Rust added core::sync::atomic::spin_loop_hint in 1.24. It would probably make sense to use that.

from spin-rs.

It appears @tmccombs's suggestion has been adopted in 481c913 (thanks!). This issue can be closed.

from spin-rs.

I don't know lot about other architectures, but it looks like Linux uses cpu_relax with ARM. Unlike me, they might know what they are doing.

Macros form https://android.googlesource.com/.../arm/include/asm/spinlock.h#67 look like this:

#define arch_spin_is_locked(x)		((x)->lock != 0)
#define arch_spin_unlock_wait(lock) \
    do { while (arch_spin_is_locked(lock)) cpu_relax(); } while (0)

from spin-rs.

This code snippet here and this ARM developer page suggest that asm!("yield" :::: "volatile"); would be the way to go on AArch64, i.e. 64-bit ARMv8. This read here might be of interest as well.

6. Conclusions

• We propose to add onSpinWait() intrinsic and SpinPause implementation in (Linux) AArch64 port with yield instruction.
• It is harmless for CPUs that implement the instruction as NOP.
• It should be benefitial for CPUs that have SMT or are able to partially shut down or pause for a while.
• It may be considered to emit several nop instead of yield where latter works as nop to decrease memory pressure without loosing throughput.
  • A concrete latency estimate for a potential implementation that should not cause any performance degradation for Java is around 16 cycles.

from spin-rs.