[PATCH] cpumask: Remove unnecessary cpumask_nth_andnot()

[PATCH] cpumask: Remove unnecessary cpumask_nth_andnot()

[Shaopeng Tan]

Commit 94f753143028("x86/resctrl: Optimize cpumask_any_housekeeping()")
switched the only user of cpumask_nth_andnot() to other cpumask functions,
but left the function cpumask_nth_andnot() unused, delete it.

Signed-off-by: Shaopeng Tan <tan.shaopeng@jp.fujitsu.com>
---
 include/linux/cpumask.h | 16 ----------------
 1 file changed, 16 deletions(-)

diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
index 7ae80a7ca81e..498790f74fa8 100644
--- a/include/linux/cpumask.h
+++ b/include/linux/cpumask.h
@@ -546,22 +546,6 @@ unsigned int cpumask_nth_and(unsigned int cpu, const struct cpumask *srcp1,
 				small_cpumask_bits, cpumask_check(cpu));
 }
 
-/**
- * cpumask_nth_andnot - get the Nth cpu set in 1st cpumask, and clear in 2nd.
- * @srcp1: the cpumask pointer
- * @srcp2: the cpumask pointer
- * @cpu: the Nth cpu to find, starting from 0
- *
- * Return: >= nr_cpu_ids if such cpu doesn't exist.
- */
-static __always_inline
-unsigned int cpumask_nth_andnot(unsigned int cpu, const struct cpumask *srcp1,
-							const struct cpumask *srcp2)
-{
-	return find_nth_andnot_bit(cpumask_bits(srcp1), cpumask_bits(srcp2),
-				small_cpumask_bits, cpumask_check(cpu));
-}
-
 /**
  * cpumask_nth_and_andnot - get the Nth cpu set in 1st and 2nd cpumask, and clear in 3rd.
  * @srcp1: the cpumask pointer
-- 
2.43.5

[PATCH] fs/resctrl: Optimize code in rdt_get_tree()

[Shaopeng Tan]

schemata_list_destroy() has to be called if schemata_list_create() fails.

rdt_get_tree() calls schemata_list_destroy() in two different ways: 
directly if schemata_list_create() itself fails and 
on the exit path via the out_schemata_free goto label.

Remove schemata_list_destroy() call on schemata_list_create() failure.
Use existing out_schemata_free goto label instead.

Signed-off-by: Shaopeng Tan <tan.shaopeng@jp.fujitsu.com>
---
 fs/resctrl/rdtgroup.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/fs/resctrl/rdtgroup.c b/fs/resctrl/rdtgroup.c
index 1beb124e25f6..592d4f69fce9 100644
--- a/fs/resctrl/rdtgroup.c
+++ b/fs/resctrl/rdtgroup.c
@@ -2608,10 +2608,8 @@ static int rdt_get_tree(struct fs_context *fc)
 		goto out_root;
 
 	ret = schemata_list_create();
-	if (ret) {
-		schemata_list_destroy();
-		goto out_ctx;
-	}
+	if (ret)
+		goto out_schemata_free;
 
 	ret = closid_init();
 	if (ret)
@@ -2683,7 +2681,6 @@ static int rdt_get_tree(struct fs_context *fc)
 	closid_exit();
 out_schemata_free:
 	schemata_list_destroy();
-out_ctx:
 	rdt_disable_ctx();
 out_root:
 	rdtgroup_destroy_root();
-- 
2.43.5

Re: [PATCH] fs/resctrl: Optimize code in rdt_get_tree()

[James Morse]

Hello!

On 13/06/2025 08:37, Shaopeng Tan wrote:
> schemata_list_destroy() has to be called if schemata_list_create() fails.
> 
> rdt_get_tree() calls schemata_list_destroy() in two different ways: 
> directly if schemata_list_create() itself fails and 
> on the exit path via the out_schemata_free goto label.
> 
> Remove schemata_list_destroy() call on schemata_list_create() failure.
> Use existing out_schemata_free goto label instead.

Reviewed-by: James Morse <james.morse@arm.com>


Thanks,

James

Re: [PATCH] fs/resctrl: Optimize code in rdt_get_tree()

[Reinette Chatre]

