#asahi · 2026-07-13 08:07
PaulFertserYep guys, talk to megi at https://pine64.org/community/#chat_platforms e.g. on #pinebook channel (irc.pine64.org network) or e-mail, I'm just the messenger here :)#asahi · 2026-07-13 05:20
PaulFertserHi there. A prominent Pine64 developer has recently released a new library to get VA-API working over V4L2-requests. https://xff.cz/git/libva-v4l2_request/ . Should be relevant with your new AVC custom firmware. More details here https://xnux.eu/log/102.html .#asahi · 2025-04-27 17:29
PaulFertserumaga: having both should be fine, you select which one to use while logging in.#asahi · 2025-04-26 19:47
PaulFertserIf you're unsure about your ALSA lib configuration.#asahi · 2025-04-26 19:47
PaulFertserthe_eye: the next step would be trying something like `aplay -vvv /usr/share/sounds/alsa/Noise.wav` so that you see what device is opened and how exactly etc.#asahi · 2025-04-26 19:33
PaulFertserIn home directory it's looking for ~/.asoundrc and ~/.config/alsa/asoundrc so `find ~/ -name asound.conf` wouldn't show those.#asahi · 2025-04-26 19:32
PaulFertserthe_eye: `strace -e access,openat speaker-test` will show you if any additional configs are used from anywhere. #asahi · 2025-04-17 14:54
PaulFertserSome people had bad luck with missing initramfs after upgrade toughh.#asahi · 2025-04-15 17:44
PaulFertserWell, defaults can be changed when they do not match your preferences. E.g. I always remove "quiet" from default Linux command line.#asahi · 2025-04-15 17:43
PaulFertserWho knows why. There should have been an error emitted during upgrade but guess it got scrolled up too far or something. If you kept the full transcript of that upgrade session probably someone could answer why it happened.#asahi · 2025-04-15 17:42
PaulFertserIt's your choice to have the GRUB menu hidden unless Esc is pressed at the right time.#asahi · 2025-04-15 17:41
PaulFertserNothing was damaged, old kernel and initramfs were preserved and available in GRUB menu, nothing dangerous there.#asahi · 2025-04-15 17:38
PaulFertserYou need initramfs there to boot obviously :)#asahi · 2025-04-15 17:38
PaulFertserThat is really odd, I wonder if the filename is correct in the grub config itself?#asahi · 2025-04-15 17:36
PaulFertserBTW, someone recently reported a similar issue due to out-of-tree evdi driver, it was somehow preventing initramfs generation.#asahi · 2025-04-13 20:33
PaulFertserI'm not sure if I should be happy about being involved in a backend development using C++ rather than JS or not :)#asahi · 2025-04-13 20:30
PaulFertserBut react runs on browser side, so on every web site user machine.#asahi · 2025-04-13 20:29
PaulFertserThank you for sharing evdi experience. Some people complained here about issues, good to know is probably no longer the case and a sane option when latency on external display doesn't matter that much.#asahi · 2025-04-13 20:28
PaulFertserquaff: just wow. This web thing really got out of control :)#asahi · 2025-04-13 20:25
PaulFertserquaff: I'm a bit curious what kind of job requires 32+ GiB RAM if you can share a glimpse into that.#asahi · 2025-04-13 20:25
PaulFertserquaff: so the lag wasn't annoying for regular tasks? Did you probably have sound stuttering while external display was updating?#asahi · 2025-04-13 20:19
PaulFertserquaff: what's your impression so far about evdi?#asahi · 2025-04-13 10:39
PaulFertserAt some point I was running a cellphone interface in that lisp virtual machine on ARMv4T with just 128 MiB SDRAM and also edited some Haskell code, it was all snappy enough. So not all of them are /that/ demanding.#asahi · 2025-04-13 10:27
PaulFertserrunning on a JavaScript virtual machine in parallel to nested VMs (emulated WinAPI over emulated x86 over arm64 4k Linux over arm64 16k Linux over hypervisor), a bunch of docker containers (easy and reliable to deploy) and other containers (Flatpacks and appimages because why bother packaging properly)? Is that your dream technology world?..#asahi · 2025-04-13 10:26
PaulFertserNow imagine I'm both ironic and serious at the same time, what would you call /that/ and what emoticon should I use? Because, seriously, coreboot is great, me_cleaner is great, I honestly would be happy to see more people using and contributing to it rather than giving money to apple for their shiny new toys. Do you really want all that DDR5 bandwidth because you "need" to run a text editor#asahi · 2025-04-13 10:13
PaulFertserYou meet the nicest people on a Honda^W^W OpenBSD forum ;)#asahi · 2025-04-13 09:55
PaulFertserSo probably same if not worse deal with the firmware versions, changing their APIs, different partially independent issues with wifi firmware, bluetooth firmware, soundcard firmware...#asahi · 2025-04-13 09:52
