On Sat, Mar 07, 2020 at 02:02:25AM +0100, Thomas Gleixner wrote: > "Paul E. McKenney" <paulmck@xxxxxxxxxx> writes: > >> In #2c RCU is eventually not watching, but as that state cannot schedule > >> anyway there is no point to worry about it so it has to invoke > >> rcu_irq_enter() before running that code. This can be optimized, but this > >> will be done as an extra step in course of the entry code consolidation > >> work. > > > > In other words, any needed rcu_irq_enter() and rcu_irq_exit() are added > > in one of the entry-code consolidation patches, and patch below depends > > on that patch, correct? > > No. The patch itself is already correct when applied to mainline. It has > no dependencies. > > It invokes rcu_irq_enter()/exit() for the case (#2c) where it is > relevant. All other case are already RCU safe today. > > The fact that the invocation is misplaced is a different story and yes, > that is part of the entry code cleanup along with some optimization > which are possible once the entry voodoo is out of ASM and adjustable > for a particular entry point in C. The weekend clearly did not come a moment too soon for me, did it? :-/ Thanx, Paul
