New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 595823 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
Closed: Mar 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug



Sign in to add a comment

UNADDRESSABLE ACCESS in SignalTest

Project Member Reported by osh...@chromium.org, Mar 17 2016

Issue description

Started on this build:

https://build.chromium.org/p/chromium.memory.fyi/builders/Windows%20Content%20Browser%20%28DrMemory%29/builds/6897

looks like these two CLs are culprit:

https://codereview.chromium.org/1809553002 (blink: Remove unused or rarely used methods from WebGraphicsContext3D)
https://codereview.chromium.org/1807813002 (Remove WebGraphicsContext3D dependency in gpu_context_tests.h)

danakj@, since this is ANADDRESSABLE ACCESS, I'll revert them unless I hear from you soon. (I'll ping you)
Let me know if you're already working on a fix, or you think your CL is unrelated.

log:
https://build.chromium.org/p/chromium.memory.fyi/builders/Windows%20Content%20Browser%20%28DrMemory%29/builds/6897/steps/memory%20test%3A%20content_browsertests/logs/stdio

UNADDRESSABLE ACCESS: reading 0x00000000-0x00000004 4 byte(s)
# 0 `anonymous namespace'::SignalTest_InvalidSignalSyncTokenTest_Test::RunTestOnMainThread [content\common\gpu\client\gpu_context_tests.h:67]
# 1 content::ContentBrowserTest::RunTestOnMainThreadLoop                       [content\public\test\content_browser_test.cc:134]
# 2 content::BrowserTestBase::ProxyRunTestOnMainThreadLoop                     [content\public\test\browser_test_base.cc:309]
# 3 content::ShellBrowserMainParts::PreMainMessageLoopRun                      [content\shell\browser\shell_browser_main_parts.cc:171]
# 4 content.dll!content::BrowserMainLoop::PreMainMessageLoopRun                [content\browser\browser_main_loop.cc:942]
# 5 content.dll!base::internal::Invoker<>::Run                                 [base\bind_internal.h:352]
# 6 content.dll!content::StartupTaskRunner::RunAllTasksNow                     [content\browser\startup_task_runner.cc:45]
# 7 content.dll!content::BrowserMainLoop::CreateStartupTasks                   [content\browser\browser_main_loop.cc:817]
# 8 content.dll!content::BrowserMainRunnerImpl::Initialize                     [content\browser\browser_main_runner.cc:137]
# 9 ShellBrowserMain                                                           [content\shell\browser\shell_browser_main.cc:23]
#10 content::ShellMainDelegate::RunProcess                                     [content\shell\app\shell_main_delegate.cc:275]
#11 content.dll!content::RunNamedProcessTypeMain                               [content\app\content_main_runner.cc:386]
#12 content.dll!content::ContentMainRunnerImpl::Run                            [content\app\content_main_runner.cc:772]
#13 content.dll!content::ContentMain                                           [content\app\content_main.cc:19]
#14 content::BrowserTestBase::SetUp                                            [content\public\test\browser_test_base.cc:282]
#15 content::ContentBrowserTest::SetUp                                         [content\public\test\content_browser_test.cc:92]
#16 testing::internal::HandleExceptionsInMethodIfSupported<>                   [testing\gtest\src\gtest.cc:2458]
Note: @0:00:18.305 in thread 1900
Note: instruction: mov    (%eax) -> %esi
The report came from the `SignalTest.InvalidSignalSyncTokenTest` test.
Suppression (error hash=#025835F7B91A8833#):
For more info on using suppressions see http://dev.chromium.org/developers/how-tos/using-drmemory#TOC-Suppressing-error-reports-from-the-
{
UNADDRESSABLE ACCESS
name=<insert_a_suppression_name_here>
*!`anonymous namespace'::SignalTest_InvalidSignalSyncTokenTest_Test::RunTestOnMainThread
*!content::ContentBrowserTest::RunTestOnMainThreadLoop
*!content::BrowserTestBase::ProxyRunTestOnMainThreadLoop
*!content::ShellBrowserMainParts::PreMainMessageLoopRun
content.dll!content::BrowserMainLoop::PreMainMessageLoopRun
content.dll!base::internal::Invoker<>::Run
content.dll!content::StartupTaskRunner::RunAllTasksNow
content.dll!content::BrowserMainLoop::CreateStartupTasks
content.dll!content::BrowserMainRunnerImpl::Initialize
*!ShellBrowserMain
*!content::ShellMainDelegate::RunProcess
content.dll!content::RunNamedProcessTypeMain
content.dll!content::ContentMainRunnerImpl::Run
content.dll!content::ContentMain
*!content::BrowserTestBase::SetUp
*!content::ContentBrowserTest::SetUp
*!testing::internal::HandleExceptionsInMethodIfSupported<>
}
 

Comment 1 by danakj@chromium.org, Mar 17 2016

Cc: zmo@chromium.org piman@chromium.org

Comment 2 by danakj@chromium.org, Mar 17 2016

Status: Started (was: Assigned)

Comment 3 by danakj@chromium.org, Mar 17 2016

Cc: georgesak@chromium.org danakj@chromium.org
 Issue 595832  has been merged into this issue.

Comment 4 by danakj@chromium.org, Mar 17 2016

Oh I think I know what's going on. I removed some early outs here: https://codereview.chromium.org/1807813002/diff/1/content/common/gpu/client/gpu_context_tests.h?context=&column_width=80&tab_spaces=8

I was like "let's see if it breaks?" i hate test early outs.

https://bugs.chromium.org/p/chromium/issues/detail?id=595832 is fun and interesting. SignalTest.BasicSignalQueryTest has a backtrace that includes BrowserGpuChannelHostFactoryTest_DISABLED_GrContextKeepsGpuChannelAlive_Test::RunTestOnMainThread.

Why is that test running some other test?

That said, the DISABLED in the middle of the string is clearly wrong and not doing what it was meant to do.

Disabling tests with macros continues to be terrible.

Comment 5 by danakj@chromium.org, Mar 17 2016

https://codereview.chromium.org/1812033002/ fixes the SignalTests
Project Member

Comment 6 by bugdroid1@chromium.org, Mar 17 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1bbe2d3a6aa7e138eb968dafd058efb8c440a84d

commit 1bbe2d3a6aa7e138eb968dafd058efb8c440a84d
Author: danakj <danakj@chromium.org>
Date: Thu Mar 17 21:40:18 2016

Restore earlyouts in SignalTests on Windows for IPC flavour of the tests

These tests fail to set up correctly on Windows, so running the tests
just crashes. Restore the early outs with an ifdef for windows so we
document where the early out will happen, along with comments.

R=kbr@chromium.org, piman@chromium.org
BUG= 595823 ,  584497 

Review URL: https://codereview.chromium.org/1812033002

Cr-Commit-Position: refs/heads/master@{#381802}

[modify] https://crrev.com/1bbe2d3a6aa7e138eb968dafd058efb8c440a84d/content/browser/gpu/gpu_ipc_browsertests.cc
[modify] https://crrev.com/1bbe2d3a6aa7e138eb968dafd058efb8c440a84d/content/common/gpu/client/gpu_context_tests.h

Comment 7 by danakj@chromium.org, Mar 17 2016

Status: Fixed (was: Started)

Sign in to add a comment