On February 28th, I delivered a talk about “The Current State of RISC-V Linux Kernel Open Source Upstream” at the 2025 China RISC-V Ecosystem Conference. Drawing from nearly a decade of experience in open-source technology, I have worked extensively to align China’s chip industry with global open-source communities.
In my talk, I felt it important to highlight the importance of upstreaming support for RISC-V to the Linux kernel and am urging vendors from China’s RISC-V industry and academia to take a more active role in contributing support for RISC-V to the open-source ecosystem.
My session was structured into four key sections:
- Why Care About Open Source, Why Upstream Linux Kernel Code?
- From Experimental to Mainstream: RISC-V’s Path in Linux
- RISC-V Growth in China
- Lessons from the Arm Ecosystem
Why Open Source Upstreaming Matters for RISC-V Business
The Commercial Value Pyramid of Open Source Contributions
In my view, the commercial value of open source participation forms a pyramid. At its base is the broad compatibility with Linux distributions and application ecosystems. Whether it’s desktops, automotive systems, or cloud computing, all Linux distributions require full support in the mainline kernel for underlying hardware. (As noted in Debian and CentOS Automotive Stream Distribution).
Cost Savings Through Community Collaboration
The next level is about collaboration (in other words, savings in R&D man-months). With the code being merged upstream, vendors can join the community in maintaining and upgrading the mainline kernel. This collaboration provides timely performance and security updates without the need for extra engineering resources to backport and test fixes. For companies, this means significant savings in R&D man-months and a reduction of risk in addition to a more secure product for customers. Moreover, a successful mainline integration for one System on a Chip (SoC) family paves the way for easier upstreaming of subsequent SoC products from the same family based on similar IP or bus architectures, resulting in additional compelling benefits such as shortening development cycles and further reducing long-term R&D costs.
Finally, at the top of the pyramid is global market trust (in other words, sales numbers). With the strong foundation built by the previous layers, an SoC vendor’s products gain wider international credibility. Especially in terms of security and compliance. Downstream product vendors tend to favor suppliers with mature open source solutions—a trend that is especially evident in Android, AI and LLMs, and cloud servers.
Through these three layers, vendors can achieve a complete commercial value loop from embracing open source.
From Experimental to Mainstream: RISC-V’s Path in Linux
I reviewed the progress of RISC-V chip support in the Linux kernel. In my opinion, there are several key milestones.
RISC-V’s Journey in the Linux Kernel: Key Milestones
The earliest RISC-V kernel support came from SiFive’s HiFive Unleashed development board built on the FU540 SoC. This support was officially merged into Linux 4.15 in January 2018 after nine rounds of review [3], covering CPU support, interrupt handling, and memory management. That was the moment when RISC-V officially became a supported CPU architecture in the Linux kernel.
But even at that point, things were far from complete. Palmer Dabbelt, in the v4.15 merge commits (dated November 15, 2017) [3], wrote:
“It’s a bit hard to actually see anything happen because there are no device drivers yet.”
Mainstream Recognition: Debian 12 Support and Beyond
Another big moment came in June 2023, when Debian 12 “Bookworm” was released with Linux kernel v6.1. This was a game-changer because riscv64 became one of Debian’s nine officially supported architectures. Suddenly, RISC-V was no longer just an experimental platform—it was a real option for Debian and Linux users.
Current Support Status in the Linux Kernel
China has been a hotspot for RISC-V development. Since 2017, I’ve seen RISC-V expand from research institutions (such as ICTCAS) to major companies like Alibaba’s Damo Xuantie (玄铁), SpacemiT K1 (进迭时空).
RISC-V Adoption Timeline in Chinese Tech
The pace of RISC-V SoC development and upstreaming from 2020 to 2025 has been making steady progress.
If you look at the upstreaming timeline, you’ll see that every year, companies have been adding more foundational support. But there’s still a gap—a fully upstreamed RISC-V SoC, with all peripherals, is still missing. Many development boards have been released, DTS files have been added into the kernel mainline. But when you try to run mainline Linux on them, you’ll find missing features. USB, PCIe, GPU, and multimedia such as display — these are often incomplete and still rely on vendor-specific downstream kernels (a.k.a. vendor trees).
Challenges in Full SoC Support and Upstreaming
RISC-V SoCs are making steady progress upstream, but the reality is that full SoC support is still not there yet! Closing this gap is critical for RISC-V’s long-term success in the Linux ecosystem.
Learning from Arm’s Ecosystem Development
Drawing from my experiences, I compared RISC-V’s challenges to the Arm ecosystem’s past struggles with fragmentation, driven by varied licensing and poor coordination. I referenced Linus Torvalds and his 2011 critique of Arm, which spurred a shift toward community collaboration and upstream-first approaches after 2012. However, I noted that Arm’s transformation required years of pressure, platform support, and a cultural shift among vendors.
RISC-V faces similar hurdles—vendor inconsistencies, slow mainline adoption, and limited unity—but benefits from its open Instruction Set Architecture (ISA), enabling wider collaboration. RISC-V International (RVI) has stepped in to coordinate collaboration around ISA and non-ISA definition and standardization. However, in the kernel space, that’s not enough. I recommend that RISC-V adopt Arm’s lessons, such as early collaboration, standardized practices, and a “mainline-first” mindset, to streamline upstreaming and avoid prolonged setbacks. I conclude that “Kernel mainlining is the ‘only path’ to success for the RISC-V industry.”
The Path Forward for RISC-V Kernel Upstreaming
It is important not to underestimate the critical role of upstreaming in unlocking RISC-V’s potential. RISCstar is dedicated to this effort. This commitment to enhancing RISC-V Linux kernel support sets the stage for broader adoption of this groundbreaking architecture. At RISCstar we’ve been working with our partners to embrace the upstream-first strategies and we invite you to join us in contributing to this open-source effort.
Talk with an Upstreaming Expert
The RISCstar team has extensive experience collaborating with global open-source communities and can help you develop and implement a strategy that reduces long-term costs.
