Re: [PATCH next v2 3/5] locking/osq_lock: Use node->prev_cpu instead of saving node->prev.

[Waiman Long <longman@redhat.com>]

On 12/31/23 16:54, David Laight wrote:
> node->prev is only used to update 'prev' in the unlikely case
> of concurrent unqueues.
> This can be replaced by a check for node->prev_cpu changing
> and then calling decode_cpu() to get the changed 'prev' pointer.
>
> node->cpu (or more particularly) prev->cpu is only used for the
> osq_wait_next() call in the unqueue path.
> Normally this is exactly the value that the initial xchg() read
> from lock->tail (used to obtain 'prev'), but can get updated
> by concurrent unqueues.
>
> Both the 'prev' and 'cpu' members of optimistic_spin_node are
> now unused and can be deleted.
>
> Signed-off-by: David Laight <david.laight@aculab.com>
> ---
>   kernel/locking/osq_lock.c | 31 +++++++++++++++++--------------
>   1 file changed, 17 insertions(+), 14 deletions(-)
>
> diff --git a/kernel/locking/osq_lock.c b/kernel/locking/osq_lock.c
> index eb8a6dfdb79d..27324b509f68 100644
> --- a/kernel/locking/osq_lock.c
> +++ b/kernel/locking/osq_lock.c
> @@ -13,9 +13,8 @@
>    */
>   
>   struct optimistic_spin_node {
> -	struct optimistic_spin_node *next, *prev;
> +	struct optimistic_spin_node *next;
>   	int locked;    /* 1 if lock acquired */
> -	int cpu;       /* encoded CPU # + 1 value */
>   	int prev_cpu;  /* encoded CPU # + 1 value */
>   };
>   
> @@ -91,10 +90,9 @@ bool osq_lock(struct optimistic_spin_queue *lock)
>   	struct optimistic_spin_node *node = this_cpu_ptr(&osq_node);
>   	struct optimistic_spin_node *prev, *next;
>   	int curr = encode_cpu(smp_processor_id());
> -	int old;
> +	int prev_cpu;
>   
>   	node->next = NULL;
> -	node->cpu = curr;
>   
>   	/*
>   	 * We need both ACQUIRE (pairs with corresponding RELEASE in
> @@ -102,13 +100,12 @@ bool osq_lock(struct optimistic_spin_queue *lock)
>   	 * the node fields we just initialised) semantics when updating
>   	 * the lock tail.
>   	 */
> -	old = atomic_xchg(&lock->tail, curr);
> -	if (old == OSQ_UNLOCKED_VAL)
> +	prev_cpu = atomic_xchg(&lock->tail, curr);
> +	if (prev_cpu == OSQ_UNLOCKED_VAL)
>   		return true;
>   
> -	node->prev_cpu = old;
> -	prev = decode_cpu(old);
> -	node->prev = prev;
> +	node->prev_cpu = prev_cpu;
> +	prev = decode_cpu(prev_cpu);
>   	node->locked = 0;
>   
>   	/*
> @@ -174,9 +171,16 @@ bool osq_lock(struct optimistic_spin_queue *lock)
>   
>   		/*
>   		 * Or we race against a concurrent unqueue()'s step-B, in which
> -		 * case its step-C will write us a new @node->prev pointer.
> +		 * case its step-C will write us a new @node->prev_cpu value.
>   		 */
> -		prev = READ_ONCE(node->prev);
> +		{
> +			int new_prev_cpu = READ_ONCE(node->prev_cpu);
> +
> +			if (new_prev_cpu == prev_cpu)
> +				continue;
> +			prev_cpu = new_prev_cpu;
> +			prev = decode_cpu(prev_cpu);
> +		}

Just a minor nit. It is not that common in the kernel to add another 
nesting level just to reduce the scope of  new_prev_cpu auto variable.

Anyway,

Reviewed-by: Waiman Long <longman@redhat.com>