for a normal motherboard and CPU (thankfully I can get used MSI DDR4 board which doesn't have BootGuard and 14Gen i5 for roughly ~300EUR and it's going to be 20% faster than the current setup as well).
This concludes a 3 year-old experiment, if you're thinking of buying cheap pre-production hardware that fell from the back of the truck in China: Don't. It's funny at first, but save yourself the headache.
https://asdf.donotsta.re/media/7726d96e3cbc92b1ccbd7679b213c58bf9f62ba2003043ac17b200d96e05251f.jpg

2026-05-10 18:49:08 CEST

age verification failed :akko_giggle:
https://asdf.donotsta.re/media/83ca9c7aad88cee0583436bf7bdde0d0fe5bf4a5e6a9d115fb2005bb656a1ff5.jpg

2026-05-09 15:46:46 CEST

> Remove the GPU to see if crashes are related to it
> Even more issues, text fringing, artifacts, monitors blinking at random, incorrect shadows
> :acat_huh:
> [ 38.114190] i915 0000:00:02.0: [drm] *ERROR* This is a pre-production stepping. It may not be fully functional.
> Oh yeah, forgot about that...

2026-05-04 23:01:29 CEST

Idea for a drinking game: take a sip everytime AMDGPU craps itself :akko_giggle:
This has been annoying me since kernel 6.17, was better on 6.19 and is now annoying me again on 7.0:
```
[14079.109651] amdgpu 0000:03:00.0: [drm] *ERROR* flip_done timed out
[14079.109659] amdgpu 0000:03:00.0: [drm] *ERROR* [CRTC:367:crtc-1] commit wait timed out
[14089.349239] amdgpu 0000:03:00.0: [drm] *ERROR* flip_done timed out
[14089.349247] amdgpu 0000:03:00.0: [drm] *ERROR* [CONNECTOR:393:HDMI-A-1] commit wait timed out
[14099.588765] amdgpu 0000:03:00.0: [drm] *ERROR* flip_done timed out
[14099.588778] amdgpu 0000:03:00.0: [drm] *ERROR* [PLANE:148:plane-2] commit wait timed out
```
"Fix" for this is simple, but still annoying (turns monitor off, waits 20 seconds (which is just enough to cause timeout) and turns it back on):
`ddccontrol -r 0xd6 -w 4 dev:/dev/i2c-4; sleep 20; dccontrol -r 0xd6 -w 1 dev:/dev/i2c-4`
I guess it's related to having monitors with mixed refresh rates, one of monitors randomly just freezes, I have to bring terminal to one that's not frozen and reset the monitor with 20-second delay. *sigh*
Really ironic that the only cards that just... work on Linux are iGPUs from intel (their dGPUs are pretty stable these days but their power management is awful)

2026-05-02 02:48:25 CEST

- reply

huh? How would messing with flash break the USB-C? :acat_confused:

2026-05-02 02:43:50 CEST

- reply

On Gentoo ugRD is pretty comfy and it's easy to set this up, but easiest option by far would be serial console (even TTL and pasing console=ttyUSB0,921600n8) and having a "serial orchestrator".
I would love to get/make something similar to what Shelby showed in his video about DECWriter, you'd just have a single system with multiple serial ports and you could manage your systems without an expensive KVM. If your system is running coreboot (like vast majority of my hardware), you would be able to reboot the system, change boot configuration, change some settings (if CFR is implemented and using EDK2) and boot Linux with full serial

2026-04-30 16:43:38 CEST

- reply

I've been procrastinating making a blog for 5 years :throwing-hands-up-crying:
Deploying nginx, getting familiar with hugo (and picking a theme, so many options...) or writing something by hand, protecting against LLM scraping these days just sounds like a lot of work

