Re: Use of newer locking primitives...

[Paul Durrant <xadimgnik@xxxxxxxxx>]

On 30/08/2022 12:35, Martin Harvey wrote:
> Hello folks.
>
> Has anyone taken it upon themselves to try using higher performance 
> locking primitives, (i.e. Queued spinlocks) in our drivers? Anyone see 
> any issues where we depend on the implementation of locking primitives, 
> or could we just do a “drop in” (i.e. search and replace) replacement?
We have various wrappers around locks in a lot of places so a drop-in 
replacement might be doable. I did think about using in-stack queued 
spinlocks at one point but the extra argument makes that a pain. Still, 
probably worth doing if anyone has the time and motivation.
  Paul