diff mbox series

[4/6] sched: Add get_current_state()

Message ID 20210602133040.461908001@infradead.org (mailing list archive)
State New, archived
Headers show
Series sched: Cleanup task_struct::state | expand

Commit Message

Peter Zijlstra June 2, 2021, 1:12 p.m. UTC
Remove yet another few p->state accesses.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 block/blk-mq.c        |    2 +-
 include/linux/sched.h |    2 ++
 kernel/freezer.c      |    2 +-
 kernel/sched/core.c   |    6 +++---
 4 files changed, 7 insertions(+), 5 deletions(-)

Comments

Mathieu Desnoyers June 2, 2021, 2:01 p.m. UTC | #1
----- On Jun 2, 2021, at 9:12 AM, Peter Zijlstra peterz@infradead.org wrote:

> Remove yet another few p->state accesses.

[...]

> 
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -212,6 +212,8 @@ struct task_group;
> 
> #endif
> 
> +#define get_current_state()	READ_ONCE(current->state)

Why use a macro rather than a static inline here ?

Thanks,

Mathieu
Peter Zijlstra June 2, 2021, 2:12 p.m. UTC | #2
On Wed, Jun 02, 2021 at 10:01:29AM -0400, Mathieu Desnoyers wrote:
> ----- On Jun 2, 2021, at 9:12 AM, Peter Zijlstra peterz@infradead.org wrote:
> 
> > Remove yet another few p->state accesses.
> 
> [...]
> 
> > 
> > --- a/include/linux/sched.h
> > +++ b/include/linux/sched.h
> > @@ -212,6 +212,8 @@ struct task_group;
> > 
> > #endif
> > 
> > +#define get_current_state()	READ_ONCE(current->state)
> 
> Why use a macro rather than a static inline here ?

Mostly to be consistent, all that state stuff is macros. I suppose we
could try and make them inlines at the end or so -- if the header maze
allows.
Will Deacon June 2, 2021, 3:02 p.m. UTC | #3
On Wed, Jun 02, 2021 at 03:12:29PM +0200, Peter Zijlstra wrote:
> Remove yet another few p->state accesses.
> 
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> ---
>  block/blk-mq.c        |    2 +-
>  include/linux/sched.h |    2 ++
>  kernel/freezer.c      |    2 +-
>  kernel/sched/core.c   |    6 +++---
>  4 files changed, 7 insertions(+), 5 deletions(-)

I think you can include kernel/kcsan/report.c here too.

With that:

Acked-by: Will Deacon <will@kernel.org>

Will
diff mbox series

Patch

--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -3891,7 +3891,7 @@  int blk_poll(struct request_queue *q, bl
 
 	hctx->poll_considered++;
 
-	state = current->state;
+	state = get_current_state();
 	do {
 		int ret;
 
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -212,6 +212,8 @@  struct task_group;
 
 #endif
 
+#define get_current_state()	READ_ONCE(current->state)
+
 /* Task command name length: */
 #define TASK_COMM_LEN			16
 
--- a/kernel/freezer.c
+++ b/kernel/freezer.c
@@ -58,7 +58,7 @@  bool __refrigerator(bool check_kthr_stop
 	/* Hmm, should we be allowed to suspend when there are realtime
 	   processes around? */
 	bool was_frozen = false;
-	long save = current->state;
+	unsigned int save = get_current_state();
 
 	pr_debug("%s entered refrigerator\n", current->comm);
 
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -8273,15 +8273,15 @@  static inline int preempt_count_equals(i
 
 void __might_sleep(const char *file, int line, int preempt_offset)
 {
+	unsigned int state = get_current_state();
 	/*
 	 * Blocking primitives will set (and therefore destroy) current->state,
 	 * since we will exit with TASK_RUNNING make sure we enter with it,
 	 * otherwise we will destroy state.
 	 */
-	WARN_ONCE(current->state != TASK_RUNNING && current->task_state_change,
+	WARN_ONCE(state != TASK_RUNNING && current->task_state_change,
 			"do not call blocking ops when !TASK_RUNNING; "
-			"state=%lx set at [<%p>] %pS\n",
-			current->state,
+			"state=%x set at [<%p>] %pS\n", state,
 			(void *)current->task_state_change,
 			(void *)current->task_state_change);