Added test to ensure context restoration due to GC doesn't crash (Take 2)

Added test to ensure context restoration due to GC doesn't crash (Take 2)

Original commit here: https://codereview.chromium.org/251443006/. Was reverted because it wasn't triggering GC agressively enough, causing timeouts on the GPU bots.

BUG=365747
R=kbr@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=267927

[bajones, 2014-04-29 21:59:54]

PTAL. Tested more thoroughly on Mac and Windows this time through. The core
problem is simply that triggering GC is unreliable. :P While this version feels
messy to me, at least it's been pretty consistent about triggering a GC within a
few seconds on all my test runs.

[Ken Russell (switch to Gerrit), 2014-04-29 22:01:52]

Let's see how the GPU try runs go.

Unfortunately there's an active outage of linux_gpu at the moment. Let's hold
off committing until we can get a clean run on it.

[bajones, 2014-04-29 22:02:44]

On 2014/04/29 22:01:52, Ken Russell wrote:
> Let's see how the GPU try runs go.
> 
> Unfortunately there's an active outage of linux_gpu at the moment. Let's hold
> off committing until we can get a clean run on it.

SGTM

[bajones, 2014-04-29 23:43:53]

GPU runs came back green. Any feedback before landing?

[bajones, 2014-04-29 23:44:43]

On 2014/04/29 23:43:53, bajones wrote:
> GPU runs came back green. Any feedback before landing?

Actually, I take that back. Not sure what happened to win_gpu_triggered_tests.

[Ken Russell (switch to Gerrit), 2014-04-30 00:28:15]

On 2014/04/29 23:44:43, bajones wrote:
> On 2014/04/29 23:43:53, bajones wrote:
> > GPU runs came back green. Any feedback before landing?
> 
> Actually, I take that back. Not sure what happened to win_gpu_triggered_tests.

What did happen? Was there a failure and did you retry it?

The runs look good. LGTM

[bajones, 2014-04-30 00:32:52]

On 2014/04/30 00:28:15, Ken Russell wrote:
> What did happen? Was there a failure and did you retry it?

I'm still not sure. I was looking into it for a bit but then the result just
magically appeared in the trybot results list. No idea what happened.

[bajones, 2014-04-30 00:32:58]

The CQ bit was checked by bajones@chromium.org

[commit-bot: I haz the power, 2014-04-30 00:34:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bajones@chromium.org/263523002/1

[commit-bot: I haz the power, 2014-04-30 01:26:44]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-04-30 01:26:45]

Try jobs failed on following builders:
  win_chromium_rel on tryserver.chromium

[Ken Russell (switch to Gerrit), 2014-04-30 01:43:30]

On 2014/04/30 01:26:45, I haz the power (commit-bot) wrote:
> Try jobs failed on following builders:
>   win_chromium_rel on tryserver.chromium

FYI, this test was just disabled in https://codereview.chromium.org/264433003 .

[bajones, 2014-04-30 01:51:28]

The CQ bit was checked by bajones@chromium.org

[commit-bot: I haz the power, 2014-04-30 01:53:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bajones@chromium.org/263523002/1

[commit-bot: I haz the power, 2014-04-30 03:43:40]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-04-30 03:43:40]

Try jobs failed on following builders:
  linux_chromium_chromeos_rel on tryserver.chromium

[bajones, 2014-04-30 05:30:46]

The CQ bit was checked by bajones@chromium.org

[commit-bot: I haz the power, 2014-04-30 05:31:21]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bajones@chromium.org/263523002/1

[commit-bot: I haz the power, 2014-04-30 10:47:36]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-04-30 10:47:36]

Failed to apply patch for content/test/gpu/gpu_tests/context_lost.py:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  patching file content/test/gpu/gpu_tests/context_lost.py
  Hunk #1 succeeded at 136 with fuzz 2 (offset 36 lines).
  Hunk #2 FAILED at 145.
  Hunk #3 FAILED at 155.
  2 out of 3 hunks FAILED -- saving rejects to file
content/test/gpu/gpu_tests/context_lost.py.rej

Patch:       content/test/gpu/gpu_tests/context_lost.py
Index: content/test/gpu/gpu_tests/context_lost.py
diff --git a/content/test/gpu/gpu_tests/context_lost.py
b/content/test/gpu/gpu_tests/context_lost.py
index
d920e9b9931dbc1a95a26e2c28af388a4b8de88f..4c3508d1a9d05fc3bfa61143ca01870fa03a5d99
100644
--- a/content/test/gpu/gpu_tests/context_lost.py
+++ b/content/test/gpu/gpu_tests/context_lost.py
@@ -100,6 +100,28 @@ class _ContextLostValidator(page_test.PageTest):
           raise page_test.Failure(
             'Test failed (context not restored properly?)')
 
+    elif page.force_garbage_collection:
+      # Try to corce GC to clean up any contexts not attached to the page.
+      # This method seem unreliable, so the page will also attempt to force
+      # GC through excessive allocations.
+      tab.CollectGarbage()
+      completed = False
+      try:
+        print "Waiting for page to finish."
+        util.WaitFor(lambda: tab.EvaluateJavaScript(
+            'window.domAutomationController._finished'), wait_timeout)
+        completed = True
+      except util.TimeoutException:
+        pass
+
+      if not completed:
+        raise page_test.Failure(
+            'Test didn\'t complete (no context restored event?)')
+      if not tab.EvaluateJavaScript(
+        'window.domAutomationController._succeeded'):
+        raise page_test.Failure(
+          'Test failed (context not restored properly?)')
+
 class ContextLost(test_module.Test):
   enabled = True
   test = _ContextLostValidator
@@ -123,6 +145,7 @@ class ContextLost(test_module.Test):
           ],
           'kill_gpu_process': True,
           'number_of_gpu_process_kills': 1,
+          'force_garbage_collection': False
         },
         {
           'name': 'ContextLost.WebGLContextLostFromLoseContextExtension',
@@ -133,7 +156,20 @@ class ContextLost(test_module.Test):
             { 'action': 'wait',
               'javascript': 'window.domAutomationController._finished' }
           ],
-          'kill_gpu_process': False
+          'kill_gpu_process': False,
+          'force_garbage_collection': False
+        },
+        {
+          'name': 'ContextLost.WebGLContextLostFromQuantity',
+          'url': 'file://webgl.html?query=forced_quantity_loss',
+          'script_to_evaluate_on_commit': harness_script,
+          'navigate_steps': [
+            { 'action': 'navigate' },
+            { 'action': 'wait',
+              'javascript': 'window.domAutomationController._loaded' }
+          ],
+          'kill_gpu_process': False,
+          'force_garbage_collection': True
         },
       ]
     }

[bajones, 2014-05-02 20:46:50]

The CQ bit was checked by bajones@chromium.org

[commit-bot: I haz the power, 2014-05-02 20:47:07]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/bajones@chromium.org/263523002/20001

[commit-bot: I haz the power, 2014-05-02 21:07:40]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-05-02 21:07:40]

Try jobs failed on following builders:
  mac_chromium_rel on tryserver.chromium

[bajones, 2014-05-02 21:34:51]

Committed patchset #2 manually as r267927 (presubmit successful).