Re: [PATCH v4 06/20] sched/core: allow only preferred CPUs in is_cpu_allowed

[Shrikanth Hegde <sshegde@linux.ibm.com>]

Hi Prateek, Yury,

On 6/18/26 9:52 AM, Shrikanth Hegde wrote:
> Hi Prateek.
> 
> On 6/18/26 9:19 AM, K Prateek Nayak wrote:
>> Hello Shrikanth,
>>
>> On 6/17/2026 11:11 PM, Shrikanth Hegde wrote:
>>> --- a/include/linux/sched.h
>>> +++ b/include/linux/sched.h
>>> @@ -1657,6 +1657,7 @@ struct task_struct {
>>>   #ifdef CONFIG_UNWIND_USER
>>>       struct unwind_task_info        unwind_info;
>>>   #endif
>>> +    int                has_preferred_cpu_state;
>>
>> Since this only really needs 2 bits, perhaps you can use the "u8 __pad"
>> in the task struct? ...
>>
>>>       /* CPU-specific state of this task: */
>>>       struct thread_struct        thread;
>>
>> [...snip...]
>>
>>> @@ -3549,6 +3568,14 @@ static int select_fallback_rq(int cpu, struct 
>>> task_struct *p)
>>>       enum { cpuset, possible, fail } state = cpuset;
>>>       int dest_cpu;
>>> +    /*
>>> +     * Cache value whether task's affinity spans preferred CPUs.
>>> +     * This helps to avoid repeating the same for each CPU
>>> +     * later in the loop. Encode call to is_cpu_allowed coming
>>> +     * via select_fallback_rq.
>>> +     */
>>> +    p->has_preferred_cpu_state = task_has_preferred_cpus(p) << 8 | 0x1;
>>
>> ... Or maybe it can be an s8 and this indicator can be a tri-state like:
>>
>>   -1:  Cached; preferred CPUs exists
>>    0:  preferred CPUs not cached
>>    1:  Cached; preferred CPUs don't exist
>>
> 
> Yes. Current encoding has.
> 
> 1. call came from select_fallback_rq and preferred CPUs exists
> 2. call came from select_fallback_rq and preferred CPUs doesn't exists
> 3. Call didn't come from select_fallback_rq and evaluate the 
> cpumask_interesect.
> 
>> and then the comparison can simply be:
>>
>>      if (cached)
>>          return cached < 0;
>>
>> ppc64le and arm64 seems to generates slightly smaller code for "< 0"
>> check compared to the current scheme of (& + >>) in
>> task_has_preferred_cpus().
>>
> 
> Yes, this seems better. I will give it a try.

I guess below captures the discussion on this patch.
Other way is pass it as local variable to is_cpu_allowed.

---

diff --git a/include/linux/sched.h b/include/linux/sched.h
index e435f3073ffc..8c2fe953f19b 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1658,7 +1658,7 @@ struct task_struct {
  #ifdef CONFIG_UNWIND_USER
         struct unwind_task_info         unwind_info;
  #endif
-       int                             has_preferred_cpu_state;
+       s8                              has_preferred_cpu_state;
  
         /* CPU-specific state of this task: */
         struct thread_struct            thread;
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 9342aae315ca..26c82cd0bcec 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2504,7 +2504,7 @@ static inline bool rq_has_pinned_tasks(struct rq *rq)
   */
  static inline bool is_cpu_allowed(struct task_struct *p, int cpu)
  {
-       bool task_check_preferred_cpu = false;
+       bool task_check_preferred_cpu;
  
         /* When not in the task's cpumask, no point in looking further. */
         if (!task_allowed_on_cpu(p, cpu))
@@ -3573,12 +3573,12 @@ static int select_fallback_rq(int cpu, struct task_struct *p)
         int dest_cpu;
  
         /*
-        * Cache value whether task's affinity spans preferred CPUs.
+        * Cache the value whether task's affinity spans preferred CPUs.
          * This helps to avoid repeating the same for each CPU
          * later in the loop. Encode call to is_cpu_allowed coming
          * via select_fallback_rq.
          */
-       p->has_preferred_cpu_state = task_has_preferred_cpus(p) << 8 | 0x1;
+       p->has_preferred_cpu_state = task_has_preferred_cpus(p) ? 1 : -1;
  
         /*
          * If the node that the CPU is on has been offlined, cpu_to_node()
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index f3814099cc0b..da8134568229 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -4220,20 +4220,20 @@ DEFINE_CLASS_IS_UNCONDITIONAL(sched_change)
  #include "ext.h"
  
  /*
- * has_preferred_cpu_state is encoding two bits of information.
- * First Byte is to encode where the call to is_cpu_allowed coming from.
- * Second Byte is to encode the intersection of task affinity
- * and cpu_preferred_mask.
+ * has_preferred_cpu_state could have the value cached from
+ * select_fallback_rq. It is set/cleared while holding pi_lock
+ * and irq disabled.
   *
- * If 1st Byte is set, call to is_cpu_allowed coming from select_fallback_rq.
- * That helps to avoid repeated calculation keeping time complexity same.
+ *  1: Cached and preferred CPUs exists in task's affinity.
+ *  0: Not cached and need to evaluate.
+ * -1: Cached and preferred CPU doesn't exits task's affinity
   */
  static inline bool task_has_preferred_cpus(struct task_struct *p)
  {
-       int cached_value = p->has_preferred_cpu_state;
+       int cached = p->has_preferred_cpu_state;
  
-       if (cached_value & 0x1)
-               return p->has_preferred_cpu_state >> 8;
+       if (cached)
+               return cached > 0;
         else
                 return cpumask_intersects(p->cpus_ptr, cpu_preferred_mask);
  }