Hi Shaopeng,

On 6/13/25 12:37 AM, Shaopeng Tan wrote:
> schemata_list_destroy() has to be called if schemata_list_create() fails.
> 
> rdt_get_tree() calls schemata_list_destroy() in two different ways: 
> directly if schemata_list_create() itself fails and 
> on the exit path via the out_schemata_free goto label.
> 
> Remove schemata_list_destroy() call on schemata_list_create() failure.
> Use existing out_schemata_free goto label instead.
> 
> Signed-off-by: Shaopeng Tan <tan.shaopeng@jp.fujitsu.com>
> ---
>  fs/resctrl/rdtgroup.c | 7 ++-----
>  1 file changed, 2 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/resctrl/rdtgroup.c b/fs/resctrl/rdtgroup.c
> index 1beb124e25f6..592d4f69fce9 100644
> --- a/fs/resctrl/rdtgroup.c
> +++ b/fs/resctrl/rdtgroup.c
> @@ -2608,10 +2608,8 @@ static int rdt_get_tree(struct fs_context *fc)
>  		goto out_root;
>  
>  	ret = schemata_list_create();
> -	if (ret) {
> -		schemata_list_destroy();
> -		goto out_ctx;
> -	}
> +	if (ret)
> +		goto out_schemata_free;
>  
>  	ret = closid_init();
>  	if (ret)
> @@ -2683,7 +2681,6 @@ static int rdt_get_tree(struct fs_context *fc)
>  	closid_exit();
>  out_schemata_free:
>  	schemata_list_destroy();
> -out_ctx:
>  	rdt_disable_ctx();
>  out_root:
>  	rdtgroup_destroy_root();

Could you please resubmit this independently? That is, not in an
email thread with the cpumask patch?

With that:
| Reviewed-by: Reinette Chatre <reinette.chatre@intel.com>

Thank you very much.

Reinette

[PATCH] fs/resctrl: Optimize code in rdt_get_tree()

[Shaopeng Tan]

schemata_list_destroy() has to be called if schemata_list_create() fails.

rdt_get_tree() calls schemata_list_destroy() in two different ways:
directly if schemata_list_create() itself fails and
on the exit path via the out_schemata_free goto label.

Remove schemata_list_destroy() call on schemata_list_create() failure.
Use existing out_schemata_free goto label instead.

Signed-off-by: Shaopeng Tan <tan.shaopeng@jp.fujitsu.com>
Reviewed-by: Reinette Chatre <reinette.chatre@intel.com>
Reviewed-by: James Morse <james.morse@arm.com>
---
 fs/resctrl/rdtgroup.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/fs/resctrl/rdtgroup.c b/fs/resctrl/rdtgroup.c
index 1beb124e25f6..592d4f69fce9 100644
--- a/fs/resctrl/rdtgroup.c
+++ b/fs/resctrl/rdtgroup.c
@@ -2608,10 +2608,8 @@ static int rdt_get_tree(struct fs_context *fc)
 		goto out_root;
 
 	ret = schemata_list_create();
-	if (ret) {
-		schemata_list_destroy();
-		goto out_ctx;
-	}
+	if (ret)
+		goto out_schemata_free;
 
 	ret = closid_init();
 	if (ret)
@@ -2683,7 +2681,6 @@ static int rdt_get_tree(struct fs_context *fc)
 	closid_exit();
 out_schemata_free:
 	schemata_list_destroy();
-out_ctx:
 	rdt_disable_ctx();
 out_root:
 	rdtgroup_destroy_root();
-- 
2.43.5

Re: [PATCH] cpumask: Remove unnecessary cpumask_nth_andnot()

[Yury Norov]

On Fri, Jun 13, 2025 at 04:37:30PM +0900, Shaopeng Tan wrote:
> Commit 94f753143028("x86/resctrl: Optimize cpumask_any_housekeeping()")
> switched the only user of cpumask_nth_andnot() to other cpumask functions,
> but left the function cpumask_nth_andnot() unused, delete it.
> 
> Signed-off-by: Shaopeng Tan <tan.shaopeng@jp.fujitsu.com>

This makes find_nth_andnot_bit() unused as well. Can you send a v2
that removes both? Also, the following patch in the series doesn't
look relevant. Can you send it separately?

Thanks,
Yury

> ---
>  include/linux/cpumask.h | 16 ----------------
>  1 file changed, 16 deletions(-)
> 
> diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
> index 7ae80a7ca81e..498790f74fa8 100644
> --- a/include/linux/cpumask.h
> +++ b/include/linux/cpumask.h
> @@ -546,22 +546,6 @@ unsigned int cpumask_nth_and(unsigned int cpu, const struct cpumask *srcp1,
>  				small_cpumask_bits, cpumask_check(cpu));
>  }
>  
> -/**
> - * cpumask_nth_andnot - get the Nth cpu set in 1st cpumask, and clear in 2nd.
> - * @srcp1: the cpumask pointer
> - * @srcp2: the cpumask pointer
> - * @cpu: the Nth cpu to find, starting from 0
> - *
> - * Return: >= nr_cpu_ids if such cpu doesn't exist.
> - */
> -static __always_inline
> -unsigned int cpumask_nth_andnot(unsigned int cpu, const struct cpumask *srcp1,
> -							const struct cpumask *srcp2)
> -{
> -	return find_nth_andnot_bit(cpumask_bits(srcp1), cpumask_bits(srcp2),
> -				small_cpumask_bits, cpumask_check(cpu));
> -}
> -
>  /**
>   * cpumask_nth_and_andnot - get the Nth cpu set in 1st and 2nd cpumask, and clear in 3rd.
>   * @srcp1: the cpumask pointer
> -- 
> 2.43.5