Skip to contentSkip to conversation

cpuidle-pseries: Fix CEDE latency conversion from tb to us

PATCH Applied Gautham R. Shenoy 3 participants 3 messages Sep 3, 2020 – Sep 10, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 cpuidle-pseries: Fix CEDE latency conversion from tb to us

From: "Gautham R. Shenoy" <[email protected]>

commit d947fb4c965c ("cpuidle: pseries: Fixup exit latency for CEDE(0)") sets the exit latency of CEDE(0) based on the latency values of the Extended CEDE states advertised by the platform. The values advertised by the platform are in timebase ticks. However the cpuidle framework requires the latency values in microseconds.

If the tb-ticks value advertised by the platform correspond to a value smaller than 1us, during the conversion from tb-ticks to microseconds, in the current code, the result becomes zero. This is incorrect as it puts a CEDE state on par with the snooze state.

This patch fixes this by rounding up the result obtained while converting the latency value from tb-ticks to microseconds. It also prints a warning in case we discover an extended-cede state with wakeup latency to be 0. In such a case, ensure that CEDE(0) has a non-zero wakeup latency.

Fixes commit d947fb4c965c ("cpuidle: pseries: Fixup exit latency for

CEDE(0)")

---
v1-->v2: Added a warning if a CEDE state has 0 wakeup latency (Suggested by Joel Stanley)
Also added code to ensure that CEDE(0) has a non-zero wakeup latency.

 drivers/cpuidle/cpuidle-pseries.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)
drivers/cpuidle/cpuidle-pseries.c +11 -4
index ff6d99e..a2b5c6f 100644
@@ -361,7 +361,10 @@ static void __init fixup_cede0_latency(void)
 	for (i = 0; i < nr_xcede_records; i++) {
 		struct xcede_latency_record *record = &payload->records[i];
 		u64 latency_tb = be64_to_cpu(record->latency_ticks);
-		u64 latency_us = tb_to_ns(latency_tb) / NSEC_PER_USEC;
+		u64 latency_us = DIV_ROUND_UP_ULL(tb_to_ns(latency_tb), NSEC_PER_USEC);
+
+		if (latency_us == 0)
+			pr_warn("cpuidle: xcede record %d has an unrealistic latency of 0us.\n", i);
 
 		if (latency_us < min_latency_us)
 			min_latency_us = latency_us;
@@ -378,10 +381,14 @@ static void __init fixup_cede0_latency(void)
 	 * Perform the fix-up.
 	 */
 	if (min_latency_us < dedicated_states[1].exit_latency) {
-		u64 cede0_latency = min_latency_us - 1;
+		/*
+		 * We set a minimum of 1us wakeup latency for cede0 to
+		 * distinguish it from snooze
+		 */
+		u64 cede0_latency = 1;
 
-		if (cede0_latency <= 0)
-			cede0_latency = min_latency_us;
+		if (min_latency_us > cede0_latency)
+			cede0_latency = min_latency_us - 1;
 
 		dedicated_states[1].exit_latency = cede0_latency;
 		dedicated_states[1].target_residency = 10 * (cede0_latency);
Signature · 1 line
1.9.4
Reviewed-by on v2 1/1
#1 lore raw
Quoted 19 lines · Gautham R Shenoy

From: "Gautham R. Shenoy" <[email protected]>

commit d947fb4c965c ("cpuidle: pseries: Fixup exit latency for CEDE(0)") sets the exit latency of CEDE(0) based on the latency values of the Extended CEDE states advertised by the platform. The values advertised by the platform are in timebase ticks. However the cpuidle framework requires the latency values in microseconds.

If the tb-ticks value advertised by the platform correspond to a value smaller than 1us, during the conversion from tb-ticks to microseconds, in the current code, the result becomes zero. This is incorrect as it puts a CEDE state on par with the snooze state.

This patch fixes this by rounding up the result obtained while converting the latency value from tb-ticks to microseconds. It also prints a warning in case we discover an extended-cede state with wakeup latency to be 0. In such a case, ensure that CEDE(0) has a non-zero wakeup latency.

