The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/third_party/autotest/+/f7aa28887f190e17043c3d21a232c8b446fc05b6

commit f7aa28887f190e17043c3d21a232c8b446fc05b6
Author: Ben Kwa <kenobi@google.com>
Date: Sat Nov 18 04:18:31 2017

[autotest] Fix lxc_cleanup.

- Update lxc_cleanup.py to use container.id, not container.get_id.
- Fix a typo (hqe vs hqes).

BUG= chromium:786219 
TEST=Create local containers with bogus PIDs, run lxc_cleanup.py and
     watch it clean those containers up.

Change-Id: I278d06d70557679dbc1c264cb2fa611441e42dac
Reviewed-on: https://chromium-review.googlesource.com/775675
Commit-Ready: Ben Kwa <kenobi@chromium.org>
Tested-by: Ben Kwa <kenobi@chromium.org>
Reviewed-by: Ilja H. Friedel <ihf@chromium.org>
Reviewed-by: Ben Kwa <kenobi@chromium.org>

[modify] https://crrev.com/f7aa28887f190e17043c3d21a232c8b446fc05b6/site_utils/lxc_cleanup.py

The version requested to be installed is supported on xenial.  build116-b1 is currently running trusty.

o.k to migrate build16-b1 to xenial?

s/build116-b1/build115-b1 above.

Yes, thanks Peter, upgrading build115-b1 to Xenial sounds good.

Do we have other bots running Xenial?
I recall a few months ago there was an issue with different version of protobuf on it than the one chrome compiles with. We might run into problems of chromium compatibility with xenial if it is not tested on other bots.

There is https://uberchromegw.corp.google.com/i/chromium.gpu.fyi/builders/Linux%20Release%20(Intel%20HD%20630) which is running zesty and it seems o.k ?

Cool.

The latest is 17.40 but it did not load.  See the Xorg.0.log snippet below.

However, the previous version, 17:30 does work.  That is what is installed.
It will be effective starting with https://uberchromegw.corp.google.com/i/chromium.gpu.fyi/builders/Linux%20Release%20%28AMD%20R7%20240%29/builds/3455

From build115-b1:/var/log/Xorg.0.log

[     7.893] (II) Loading /opt/amdgpu-pro/lib/xorg/modules/extensions/libglx.so
[     7.894] (II) Module glx: vendor="X.Org Foundation"
[     7.894]    compiled for 1.19.0, module version = 1.0.0
[     7.894]    ABI class: X.Org Server Extension, version 10.0
[     7.894] (EE) module ABI major version (10) doesn't match the server's version (9)
[     7.894] (II) UnloadModule: "glx"
[     7.894] (II) Unloading glx
[     7.894] (EE) Failed to load module "glx" (module requirement mismatch, 0)
[     7.894] (II) Applying OutputClass "amdgpu-pro" to /dev/dri/card0
[     7.894]    loading driver: amdgpu
[     7.894] (II) Applying OutputClass "AMDgpu" to /dev/dri/card0
[     7.894]    loading driver: amdgpu
[     7.894] (==) Matched amdgpu as autoconfigured driver 0
[     7.894] (==) Matched amdgpu as autoconfigured driver 1
[     7.894] (==) Matched ati as autoconfigured driver 2
[     7.894] (==) Matched ati as autoconfigured driver 3
[     7.894] (==) Matched modesetting as autoconfigured driver 4
[     7.894] (==) Matched fbdev as autoconfigured driver 5
[     7.894] (==) Matched vesa as autoconfigured driver 6
[     7.894] (==) Assigned the driver to the xf86ConfigLayout
[     7.894] (II) LoadModule: "amdgpu"
[     7.894] (II) Loading /opt/amdgpu-pro/lib/xorg/modules/drivers/amdgpu_drv.so
[     7.894] (II) Module amdgpu: vendor="X.Org Foundation"
[     7.894]    compiled for 1.19.3, module version = 1.3.99
[     7.894]    Module class: X.Org Video Driver
[     7.894]    ABI class: X.Org Video Driver, version 23.0
[     7.894] (EE) module ABI major version (23) doesn't match the server's version (20)
[     7.894] (II) UnloadModule: "amdgpu"
[     7.894] (II) Unloading amdgpu
[     7.894] (EE) Failed to load module "amdgpu" (module requirement mismatch, 0)

Thanks very much Peter.

If AMD's latest proprietary driver is kind of broken then maybe we should bite the bullet and switch to the open-source one later. But let's see how the upgrade looks first.

Attempting to CC: Paul Blinzer from AMD so he can see the xorg log.

 Issue 787947  has been merged into this issue.

