Will RISCV avoid the linux mainlining mess that ARM had?


#1

If I understand, the issue is that unlike x86, ARM did not standardised the board support package. And thus mainlining of device support in linux is hampered by the fragmented hardware device drivers.

Has there been an active effort to avoid this curse on the RISCV ecosystem? At least for chips powerful enough to run linux.


(Paul Walmsley) #2

Hi. We’re definitely planning to maintain our current positive relationship with the upstream Linux maintainers (and with upstream open source projects in general). We’d also like to make sure that Linux supports as many RISC-V devices, emulators, and architectural features as makes sense. There’s no doubt that figuring out how to cleanly support the vast range of variation and capability in modern non-x86 hardware has been a challenge, historically, in the upstream Linux community. The good news is that many of the folks working on Linux RISC-V code have been through the ARM and ARM64 upstreaming process. We’re following those same guidelines and rules as we work on RISC-V Linux code. We’re also working on ways to take some of those lessons and work with hardware designers to minimize the load on the Linux upstreaming process.

These are still early days for RISC-V software support, but I think the outlook is very optimistic. And we hope you’ll get involved and contribute! Much of the initial Linux work was done on freely-available emulators, so there’s a very low barrier to joining in and helping us develop outstanding support - not only in Linux, but across the vast world of open-source software that folks would like to run on RISC-V and other open-hardware devices. Here’s hoping you’ll join in!