timekeeping: Annotate ktime_get_boot_fast_ns() with data_race()
authorThomas Gleixner <tglx@linutronix.de>
Fri, 15 Apr 2022 09:19:35 +0000 (11:19 +0200)
committerThomas Gleixner <tglx@linutronix.de>
Mon, 2 May 2022 12:00:20 +0000 (14:00 +0200)
Accessing timekeeper::offset_boot in ktime_get_boot_fast_ns() is an
intended data race as the reader side cannot synchronize with a writer and
there is no space in struct tk_read_base of the NMI safe timekeeper.

Mark it so.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20220415091920.956045162@linutronix.de
kernel/time/timekeeping.c

index 2c22023..3479804 100644 (file)
@@ -528,7 +528,7 @@ u64 notrace ktime_get_boot_fast_ns(void)
 {
        struct timekeeper *tk = &tk_core.timekeeper;
 
-       return (ktime_get_mono_fast_ns() + ktime_to_ns(tk->offs_boot));
+       return (ktime_get_mono_fast_ns() + ktime_to_ns(data_race(tk->offs_boot)));
 }
 EXPORT_SYMBOL_GPL(ktime_get_boot_fast_ns);