Re: [PATCH] io-wq: Fix io_wq_worker_affinity()

[Jens Axboe]

On 4/8/21 3:44 AM, Peter Zijlstra wrote:
> 
> Do not include private headers and do not frob in internals.
> 
> On top of that, while the previous code restores the affinity, it
> doesn't ensure the task actually moves there if it was running,
> leading to the fun situation that it can be observed running outside
> of its allowed mask for potentially significant time.
> 
> Use the proper API instead.

Applied, thanks Peter.

-- 
Jens Axboe

[PATCH] io-wq: Fix io_wq_worker_affinity()

[Peter Zijlstra]

Do not include private headers and do not frob in internals.

On top of that, while the previous code restores the affinity, it
doesn't ensure the task actually moves there if it was running,
leading to the fun situation that it can be observed running outside
of its allowed mask for potentially significant time.

Use the proper API instead.

Signed-off-by: Peter Zijlstra (Intel) 
---
 fs/io-wq.c |   11 ++-
 1 file changed, 2 insertions(+), 9 deletions(-)

--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -17,7 +17,6 @@
 #include 
 #include 
 
-#include "../kernel/sched/sched.h"
 #include "io-wq.h"
 
 #define WORKER_IDLE_TIMEOUT(5 * HZ)
@@ -1098,14 +1097,8 @@ void io_wq_put_and_exit(struct io_wq *wq
 
 static bool io_wq_worker_affinity(struct io_worker *worker, void *data)
 {
-   struct task_struct *task = worker->task;
-   struct rq_flags rf;
-   struct rq *rq;
-
-   rq = task_rq_lock(task, &rf);
-   do_set_cpus_allowed(task, cpumask_of_node(worker->wqe->node));
-   task->flags |= PF_NO_SETAFFINITY;
-   task_rq_unlock(rq, task, &rf);
+   set_cpus_allowed_ptr(worker->task, cpumask_of_node(worker->wqe->node));
+
return false;
 }