Reviewing Valentin's patchset made me stare at some memory barriers
on the way. Here is some removal proposal. Some may be beneficial on
runtime (fqs snapshot with potentially as many smp_mb() as the number
of online CPUs for each GP). Some happen on more rare path. In any
case they clarify code reviews so we don't stumble upon mysterious
barriers.
Thanks.
Frederic Weisbecker (6):
rcu: Remove full ordering on second EQS snapshot
rcu: Remove superfluous full memory barrier upon first EQS snapshot
rcu/exp: Remove superfluous full memory barrier upon first EQS
snapshot
rcu: Remove full memory barrier on boot time eqs sanity check
rcu: Remove full memory barrier on RCU stall printout
rcu/exp: Remove redundant full memory barrier at the end of GP
.../Tree-RCU-Memory-Ordering.rst | 6 +++---
kernel/rcu/tree.c | 21 +++++++------------
kernel/rcu/tree_exp.h | 16 +++++++++++---
kernel/rcu/tree_stall.h | 4 ++--
4 files changed, 26 insertions(+), 21 deletions(-)
--
2.44.0
