Re: [PATCH v4 1/1] cpuset: relax the overlap check for cgroup-v2

Hi, Longman,

On Tue, 18 Nov 2025 14:53:27 -0500, Longman wrote:

On 11/16/25 8:57 PM, Sun Shaojie wrote:

...

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 52468d2c178a..0fd803612513 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -580,35 +580,56 @@ static inline bool cpusets_are_exclusive(struct cpuset *cs1, struct cpuset *cs2) /**

• cpus_excl_conflict - Check if two cpusets have exclusive CPU conflicts

• • @cs1: first cpuset to check
• • @cs2: second cpuset to check

• • @cs1: current cpuset to check
• • @cs2: cpuset involved in the check
  • Returns: true if CPU exclusivity conflict exists, false otherwise
  • Conflict detection rules:

• • 1. If either cpuset is CPU exclusive, they must be mutually exclusive

• • For cgroup-v1:

• • see cpuset1_cpus_excl_conflict()
    

• • For cgroup-v2:
• • 1. If cs1 is exclusive, cs1 and cs2 must be mutually exclusive
  • 2. exclusive_cpus masks cannot intersect between cpusets

• • 3. The allowed CPUs of one cpuset cannot be a subset of another's exclusive CPUs

• • 3. If cs2 is exclusive, cs2's allowed CPUs cannot be a subset of cs1's exclusive CPUs
• • 4. if cs1 and cs2 are not exclusive, the allowed CPUs of one cpuset cannot be a subset
• • of another's exclusive CPUs
  */ static inline bool cpus_excl_conflict(struct cpuset *cs1, struct cpuset *cs2)

As cs1 and cs2 is going to be handled differently, their current naming will make it hard to understand why they are treated differently. I will recommended changing the parameter name to "trial, sibling" as the caller call it with "cpus_excl_conflict(trial, c)" where trial is the new cpuset data to be tested and sibling is one of its sibling cpusets. It has to be clearly document what each parameter is for and the fact that swapping the parameters will cause it to return incorrect result.

...

{

• /* If either cpuset is exclusive, check if they are mutually exclusive */
• if (is_cpu_exclusive(cs1) || is_cpu_exclusive(cs2))

• /* For cgroup-v1 */
• if (!cpuset_v2())

• return cpuset1_cpus_excl_conflict(cs1, cs2);
  

• /* If cs1 are exclusive, check if they are mutually exclusive */
• if (is_cpu_exclusive(cs1)) return !cpusets_are_exclusive(cs1, cs2);

Code change like the following can eliminate the need to introduce a new cpuset1_cpus_excl_conflict() helper.

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index ec8bebc66469..201c70fb7401 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -599,9 +599,15 @@ static inline bool cpusets_are_exclusive(struct cpuset *cs1, struct cpuset *cs2) */ static inline bool cpus_excl_conflict(struct cpuset *cs1, struct cpuset *cs2) {

•   /* If either cpuset is exclusive, check if they are mutually 
  

exclusive */

•   if (is_cpu_exclusive(cs1) || is_cpu_exclusive(cs2))
  

•           return !cpusets_are_exclusive(cs1, cs2);
  

•   /*
  

•    * If trial is exclusive or sibling is exclusive & in v1,
  

•    * check if they are mutually exclusive
  

•    */
  

•   if (is_cpu_exclusive(trial) || (!cpuset_v2() && 
  

is_cpu_exclusive(sibling)))

•           return !cpusets_are_exclusive(trial, sibling);
  

•   if (!cpuset_v2())
  

•           return false;   /* The checking below is irrelevant to 
  

cpuset v1 */

    /* Exclusive_cpus cannot intersect */
    if (cpumask_intersects(cs1->exclusive_cpus, cs2->exclusive_cpus))

Thank you very much for your guidance and suggestions on the code.

I've updated patch v5 with some new ideas and look forward to your feedback.

patch v5 : https://lore.kernel.org/cgroups/20251119105749.1385946-1-sunshaojie@kylinos....

Thanks, Sun Shaojie