Re: [PATCH 3/3] rcu: Comment on the extraneous delta test on rcu_seq_done_exact()
From: Paul E. McKenney
Date: Wed Mar 26 2025 - 18:37:53 EST
On Mon, Mar 24, 2025 at 01:01:55PM -0400, Joel Fernandes wrote:
> From: Frederic Weisbecker <frederic@xxxxxxxxxx>
>
> The numbers used in rcu_seq_done_exact() lack some explanation behind
> their magic. Especially after the commit:
>
> 85aad7cc4178 ("rcu: Fix get_state_synchronize_rcu_full() GP-start detection")
>
> which reported a subtle issue where a new GP sequence snapshot was taken
> on the root node state while a grace period had already been started and
> reflected on the global state sequence but not yet on the root node
> sequence, making a polling user waiting on a wrong already started grace
> period that would ignore freshly online CPUs.
>
> The fix involved taking the snaphot on the global state sequence and
> waiting on the root node sequence. And since a grace period is first
> started on the global state and only afterward reflected on the root
> node, a snapshot taken on the global state sequence might be two full
> grace periods ahead of the root node as in the following example:
>
> rnp->gp_seq = rcu_state.gp_seq = 0
>
> CPU 0 CPU 1
> ----- -----
> // rcu_state.gp_seq = 1
> rcu_seq_start(&rcu_state.gp_seq)
> // snap = 8
> snap = rcu_seq_snap(&rcu_state.gp_seq)
> // Two full GP differences
> rcu_seq_done_exact(&rnp->gp_seq, snap)
> // rnp->gp_seq = 1
> WRITE_ONCE(rnp->gp_seq, rcu_state.gp_seq);
>
> Add a comment about those expectations and to clarify the magic within
> the relevant function.
>
> Note that the issue arises mainly with the use of rcu_seq_done_exact()
> which has a much tigher guardband (of 2 GPs) to ensure the false-negative
> window of the API during wraparound is limited to just 2 GPs.
> rcu_seq_done() does not have such strict requirements, however its large
> false-negative window of ULONG_MAX/2 is not ideal for the polling API.
> However, this also means care is needed to ensure the guardband is as
> large as needed to avoid the example scenario describe above which a
> warning added in an earlier patch does.
>
> [ Comment wordsmithing by Joel ]
>
> Signed-off-by: Frederic Weisbecker <frederic@xxxxxxxxxx>
> Reviewed-by: Paul E. McKenney <paulmck@xxxxxxxxxx>
Looks good, and I stand by my Reviewed-by. ;-)
Thanx, Paul
> Signed-off-by: Joel Fernandes <joelagnelf@xxxxxxxxxx>
> ---
> kernel/rcu/rcu.h | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> diff --git a/kernel/rcu/rcu.h b/kernel/rcu/rcu.h
> index 5e1ee570bb27..db63f330768c 100644
> --- a/kernel/rcu/rcu.h
> +++ b/kernel/rcu/rcu.h
> @@ -160,6 +160,15 @@ static inline bool rcu_seq_done(unsigned long *sp, unsigned long s)
> * Given a snapshot from rcu_seq_snap(), determine whether or not a
> * full update-side operation has occurred, but do not allow the
> * (ULONG_MAX / 2) safety-factor/guard-band.
> + *
> + * The token returned by get_state_synchronize_rcu_full() is based on
> + * rcu_state.gp_seq but it is tested in poll_state_synchronize_rcu_full()
> + * against the root rnp->gp_seq. Since rcu_seq_start() is first called
> + * on rcu_state.gp_seq and only later reflected on the root rnp->gp_seq,
> + * it is possible that rcu_seq_snap(rcu_state.gp_seq) returns 2 full grace
> + * periods ahead of the root rnp->gp_seq. To prevent false-positives with the
> + * full polling API that a wrap around instantly completed the GP, when nothing
> + * like that happened, adjust for the 2 GPs in the ULONG_CMP_LT().
> */
> static inline bool rcu_seq_done_exact(unsigned long *sp, unsigned long s)
> {
> --
> 2.43.0
>