2026-04-30 06:11:37 CEST

This might be a hot take, but I will never understand why people new to hacking on mobile/embedded devices insist on trying to use vendor-provided sources.
Even if you get a kernel source from $vendor (which is not a given, welcome to the magical land of GPL violation) you will run into numerous issues even trying to compile it, only to end up with a pile of legacy crap that will never* work correctly with upstream userspace (unless you resort to using Ubuntu Touch or Droidian, which AFAIK use halium and can use Android HALs).
Seriously, if you're new to this and want to learn how to do it *properly* on a completely unsupported SoC, you should:
1. Dump downstream devicetree from running device. It's much easier to understand than vendor sources once you understand how it works: `dtc -I dtb /sys/firmware/fdt -O dts > /storage/emulated/0/Downloads/dumped.dts` (and pull it using adb).
2. Locate UART pins on your device (beware that most ARM SoCs use 1.8V or under) and what's the address for that (for example on MT6789/MT8781 it's `0x11002000`). Using simple-framebuffer is a last resort but you really want functional UART for early bringup.
3. Write initial upstream devicetree and minimal driver set. You more-or-less need the following to get into initramfs:
- Clocks
- Core cluster definition (in DT)
- Global Interrupt Controller (usually GICv3 unless it's Apple, then AIC)
- Pinctrl
- UART
*(Hopefully not forgetting about anything?)*
4. Go from there (drivers for interconnect, i2c/spi etc).
Using kernel trees for the same SoC (even if it's a different vendor) is a protip, it's all $soc_vendor's BSP after all. Or if you're in really deep, throwing kernel modules at ghidra is also an option :akko_shrug:

2026-04-21 12:52:24 CEST

- reply

Form-factor looks like standard miniPCI, but pinout looks custom... probably removed from some ancient ISP-provided router/modem? I wouldn't stick that into a computer though

2026-04-19 02:31:53 CEST

- reply

my end-goal is to hang those monitors on the wall and have a shelf holding hotair/bench PSU, so I would be able to fold the desk when it's not in use and have more space.
All I have is an 18m2 (sqm) room, which is decent for someone living alone but it requires management system to avoid cluttering it and there's some furniture I can't get rid of (like gigantic standing desk) because it belongs to flat's owner
This dual-monitor arm was *really* cheap (~14EUR) but I can mount in on a wall and 3D-print upper reinforcement (once I receive the printer) and just... put them on a wall. I will need help of someone who has muscles and owns a proper electric drill, but my brother volunteered to do this for me :neocat_heart:

2026-04-19 01:38:35 CEST

Amazing IKEA quality, good thing I noticed it before monitors slammed the keyboard :neocat_shocked:
https://asdf.donotsta.re/media/be15b47d1f1fc47caf0e1d9dd2ae8d6c05e321397c37062d78ff3cdb93d0d5a5.jpg

2026-04-14 20:23:11 CEST

- reply

well, this package doesn't seem to offer Linux binaries :blobcatblep:
(It's fine, it's long overdue anyway. I bypassed stuff manually by installing VxKex, patching DLLs which gave me much more up-to-date tools like Python 3.14 but it's becoming more and more unfeasible)
https://asdf.donotsta.re/media/1f7425bb6dd6b939ab7ffa1ed14c1e609aa1f1e73a1e4fd7e6f71ed7c64dd6fe.jpg

2026-04-14 06:35:23 CEST

- reply

when I worked in enterprise we had machines with JBODs, so you had classic sd[a-z] and then... kernel started counting again, so you had stuff like /dev/sdayy

2026-04-14 01:32:28 CEST

- reply

RIP parition table :neocat_floof_explode:
https://asdf.donotsta.re/media/240ad21f586171dd333d78e8462610edec9ab62c7a15eded587159f26fb0d2e9.jpg

2026-04-12 10:06:55 CEST

Eh, EL1? That's kinda lame, I'm pretty sure we can boot in EL2 :akko_thonk:
https://asdf.donotsta.re/media/4b071731f9f31ef684e3a1c0b8c36d25a4a3821eed78f0c0a746a8a734e6110b.jpg

