Use PAUSE instruction?

[birkenfeld]

In obtain_lock, would it make sense (on x86) to use the PAUSE instruction inside the loop? Looking at the Intel manuals, it seems to be designed for this kind of situation.

[mvdnes]

I am unsure what this instruction does. Can you provide a resource which has some details?

[birkenfeld]

(Note that I'm not at all an expert. Just trying to learn something.)

In Linux, the cpu_relax function used in spinlocks calls rep; nop which is the same as PAUSE and says it's "good thing to insert into busy-wait loops".

This quote is from the Intel x86 developer's manual:

Improves the performance of spin-wait loops. When executing a “spin-wait loop,” processors will suffer a severe performance penalty when exiting the loop because it detects a possible memory order violation. The PAUSE instruction provides a hint to the processor that the code sequence is a spin-wait loop. The processor uses this hint to avoid the memory order violation in most situations, which greatly improves processor performance. For this reason, it is recommended that a PAUSE instruction be placed in all spin-wait loops.
An additional function of the PAUSE instruction is to reduce the power consumed by a processor while executing a spin loop. A processor can execute a spin-wait loop extremely quickly, causing the processor to consume a lot of power while it waits for the resource it is spinning on to become available. Inserting a pause instruction in a spin-wait loop greatly reduces the processor’s power consumption.

[mvdnes]

Hmm seems like a good idea then. If you or anyone else would submit a pull request I would be happy to merge it.

[mvdnes]

Thanks! 👍

[mstewartgallus]

I'm thinking of putting this functionality in the std lib itself rust-lang/rust#40537