DM: make GPU tasks multithreaded again.  Big refactor.

The main meat of things is in SkThreadPool.  We can now give SkThreadPool a
type for each thread to create and destroy on its local stack.  It's TLS
without going through SkTLS.

I've split the DM tasks into CpuTasks that run on threads with no TLS, and
GpuTasks that run on threads with a thread local GrContextFactory.

The old CpuTask and GpuTask have been renamed to CpuGMTask and GpuGMTask.

Upshot: default run of out/Debug/dm goes from ~45 seconds to ~20 seconds.

BUG=skia:
R=bsalomon@google.com, mtklein@google.com, reed@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/179233005

git-svn-id: http://skia.googlecode.com/svn/trunk@13632 2bbb7eff-a529-9590-31e7-b0007b416f81
diff --git a/dm/DMTask.cpp b/dm/DMTask.cpp
index d26971c..1c4cc25 100644
--- a/dm/DMTask.cpp
+++ b/dm/DMTask.cpp
@@ -1,43 +1,22 @@
 #include "DMTask.h"
-
 #include "DMTaskRunner.h"
-#include "DMUtil.h"
-#include "SkBitmap.h"
-#include "SkCommandLineFlags.h"
 
 namespace DM {
 
 Task::Task(Reporter* reporter, TaskRunner* taskRunner)
-    : fReporter(reporter), fTaskRunner(taskRunner), fDepth(0) {
+    : fReporter(reporter)
+    , fTaskRunner(taskRunner)
+    , fDepth(0) {
     fReporter->start();
 }
 
 Task::Task(const Task& parent)
-    : INHERITED(parent)
-    , fReporter(parent.fReporter)
+    : fReporter(parent.fReporter)
     , fTaskRunner(parent.fTaskRunner)
-    , fDepth(parent.depth()+1) {
+    , fDepth(parent.depth() + 1) {
     fReporter->start();
 }
 
-Task::~Task() {}
-
-void Task::run() {
-    if (!this->shouldSkip()) {
-        this->draw();
-    }
-    fReporter->finish(this->name());
-    delete this;
-}
-
-void Task::spawnChild(Task* task) {
-    if (!task->usesGpu()) {
-        fTaskRunner->add(task);
-    } else {
-        SkDEBUGFAIL("Sorry, we can't spawn GPU tasks. :(  See comment in TaskRunner::wait().");
-    }
-}
-
 void Task::fail(const char* msg) {
     SkString failure(this->name());
     if (msg) {
@@ -46,8 +25,35 @@
     fReporter->fail(failure);
 }
 
-GrContextFactory* Task::getGrContextFactory() const {
-    return fTaskRunner->getGrContextFactory();
+void Task::finish() {
+    fReporter->finish(this->name());
 }
 
+void Task::spawnChild(CpuTask* task) {
+    fTaskRunner->add(task);
+}
+
+CpuTask::CpuTask(Reporter* reporter, TaskRunner* taskRunner) : Task(reporter, taskRunner) {}
+CpuTask::CpuTask(const Task& parent) : Task(parent) {}
+
+void CpuTask::run() {
+    if (!this->shouldSkip()) {
+        this->draw();
+    }
+    this->finish();
+    SkDELETE(this);
+}
+
+GpuTask::GpuTask(Reporter* reporter, TaskRunner* taskRunner) : Task(reporter, taskRunner) {}
+
+void GpuTask::run(GrContextFactory& factory) {
+    if (!this->shouldSkip()) {
+        this->draw(&factory);
+    }
+    this->finish();
+    SkDELETE(this);
+}
+
+
+
 }  // namespace DM