Skip to contentSkip to conversation

cpuidle: teo: Fix selection when idle state 0 is disabled

PATCH In review Rafael J. Wysocki 3 participants 5 messages Jul 30, 2021 – Aug 3, 2021 CPU idle time management framework

lore
Share & tools
lore
#0 lore raw

cover cpuidle: teo: Fix selection when idle state 0 is disabled

Hi,

Patch [1/2] fixes idle state selection in the teo governor when idle state 0 is disabled (which is broken after recent changes in that governor) and patch [2/2] renames two local variables on top of that change.

Please see the patch changelogs for details.

Thanks!

#1 lore raw

PATCH 1/2 cpuidle: teo: Fix alternative idle state lookup

From: Rafael J. Wysocki <[email protected]>

There are three mistakes in the loop in teo_select() that is looking for an alternative candidate idle state. First, it should walk all of the idle states shallower than the current candidate one, including all of the disabled ones, but it terminates after the first enabled idle state. Second, it should not terminate its last step if idle state 0 is disabled (which is related to the first issue). Finally, it may return the current alternative candidate idle state prematurely if the time span criterion is not met by the idle state under consideration at the moment.

To address the issues mentioned above, make the loop in question walk all of the idle states shallower than the current candidate idle state all the way down to idle state 0 and rearrange the checks in it.

Fixes 77577558f25d ("cpuidle: teo: Rework most recent idle duration values treatment")
---
 drivers/cpuidle/governors/teo.c |   40 +++++++++++++++++++++++++++-------------
 1 file changed, 27 insertions(+), 13 deletions(-)
drivers/cpuidle/governors/teo.c +27 -13
Index: linux-pm/drivers/cpuidle/governors/teo.c
===================================================================
@@ -397,32 +397,46 @@ static int teo_select(struct cpuidle_dri
 		intercept_sum = 0;
 		recent_sum = 0;
 
-		for (i = idx - 1; i >= idx0; i--) {
+		for (i = idx - 1; i >= 0; i--) {
 			struct teo_bin *bin = &cpu_data->state_bins[i];
 			s64 span_ns;
 
 			intercept_sum += bin->intercepts;
 			recent_sum += bin->recent;
 
+			span_ns = teo_middle_of_bin(i, drv);
+
+			if ((!alt_recent || 2 * recent_sum > idx_recent_sum) &&
+			    (!alt_intercepts ||
+			     2 * intercept_sum > idx_intercept_sum)) {
+				if (teo_time_ok(span_ns) &&
+				    !dev->states_usage[i].disable) {
+					idx = i;
+					duration_ns = span_ns;
+				} else {
+					/*
+					 * The current state is too shallow or
+					 * disabled, so take the first enabled
+					 * deeper state with suitable time span.
+					 */
+					idx = last_enabled_idx;
+					duration_ns = last_enabled_span_ns;
+				}
+				break;
+			}
+
 			if (dev->states_usage[i].disable)
 				continue;
 
-			span_ns = teo_middle_of_bin(i, drv);
 			if (!teo_time_ok(span_ns)) {
 				/*
-				 * The current state is too shallow, so select
-				 * the first enabled deeper state.
+				 * The current state is too shallow, but if an
+				 * alternative candidate state has been found,
+				 * it may still turn out to be a better choice.
 				 */
-				duration_ns = last_enabled_span_ns;
-				idx = last_enabled_idx;
-				break;
-			}
+				if (last_enabled_idx != idx)
+					continue;
 
-			if ((!alt_recent || 2 * recent_sum > idx_recent_sum) &&
-			    (!alt_intercepts ||
-			     2 * intercept_sum > idx_intercept_sum)) {
-				idx = i;
-				duration_ns = span_ns;
 				break;
 			}
 
#2 lore raw

PATCH 2/2 cpuidle: teo: Rename two local variables in teo_select()

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Rename two local variables in teo_select() so that their names better reflect their purpose.

No functional impact.

---
 drivers/cpuidle/governors/teo.c |   14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/cpuidle/governors/teo.c +7 -7
Index: linux-pm/drivers/cpuidle/governors/teo.c
===================================================================
@@ -382,8 +382,8 @@ static int teo_select(struct cpuidle_dri
 	alt_intercepts = 2 * idx_intercept_sum > cpu_data->total - idx_hit_sum;
 	alt_recent = idx_recent_sum > NR_RECENT / 2;
 	if (alt_recent || alt_intercepts) {
-		s64 last_enabled_span_ns = duration_ns;
-		int last_enabled_idx = idx;
+		s64 first_suitable_span_ns = duration_ns;
+		int first_suitable_idx = idx;
 
 		/*
 		 * Look for the deepest idle state whose target residency had
@@ -419,8 +419,8 @@ static int teo_select(struct cpuidle_dri
 					 * disabled, so take the first enabled
 					 * deeper state with suitable time span.
 					 */
-					idx = last_enabled_idx;
-					duration_ns = last_enabled_span_ns;
+					idx = first_suitable_idx;
+					duration_ns = first_suitable_span_ns;
 				}
 				break;
 			}
@@ -434,14 +434,14 @@ static int teo_select(struct cpuidle_dri
 				 * alternative candidate state has been found,
 				 * it may still turn out to be a better choice.
 				 */
-				if (last_enabled_idx != idx)
+				if (first_suitable_idx != idx)
 					continue;
 
 				break;
 			}
 
-			last_enabled_span_ns = span_ns;
-			last_enabled_idx = i;
+			first_suitable_span_ns = span_ns;
+			first_suitable_idx = i;
 		}
 	}
 
#3 lore raw
In reply to Rafael J. Wysocki
Quoted 8 lines · Rafael J. Wysocki

Hi,

Patch [1/2] fixes idle state selection in the teo governor when idle state 0 is disabled (which is broken after recent changes in that governor) and patch [2/2] renames two local variables on top of that change.

Please see the patch changelogs for details.

Hi Rafael,

I tested this version of the patches, and the idle state 0 disabled issue has been fixed.

... Doug

#4 lore raw
Quoted 9 lines · Doug Smythies
Quoted 7 lines · Rafael J. Wysocki

Hi,

Patch [1/2] fixes idle state selection in the teo governor when idle state 0 is disabled (which is broken after recent changes in that governor) and patch [2/2] renames two local variables on top of that change.

Quoted 6 lines
Quoted 1 line

Please see the patch changelogs for details.

Hi Rafael,

I tested this version of the patches, and the idle state 0 disabled issue has been fixed.

Thanks for the confirmation!