Hello milkv community
The idea of making a risc-v desktop was revolutionary. Iam talking about milk-v pioneer and the endless possibilities to experiment. As a freebsd fanboy i came across this distribution CheriBSD and i was wondering if anybody manage to boot or experiment with it, as it looks a fully capable wayland+KDE enviroment… Thank you
The answer is no, cheri is a hardware architecture extension, extra bits for access management.
Harvard just released their cheri risc-v document last year, so nobody implemented any cheri in RISC-V except for maybe in an FPGA to test
As bruce_loco said, it is not possible because the pioneer’s sg2042 does not implement the cheri extensions. Codasip is the only one currently providing a commercially available implementation RISC-V CHERI security - Codasip but that’s not for end-users.
Even if you could get it, I doubt it would run CheriBSD, let alone a desktop environment on top. RiscV+CHERI chips will not be suitable for desktops for at least a couple of years. I would first expect custom designs to be used in IoT, automotive, and the likes. Probably also not with CheriBSD but something like CheriFreeRTOS. Then maybe servers based on CheriBSD, which could also run desktop environments.
Which document are you referring to?
Capability Hardware
Enhanced RISC Instructions:
CHERI Instruction-Set Architecture
(Version 9)
https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-987.pdf
There are some studies that reviewed porting efforts for bsd, linux, some apps like QT, mail, etc
Depending on the stringency of the compiler flags, it makes it problematic to do even the basic cheri protections.
A simple build run will display a bulk load of programs. Some programs can run in native cheri, or oblivious to cheri, as long as the OS is wise. However then some program functionality might fail and the OS will cause the program to crash.
What will happen is people will do BSD and Linux in CHERI QEMU for Arm for example and then the portability should be extendable across other architectures(from a source build perspective).