On Thu, Mar 28, 2024 at 03:53:30PM +0800, Abel Wu wrote:
> > +static int schedule_flush_migrate_mm(void)
> > +{
> > + struct callback_head *flush_cb;
> > +
> > + flush_cb = kzalloc(sizeof(*flush_cb), GFP_KERNEL);
> > + if (!flush_cb)
> > + return -ENOMEM;
> > +
> > + flush_cb->func = flush_migrate_mm_task_workfn;
> > + if (task_work_add(current, flush_cb, TWA_RESUME))
> > + kfree(flush_cb);
>
> It seems we will lose track of flush_cb and causes memleak here. Did I miss
> anything?
Oops, yeah, the work item needs to free itself. Thanks for spotting it.
--
tejun
