kernel: Add Z_IS_TIMEOUT_RELATIVE() macro

Introduces the Z_IS_TIMEOUT_RELATIVE() macro to help ensure that
checking for relative/absolute timeouts is consistent. Using this
macro also helps ensure that we get the correct behavior when using
32-bit timeouts (CONFIG_TIMEOUT_64BIT=n).

Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>

drivers/timer/nrf_grtc_timer.c

@@ -300,8 +300,7 @@ uint64_t z_nrf_grtc_timer_get_ticks(k_timeout_t t)
 
 	grtc_ticks = t.ticks * CYC_PER_TICK;
 	abs_ticks = Z_TICK_ABS(t.ticks);
-	if (abs_ticks < 0) {
+	if (Z_IS_TIMEOUT_RELATIVE(t)) {
-		/* relative timeout */
 		return (grtc_ticks > (int64_t)COUNTER_SPAN) ?
 			-EINVAL : (curr_time + grtc_ticks);
 	}

drivers/timer/nrf_rtc_timer.c

@@ -208,8 +208,7 @@ uint64_t z_nrf_rtc_timer_get_ticks(k_timeout_t t)
 	} while (curr_time != z_nrf_rtc_timer_read());
 
 	abs_ticks = Z_TICK_ABS(t.ticks);
-	if (abs_ticks < 0) {
+	if (Z_IS_TIMEOUT_RELATIVE(t)) {
-		/* relative timeout */
 		return (t.ticks > COUNTER_SPAN) ?
 			-EINVAL : (curr_time + t.ticks);
 	}

include/zephyr/sys_clock.h

@@ -149,6 +149,13 @@ typedef struct {
  */
 #define Z_TICK_ABS(t) (K_TICKS_FOREVER - 1 - (t))
 
+/* Test for relative timeout */
+#if CONFIG_TIMEOUT_64BIT
+#define Z_IS_TIMEOUT_RELATIVE(timeout)  (Z_TICK_ABS((timeout).ticks) < 0)
+#else
+#define Z_IS_TIMEOUT_RELATIVE(timeout)  true
+#endif
+
 /* added tick needed to account for tick in progress */
 #define _TICK_ALIGN 1
 

kernel/nothread.c

@@ -36,7 +36,7 @@ int32_t z_impl_k_sleep(k_timeout_t timeout)
 	}
 
 	ticks = timeout.ticks;
-	if (Z_TICK_ABS(ticks) <= 0) {
+	if (Z_IS_TIMEOUT_RELATIVE(timeout)) {
 		/* ticks is delta timeout */
 		ticks_to_wait = ticks;
 	} else {

kernel/sched.c

@@ -1067,27 +1067,26 @@ static inline void z_vrfy_k_yield(void)
 #include <zephyr/syscalls/k_yield_mrsh.c>
 #endif /* CONFIG_USERSPACE */
 
-static int32_t z_tick_sleep(k_ticks_t ticks)
+static int32_t z_tick_sleep(k_timeout_t timeout)
 {
 	uint32_t expected_wakeup_ticks;
 
 	__ASSERT(!arch_is_in_isr(), "");
 
-	LOG_DBG("thread %p for %lu ticks", _current, (unsigned long)ticks);
+	LOG_DBG("thread %p for %lu ticks", _current, (unsigned long)timeout.ticks);
 
-	/* wait of 0 ms is treated as a 'yield' */
+	/* K_NO_WAIT is treated as a 'yield' */
-	if (ticks == 0) {
+	if (K_TIMEOUT_EQ(timeout, K_NO_WAIT)) {
 		k_yield();
 		return 0;
 	}
 
-	if (Z_TICK_ABS(ticks) <= 0) {
+	if (Z_IS_TIMEOUT_RELATIVE(timeout)) {
-		expected_wakeup_ticks = ticks + sys_clock_tick_get_32();
+		expected_wakeup_ticks = timeout.ticks + sys_clock_tick_get_32();
 	} else {
-		expected_wakeup_ticks = Z_TICK_ABS(ticks);
+		expected_wakeup_ticks = Z_TICK_ABS(timeout.ticks);
 	}
 
