suppress unit_tests deliberate null deref

tools/valgrind/drmemory/suppressions.txt

 ############################
 # Known reports on the third party we have no control over.
 UNINITIALIZED READ
 ...
 *!deflate_*
 *!MOZ_Z_deflate
 
 # http://code.google.com/p/drmemory/issues/detail?id=406
 # TODO(timurrrr): check if these frames change when NT_SYMBOLS are present.
 LEAK
@@ skipping 696 matching lines @@
 # function is small, little risk for false negative in rest of it
 # TODO(bruening): switch to annotation once have support for that
 chrome.dll!HandleRendererErrorTestParameters
 
 UNADDRESSABLE ACCESS
 name=NPAPITesterBase.NoHangIfInitCrashes deliberate crash
 # function is small, little risk for false negative in rest of it
 # TODO(bruening): switch to annotation once have support for that
 npapi_test_plugin.dll!NPAPIClient::PluginClient::Initialize
 
+# This test intentionally leaks an object and checks that it's never deleted
+LEAK

[Timur Iskhodzhanov, 2011/11/18 20:05:40]

can you add name= lines to these suppressions?

[Derek Bruening, 2011/12/07 14:53:59]

meant to do this next time I changed this file but hasn't happened yet.  Reid if
you touch this file before me can you add names.

[Timur Iskhodzhanov, 2011/12/07 15:01:29]

I already did that.

On 2011/12/07 14:53:59, Derek Bruening wrote:

+...
+*!BrowserThreadTest_NotReleasedIfTargetThreadNonExistent_Test::TestBody
+
+# This test has a death test at the end that derefences a NULL pointer.
+UNADDRESSABLE ACCESS
+...
+unit_tests.exe!BrowserAboutHandlerTest_WillHandleBrowserAboutURL_Test::TestBody

[Timur Iskhodzhanov, 2011/12/07 13:32:16]

Do you remember what was the stack?
This one matches the reports in http://crbug.com/96010 which is a different
issue

[Derek Bruening, 2011/12/07 14:53:59]

This diff came straight from Reid and I just posted it as he didn't have commit
access at the time.

Reid did you analyze this test and see a "death test"?  b/c 96010 is a real bug
not a "death test" it seems.

[Reid Kleckner, 2011/12/07 15:04:23]

Right:
http://codereview.chromium.org/8585046/
The last line of the test is:
  EXPECT_DEATH(HandleNonNavigationAboutURL(url), "");

Unfortunately I didn't link to any of the failing builds I was looking at at the
time.

+
 ############################
 # Benign issues in Chromium
 
 # We deliberately leak different kinds of Histograms to avoid data races.
 LEAK
 ...
 *!operator new*
 ...
 *!*::*Histogram::FactoryGet
 
@@ skipping 24 matching lines @@
 *!notifier::Login::StartConnection
 *!sync_notifier::InvalidationNotifier::UpdateCredentials
 *!sync_notifier::NonBlockingInvalidationNotifier::Core::UpdateCredentials
 
 LEAK
 ...
 *!TestURLRequestContext::TestURLRequestContext
 *!TestURLRequestContextGetter::GetURLRequestContext
 *!URLFetcher::Core::StartURLRequest
 
-# This test intentionally leaks an object and checks that it's never deleted
-LEAK
-...
-*!BrowserThreadTest_NotReleasedIfTargetThreadNonExistent_Test::TestBody
-
 WARNING
 # http://code.google.com/p/chromium/issues/detail?id=72463
 name=Ci#72463: prefetches in generated MemCopy
 instruction=prefetch*
 <not in a module>                  
 chrome.dll!v8::internal::CopyChars*
 
 UNADDRESSABLE ACCESS
 # http://code.google.com/p/drmemory/issues/detail?id=582
 name=bizarre cl-generated read-beyond-TOS
@@ skipping 152 matching lines @@
 *!operator new
 *!views::NativeWidgetWin::OnCreate
 
 # http://crbug.com/96010
 UNADDRESSABLE ACCESS
 ...
 *!NotificationService::Notify
 
 # Empty line at EOF - workaround for
 # http://code.google.com/p/drmemory/issues/detail?id=41