The latest development version of this page may be more current than this released 3.7.0 version.

Barriers API

group barrier_apis

Since

3.4

Version

0.1.0

Functions

ALWAYS_INLINE static void barrier_dmem_fence_full(void)

Full/sequentially-consistent data memory barrier.

This routine acts as a synchronization fence between threads and prevents re-ordering of data accesses instructions across the barrier instruction.

ALWAYS_INLINE static void barrier_dsync_fence_full(void)

Full/sequentially-consistent data synchronization barrier.

This routine acts as a synchronization fence between threads and prevents re-ordering of data accesses instructions across the barrier instruction like barrier_dmem_fence_full(), but has the additional effect of blocking execution of any further instructions, not just loads or stores, or both, until synchronization is complete.

Note

When not supported by hardware or architecture, this instruction falls back to a full/sequentially-consistent data memory barrier.

ALWAYS_INLINE static void barrier_isync_fence_full(void)

Full/sequentially-consistent instruction synchronization barrier.

This routine is used to guarantee that any subsequent instructions are fetched and to ensure any previously executed context-changing operations, such as writes to system control registers, have completed by the time the routine completes. In hardware terms, this might mean that the instruction pipeline is flushed, for example.

Note

When not supported by hardware or architecture, this instruction falls back to a compiler barrier.