This function is concerned with the long-term cpu mask, not the
transitory mask the task might have while migrate disabled. Before
this patch, if a task was migrate disabled at the time
__set_cpus_allowed_ptr() was called, and the new mask happened to be
equal to the cpu that the task was running on, then the mask update
would be lost.
Signed-off-by: Scott Wood <[email protected]>
---
kernel/sched/core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index c3407707e367..6e643d656d71 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1218,7 +1218,7 @@ static int __set_cpus_allowed_ptr(struct task_struct *p,
goto out;
}
- if (cpumask_equal(p->cpus_ptr, new_mask))
+ if (cpumask_equal(&p->cpus_mask, new_mask))
goto out;
if (!cpumask_intersects(new_mask, cpu_valid_mask)) {
--
1.8.3.1
On 2019-07-27 00:56:32 [-0500], Scott Wood wrote:
> This function is concerned with the long-term cpu mask, not the
> transitory mask the task might have while migrate disabled. Before
> this patch, if a task was migrate disabled at the time
> __set_cpus_allowed_ptr() was called, and the new mask happened to be
> equal to the cpu that the task was running on, then the mask update
> would be lost.
lost as in "would not be carried out" I assume.
> Signed-off-by: Scott Wood <[email protected]>
> ---
> kernel/sched/core.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index c3407707e367..6e643d656d71 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -1218,7 +1218,7 @@ static int __set_cpus_allowed_ptr(struct task_struct *p,
> goto out;
> }
>
> - if (cpumask_equal(p->cpus_ptr, new_mask))
> + if (cpumask_equal(&p->cpus_mask, new_mask))
> goto out;
>
> if (!cpumask_intersects(new_mask, cpu_valid_mask)) {
Sebastian
On Tue, 2019-09-17 at 16:57 +0200, Sebastian Andrzej Siewior wrote:
> On 2019-07-27 00:56:32 [-0500], Scott Wood wrote:
> > This function is concerned with the long-term cpu mask, not the
> > transitory mask the task might have while migrate disabled. Before
> > this patch, if a task was migrate disabled at the time
> > __set_cpus_allowed_ptr() was called, and the new mask happened to be
> > equal to the cpu that the task was running on, then the mask update
> > would be lost.
>
> lost as in "would not be carried out" I assume.
Right. The old mask would be restored upon migrate_enable() even though
that's no longer the policy that was requested.
-Scott