It's not clear to me from the comments here what status is, but I can tell you it broke at https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Release%20%28AMD%20R7%20240%29/builds/3455

Can we switch back to something that works?

Since there isn't anything private in these logs, un-restricting the view so our colleagues from AMD can see it.

Frank, you're right, this driver looks pretty badly broken. Here's a more permanent log:

https://ci.chromium.org/buildbot/chromium.gpu.fyi/Linux%20Release%20(AMD%20R7%20240)/3455

Judging from the many test failures as well as examining some of the logs:
WebglConformance_conformance2_textures_video_tex_3d_r16f_red_float
WebglConformance_conformance2_textures_video_tex_3d_r16f_red_half_float
WebglConformance_conformance2_textures_video_tex_3d_r8_red_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_r8ui_red_integer_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_rg8_rg_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_rg8ui_rg_integer_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_rgb565_rgb_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_rgb565_rgb_unsigned_short_5_6_5
WebglConformance_conformance2_textures_video_tex_3d_rgb5_a1_rgba_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_rgb5_a1_rgba_unsigned_short_5_5_5_1
WebglConformance_conformance2_textures_video_tex_3d_rgb9_e5_rgb_float
WebglConformance_conformance2_textures_video_tex_3d_rgb9_e5_rgb_half_float
WebglConformance_conformance2_textures_video_tex_3d_rgba4_rgba_unsigned_byte
WebglConformance_conformance2_textures_video_tex_3d_rgba4_rgba_unsigned_short_4_4_4_4
WebglConformance_deqp_functional_gles3_fbocolorbuffer_tex3d_00
WebglConformance_deqp_functional_gles3_fbocolorbuffer_tex3d_01
WebglConformance_deqp_functional_gles3_fbocolorbuffer_tex3d_02
WebglConformance_deqp_functional_gles3_fbocolorbuffer_tex3d_03
WebglConformance_deqp_functional_gles3_fbocolorbuffer_tex3d_04
WebglConformance_deqp_functional_gles3_fbocolorbuffer_tex3d_05
WebglConformance_deqp_functional_gles3_shadertexturefunction_texelfetch
WebglConformance_deqp_functional_gles3_shadertexturefunction_texturegradoffset
WebglConformance_deqp_functional_gles3_shadertexturefunction_texturelod
WebglConformance_deqp_functional_gles3_shadertexturefunction_texturelodoffset
WebglConformance_deqp_functional_gles3_shadertexturefunction_textureoffset
WebglConformance_deqp_functional_gles3_shadertexturefunction_textureproj
WebglConformance_deqp_functional_gles3_shadertexturefunction_textureprojgrad
WebglConformance_deqp_functional_gles3_shadertexturefunction_textureprojgradoffset
WebglConformance_deqp_functional_gles3_shadertexturefunction_textureprojlod
WebglConformance_deqp_functional_gles3_shadertexturefunction_textureprojlodoffset

It looks like something's broken with 3D texture sampling in this driver.

The angle_end2end_tests failure to start:
X Error of failed request:  GLXBadContext
  Major opcode of failed request:  155 (GLX)
  Minor opcode of failed request:  4 (X_GLXDestroyContext)
  Serial number of failed request:  61
  Current serial number in output stream:  64


is concerning.

It looks like something's broken in EXT_blend_func_extended too judging from these failures in gl_tests:
TranslatorVariants/EXTBlendFuncExtendedDrawTest.ESSL1FragColor/0
TranslatorVariants/EXTBlendFuncExtendedDrawTest.ESSL1FragColor/1


The failure of GLSurfaceEGLTest.SurfaceFormatTest in gl_unittests is also new:
[ RUN      ] GLSurfaceEGLTest.SurfaceFormatTest
[31276:31276:1122/114445.035322:11189741074:ERROR:gl_surface_egl.cc(464)] No suitable EGL configs found.
[31276:31276:1122/114445.035356:11189741106:ERROR:gl_context_egl.cc(81)] eglGetConfigAttrib failed with error EGL_BAD_CONFIG
[31276:31276:1122/114445.035371:11189741120:FATAL:scoped_refptr.h(211)] Check failed: ptr_. 
#0 0x0000004f4f2c base::debug::StackTrace::StackTrace()
#1 0x0000004f27dc logging::LogMessage::~LogMessage()
#2 0x0000003921c8 gl::GLSurfaceEGL::InitializeOneOff()
#3 0x00000055bf5b gl::init::InitializeGLOneOffPlatform()
#4 0x000000557b8a gl::init::InitializeGLOneOffImplementation()
#5 0x0000004e8cf8 gl::GLSurfaceTestSupport::InitializeOneOffImplementation()
#6 0x0000002d4530 gl::(anonymous namespace)::GLSurfaceEGLTest_SurfaceFormatTest_Test::TestBody()


This test was running in the last known good build:
https://ci.chromium.org/buildbot/chromium.gpu.fyi/Linux%20Release%20(AMD%20R7%20240)/3452


Peter, would it be possible to leave this machine running Xenial and still install the older driver that was running on this machine previously? If not then let me know and we'll figure out the path forward. (Frank, I would prefer not to revert back to the machine's old OS because that is a dead end.) Thanks.

