On 2025/11/14 14:33, Sun Shaojie wrote:
> On 2025/11/14 08:50, Chen Ridong Wrote:
>> On 2025/11/13 21:14, Sun Shaojie wrote:
>>> ...
>>> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
>>> index 52468d2c178a..3240b3ab5998 100644
>>> --- a/kernel/cgroup/cpuset.c
>>> +++ b/kernel/cgroup/cpuset.c
>>> @@ -592,8 +592,13 @@ 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))
>>> + /* If both cpusets are exclusive, check if they are mutually exclusive
>>> */
>>> + if (is_cpu_exclusive(cs1) && is_cpu_exclusive(cs2))
>>> + return !cpusets_are_exclusive(cs1, cs2);
>>> +
>>> + /* In cgroup-v1, if either cpuset is exclusive, check if they are
>>> mutually exclusive */
>>> + if (!is_in_v2_mode() &&
>>> + (is_cpu_exclusive(cs1) != is_cpu_exclusive(cs2)))
>>> return !cpusets_are_exclusive(cs1, cs2);
>>>
>>
>> I prefer adding a helper function in the cpuset-v1.c file, similar to
>> cpus_excl_conflict_legacy().
>>
>> For cpuset v1, it can simply return cpus_excl_conflict_legacy(). It seems
>> that other rules are not
>> relevant to v1.
>>
>>> /* Exclusive_cpus cannot intersect */
>
> Hi, Ridong,
>
> Thank you for the suggestion.I will update the patch accordingly.
>
If we are ready to relax this rule, adding the v1 logic in
cpuset1_validate_change might be
appropriate. However, as I mentioned in my reply to Michal, I believe further
discussion is needed.
--
Best regards,
Ridong
