Re: [patch 16/19] sched/mmcid: Provide CID ownership mode fixup functions

[Peter Zijlstra <peterz@infradead.org>]

On Wed, Oct 15, 2025 at 07:29:54PM +0200, Thomas Gleixner wrote:

> +static bool mm_cid_fixup_task_to_cpu(struct task_struct *t, struct mm_struct *mm)
> +{
> +	/* Remote access to mm::mm_cid::pcpu requires rq_lock */
> +	guard(task_rq_lock)(t);
> +	if (t->mm != mm)
> +		return false;

I'm slightly confused about this one; I'm assuming it is a double check
of mm for the VM_CLONE case below, once before lock once after. Except,
none of the scheduler locks actually serialize t->mm ...

IIRC the only way to actually change ->mm is exec(), and that is under
task_lock().

> +	if (cid_on_task(t->mm_cid.cid)) {
> +		/* If running on the CPU, transfer the CID, otherwise drop it */
> +		if (task_rq(t)->curr == t)
> +			mm_cid_transfer_to_cpu(t, per_cpu_ptr(mm->mm_cid.pcpu, task_cpu(t)));
> +		else
> +			mm_unset_cid_on_task(t);
> +	}
> +	return true;
> +}
> +
> +static void __maybe_unused mm_cid_fixup_tasks_to_cpus(void)
> +{
> +	struct mm_struct *mm = current->mm;
> +	struct task_struct *p, *t;
> +	unsigned int users;
> +
> +	/*
> +	 * This can obviously race with a concurrent affinity change, which
> +	 * increases the number of allowed CPUs for this mm, but that does
> +	 * not affect the mode and only changes the CID constraints. A
> +	 * possible switch back to per task mode happens either in the
> +	 * deferred handler function or in the next fork()/exit().
> +	 *
> +	 * The caller has already transferred. The newly incoming task is
> +	 * already accounted for, but not yet visible.
> +	 */
> +	users = mm->mm_cid.users - 2;
> +	if (!users)
> +		return;
> +
> +	guard(rcu)();
> +	for_other_threads(current, t) {
> +		mm_cid_fixup_task_to_cpu(t, mm);
> +		users--;
> +	}
> +
> +	if (!users)
> +		return;
> +
> +	/* Happens only for VM_CLONE processes. */

Right, sorry for reminding you about this :-)

> +	for_each_process_thread(p, t) {
> +		if (t == current || t->mm != mm)
> +			continue;
> +		if (mm_cid_fixup_task_to_cpu(t, mm)) {
> +			if (--users == 0)
> +				return;
> +		}
> +	}
> +}