2026-04-11 21:36:39 CEST

Welp, looks like upstream changes to uclass broke mediatek ufs driver, so that will need some debugging...
But it works on older U-Boot tree, so I don't have to deal with annoying LK. Since I can now boot this phone using U-Boot's EFI implementation, it's time to pet some penguins :blobcat3c:
https://asdf.donotsta.re/media/26632f9a4baab4ddd983c36c305d55b16dfe866c5d038f7ecb36fd5bbdd8d743.jpg

2026-04-10 23:13:10 CEST

- reply

lol, no

2026-04-09 03:02:06 CEST

Power button on this phone feels pretty nice :akko_giggle:
I got annoyed because buttons on this phone have terrible quality and don't register presses when case is off.
While poking the board with oscilloscope to find UART I found a testpad that was shorting to ground when I pressed the power button, so here we go.
Also found that shorting two pins around the battery connector makes phone boot with only USB attached (both testpads had same voltage with battery connected and USB cable plugged in) so I can attach a MCU to toggle power on/off and turn phone on/off by setting pins high or low, I should have some octocouplers in parts drawer.
Sadly still no UART, I *really* don't want to deploy a Windows VM just to use some Chinese software to get schematics for this but it should be much easier to find it now.
Someone sent me memory dump from that software and it looks like it opens a tunnel to China (geoblock), authenticates over HTTP with cookie and downloads PDFs from "http://down.dzkj16888.com/down/51/pdf/$VENDOR/$MODEL/$FILE.pdf". I probably could pwn it and archive all pdfs they host on that server but I don't have enough spoons :akko_tired:
https://asdf.donotsta.re/media/91be3a55b34c556d9c53c6aff2dc51ecdfbd5411f4760ad6087a55c6a2423147.jpg

2026-04-08 22:08:41 CEST

- reply

WAIT, this looks remarkably simple...
Since it's just an I2C bus, we could pinmux it in coreboot and write a driver to forward console output to it (similar to how it's done with LPSS) and grab it over DVI/HDMI/DP with something like a Pi Pico. It would make it possible to develop/debug boards without RS232 or even laptops without opening the chassis :neocat_hyper:

2026-04-08 21:59:26 CEST

- reply

I think it's a bit more complicated with HDMI, but you can access i2c on GPUs just fine (it's used for DDC). I think I only tested this on Intel iGPUs so it's good to know it works on (at least older) AMD APUs as well.
In fact, Google did some cursed stuff in 2013 where they hooked up an input device (either touchscreen or touchpad, I don't remember) to iGPU's i2c bus (I think they did it in LINK)

2026-04-02 20:08:23 CEST

- reply

I mean, that's... pretty standard? If you don't install GApps on LineageOS it's the same exact story

2026-04-01 21:30:54 CEST

Can you tell that this driver was written by catgirls? :neocat_pat_floof:
https://asdf.donotsta.re/media/e93e44e7e231843b8e5c0bbe09433a0d9d48f492bb4c49a5ea875bf9b8c7bf1e.jpg

2026-03-28 10:30:31 CET

- reply

I would say that Strasbourg was the prettiest place I've lived in so far overall and one of prettiest places I've ever been to tbh
https://asdf.donotsta.re/media/2eda1934f926b190ee92d0cdc2ca2aacb892432b44f8ba6f929d140bd6e60af0.jpg
https://asdf.donotsta.re/media/58bb1e9079dfc8e1cb54f87574f3f9994aac8015dd81e02531a4b78a52b8ef1b.jpg

2026-03-26 00:51:43 CET

I guess it's time to find out if thermal protection is working
https://asdf.donotsta.re/media/6393bae0167c2f98e39a9491e433b6c55745ec0780b0f284f989ecef5850e88f.jpg

2026-03-24 09:24:18 CET

- reply

I suspected that was the case to be honest :blobcatblep:

2026-03-24 02:14:09 CET

- reply

> lovely little nugget but it sucks that it only has 8GB of RAM
Uhh, my pyxis has 6GB of RAM and I managed to run docker + fex + steam + windows game on it