-	k_timeout_t timeout = Z_TIMEOUT_TICKS(ticks);
 	k_spinlock_key_t key = k_spin_lock(&_sched_spinlock);
 
 #if defined(CONFIG_TIMESLICING) && defined(CONFIG_SWAP_NONATOMIC)
@@ -1103,7 +1102,8 @@ static int32_t z_tick_sleep(k_ticks_t ticks)
 	uint32_t left_ticks = expected_wakeup_ticks - sys_clock_tick_get_32();
 
 	/* To handle a negative value correctly, once type-cast it to signed 32 bit */
-	ticks = (k_ticks_t)(int32_t)left_ticks;
+	k_ticks_t ticks = (k_ticks_t)(int32_t)left_ticks;
+
 	if (ticks > 0) {
 		return ticks;
 	}
@@ -1119,9 +1119,7 @@ int32_t z_impl_k_sleep(k_timeout_t timeout)
 
 	SYS_PORT_TRACING_FUNC_ENTER(k_thread, sleep, timeout);
 
-	ticks = timeout.ticks;
+	ticks = z_tick_sleep(timeout);
-
-	ticks = z_tick_sleep(ticks);
 
 	/* k_sleep() still returns 32 bit milliseconds for compatibility */
 	int64_t ms = K_TIMEOUT_EQ(timeout, K_FOREVER) ? K_TICKS_FOREVER :
@@ -1146,7 +1144,7 @@ int32_t z_impl_k_usleep(int32_t us)
 	SYS_PORT_TRACING_FUNC_ENTER(k_thread, usleep, us);
 
 	ticks = k_us_to_ticks_ceil64(us);
-	ticks = z_tick_sleep(ticks);
+	ticks = z_tick_sleep(Z_TIMEOUT_TICKS(ticks));
 
 	int32_t ret = k_ticks_to_us_ceil64(ticks);
 

kernel/timeout.c

@@ -117,13 +117,12 @@ void z_add_timeout(struct _timeout *to, _timeout_func_t fn,
 	K_SPINLOCK(&timeout_lock) {
 		struct _timeout *t;
 
-		if (IS_ENABLED(CONFIG_TIMEOUT_64BIT) &&
+		if (Z_IS_TIMEOUT_RELATIVE(timeout)) {
-		    (Z_TICK_ABS(timeout.ticks) >= 0)) {
+			to->dticks = timeout.ticks + 1 + elapsed();
+		} else {
 			k_ticks_t ticks = Z_TICK_ABS(timeout.ticks) - curr_tick;
 
 			to->dticks = MAX(1, ticks);
-		} else {
-			to->dticks = timeout.ticks + 1 + elapsed();
 		}
 
 		for (t = first(); t != NULL; t = next(t)) {
@@ -310,10 +309,10 @@ k_timepoint_t sys_timepoint_calc(k_timeout_t timeout)
 	} else {
 		k_ticks_t dt = timeout.ticks;
 
-		if (IS_ENABLED(CONFIG_TIMEOUT_64BIT) && Z_TICK_ABS(dt) >= 0) {
+		if (Z_IS_TIMEOUT_RELATIVE(timeout)) {
-			timepoint.tick = Z_TICK_ABS(dt);
-		} else {
 			timepoint.tick = sys_clock_tick_get() + MAX(1, dt);
+		} else {
+			timepoint.tick = Z_TICK_ABS(dt);
 		}
 	}
 

kernel/timer.c

@@ -166,8 +166,13 @@ void z_impl_k_timer_start(struct k_timer *timer, k_timeout_t duration,
 	 * argument the same way k_sleep() does), but historical.  The
 	 * timer_api test relies on this behavior.
 	 */
-	if (Z_TICK_ABS(duration.ticks) < 0) {
+	if (Z_IS_TIMEOUT_RELATIVE(duration)) {
-		duration.ticks = MAX(duration.ticks - 1, 0);
+		/* For the duration == K_NO_WAIT case, ensure that behaviour
+		 * is consistent for both 32-bit k_ticks_t which are unsigned
+		 * and 64-bit k_ticks_t which are signed.
+		 */
+		duration.ticks = MAX(1, duration.ticks);
+		duration.ticks = duration.ticks - 1;
 	}
 
 	(void)z_abort_timeout(&timer->timeout);