On Tue, Feb 24, 2026 at 11:17:12PM +0800, Kunwu Chan wrote:
> Replace the incomplete phrase "all other RCU" with the more precise
> "all other RCU-deferred objects", clarifying that what gets blocked
I made this be "all other RCU-protected objects", but otherwise took
all five patches as-is.
Good eyes, and thank you very much!
Thanx, Paul
> from being freed are objects awaiting RCU grace-period reclamation,
> not RCU itself.
>
> Signed-off-by: Kunwu Chan <[email protected]>
> ---
> defer/whichtochoose.tex | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/defer/whichtochoose.tex b/defer/whichtochoose.tex
> index 139c9636..ff4545bc 100644
> --- a/defer/whichtochoose.tex
> +++ b/defer/whichtochoose.tex
> @@ -118,8 +118,8 @@ The ``Duration of Protection'' row describes constraints
> (if any) on how
> long a period of time a user may protect a given object.
> Reference counting and hazard pointers can both protect objects for
> extended time periods with no untoward side effects, but
> -maintaining an RCU reference to even one object prevents all other RCU
> -from being freed.
> +maintaining an RCU reference to even one object prevents all other
> +RCU-deferred objects from being freed.
> RCU readers must therefore be relatively short in order to avoid running
> the system out of memory, with special-purpose implementations such
> as SRCU, Tasks RCU, and Tasks Trace RCU being exceptions to this rule.
> --
> 2.25.1
>
