drm_hwcomposer: Choose client layer range to achieve minimal GPU load

1. Allow CLIENT layer at any position.
2. Check all possible layer combination and choose most efficient
   based on total pixel operations value.

Use case 1 - Layer count is greater than available planes
Table shows difference before and after this commit for 4 or 2 planes cases:

```
z_order   layer      size        pixels  before(4)  after(4) before(2) after(2)
-
0         Wallpaper  1960x1080   2MiP    CLIENT     DEVICE   CLIENT    DEVICE
1         Launcher   1960x1080   2MiP    CLIENT     DEVICE   CLIENT    CLIENT
2         Status bar 1960x80     0.15MiP DEVICE     DEVICE   CLIENT    CLIENT
3         Nav. bar   1960x80     0.15MiP DEVICE     CLIENT   CLIENT    CLIENT
4         Cursor     24x24       576P    DEVICE     CLIENT   DEVICE    CLIENT

Total pixels merged by CLIENT (GPU)      4MiP  ->   0.15MiP  4.3MiP -> 2.3MiP
```

Use case 2 - Unsupported layer type in the middle (Dim layer, etc):
Table shows difference before and after this commit for 4 or 2 planes cases:

```
z_order   layer      size        pixels  before(4)  after(4) before(2) after(2)
-
0         App        1960x1080   2MiP    CLIENT     DEVICE   CLIENT    DEVICE
1         Status bar 1960x80     0.15MiP CLIENT     DEVICE   CLIENT    CLIENT
2         Nav. bar   1960x80     0.15MiP CLIENT     DEVICE   CLIENT    CLIENT
3         Dim layer  1960x1080   2MiP    CLIENT     CLIENT   CLIENT    CLIENT
4         Dialog     1000x500    0.5MiP  DEVICE     CLIENT   DEVICE    CLIENT

Total pixels merged by CLIENT (GPU)      4.3MiP ->  2.5MiP   4.3MiP -> 2.8MiP
```

Signed-off-by: Roman Stratiienko <roman.stratiienko@globallogic.com>

2 files changed, 66 insertions, 16 deletions

diff --git a/drmhwctwo.cpp b/drmhwctwo.cpp
index fa25d6b..b743333 100644
--- a/drmhwctwo.cpp
+++ b/drmhwctwo.cpp
@@ -857,6 +857,30 @@ HWC2::Error DrmHwcTwo::HwcDisplay::SetVsyncEnabled(int32_t enabled) {
   return HWC2::Error::None;
 }
 
+uint32_t DrmHwcTwo::HwcDisplay::CalcPixOps(
+    std::map<uint32_t, DrmHwcTwo::HwcLayer *> &z_map, size_t first_z,
+    size_t size) {
+  uint32_t pixops = 0;
+  for (std::pair<const uint32_t, DrmHwcTwo::HwcLayer *> &l : z_map) {
+    if (l.first >= first_z && l.first < first_z + size) {
+      hwc_rect_t df = l.second->display_frame();
+      pixops += (df.right - df.left) * (df.bottom - df.top);
+    }
+  }
+  return pixops;
+}
+
+void DrmHwcTwo::HwcDisplay::MarkValidated(
+    std::map<uint32_t, DrmHwcTwo::HwcLayer *> &z_map, size_t client_first_z,
+    size_t client_size) {
+  for (std::pair<const uint32_t, DrmHwcTwo::HwcLayer *> &l : z_map) {
+    if (l.first >= client_first_z && l.first < client_first_z + client_size)
+      l.second->set_validated_type(HWC2::Composition::Client);
+    else
+      l.second->set_validated_type(HWC2::Composition::Device);
+  }
+}
+
 HWC2::Error DrmHwcTwo::HwcDisplay::ValidateDisplay(uint32_t *num_types,
                                                    uint32_t *num_requests) {
   supported(__func__);
@@ -871,39 +895,61 @@ HWC2::Error DrmHwcTwo::HwcDisplay::ValidateDisplay(uint32_t *num_types,
   if (avail_planes < layers_.size())
     avail_planes--;
 
