Filter
Exclude
Time range
-
Near
ShiniNet retweeted
Jun 13
zmk-layout-shift v2を公開した Kconfigベースの設定からdevicetreeベースに変更して、複数レイアウトを同時に使用できるようになった github.com/kot149/zmk-layout…
1
4
14
1,289
Jun 13
devicetreeベースになったことでユーザーがモジュール本体をいじらなくてもレイアウトを追加できるようになった あと細かい部分でいうとパフォーマンス改善、テスト実装で堅牢になった、曖昧だった名前を明確な名前に変更したなど
3
239
of(devicetree)前提のframeworkをむりやりfwnodeに書き直してtry-and-errorで突破できないか試行錯誤してたが、全くACPIを取ってこないので辿ってたら元からfwnodeが空でofだけコピーする嫌なコードだった
1
70
now, if it were all parens and semicolons in place of commas, you'd know it was high quality devicetree
sign of high-quality code
4
133
And millions will be heartbroken when they realize they won't be able to install Linux. One devicetree per chip per manufacturer per model that a developer needs to make. Every generation. ditching x86 and ACPI will be the end of software freedom as we know it today
2
2
44
1,402
May 29
最近はYoctoいじってdevicetreeいじってゴニョゴニョする楽な仕事ばかりしているが この仕事派遣労働者が1人でやってていいもんなんか? と思ったり
1
2
223
What’s new in Zephyr this week? 🎙️ @kartben & Fabio cover remote build/local flash workflows, a Zephyr-native Meshtastic stack, devicetree updates, new drivers, DALI lighting support, new boards, the Developer Survey 2026, WG updates, & upcoming meetups. hubs.la/Q04h_ZLP0

1
4
488
自作ZMK ModuleのConfigをどう持つか悩み中。一般的にはKConfigがSW用で、devicetreeがHW用らしい。devicetreeだと.keymapでも編集できるんだよなぁ。両方にするかなぁ。 そして、KConfigの説明とかhelpは、ビルド時に-t guiconfigとか-t menuconfigを指定したら設定画面が出て表示されるのね。
2
4
393
他妈的,它亮了 屏幕上跳出“CUDA available: True”那行字的时候,林宇的手在发抖。 不是激动的那种抖。是那种——手心全是汗,拇指按着触摸板,按了三下才点到确认窗口。旁边那台RTX 5090像个外接的银色器官,风扇还没转,安静得像在嘲笑他。 “不可能。”他妈的,所有人都说不可能。 苹果说Apple Silicon不支持外接显卡。英伟达说驱动只给x86。Reddit上的人说别想了,省省吧。他在那篇帖子里翻了两个小时,底下全是“LOL”、“RIP”、“just buy a PC bro”。他关掉浏览器的时候,手指关节咔咔响,像在捏碎什么。不是愤怒,是——说不上来,就是那种“你们等着”的感觉。 三个月前他还在用一台2019年的Intel Mac,那台机器接eGPU就跟喝水一样自然。卖掉换M4的时候,他站在苹果店里,店员说“先生,这个不支持外接显卡哦”,语气轻松得像在说今天天气不错。 他当时笑了笑。没反驳。心想:你懂个屁。 回家的路上他一直在想那台旧Mac。每次跑LLM推理,风扇起飞,提示处理慢得像在泥里走路。换了M4之后快了一点点,但也就是一点点。跑个7B模型,提示一长,还是卡。他查过,这是Apple Silicon的硬伤——统一内存带宽就那么回事,你拿它没办法。 除非—— 他把TinyGPU的源码又翻了一遍。那个项目是台湾一个家伙写的,文档里全是繁体字,有些地方还夹着粤语,“搞呢啲嘢真係好撚烦”。他看不太懂,但那个人的语气他懂。那种一边骂一边不放弃的劲儿。 虚拟机GPU直通的部分最难搞。他试了三十多次,每次都是黑屏或者kernel panic。有一次凌晨三点,屏幕突然亮了。不是系统界面——是报错。全红的文字,像血一样糊在屏幕上。他盯着那些字看了很久,然后把手边的啤酒喝完了。 也不是伤心。就是——啊,原来是这样。 第二天他改了devicetree的配置,重新编译。这次没报错。什么都没显示。他等了一分钟,两分钟,手指在键盘上悬着。然后他输入了那个命令。 nvidia-smi。 屏幕上跳出一行字:RTX 5090,24576 MiB,0 MiB used。 他愣了三秒。然后开始跑LLM。 提示处理速度从之前的40 tokens每秒飙到了120。那种感觉就像——就像你开了十年的破车突然换了引擎。不是快,是轻盈。是那种“我可以试着跑点大的”的冲动。 窗外天快亮了。他靠在椅背上,听到自己的心跳,还有显卡风扇刚刚启动的声音。 嗡—— 很轻。像在说:我知道你等很久了。
6
1,236
iSkorpion - iOSTool v7.9 released. - Added Live Support/Chat within the app (requires an account on the website) - Fixed bugs & removed “Supervised by” - updated telegram links New Group: t.me/iSkorpionTool #MDM #icloud #MobileGestalt #DeviceTree #SetupApp
1
1
4
888
But that does not mean that the problem is DeviceTree, the problem is that laptop OEMs couldn't care less about anything non windows.
