Re: [RFC PATCH] mm/damon: remove damon_lock

[SeongJae Park <sj@kernel.org>]

Thank you for this patch, Alex!

On Wed, 10 Nov 2021 19:47:21 +0800 alexs@kernel.org wrote:

> From: Alex Shi <alexs@kernel.org>
> 
> Variable nr_running_ctxs guards by damon_lock, but a lock for a int
> variable seems a bit heavy, a atomic_t is enough.

The lock is not only for protecting nr_running_ctxs, but also for avoiding
different users concurrently executing damon_start(), because that could allow
the users interfering others.

> 
> Signed-off-by: Alex Shi <alexs@kernel.org>
> Cc: SeongJae Park <sj@kernel.org> 
> Cc: Andrew Morton <akpm@linux-foundation.org> 
> Cc: linux-mm@kvack.org 
> Cc: linux-kernel@vger.kernel.org 
> ---
>  include/linux/damon.h |  1 -
>  mm/damon/core.c       | 31 +++++--------------------------
>  mm/damon/dbgfs.c      |  8 +++++---
>  3 files changed, 10 insertions(+), 30 deletions(-)
> 
> diff --git a/include/linux/damon.h b/include/linux/damon.h
> index b4d4be3cc987..e5dcc6336ef2 100644
> --- a/include/linux/damon.h
> +++ b/include/linux/damon.h
> @@ -453,7 +453,6 @@ int damon_set_attrs(struct damon_ctx *ctx, unsigned long sample_int,
>  		unsigned long min_nr_reg, unsigned long max_nr_reg);
>  int damon_set_schemes(struct damon_ctx *ctx,
>  			struct damos **schemes, ssize_t nr_schemes);
> -int damon_nr_running_ctxs(void);
>  
>  int damon_start(struct damon_ctx **ctxs, int nr_ctxs);
>  int damon_stop(struct damon_ctx **ctxs, int nr_ctxs);
> diff --git a/mm/damon/core.c b/mm/damon/core.c
> index c381b3c525d0..e821e36d5c10 100644
> --- a/mm/damon/core.c
> +++ b/mm/damon/core.c
[...]
> @@ -437,19 +422,15 @@ int damon_start(struct damon_ctx **ctxs, int nr_ctxs)
>  	int i;
>  	int err = 0;
>  
> -	mutex_lock(&damon_lock);
> -	if (nr_running_ctxs) {
> -		mutex_unlock(&damon_lock);
> +	if (atomic_read(&nr_running_ctxs))
>  		return -EBUSY;
> -	}
>  
>  	for (i = 0; i < nr_ctxs; i++) {
>  		err = __damon_start(ctxs[i]);
>  		if (err)
>  			break;
> -		nr_running_ctxs++;
> +		atomic_inc(&nr_running_ctxs);
>  	}
> -	mutex_unlock(&damon_lock);
>  
>  	return err;
>  }

This would let multiple concurrent threads seeing nr_running_ctxs of zero and
therefore proceed together.


Thanks,
SJ