2026-03-22 23:58:13 CET

- reply

this just... works in Plasma Mobile on my phone?
There's a toggle in display settings if I want to use ALS to adjust brightness automatically... and it works the same as on Android (poorly because Xiaomi didn't calibrate the sensor correctly, but it does).
I guess you'd need to check how it's connected and if there are drivers in mainline for that particular sensor in your laptop

2026-03-20 03:46:38 CET

- reply

actually kinda wondering, how does that work with libc rootfs/musl host?
I did extremely cursed sh^H^H stuff with stealing /usr/lib and /usr/lib64 from fedora, placing it in /usr/local/lib and /usr/local/lib64 and using patchelf on software compiled for libc.
*(Yes, I know I'm going to hell for this)*
```
/tmp/Geekbench-6.6.0-LinuxARMPreview # patchelf --set-interpreter /usr/local/lib/ld-linux-aarch64.so.1 ./geekbench_aarch64
/tmp/Geekbench-6.6.0-LinuxARMPreview # patchelf --set-rpath /usr/local/lib:/usr/local/lib64 ./geekbench_aarch64
/tmp/Geekbench-6.6.0-LinuxARMPreview # ldd ./geekbench_aarch64
/usr/local/lib/ld-linux-aarch64.so.1 (0xffffbc1e0000)
libdl.so.2 => /usr/local/lib/ld-linux-aarch64.so.1 (0xffffbc1e0000)
libpthread.so.0 => /usr/local/lib/ld-linux-aarch64.so.1 (0xffffbc1e0000)
libm.so.6 => /usr/local/lib/ld-linux-aarch64.so.1 (0xffffbc1e0000)
libgcc_s.so.1 => /usr/local/lib64/libgcc_s.so.1 (0xffffb7b7d000)
libc.so.6 => /usr/local/lib/ld-linux-aarch64.so.1 (0xffffbc1e0000)
ld-linux-aarch64.so.1 => /usr/local/lib/ld-linux-aarch64.so.1 (0xffffb7b3b000)
[...]
```
```
/tmp/Geekbench-6.6.0-LinuxARMPreview # ./geekbench_aarch64
Geekbench 6.6.0 Preview : https://www.geekbench.com/
Geekbench 6 for Linux/AArch64 is a preview build. Preview builds require an
active Internet connection and automatically upload benchmark results to the
Geekbench Browser.
System Information
Operating System postmarketOS edge
Kernel Linux 6.18.5-sdm670 aarch64
Model qcom Xiaomi Mi 9 Lite (xiaomi-pyxis)
Motherboard qcom Xiaomi Mi 9 Lite (xiaomi-pyxis)
CPU Information
Name Qualcomm ARMv8
Topology 1 Processor, 8 Cores
Identifier ARM implementer 81 architecture 8 variant 6 part 2050 revision 13
Base Frequency 1.71 GHz
Instruction Sets neon aes sha1 sha2 neon-fp16
Memory Information
Size 5.41 GB
```
https://browser.geekbench.com/v6/cpu/17168133

2026-03-20 00:31:14 CET

I really appreciate folks working on Plasma Mobile though, this is really comfy (pretty much my daily-driver now) :neocat_floof:
(Yes, I still need to fix internal audio, wifi and cameras, but they're not deal-breakers for me especially since I can use bluetooth in the meantime)
https://asdf.donotsta.re/media/2eb27c14949537fcb9f0cfd73bf188f2e2bc011763858b5fb28ad3d28bd8fe28.pmosftw

2026-03-19 22:33:47 CET

- reply

probably would run, I revised my instructions for doing this in Docker on pmOS wiki. Building FEX on xiaomi-pyxis didn't take that long and SoC didn't even get to 70*C :blobcat3c:
If you edit devicetree to enable dual-role and mode-switch on usb node, it shooooould work in host mode if you use powered USB hub and do manual modeswitch.
SDM670/SDM710 are cut-down versions of SDM845 so if it works here, I assume it would work on SDM845 as well: https://wiki.postmarketos.org/wiki/Qualcomm_Snapdragon_710_(SDM710)#USB
https://asdf.donotsta.re/media/3196b000ded40196ccaebacd84802ee56e5d50125a1e23697896949a55e2a827.jpg

