From: Guo Ren <guoren@xxxxxxxxxxxxxxxxx> Optimize arch_xchg|cmpxchg|cmpxchg_local with ASM acquire|release instructions instead of previous C based. The generic atomic.h used cmpxchg to implement the atomic operations, it will cause daul loop to reduce the forward guarantee. The patch implement csky custom atomic operations with ldex/stex instructions for the best performance. Changes in V4: - Remove RELEASE_FENCE in ldex/stex loop by Buoqun's advice Changes in V3: - Add arch_atomic_(fetch_add_unless, inc_unless_negative, dec_unless_positive, dec_if_positive) Changes in V2: - Fixup use of acquire + release for barrier semantics by Rutland. Guo Ren (3): csky: atomic: Optimize cmpxchg with acquire & release csky: atomic: Add custom atomic.h implementation csky: atomic: Add conditional atomic operations' optimization arch/csky/include/asm/atomic.h | 237 ++++++++++++++++++++++++++++++++ arch/csky/include/asm/barrier.h | 11 +- arch/csky/include/asm/cmpxchg.h | 64 ++++++++- 3 files changed, 304 insertions(+), 8 deletions(-) create mode 100644 arch/csky/include/asm/atomic.h -- 2.25.1
