Re: [PATCH -next RFC 01/11] cpuset: move the root cpuset write check earlier

[Chen Ridong <chenridong@huaweicloud.com>]

On 2025/8/30 3:21, Waiman Long wrote:
> On 8/28/25 8:56 AM, Chen Ridong wrote:
>> From: Chen Ridong <chenridong@huawei.com>
>>
>> The 'cpus' or 'mems' lists of the top_cpuset cannot be modified.
>> This check can be moved before acquiring any locks as a common code
>> block to improve efficiency and maintainability.
>>
>> Signed-off-by: Chen Ridong <chenridong@huawei.com>
>> ---
>>   kernel/cgroup/cpuset.c | 22 +++++++++-------------
>>   1 file changed, 9 insertions(+), 13 deletions(-)
>>
>> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
>> index a78ccd11ce9b..7e6a40e361ea 100644
>> --- a/kernel/cgroup/cpuset.c
>> +++ b/kernel/cgroup/cpuset.c
>> @@ -337,6 +337,11 @@ static inline bool cpuset_v2(void)
>>           cgroup_subsys_on_dfl(cpuset_cgrp_subsys);
>>   }
>>   +static inline bool cpuset_is_root(struct cpuset *cs)
>> +{
>> +    return (cs == &top_cpuset);
>> +}
>> +
>>   /*
>>    * Cgroup v2 behavior is used on the "cpus" and "mems" control files when
>>    * on default hierarchy or when the cpuset_v2_mode flag is set by mounting
>> @@ -2334,10 +2339,6 @@ static int update_cpumask(struct cpuset *cs, struct cpuset *trialcs,
>>       bool force = false;
>>       int old_prs = cs->partition_root_state;
>>   -    /* top_cpuset.cpus_allowed tracks cpu_active_mask; it's read-only */
>> -    if (cs == &top_cpuset)
>> -        return -EACCES;
>> -
>>       /*
>>        * An empty cpus_allowed is ok only if the cpuset has no tasks.
>>        * Since cpulist_parse() fails on an empty mask, we special case
>> @@ -2783,15 +2784,6 @@ static int update_nodemask(struct cpuset *cs, struct cpuset *trialcs,
>>   {
>>       int retval;
>>   -    /*
>> -     * top_cpuset.mems_allowed tracks node_stats[N_MEMORY];
>> -     * it's read-only
>> -     */
>> -    if (cs == &top_cpuset) {
>> -        retval = -EACCES;
>> -        goto done;
>> -    }
>> -
>>       /*
>>        * An empty mems_allowed is ok iff there are no tasks in the cpuset.
>>        * Since nodelist_parse() fails on an empty mask, we special case
>> @@ -3257,6 +3249,10 @@ ssize_t cpuset_write_resmask(struct kernfs_open_file *of,
>>       struct cpuset *trialcs;
>>       int retval = -ENODEV;
>>   +    /* root is read-only */
>> +    if (cpuset_is_root(cs))
>> +        return -EACCES;
>> +
>>       buf = strstrip(buf);
>>       cpuset_full_lock();
>>       if (!is_cpuset_online(cs))
> 
> The (cs == &top_cpuset) check is pretty straight forward. So if the cpuset_is_root() helper is only
> used once, I don't think we need a new helper here.
> 
> Cheers,
> Longman

Thanks you Longman,

There are several places where cs == &top_cpuset could be replaced with the helper function
cpuset_is_root. However, since this series focuses on refactoring the CPU settings, I did not make
those replacements. Would it be acceptable to submit a separate patch that introduces this helper?

-- 
Best regards,
Ridong