Pack preloaded framework assets in a texture atlas

When the Android runtime starts, the system preloads a series of assets
in the Zygote process. These assets are shared across all processes.
Unfortunately, each one of these assets is later uploaded in its own
OpenGL texture, once per process. This wastes memory and generates
unnecessary OpenGL state changes.

This CL introduces an asset server that provides an atlas to all processes.

Note: bitmaps used by skia shaders are *not* sampled from the atlas.
It's an uncommon use case and would require extra texture transforms
in the GL shaders.

WHAT IS THE ASSETS ATLAS

The "assets atlas" is a single, shareable graphic buffer that contains
all the system's preloaded bitmap drawables (this includes 9-patches.)
The atlas is made of two distinct objects: the graphic buffer that
contains the actual pixels and the map which indicates where each
preloaded bitmap can be found in the atlas (essentially a pair of
x and y coordinates.)

HOW IS THE ASSETS ATLAS GENERATED

Because we need to support a wide variety of devices and because it
is easy to change the list of preloaded drawables, the atlas is
generated at runtime, during the startup phase of the system process.

There are several steps that lead to the atlas generation:

1. If the device is booting for the first time, or if the device was
updated, we need to find the best atlas configuration. To do so,
the atlas service tries a number of width, height and algorithm
variations that allows us to pack as many assets as possible while
using as little memory as possible. Once a best configuration is found,
it gets written to disk in /data/system/framework_atlas

2. Given a best configuration (algorithm variant, dimensions and
number of bitmaps that can be packed in the atlas), the atlas service
packs all the preloaded bitmaps into a single graphic buffer object.

3. The packing is done using Skia in a temporary native bitmap. The
Skia bitmap is then copied into the graphic buffer using OpenGL ES
to benefit from texture swizzling.

HOW PROCESSES USE THE ATLAS

Whenever a process' hardware renderer initializes its EGL context,
it queries the atlas service for the graphic buffer and the map.

It is important to remember that both the context and the map will
be valid for the lifetime of the hardware renderer (if the system
process goes down, all apps get killed as well.)

Every time the hardware renderer needs to render a bitmap, it first
checks whether the bitmap can be found in the assets atlas. When
the bitmap is part of the atlas, texture coordinates are remapped
appropriately before rendering.

Change-Id: I8eaecf53e7f6a33d90da3d0047c5ceec89ea3af0
diff --git a/libs/hwui/PatchCache.cpp b/libs/hwui/PatchCache.cpp
index 62e38d3..5fa75e9 100644
--- a/libs/hwui/PatchCache.cpp
+++ b/libs/hwui/PatchCache.cpp
@@ -16,8 +16,10 @@
 
 #define LOG_TAG "OpenGLRenderer"
 
+#include <utils/JenkinsHash.h>
 #include <utils/Log.h>
 
+#include "Caches.h"
 #include "PatchCache.h"
 #include "Properties.h"
 
@@ -28,107 +30,107 @@
 // Constructors/destructor
 ///////////////////////////////////////////////////////////////////////////////
 
-PatchCache::PatchCache(): mMaxEntries(DEFAULT_PATCH_CACHE_SIZE) {
-}
-
-PatchCache::PatchCache(uint32_t maxEntries): mMaxEntries(maxEntries) {
+PatchCache::PatchCache(): mCache(LruCache<PatchDescription, Patch*>::kUnlimitedCapacity) {
+    char property[PROPERTY_VALUE_MAX];
+    if (property_get(PROPERTY_PATCH_CACHE_SIZE, property, NULL) > 0) {
+        INIT_LOGD("  Setting patch cache size to %skB", property);
+        mMaxSize = KB(atoi(property));
+    } else {
+        INIT_LOGD("  Using default patch cache size of %.2fkB", DEFAULT_PATCH_CACHE_SIZE);
+        mMaxSize = KB(DEFAULT_PATCH_CACHE_SIZE);
+    }
+    mSize = 0;
 }
 
 PatchCache::~PatchCache() {
     clear();
 }
 
+void PatchCache::init(Caches& caches) {
+    glGenBuffers(1, &mMeshBuffer);
+    caches.bindMeshBuffer(mMeshBuffer);
+    caches.resetVertexPointers();
+
+    glBufferData(GL_ARRAY_BUFFER, mMaxSize, NULL, GL_DYNAMIC_DRAW);
+}
+
 ///////////////////////////////////////////////////////////////////////////////
 // Caching
 ///////////////////////////////////////////////////////////////////////////////
 
-int PatchCache::PatchDescription::compare(
-        const PatchCache::PatchDescription& lhs, const PatchCache::PatchDescription& rhs) {
-    int deltaInt = lhs.bitmapWidth - rhs.bitmapWidth;
-    if (deltaInt != 0) return deltaInt;
+hash_t PatchCache::PatchDescription::hash() const {
+    uint32_t hash = JenkinsHashMix(0, android::hash_type(mPatch));
+    hash = JenkinsHashMix(hash, mBitmapWidth);
+    hash = JenkinsHashMix(hash, mBitmapHeight);
+    hash = JenkinsHashMix(hash, mPixelWidth);
+    hash = JenkinsHashMix(hash, mPixelHeight);
+    return JenkinsHashWhiten(hash);
+}
 
