Widen CPU affinity set.

fml/cpu_affinity.cc

@@ -58,6 +58,8 @@ CPUSpeedTracker::CPUSpeedTracker(std::vector<CpuIndexAndSpeed> data)
     }
     if (data.speed == min_speed_value) {
       efficiency_.push_back(data.index);
+    } else {
+      not_efficiency_.push_back(data.index);
     }
   }
 
@@ -77,6 +79,8 @@ const std::vector<size_t>& CPUSpeedTracker::GetIndices(
       return efficiency_;
     case CpuAffinity::kNotPerformance:
       return not_performance_;
+    case CpuAffinity::kNotEfficiency:
+      return not_efficiency_;
   }
 }
 

fml/cpu_affinity.h

@@ -26,6 +26,9 @@ enum class CpuAffinity {
 
   /// @brief Request affinity for all non-performance cores.
   kNotPerformance,
+
+  /// @brief Request affinity for all non-efficiency cores.
+  kNotEfficiency,
 };
 
 /// @brief Request count of efficiency cores.
@@ -79,6 +82,7 @@ class CPUSpeedTracker {
   std::vector<size_t> efficiency_;
   std::vector<size_t> performance_;
   std::vector<size_t> not_performance_;
+  std::vector<size_t> not_efficiency_;
 };
 
 /// @note Visible for testing.

fml/cpu_affinity_unittests.cc

@@ -29,6 +29,9 @@ TEST(CpuAffinity, NormalSlowMedFastCores) {
   ASSERT_EQ(tracker.GetIndices(CpuAffinity::kNotPerformance).size(), 2u);
   ASSERT_EQ(tracker.GetIndices(CpuAffinity::kNotPerformance)[0], 0u);
   ASSERT_EQ(tracker.GetIndices(CpuAffinity::kNotPerformance)[1], 1u);
+  ASSERT_EQ(tracker.GetIndices(CpuAffinity::kNotEfficiency).size(), 2u);
+  ASSERT_EQ(tracker.GetIndices(CpuAffinity::kNotEfficiency)[0], 1u);
+  ASSERT_EQ(tracker.GetIndices(CpuAffinity::kNotEfficiency)[1], 2u);
 }
 
 TEST(CpuAffinity, NoCpuData) {

shell/platform/android/android_shell_holder.cc

@@ -49,14 +49,14 @@ static void AndroidPlatformThreadConfigSetter(
       break;
     }
     case fml::Thread::ThreadPriority::kDisplay: {
-      fml::RequestAffinity(fml::CpuAffinity::kPerformance);
+      fml::RequestAffinity(fml::CpuAffinity::kNotEfficiency);
       if (::setpriority(PRIO_PROCESS, 0, -1) != 0) {
         FML_LOG(ERROR) << "Failed to set UI task runner priority";
       }
       break;
     }
     case fml::Thread::ThreadPriority::kRaster: {
-      fml::RequestAffinity(fml::CpuAffinity::kPerformance);
+      fml::RequestAffinity(fml::CpuAffinity::kNotEfficiency);
       // Android describes -8 as "most important display threads, for
       // compositing the screen and retrieving input events". Conservatively
       // set the raster thread to slightly lower priority than it.