Quoted 4 lines
Fixes commit d947fb4c965c ("cpuidle: pseries: Fixup exit latency for

CEDE(0)")

Quoted 11 lines

---
v1-->v2: Added a warning if a CEDE state has 0 wakeup latency (Suggested by Joel Stanley)
Also added code to ensure that CEDE(0) has a non-zero wakeup latency.

 drivers/cpuidle/cpuidle-pseries.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)
drivers/cpuidle/cpuidle-pseries.c +0 -0
index ff6d99e..a2b5c6f 100644

drivers/cpuidle/cpuidle-pseries.c static void __init fixup_cede0_latency(void) · line 364

Quoted 5 lines
drivers/cpuidle/cpuidle-pseries.c +1 -1
@@ -361,7 +361,10 @@ static void __init fixup_cede0_latency(void)
 	for (i = 0; i < nr_xcede_records; i++) {
 		struct xcede_latency_record *record = &payload->records[i];
 		u64 latency_tb = be64_to_cpu(record->latency_ticks);
-		u64 latency_us = tb_to_ns(latency_tb) / NSEC_PER_USEC;
+		u64 latency_us = DIV_ROUND_UP_ULL(tb_to_ns(latency_tb), NSEC_PER_USEC);

This would fix the issue of rounding down to zero.

drivers/cpuidle/cpuidle-pseries.c static void __init fixup_cede0_latency(void) · line 367

Quoted 3 lines
drivers/cpuidle/cpuidle-pseries.c +3 -0
+
+		if (latency_us == 0)
+			pr_warn("cpuidle: xcede record %d has an unrealistic latency of 0us.\n", i);

+1 This should not happen.

Quoted 7 lines
drivers/cpuidle/cpuidle-pseries.c +0 -1
 		if (latency_us < min_latency_us)
 			min_latency_us = latency_us;
@@ -378,10 +381,14 @@ static void __init fixup_cede0_latency(void)
 	 * Perform the fix-up.
 	 */
 	if (min_latency_us < dedicated_states[1].exit_latency) {
-		u64 cede0_latency = min_latency_us - 1;

drivers/cpuidle/cpuidle-pseries.c static void __init fixup_cede0_latency(void) · line 391

Quoted 10 lines
drivers/cpuidle/cpuidle-pseries.c +7 -2
+		/*
+		 * We set a minimum of 1us wakeup latency for cede0 to
+		 * distinguish it from snooze
+		 */
+		u64 cede0_latency = 1;
 
-		if (cede0_latency <= 0)
-			cede0_latency = min_latency_us;
+		if (min_latency_us > cede0_latency)
+			cede0_latency = min_latency_us - 1;

Good checks to expect cede latency of 1us or more.

drivers/cpuidle/cpuidle-pseries.c static void __init fixup_cede0_latency(void) · line 394

Quoted 2 lines
drivers/cpuidle/cpuidle-pseries.c +0 -0
 		dedicated_states[1].exit_latency = cede0_latency;
 		dedicated_states[1].target_residency = 10 * (cede0_latency);

--Vaidy

#2 lore raw
In reply to Gautham R. Shenoy
Quoted 6 lines · Gautham R. Shenoy

commit d947fb4c965c ("cpuidle: pseries: Fixup exit latency for CEDE(0)") sets the exit latency of CEDE(0) based on the latency values of the Extended CEDE states advertised by the platform. The values advertised by the platform are in timebase ticks. However the cpuidle framework requires the latency values in microseconds.

Quoted 6 lines

If the tb-ticks value advertised by the platform correspond to a value smaller than 1us, during the conversion from tb-ticks to microseconds, in the current code, the result becomes zero. This is incorrect as it puts a CEDE state on par with the snooze state.

[...]

Applied to powerpc/fixes.

[1/1] cpuidle: pseries: Fix CEDE latency conversion from tb to us

      https://git.kernel.org/powerpc/c/1d3ee7df009a46440c58508b8819213c09503acd

cheers