tinyboot v0.4.0 Released — The API is Stable — Stable…
    Neura MarketNeura Market/Stable Diffusion
    ChatGPTChatGPTClaudeClaudeGeminiGeminiCursorCursorGrokGrokPerplexityPerplexityStable DiffusionStable Diffusion
    DeepSeekDeepSeekCoPilotCoPilotMidjourneyMidjourney
    View All Directories
    OverviewPromptsBlogVideosGuidesCoursesCommunityModelsLoRAsComfyUI WorkflowsTrending
    Stable DiffusionBlogtinyboot v0.4.0 Released — The API is Stable
    Back to Blog
    tinyboot v0.4.0 Released — The API is Stable
    rust

    tinyboot v0.4.0 Released — The API is Stable

    Aaron Qian April 23, 2026
    0 views

    Minimal Rust bootloader for CH32 MCUs. v0.4.0 adds CH32V00x support, stabilizes the API and wire protocol, and fits in 1920 bytes of system flash.


    title: tinyboot v0.4.0 Released — The API is Stable published: true description: Minimal Rust bootloader for CH32 MCUs. v0.4.0 adds CH32V00x support, stabilizes the API and wire protocol, and fits in 1920 bytes of system flash. tags: rust, embedded, bootloader, ch32

    cover_image: https://direct_url_to_image.jpg

    published_at: 2026-04-23 05:10 +0000 canonical_url: https://aaronqian.com/log/2026-04-22-tinyboot-v040-released/

    If you've been following tinyboot, you might have noticed there was no announcement for v0.3.0. That release added CH32V103 support, but things were still in flux. Crate structure was shifting, APIs were changing, the protocol was being reworked. I didn't want to write up something that'd be outdated in two weeks. I wanted to wait until the dust settled. And with v0.4.0, we have finally arrived at this point.

    Links

    • Repository: OpenServoCore/tinyboot
    • Handbook: openservocore.github.io/tinyboot
    • Changelog: CHANGELOG.md

    What is tinyboot

    tinyboot is a minimal bootloader for resource constrained MCUs. It is written in Rust, and it fits in 1920 bytes of system flash, leaving all user flash free for your application, except a small page (64 bytes on V003) of user flash to store boot metadata. It gives you CRC-validated firmware updates over UART with trial boot and automatic fallback to bootloader service mode when trials run out. The kind of safe OTA update story you'd expect from a much larger bootloader, squeezed into the constraints of a $0.22 MCU.

    It's currently focused on the CH32 family, but the core is chip-agnostic and designed to be portable. If you're interested in bringing tinyboot to another chip family, see the porting section of the handbook.

    I'm building it as part of OpenServoCore, an open-source smart servo platform. tinyboot handles the OTA updates via the existing single wire UART (Dynamixel TTL), so you don't have to tear your robot apart and open up each servo, unsolder the board just to flash a new firmware.

    Platform Support

    FamilyStatus
    CH32V003Supported
    CH32V00xNew in v0.4.0
    CH32V103Supported
    CH32X03xPlanned

    V00x Support

    The biggest addition in v0.4.0 is full support for the CH32V00x family: V002, V004, V005, V006, and V007. This is the release that matters most for OpenServoCore, because the OSC dev board runs on the CH32V006. Hardware validation for this release was done on the actual dev board.

    Getting here wasn't entirely smooth. I ran into a hardware issue where the RX line wouldn't work without driving the inverse TX_EN line, which took a scope session to figure out. But that's a story for another post. I'm just glad I was able to finally get RX to work and complete the hardware test.

    Everything Fits

    TX_EN support used to overflow system flash on some chip variants. That's not acceptable. TX_EN isn't optional for OpenServoCore. It's required for half-duplex RS-485 / DXL TTL communication.

    As of v0.4.0, all chip variants compile with TX_EN enabled and fit in system flash. The V103 was the outlier. It has a split flash layout, and I moved the UART transport into the second region to make it fit. That trick deserves its own post. Down the road, that second region also has enough room for a USB transport, so flashing via USB on V103 is a real possibility.

    Crate Restructure

    The previous three separate crates (tinyboot-ch32-boot, -app, -hal) are now merged into a single tinyboot-ch32 crate with boot, app, and hal modules. There's also a new tinyboot-ch32-rt, a minimal runtime because qingke-rt is too large for system flash.

    This wasn't a planned refactor. It was the natural result of continuous iteration during tinyboot's early development. I'd rather get the architecture right early by sacrificing API stability than lock in the wrong abstractions.

    tinyboot-ch32 is currently git-only and not published to crates.io. It depends on a git version of ch32-metapac that includes flash fixes for the V00x family that haven't been released yet.

