Re: Wait-free Hazard Pointers Using Std Atomics

On 5/26/25 19:17, Chris M. Thomasson wrote:

On 5/26/2025 2:58 PM, jseigh wrote:

and asymmetric memory barriers of course.
>
https://threadnought.wordpress.com/2025/05/26/wait-free-hazard- pointers- using-std-atomics/
>
Looks like it runs faster without the conditional branch.
>
Not sure it's a new idea.  The only wait-free version of hazard pointers
I've seen so far involves a storage to storage move instruction which
might not be available on all platforms.

 Interesting, thanks Joe! Fwiw, I think, something like FlushProcessWriteBuffers() should be in a new C++ std? Perhaps, allow a system to use DWCAS without resorting to not lock-free wrt the checks? Damn it. ;^o Ouch.
 https://learn.microsoft.com/en-us/windows/win32/api/processthreadsapi/ nf-processthreadsapi-flushprocesswritebuffers
 Sometimes I think that function was specifically designed for exotic RCU, ect...

It's basically the equivalent of linux membarrier().  I don't know
why they just focus on writes, it's a full membar AFAIK.
As far as the wait-free hazard pointer goes, I though there would
be forward progress issues with the reader threads but it's not
an issue other schemes like URCU has with quiescent states.
Joe Seigh