s390/time: correct use of store clock fast

The result of the store-clock-fast (STCKF) instruction is a bit fuzzy.
It can happen that the value stored on one CPU is smaller than the value
stored on another CPU, although the order of the stores is the other
way around. This can cause deltas of get_tod_clock() values to become
negative when they should not be.

We need to be more careful with store-clock-fast, this patch partially
reverts git commit e4b7b4238e666682555461fa52eecd74652f36bb "time:
always use stckf instead of stck if available". The get_tod_clock()
function now uses the store-clock-extended (STCKE) instruction.
get_tod_clock_fast() can be used if the fuzziness of store-clock-fast
is acceptable e.g. for wait loops local to a CPU.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

3 files changed, 9 insertions, 9 deletions

diff --git a/drivers/s390/char/sclp.c b/drivers/s390/char/sclp.c
index a3aa374..1fe2643 100644
--- a/drivers/s390/char/sclp.c
+++ b/drivers/s390/char/sclp.c
@@ -486,7 +486,7 @@ sclp_sync_wait(void)
 	timeout = 0;
 	if (timer_pending(&sclp_request_timer)) {
 		/* Get timeout TOD value */
-		timeout = get_tod_clock() +
+		timeout = get_tod_clock_fast() +
 			  sclp_tod_from_jiffies(sclp_request_timer.expires -
 						jiffies);
 	}
@@ -508,7 +508,7 @@ sclp_sync_wait(void)
 	while (sclp_running_state != sclp_running_state_idle) {
 		/* Check for expired request timer */
 		if (timer_pending(&sclp_request_timer) &&
-		    get_tod_clock() > timeout &&
+		    get_tod_clock_fast() > timeout &&
 		    del_timer(&sclp_request_timer))
 			sclp_request_timer.function(sclp_request_timer.data);
 		cpu_relax();
diff --git a/drivers/s390/cio/cio.c b/drivers/s390/cio/cio.c
index d7da67a..88e35d8 100644
--- a/drivers/s390/cio/cio.c
+++ b/drivers/s390/cio/cio.c
@@ -878,9 +878,9 @@ static void css_reset(void)
 			atomic_inc(&chpid_reset_count);
 	}
 	/* Wait for machine check for all channel paths. */
-	timeout = get_tod_clock() + (RCHP_TIMEOUT << 12);
+	timeout = get_tod_clock_fast() + (RCHP_TIMEOUT << 12);
 	while (atomic_read(&chpid_reset_count) != 0) {
-		if (get_tod_clock() > timeout)
+		if (get_tod_clock_fast() > timeout)
 			break;
 		cpu_relax();
 	}
diff --git a/drivers/s390/cio/qdio_main.c b/drivers/s390/cio/qdio_main.c
index 8ed52aa..bbd3e51 100644
--- a/drivers/s390/cio/qdio_main.c
+++ b/drivers/s390/cio/qdio_main.c
@@ -338,10 +338,10 @@ again:
 		retries++;
 
 		if (!start_time) {
-			start_time = get_tod_clock();
+			start_time = get_tod_clock_fast();
 			goto again;
 		}
-		if ((get_tod_clock() - start_time) < QDIO_BUSY_BIT_PATIENCE)
+		if (get_tod_clock_fast() - start_time < QDIO_BUSY_BIT_PATIENCE)
 			goto again;
 	}
 	if (retries) {
@@ -504,7 +504,7 @@ static int get_inbound_buffer_frontier(struct qdio_q *q)
 	int count, stop;
 	unsigned char state = 0;
 
-	q->timestamp = get_tod_clock();
+	q->timestamp = get_tod_clock_fast();
 
 	/*
 	 * Don't check 128 buffers, as otherwise qdio_inbound_q_moved
@@ -595,7 +595,7 @@ static inline int qdio_inbound_q_done(struct qdio_q *q)
 	 * At this point we know, that inbound first_to_check
 	 * has (probably) not moved (see qdio_inbound_processing).
 	 */
-	if (get_tod_clock() > q->u.in.timestamp + QDIO_INPUT_THRESHOLD) {
+	if (get_tod_clock_fast() > q->u.in.timestamp + QDIO_INPUT_THRESHOLD) {
 		DBF_DEV_EVENT(DBF_INFO, q->irq_ptr, "in done:%02x",
 			      q->first_to_check);
 		return 1;
@@ -728,7 +728,7 @@ static int get_outbound_buffer_frontier(struct qdio_q *q)
 	int count, stop;
 	unsigned char state = 0;
 
-	q->timestamp = get_tod_clock();
+	q->timestamp = get_tod_clock_fast();
 
 	if (need_siga_sync(q))
 		if (((queue_type(q) != QDIO_IQDIO_QFMT) &&