    Protocol and API Changes

    The wire protocol now uses 24-bit addresses instead of 32-bit, freeing the fourth byte for per-command flags. 24 bits is still 16MB of addressable space, more than enough for these MCUs. The standalone Flush command is gone; it's now a flag on the final Write. Cleaner on the wire, simpler in the dispatcher, improves the developer experience, and best of all no size increase due to zero-cost abstractions via Rust's union types.

    On the API side: BootMode became RunMode to separate the concept of boot flash region selection vs bootloader run mode (handoff or service). BootClient was also removed after the crates merge. Boolean parameters in the public API became semantic enums (Duplex::Half instead of half_duplex: true). Last but not least, flash lock/unlock is now scoped per operation instead of manual.

    Bug Fixes

    Two nasty half-duplex communication bugs, both found during hardware validation on the dev board:

    The dispatcher wasn't flushing the transport after sending a response. On a full-duplex UART you'd never notice, but on RS-485 / DXL TTL half-duplex the data just sits in the buffer and never goes out on the wire.

    The ring buffer wasn't resetting its head/tail pointers after flushing buffered writes to flash. The buffer was logically empty but the pointers were advanced, so subsequent writes would eventually wrap incorrectly. I didn't notice this before because I didn't do back-to-back flash commands with the tinyboot CLI.

    Docs Overhaul

    Documentation has been completely rewritten for users instead of maintainers, and a user handbook has been created. When the architecture is changing every release, writing user-facing docs is a losing game. Now that things have stabilized, it actually makes sense.

    What "Stable" Means

    To be clear, tinyboot is not production-grade. But as of v0.4.0, it's stable in the ways that matter:

    • Architecture stable: no more big crate restructures.
    • Protocol stable: no more wire format changes.
    • Features stable: all core features compile and fit on all supported chips.
    • Behavior stable: no obvious bugs.

    This means I'm shifting focus. tinyboot satisfies all of OpenServoCore's needs, and my attention is moving to rewriting the OSC firmware. I'll still maintain tinyboot, and issues and PRs are welcome, but active feature development is pausing for now.

    What's Next

    • USB transport for V103 (the second flash region has room).
    • CH32X03x support eventually.
    • But the immediate priority is OpenServoCore firmware.

    Tags

    rustembeddedbootloaderch32

    Comments

    More Blog

    View all
    Context bankruptcy: The case for strategic forgetting for AI Agentsai

    Context bankruptcy: The case for strategic forgetting for AI Agents

    Most of us have seen a coding agent fail to complete a task we know it can do. We just don't...

    J
    James O'Reilly
    Parallel Compliance Engine: Drive-to-Sheets Multi-Agent Orchestrationgooglecloud

    Parallel Compliance Engine: Drive-to-Sheets Multi-Agent Orchestration

    When building Generative AI applications, developers often encounter a massive bottleneck: sequential...

    A
    Aryan Irani
    Is It Ethical to Post and Ask About Circuits on Dev.to?discuss

    Is It Ethical to Post and Ask About Circuits on Dev.to?

    I’ve been thinking about sharing some electronic circuit posts on Dev.to — small circuits, DIY...

    C
    codebunny20
    The One-Click Exporter: AI Studio Antigravity, Probed to Its Limitsagents

    The One-Click Exporter: AI Studio Antigravity, Probed to Its Limits

    What nobody tells you about exporting your multi-agent prototype to a local workspace. Every...

    L
    leslysandra
    Guarding the till while autonomous data agents do the diggingagenticarchitect

    Guarding the till while autonomous data agents do the digging

    Autonomous agents are genuinely good at answering messy business questions. Give one an LLM and a set...

    S
    Sireesha Pulipati
    Return on Attention: Why AI Code Reviews Are Wearing Us Outai

    Return on Attention: Why AI Code Reviews Are Wearing Us Out

    PR volume went up, ticket quality didn't, and the gap got filled with LLMs on both sides of the review: bots reviewing, bots replying, bots occasionally arguing with bots about priorities that only existed in a teammate's head. Our CEO named the actual problem, and it's bigger than code review.

    C
    christine

    Stay up to date

    Get the latest Stable Diffusion prompts, rules, and resources delivered to your inbox weekly.

    Neura Market LogoNeura Market

    Discover the best AI prompts, plugins, and resources for Stable Diffusion and more.

    Content Types

    • Rules
    • Prompts
    • MCPs
    • Agents
    • Guides

    Platforms

    • ChatGPT Directory
    • Claude Directory
    • Gemini Directory
    • Cursor Directory
    • Grok Directory
    • Perplexity Directory
    • DeepSeek Directory
    • CoPilot Directory
    • Stable Diffusion Directory
    • Midjourney Directory
    • All Directories

    Resources

    • Blog
    • Documentation
    • Help Center
    • Marketplace

    Legal

    • Privacy Policy
    • Terms of Service

    © 2026 Neura Market. All rights reserved.

    |

    Not affiliated with any AI platform vendors.