1
2
11
DeviceTree will tell you what resources are there and what device names (to match drivers agains) are there. That's it. It's not capable of handling multiple versions of an interface based on declared capability support from OS, like ACPI does with _OSI and _OSC
2
2
64
DeviceTree is not as flexible, and is becoming a burden in ARM world. Definitely blocks it from being as end-user-friendly as PC. What people forget is how a new OpenFirmware machine required a new OS release with drivers and quirks despite DeviceTree
1
2
46
more of this in our TNA discord ## Hardware - **SoC:** Kendryte K230 (RISC-V, `rv64imafdcvsu` — vector ext present) - **Model string (devicetree):** `kendryte k230 nano3s` - **DRAM:** 112 MiB - **NAND:** Winbond SPI NAND, 256 MiB (block 128 KiB / page 2048 / OOB 128) - **WiFi/BT:** Realtek RTL8723DU (USB; firmware `rtl8723du_fw`) - **USB:** dwc2 OTG controller @ 0x91500000 - **Console UART:** `serial@91401000`, 115200 8N1 ## Firmware - U-Boot SPL U-Boot 2022.10 — built **2025-06-11 16:09:32 0800** - OpenSBI v0.9 - SDK string: `v1.6-20250611-161026-root-ebc36f47f86f-8074e1e1a` - Release tag: `Avalon_Nano3s_release_2025061101_4ba23e3` ## Boot flow (from log nano3s-20260506-004730) 1. SPL prints `I am wakeup!`, `SYS_CTL_INT2_RAW=0x13 SOC_WAKUP_SRC=0x10` 2. SPL banner 3. PMU coprocessor message: `Firmware run has completed 2667` 4. Loads U-Boot env from SPI NAND (Winbond detected) 5. Jumps to U-Boot proper → CPU/DRAM/MMC/serial init 6. `light on...` (status LED routine — relevant if you want a heartbeat indicator) 7. `Net: No ethernet found` (no link), then `Hit any key to stop autoboot: 0` 8. Probes USB-OTG for storage (recovery hook!) 9. Hands off to OpenSBI → kernel 10. Linux brings up syslog, klog, mdev, networking, sshd 11. UBI mounts two volumes (~11.6 MiB and ~54.2 MiB) 12. wpa_supplicant udhcpc — no DHCP lease (no link)
3
245
May 1
The recently released dtre-hax project is really cool. I tested it on a device running iOS 16.7.14. Everything works fine, reading and rewriting values ​​in DeviceTree.
5
45
3,982
4. ACPI really only a x86 standard, what else have you seen use it. Arm merely adopted it. You can do the same things with uboot and dt. I have are boards with upstream devicetree that can seamlessly boot generic arm64 iso and even give me a boot source menu
2
2
342
DeviceTree manipulation pongoOS module Can modify nearly all devicetree entries in memory straight from pongoOS. syscfg, ECID, UDID format, demotion status. Similar to MobileGestaltCache with bigger playground, but tethered :/ github.com/hanakim3945/dtre_…
1
8
53
3,716