Re: __atomic_test_and_set() and mips o32 - help wanted

On Thu, 20 Nov 2025, Jason Thorpe wrote:

> > Interesting, thanks, I didn't know. 
> 
> Here’s a link to a copy of the paper:
> 
>     https://www.cs.tufts.edu/~nr/cs257/archive/john-ellis/mutex.pdf

 Thank you!

 I've skimmed over the paper and while surely feasible this user-mode code 
proposed is inherently not MP-safe.  So while acceptable on a proprietary 
or dedicated target that is known not to ever run a given compilation of 
user software on MP, on a general-purpose OS it seems to me feasible only 
by using a vDSO to export a standard set of atomic primitives implemented 
according to actual hardware configuration being run on.

 Well, why not; go for it if you have the resources, it seems to me like 
an interesting project!

  Maciej