[PATCH] async: Use ktime_us_delta() for debug timings

[PATCH] async: Use ktime_us_delta() for debug timings

[zhoumin]

Replace microseconds_since() with ktime_us_delta() for accurate
microsecond timings.

Fixes: 22a9d645677f ("async: Asynchronous function calls to speed up kernel boot")
Signed-off-by: zhoumin <teczm@foxmail.com>
---
 kernel/async.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/kernel/async.c b/kernel/async.c
index 0e3a783dc991..e8862c39344b 100644
--- a/kernel/async.c
+++ b/kernel/async.c
@@ -80,12 +80,6 @@ static DECLARE_WAIT_QUEUE_HEAD(async_done);
 
 static atomic_t entry_count;
 
-static long long microseconds_since(ktime_t start)
-{
-	ktime_t now = ktime_get();
-	return ktime_to_ns(ktime_sub(now, start)) >> 10;
-}
-
 static async_cookie_t lowest_in_progress(struct async_domain *domain)
 {
 	struct async_entry *first = NULL;
@@ -130,7 +124,7 @@ static void async_run_entry_fn(struct work_struct *work)
 
 	pr_debug("initcall %lli_%pS returned after %lld usecs\n",
 		 (long long)entry->cookie, entry->func,
-		 microseconds_since(calltime));
+		 ktime_us_delta(ktime_get(), calltime));
 
 	/* 2) remove self from the pending queues */
 	spin_lock_irqsave(&async_lock, flags);
@@ -317,7 +311,7 @@ void async_synchronize_cookie_domain(async_cookie_t cookie, struct async_domain
 	wait_event(async_done, lowest_in_progress(domain) >= cookie);
 
 	pr_debug("async_continuing @ %i after %lli usec\n", task_pid_nr(current),
-		 microseconds_since(starttime));
+		 ktime_us_delta(ktime_get(), starttime));
 }
 EXPORT_SYMBOL_GPL(async_synchronize_cookie_domain);
 

base-commit: d135ab453a7400a34930bdda1ef4bdc8acfac80e
-- 
2.53.0

Re: [PATCH] async: Use ktime_us_delta() for debug timings

[Arjan van de Ven]

On 10/27/2026 5:01 AM, zhoumin wrote:
> Replace microseconds_since() with ktime_us_delta() for accurate
> microsecond timings.

Hi,


while there is clearly nothing wrong with the change and it is a slight 
cleanup (so "Reviewed-by:  Arjan van de Ven <arjan@linux.intel.com> "), I'm

less convinced this is reaching to the level that this change is worth 
the churn in this era of mass AI originated changes that are fixing 
actual bugs.

Obviously I'm totally fine either way and will let Andrew decide in the 
end ;-)



> base-commit: d135ab453a7400a34930bdda1ef4bdc8acfac80e