40 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3========================4membarrier() System Call5========================6 7MEMBARRIER_CMD_{PRIVATE,GLOBAL}_EXPEDITED - Architecture requirements8=====================================================================9 10Memory barriers before updating rq->curr11----------------------------------------12 13The commands MEMBARRIER_CMD_PRIVATE_EXPEDITED and MEMBARRIER_CMD_GLOBAL_EXPEDITED14require each architecture to have a full memory barrier after coming from15user-space, before updating rq->curr. This barrier is implied by the sequence16rq_lock(); smp_mb__after_spinlock() in __schedule(). The barrier matches a full17barrier in the proximity of the membarrier system call exit, cf.18membarrier_{private,global}_expedited().19 20Memory barriers after updating rq->curr21---------------------------------------22 23The commands MEMBARRIER_CMD_PRIVATE_EXPEDITED and MEMBARRIER_CMD_GLOBAL_EXPEDITED24require each architecture to have a full memory barrier after updating rq->curr,25before returning to user-space. The schemes providing this barrier on the various26architectures are as follows.27 28 - alpha, arc, arm, hexagon, mips rely on the full barrier implied by29 spin_unlock() in finish_lock_switch().30 31 - arm64 relies on the full barrier implied by switch_to().32 33 - powerpc, riscv, s390, sparc, x86 rely on the full barrier implied by34 switch_mm(), if mm is not NULL; they rely on the full barrier implied35 by mmdrop(), otherwise. On powerpc and riscv, switch_mm() relies on36 membarrier_arch_switch_mm().37 38The barrier matches a full barrier in the proximity of the membarrier system call39entry, cf. membarrier_{private,global}_expedited().40