-    deltaInt = lhs.bitmapHeight - rhs.bitmapHeight;
-    if (deltaInt != 0) return deltaInt;
-
-    if (lhs.pixelWidth < rhs.pixelWidth) return -1;
-    if (lhs.pixelWidth > rhs.pixelWidth) return +1;
-
-    if (lhs.pixelHeight < rhs.pixelHeight) return -1;
-    if (lhs.pixelHeight > rhs.pixelHeight) return +1;
-
-    deltaInt = lhs.xCount - rhs.xCount;
-    if (deltaInt != 0) return deltaInt;
-
-    deltaInt = lhs.yCount - rhs.yCount;
-    if (deltaInt != 0) return deltaInt;
-
-    deltaInt = lhs.emptyCount - rhs.emptyCount;
-    if (deltaInt != 0) return deltaInt;
-
-    deltaInt = lhs.colorKey - rhs.colorKey;
-    if (deltaInt != 0) return deltaInt;
-
-    return 0;
+int PatchCache::PatchDescription::compare(const PatchCache::PatchDescription& lhs,
+            const PatchCache::PatchDescription& rhs) {
+    return memcmp(&lhs, &rhs, sizeof(PatchDescription));
 }
 
 void PatchCache::clear() {
-    size_t count = mCache.size();
-    for (size_t i = 0; i < count; i++) {
-        delete mCache.valueAt(i);
+    glDeleteBuffers(1, &mMeshBuffer);
+    clearCache();
+    mSize = 0;
+}
+
+void PatchCache::clearCache() {
+    LruCache<PatchDescription, Patch*>::Iterator i(mCache);
+    while (i.next()) {
+        ALOGD("Delete %p", i.value());
+        delete i.value();
     }
     mCache.clear();
 }
 
-Patch* PatchCache::get(const uint32_t bitmapWidth, const uint32_t bitmapHeight,
-        const float pixelWidth, const float pixelHeight,
-        const int32_t* xDivs, const int32_t* yDivs, const uint32_t* colors,
-        const uint32_t width, const uint32_t height, const int8_t numColors) {
+const Patch* PatchCache::get(const AssetAtlas::Entry* entry,
+        const uint32_t bitmapWidth, const uint32_t bitmapHeight,
+        const float pixelWidth, const float pixelHeight, const Res_png_9patch* patch) {
 
-    int8_t transparentQuads = 0;
-    uint32_t colorKey = 0;
-
-    if (uint8_t(numColors) < sizeof(uint32_t) * 4) {
-        for (int8_t i = 0; i < numColors; i++) {
-            if (colors[i] == 0x0) {
-                transparentQuads++;
-                colorKey |= 0x1 << i;
-            }
-        }
-    }
-
-    // If the 9patch is made of only transparent quads
-    if (transparentQuads == int8_t((width + 1) * (height + 1))) {
-        return NULL;
-    }
-
-    const PatchDescription description(bitmapWidth, bitmapHeight,
-            pixelWidth, pixelHeight, width, height, transparentQuads, colorKey);
-
-    ssize_t index = mCache.indexOfKey(description);
-    Patch* mesh = NULL;
-    if (index >= 0) {
-        mesh = mCache.valueAt(index);
-    }
+    const PatchDescription description(bitmapWidth, bitmapHeight, pixelWidth, pixelHeight, patch);
+    const Patch* mesh = mCache.get(description);
 
     if (!mesh) {
-        PATCH_LOGD("New patch mesh "
-                "xCount=%d yCount=%d, w=%.2f h=%.2f, bw=%.2f bh=%.2f",
-                width, height, pixelWidth, pixelHeight, bitmapWidth, bitmapHeight);
+        Patch* newMesh = new Patch();
+        TextureVertex* vertices;
 
-        mesh = new Patch(width, height, transparentQuads);
-        mesh->updateColorKey(colorKey);
-        mesh->copy(xDivs, yDivs);
-        mesh->updateVertices(bitmapWidth, bitmapHeight, 0.0f, 0.0f, pixelWidth, pixelHeight);
-
-        if (mCache.size() >= mMaxEntries) {
-            delete mCache.valueAt(mCache.size() - 1);
-            mCache.removeItemsAt(mCache.size() - 1, 1);
+        if (entry) {
+            vertices = newMesh->createMesh(bitmapWidth, bitmapHeight,
+                    0.0f, 0.0f, pixelWidth, pixelHeight, entry->uvMapper, patch);
+        } else {
+            vertices = newMesh->createMesh(bitmapWidth, bitmapHeight,
+                    0.0f, 0.0f, pixelWidth, pixelHeight, patch);
         }
 
-        mCache.add(description, mesh);
-    } else if (!mesh->matches(xDivs, yDivs, colorKey, transparentQuads)) {
-        PATCH_LOGD("Patch mesh does not match, refreshing vertices");
-        mesh->updateVertices(bitmapWidth, bitmapHeight, 0.0f, 0.0f, pixelWidth, pixelHeight);
+        if (vertices) {
+            Caches& caches = Caches::getInstance();
+            caches.bindMeshBuffer(mMeshBuffer);
+            caches.resetVertexPointers();
+
+            // TODO: Simply remove the oldest items until we have enough room
+            // This will require to keep a list of free blocks in the VBO
+            uint32_t size = newMesh->getSize();
+            if (mSize + size > mMaxSize) {
+                clearCache();
+                glBufferData(GL_ARRAY_BUFFER, mMaxSize, NULL, GL_DYNAMIC_DRAW);
+                mSize = 0;
+            }
+
+            newMesh->offset = (GLintptr) mSize;
+            newMesh->textureOffset = newMesh->offset + gMeshTextureOffset;
+            mSize += size;
+
+            glBufferSubData(GL_ARRAY_BUFFER, newMesh->offset, size, vertices);
+
+            delete[] vertices;
+        }
+
+        mCache.put(description, newMesh);
+        return newMesh;
     }
 
     return mesh;