Le Wed, Mar 27, 2024 at 12:29:02PM +0100, Valentin Schneider a écrit : > The symbols relating to the CT_STATE part of context_tracking.state are now > all prefixed with CT_STATE. > > The RCU dynticks counter part of that atomic variable still involves > symbols with different prefixes, align them all to be prefixed with > CT_DYNTICKS, as CT_DYNTICKS_MASK already is. > > Signed-off-by: Valentin Schneider <vschneid@xxxxxxxxxx> It used to be that RCU extended quiescent state and dynticks enter/exit were coupled. But this isn't the case anymore. Nowadays RCU stops watching some time later after dynticks is entered. I wonder if we shouldn't take advantage of that cleanup for a meaning that really reflect that RCU stops watching from there. Paul what do you think? CT_EQS_IDX ? CT_RCUEQS_IDX? CT_RCUOFF_IDX? ...? Thanks.