PaulFertserhanged the wifi right away so it wasn't usable at all. With Fedora 6.12 it was loading newer ax211 firmware and it worked but not the Bluetooth part (it just wasn't replying to driver). And on top of that the internal speakers didn't work at all while 3.5 mm socket did. Go figure.#asahi · 2025-04-13 09:52
PaulFertserWell, recently I had issues simply booting an x86 laptop (some asus expertbook). Debian 6.1 booted fine but lacked Xe driver for GPU. Debian 6.12 hanged hard after initing USB host interfaces, no panic no error no nothing, and I have no clue if SysRq worked because I couldn't find SysRq. Same about Debian 6.13. But somehow Fedora 6.12 booted just fine. With 6.1 the ax211 firmware was older and#asahi · 2025-04-13 09:45
PaulFertserI thought Rob and friends care a lot about not breaking compatibility where it matters. #asahi · 2025-04-13 09:44
PaulFertserDTs are forward-compatible. So the vendor releases a new shiny server, prepares DT which works with at least some Linux version actual at the time, ships this DT with platform firmware. Newer kernel shouldn't have a problem using this older DT. And when some important changes are needed a newer DT can be used instead.#asahi · 2025-04-13 09:42
PaulFertserchaos_princess: bytecode is probably good because you can kind of RE it easier and do you own thing if needed.#asahi · 2025-04-13 09:41
PaulFertserWell, normally it should be the platform firmware but in fact usually Linux+BSD devs are much better than the vendor so you probably have to count on them keeping it sane and up-to-date and use what's on kernel.org rather than provided by the vendor.#asahi · 2025-04-13 09:40
PaulFertserThe schematics for the board.#asahi · 2025-04-13 09:39
PaulFertserOn typical Arm you have DT describing hardware, standard PSCI calls to handle power management. You just grab aarch64 image and boot it on any system, feels OK in my experience.#asahi · 2025-04-13 09:36
PaulFertserfl0_id: device trees do not provide those bytecode snippets. And that's a great thing to not have to use them.#asahi · 2025-04-13 09:35
PaulFertserchaos_princess: only that talking about laptops you know how many custom acpi-specific (also wmi) code there is in Linux with probably a hundred of model-specific quirks. And how Linux pretends to be a certain windows version when interpreting the obscure bytecode, and then on some models or UEFI versions it needs to pretend being a different windows version etc etc. Pure filth.#asahi · 2025-04-13 09:30
PaulFertserAnd Asahi devs do RE and implement APIs of the (typically) most recent stable version available so that it works on the widest range of devices possible (Apple provides newer firmware for older devices but not older firmware for newer devices).#asahi · 2025-04-13 09:28
PaulFertserI guess for someone familiar with WiFi this analogy might work: where you're used to have SoftMAC-like solutions the Apple hardware offers you FullMAC, and they change the API to those "FullMAC" devices any moment they feel like because they control both sides of the API.#asahi · 2025-04-13 09:26
PaulFertserAnd I'm not sure it's just a great advantage.#asahi · 2025-04-13 09:26
PaulFertserArm requires ACPI for their server platforms now :/#asahi · 2025-04-13 09:23
PaulFertserAnd of course those "CISC" CPUs like their microcode updates.#asahi · 2025-04-13 09:22
PaulFertserTF-A via PSCI on common Arm-based systems.#asahi · 2025-04-13 09:22
PaulFertserIBM PC mental model isn't much useful for any modern hardware, including amd64. The days of VESA VBE, APM and A20 gating are long gone, things are done very differently now. But you still depend on things outside the OS kernel even after passing control to it: ACPI tables for detecting on-board hardware and ACPI bytecode for runtime power-related actions on typical x86 platforms, and calls into#asahi · 2025-04-13 08:58
PaulFertserfeatures are likely to require firmware versions that would be current by the time those features are developed.#asahi · 2025-04-13 08:58
PaulFertserByronJackson: on a desktop you also get firmware updates, e.g. for wifi adapters (quite often), for USB printers, for GPUs etc etc. It's just that apple hardware is special and you have plenty of additional CPUs on board that you need to talk to for doing useful things. If the current Asahi feature set covers all your usecase then you won't be forced to upgrade firmware most probably. But new#asahi · 2025-04-13 08:26
PaulFertserSlick#asahi · 2025-04-13 08:23
PaulFertserIs it personalised per machine?#asahi · 2025-04-13 08:22
PaulFertserProbably if you have a friend who doesn't mind wasting SSD space on macOS the new blobs can be just copied over?#asahi · 2025-04-13 08:20
PaulFertserIf only firmware was redistributable