-  std::map<uint32_t, DrmHwcTwo::HwcLayer *, std::greater<int>> z_map;
+  std::map<uint32_t, DrmHwcTwo::HwcLayer *> z_map;
   for (std::pair<const hwc2_layer_t, DrmHwcTwo::HwcLayer> &l : layers_)
     z_map.emplace(std::make_pair(l.second.z_order(), &l.second));
 
-  uint32_t total_pixops = 0, gpu_pixops = 0;
+  uint32_t total_pixops = CalcPixOps(z_map, 0, z_map.size()), gpu_pixops = 0;
+
+  int client_start = -1, client_size = 0;
 
-  bool gpu_block = false;
   for (std::pair<const uint32_t, DrmHwcTwo::HwcLayer *> &l : z_map) {
-    hwc_rect_t df = l.second->display_frame();
-    uint32_t pixops = (df.right - df.left) * (df.bottom - df.top);
-    if (gpu_block || avail_planes == 0 ||
-        !HardwareSupportsLayerType(l.second->sf_type()) ||
+    if (!HardwareSupportsLayerType(l.second->sf_type()) ||
         !importer_->CanImportBuffer(l.second->buffer()) ||
         color_transform_hint_ != HAL_COLOR_TRANSFORM_IDENTITY) {
-      gpu_block = true;
-      gpu_pixops += pixops;
-      ++*num_types;
+      if (client_start < 0)
+        client_start = l.first;
+      client_size = (l.first - client_start) + 1;
+    }
+  }
+
+  int extra_client = (z_map.size() - client_size) - avail_planes;
+  if (extra_client > 0) {
+    int start = 0, steps;
+    if (client_size != 0) {
+      int prepend = std::min(client_start, extra_client);
+      int append = std::min(int(z_map.size() - (client_start + client_size)),
+                            extra_client);
+      start = client_start - prepend;
+      client_size += extra_client;
+      steps = 1 + std::min(std::min(append, prepend),
+                           int(z_map.size()) - (start + client_size));
     } else {
-      avail_planes--;
+      client_size = extra_client;
+      steps = 1 + z_map.size() - extra_client;
     }
 
-    total_pixops += pixops;
-    l.second->set_validated_type(gpu_block ? HWC2::Composition::Client
-                                           : HWC2::Composition::Device);
+    gpu_pixops = INT_MAX;
+    for (int i = 0; i < steps; i++) {
+      uint32_t po = CalcPixOps(z_map, start + i, client_size);
+      if (po < gpu_pixops) {
+        gpu_pixops = po;
+        client_start = start + i;
+      }
+    }
   }
 
+  MarkValidated(z_map, client_start, client_size);
+
   if (CreateComposition(true) != HWC2::Error::None) {
     ++total_stats_.failed_kms_validate_;
     gpu_pixops = total_pixops;
-    for (std::pair<const hwc2_layer_t, DrmHwcTwo::HwcLayer> &l : layers_)
-      l.second.set_validated_type(HWC2::Composition::Client);
+    client_size = z_map.size();
+    MarkValidated(z_map, 0, client_size);
   }
 
+  *num_types = client_size;
+
   total_stats_.gpu_pixops_ += gpu_pixops;
   total_stats_.total_pixops_ += total_pixops;
 
diff --git a/include/drmhwctwo.h b/include/drmhwctwo.h
index 4a10fef..babe000 100644
--- a/include/drmhwctwo.h
+++ b/include/drmhwctwo.h
@@ -202,6 +202,10 @@ class DrmHwcTwo : public hwc2_device_t {
     HWC2::Error CreateComposition(bool test);
     void AddFenceToPresentFence(int fd);
     bool HardwareSupportsLayerType(HWC2::Composition comp_type);
+    uint32_t CalcPixOps(std::map<uint32_t, DrmHwcTwo::HwcLayer *> &z_map,
+                        size_t first_z, size_t size);
+    void MarkValidated(std::map<uint32_t, DrmHwcTwo::HwcLayer *> &z_map,
+                       size_t client_first_z, size_t client_size);
 
     constexpr static size_t MATRIX_SIZE = 16;