Re: [PATCH] rcu: Use per_cpu_ptr to get the pointer of per_cpu variable

[Neeraj Upadhyay <neeraju@codeaurora.org>]

On 6/30/2021 7:38 PM, Liu Song wrote:
> From: Liu Song <liu.song11@zte.com.cn>
> 
> In rcu, most codes have used per_cpu_ptr to obtain per_cpu variable
> pointers. This patch adjusts a few codes that have not done so.
> 

Minor: Can we also add below statement?
"This patch doesn't introduce any functional change."

> Signed-off-by: Liu Song <liu.song11@zte.com.cn>
> ---
>   kernel/rcu/tasks.h      | 2 +-
>   kernel/rcu/tree.c       | 4 ++--
>   kernel/rcu/tree_stall.h | 2 +-
>   3 files changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h
> index 350ebf5051f9..0275217cb6e4 100644
> --- a/kernel/rcu/tasks.h
> +++ b/kernel/rcu/tasks.h
> @@ -875,7 +875,7 @@ static void trc_read_check_handler(void *t_in)
>   	// Allow future IPIs to be sent on CPU and for task.
>   	// Also order this IPI handler against any later manipulations of
>   	// the intended task.
> -	smp_store_release(&per_cpu(trc_ipi_to_cpu, smp_processor_id()), false); // ^^^
> +	smp_store_release(per_cpu_ptr(&trc_ipi_to_cpu, smp_processor_id()), false); // ^^^
>   	smp_store_release(&texp->trc_ipi_to_cpu, -1); // ^^^
>   }
>   
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 8e78b2430c16..5cab8e39a731 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -399,7 +399,7 @@ bool rcu_eqs_special_set(int cpu)
>   	int old;
>   	int new;
>   	int new_old;
> -	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
> +	struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu);

rcu_eqs_special_set() does not exist in rcu dev branch [1]; so this 
might need rebase.


[1]
https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git/tree/kernel/rcu/tree.c?h=dev

>   
>   	new_old = atomic_read(&rdp->dynticks);
>   	do {
> @@ -1333,7 +1333,7 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp)
>   	 */
>   	jtsq = READ_ONCE(jiffies_to_sched_qs);
>   	ruqp = per_cpu_ptr(&rcu_data.rcu_urgent_qs, rdp->cpu);
> -	rnhqp = &per_cpu(rcu_data.rcu_need_heavy_qs, rdp->cpu);
> +	rnhqp = per_cpu_ptr(&rcu_data.rcu_need_heavy_qs, rdp->cpu);
>   	if (!READ_ONCE(*rnhqp) &&
>   	    (time_after(jiffies, rcu_state.gp_start + jtsq * 2) ||
>   	     time_after(jiffies, rcu_state.jiffies_resched) ||
> diff --git a/kernel/rcu/tree_stall.h b/kernel/rcu/tree_stall.h
> index 59b95cc5cbdf..990fdb793d57 100644
> --- a/kernel/rcu/tree_stall.h
> +++ b/kernel/rcu/tree_stall.h
> @@ -349,7 +349,7 @@ static void rcu_dump_cpu_stacks(void)
>   
>   static void print_cpu_stall_fast_no_hz(char *cp, int cpu)
>   {
> -	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
> +	struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu);
>   
>   	sprintf(cp, "last_accelerate: %04lx/%04lx dyntick_enabled: %d",
>   		rdp->last_accelerate & 0xffff, jiffies & 0xffff,
>

Reviewed-by: Neeraj Upadhyay <neeraju@codeaurora.org>



Thanks
Neeraj

-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a 
member of the Code Aurora Forum, hosted by The Linux Foundation