investigate wait-free IO buffer handling

[spacejam]

Reserving a slot in an IO buffer is way too contentious right now, and we're spinning so much that a mutex would likely outperform our current approach. This is a great opportunity to evaluate the feasibility of wait-free algorithms. We may be able to attempt the "fast-path" first with a normal lock-free algorithm, falling back on a bounded-time wait-free backup to get the best of both worlds. Maybe not :) Let's find out!

reading list:

• A Methodology for Creating Fast Wait-Free Data Structures
• A Practical Wait-Free Simulation for Lock-Free Data Structures
• Obstruction-Free Synchronization: Double-Ended Queues as an Example

[jeehoonkang]

I guess you may also be interested in A wait-free queue as fast as fetch-and-add :)

[jeehoonkang]

I guess you may also be interested in A wait-free queue as fast as fetch-and-add :)

[spacejam]

@jeehoonkang ahh very cool! I'll definitely check that out :)

[spacejam]

@jeehoonkang ahh very cool! I'll definitely check that out :)