CheriBSD
cheribsd.org
cheribsd.org
Solaris has been doing this since 2016, with SPARC ADI.
[1] https://www.youtube.com/watch?v=ptSUKbKe4c4
[2] https://media.ccc.de/v/31c3_-_6574_-_en_-_saal_1_-_201412301...
[3] https://media.ccc.de/v/35c3-9647-taming_the_chaos_can_we_bui...
It is a hardware solution for software problem. Pretty much like a lot of things in hardware just to make software faster or easier.
From [1] "The Morello program started in October 2019, and will span a 5-year period."
If that’s what you intended to point out, I am sorry.
I don't know how any of your links really let us weigh ARM's commitment to CHERI going forward, which is the point of discussion. Indeed, none of those 3 links has anything to do with ARM at all, other than the first mentioning Morello.
The comment you responded to is this:
> It's not clear to me if ARM (the company) is actually interested in pursuing Cheri
Sigh.
Please don't waste more of your precious time replying to this.
It is a super interesting project though. It would be great if the hardware performance overheads were reduced so much that it became a mainstream ARM target.
We are only looking at sub 5% performance hit on an optimised ARM design. I don't see that really as performance overheads.
For many safety concepts in software development, the problem isn't technical.
And since many devs are like St. Thomas, where only if you build X with Y to prove Z, while achieving Webscale delivery, they will "maybe" adopt the technology.
Thus failing the money, and with such sceptic attitude, many good technologies fail to gain adoption.
Any strategy for improving software resilience has to account for the scale and pace of software development, as well as accommodating the skill gradient and surrounding social factors. FV as currently practiced fails at all of this.
A morello-like Apple Silicon implementation with software support in macOS might not be as blindingly obvious to end users in terms of superior behavior, but it could potentially serve as a compelling proof-of-concept example to competing vendors, demonstrating the practicality of hardware memory safety features for mass market devices.