2026-03-06 09:20:09 CET

- reply

I mean... that's what manpages are for.
I've been using Linux for 18 years at this point, so I can't imagine any kind of scenario where I would need text predictor's help with... using my operating system. I have zsh autocomplete and tab button on my keyboard for that :akko_giggle:

2026-03-06 09:11:05 CET

- reply

> another useful thing is letting ai do stuff on ur computer (eg. install xyz program) but i wouldnt trust this too much bc of injection attacks
I would rather go swimming naked in freezing pond than let text predictor anywhere near my operating system or any sort of credentials

2026-03-04 22:49:10 CET

- reply

hey, look at it this way - if we would meet our teenage selves they'd think we grew up to be pretty darn cool (at least I hope so)

2026-03-04 22:44:16 CET

- reply

real question being: Did you (also) listen to bands like Escape the Fate?
(I kept my old music collection, Eyes Set to Kill and Bring me the Horizon are still bangers tbh)

2026-03-04 16:38:37 CET

- reply

critical digital infrastructure
I noticed I wasn't getting DHCP requests from that thinclient anymore, doesn't seem to react to power presses... rest in pepperoni. I'll investigate once I get some proper sleep.
Worst-case I can migrate w2k install to this beautiful nugget or ThinkPad X230: https://www.nowatron.cz/download/245/datasheet-lexcom-3v900d.pdf
Probably will finally deploy the CI runner (Intel NUC 12 Pro) and Chromebox/LattePanda Mu (I finished the project and client told me I can keep it and have no use for it, so it would be perfect for hosting stuff for friends - 6W idle, 64GB(I think?) eMMC, 8GB LPDDR5. I re-did the coreboot port for this because DFRobot's code is utter dogpoop, so we can even do remote attestation on it with LinuxBoot sind scheisse :blobcat3c:
https://asdf.donotsta.re/media/9b7d79579825184c107485cb2e1ff6429c8372dea15a04733e2d1bc864f9068e.jpg

2026-02-03 07:57:16 CET

1. Be me
2. Turn heating off before heading to FOSDEM
3. Be on your way back
4. Realize heating's off and it's -15*C outside (compared to ~6*C in Brussels)
5. Neuron activation, desktop is online
6. SSH into your desktop at home
7. Set fans to 50% (piping to hwmon)
8. Start a full system update with emerge
I knew running Gentoo is a good idea :akko_giggle:

2026-01-30 22:11:39 CET

Welp, looks like I'm allergic to Spezi. Not sure why I'm surprised given I'm allergic to most citrus-like fruits (especially oranges) but still... sucks, no more Spezi for me :meowcry:

2026-01-26 23:45:02 CET

- reply

https://asdf.donotsta.re/media/f42d9b9ff4f0d73bbfbb7757079e24f63b684e190f525d8fb7b0d810f9f3de15.jpg

2026-01-25 16:08:09 CET

- reply

~~is this a personal attack?~~
https://asdf.donotsta.re/media/348dbde5953e47e10f553dc8abb3ba37cd9e65c71be71521d633d1536753dad7.png

2026-01-23 16:26:06 CET

I'm wondering if there are good resources on diagnosing/fixing CRTs.
Sadly search engines in 2026 are completely filled with LLM-generated nonsense that would (quite literally) kill me.
Thanks to @TadeusTaD and @BluRaf I got my hands on a decent CRT monitor few months back, after initial testing we determined that input signal is very, very dim but it works fine otherwise.
Given the operating temperature, age and era it was produced in (~2002) I suspect capacitors right off the bat so I'll need to open it and take a look at the board.
Probably will swap them all in one go as I'm planning a trip to my hometown where I can just go to the store and pick individual capacitors I need.
I don't think diagnosing it would be too difficult for someone who has a soldering iron, (cheap) oscilloscope etc, but I never worked on CRTs before and I find HV a bit... scary so I'd appreciate some resources :ablobcatbongo:

2026-01-10 22:54:14 CET

- reply

I wonder what we could cram into modern SPI chips (32MB, but memory-mapped to 16MB).
You could cram IFD (4K), ME (5M), BIOS (CBFS - coreboot w/libgfxinit + SeaBIOS) (8M) into the upped 16M boundary, giving you 16MB of SPI space just for the floppy image

2025-12-21 17:27:07 CET

My train got delayed today so I pulled out my laptop and started playing a game.
When it was time to get off the train I started panicking because I couldn't save (seems like you can only save at determined points in VA-11 Hall-A) but had to go, so I closed the lid and threw it in the backpack.
Not only system had suspended without any issues, when I pulled the laptop out at my brother's place... game resumed as well and I could continue playing. Just like that.
One of my friends owns a handheld running windows and he's upset with how buggy the suspend is, as it takes forever to suspend and games tend to crash...
I'm really impressed with how far we've come with power management and suspend in mainline Linux.
Not only this laptop lasts longer in Linux than it did with stock firmware and ChromeOS, it also suspends without any issues whatsoever.

2025-12-15 21:06:47 CET

Hot take: Industry should abandon HDMI.
DisplayPort is better in every single case (licensing, bandwidth, physical connection).
Every single time I need to use HDMI, I have some kind of issue. One of my monitors is connected using HDMI and I had to add aliases to my shell called "fucking-lg-on" and "fucking-lg-off" that writes values to i2c device (DDC) to tell the monitor to suspend.
Otherwise when kscreenlocker sends DPMS standby, monitor keeps disconnecting and re-connecting itself which sooner or later crashes my wayland session.... and that's just one example.
I've had issues with HDMI as far back as I can remember (likely circa 2014 when I finally could afford swapping burnt-out CRT from 1999 with 1080p LCD monitor).
DisplayPort however? Zero, nada.
Then of course there's HDMI Forum which I personally hate with burning passion (for obvious reasons).
I genuinely wonder *why* manufacturers of monitors and TVs stick with HDMI when DP is superior in every single way.

2025-12-12 06:32:35 CET

I finally fixed touchpad in my favorite Chromebook (KOHAKU) and annoying issue with random wakeups while testing upcoming coreboot release (24.12, scheduled for Christmas Eve).
It feels... weird to have fully-functional machine with no quirks. Every other piece of hardware I own has weird quirks requiring workarounds or is straight up broken in one way or another :blobcatgoogly2:

2025-12-02 03:47:03 CET

- reply

VRAM allocation? If you've updated the firmware in the past ~3 months you should have CFR options now to mess with those settings (screenshot is from my workstation though)
https://asdf.donotsta.re/media/c4a0d028556da305650ae5d2147ad5abc1fde54ac1033cf5d9b6ccc4ee85df8d.png

2025-11-28 18:14:50 CET

Since we're on the topic of Cellebrite: #postmarketOS is NOT vulnerable.
Among devices that police seized from my friend was Xiaomi Poco F1 (xiaomi-beryllium) running postmarketOS build I pmbootstraped in late January 2024 (without LUKS2). Police seized the device 2 weeks after I gave it to said friend.
They tried to exploit it, but gave up. Mainline Linux kernel (6.6 at the time) did not have vulnerabilities in USB HID drivers.
To mitigate further against tools such as Cellebrite, we could enable USB authentication which prevents kernel modules from loading prior to user's consent.
Problem with that approach is that Desktop Environments have authentication implemented for Thunderbolt, but not for USB.
https://usbguard.github.io/
I really would like to see SELinux implemented as well, but it would be a Herculean effort because rules would need to be hand-written for Alpine.

2025-09-08 04:10:24 CEST

- reply

damage control