P.S. Frank, I'm also asking AMD for some guidance about what to do here.

Maybe we should try Zesty instead of Xenial?

I'm inclined to request that, and also to just use the open-source AMDGPU driver (which I think is just part of Mesa) rather than the partially-closed-source AMDGPU PRO driver.

Let me get some guidance from AMD before making that request though. They've already responded to one email today so hopefully we should know what to do shortly.

In it's prior life build115-b1 was running trusty with the fglrx driver which has been depreciated in Xenial in favor of amdgpu-pro  (Canonical includes a build of amdgpu-pro in xenial).

Then there is the radeon port.  Not sure if you want to deal with that?

AMD's advice was to suppress the test failures on the newer AMDGPU PRO driver rather than try the open-source driver. :(

Unfortunately, this would be quite messy because some of the test suites are completely broken.

Peter, I'm sorry to ask, but would it be a huge burden to reimage the machine with trusty and the old driver so we can get it green again? AMD's going to look into the xorg loading issue with the 17.40 driver, so maybe we can try the update again once they've confirmed they've shipped a new driver that they've tested.

What do you think? If this is too much of a pain, I can suppress the failures.

We could also try Zesty with 17.40.
From the log "module ABI major version (23) doesn't match the server's version (20)" seems like the driver is newer than X. Maybe it would work on a newer OS.

And if it doesn't, we can go back to Trusty.

Can I do this on Monday?  

Pretty please?

Also if we revert build1115-b1 how about standing up a separate new host to test/debug the amdgpu-pro driver?

Peter, of course, don't worry about the bot now or over the holiday. Having it be red for the next couple of days is not that big a deal.

Tell you what, on Monday, could you try upgrading it to Zesty and try the 17.40 AMDGPU PRO driver again?

Thanks again for your help.

o.k I think I figured out a way around.  The 17.40 package was built for xorg 1.19 and that comes with the Xenial LTS entitlement stack.

Let me try that.

That did the trick.

chrome-bot@build115-b1:~$ modinfo amdgpu |grep version
version:        17.40.3.12

This is effective starting with https://ci.chromium.org/buildbot/chromium.gpu.fyi/Linux%20Release%20%28AMD%20R7%20240%29/3494

Fantastic, thank you Peter! The newer driver seems a lot more reliable.

Yuly, what do you think we should do about the few remaining failures on this bot? I think what we need to do is log on to it and try the angle_end2end_tests target directly, and figure out why it's getting a GLXBadContext error (and the other test runs don't).

The error looks to be coming from GLXDestroyContext. ANGLE creates and destroys each available config once on startup to determine if the config type is supported. It's possible that the tear down code is crashing, and that the other tests simply don't tear down this way.

Xlib:  extension "AMDGPU" missing on display ":0.0".
Xlib:  extension "AMDGPU" missing on display ":0.0".
Xlib:  extension "AMDGPU" missing on display ":0.0".
Xlib:  extension "AMDGPU" missing on display ":0.0".
X Error of failed request:  GLXBadContext
  Major opcode of failed request:  155 (GLX)
  Minor opcode of failed request:  4 (X_GLXDestroyContext)
                                        ^^^^^^^^^^^^^^^^^

So, angle_end2end_tests crash calling XCloseDisplay 2nd time here:
https://chromium.googlesource.com/angle/angle/+/master/util/x11/X11Window.cpp#283
With error:
X Error of failed request:  GLXBadContext
  Major opcode of failed request:  155 (GLX)
  Minor opcode of failed request:  4 (X_GLXDestroyContext)
  Serial number of failed request:  61
  Current serial number in output stream:  64

I've commented it out, but then it started crashing calling glXQueryExtension here:
https://chromium.googlesource.com/angle/angle/+/master/src/libANGLE/renderer/gl/glx/FunctionsGLX.cpp#187
With a similar error:
X Error of failed request:  GLXBadContext
  Major opcode of failed request:  155 (GLX)
  Minor opcode of failed request:  4 (X_GLXDestroyContext)
  Serial number of failed request:  102
  Current serial number in output stream:  103

I believe something is wrong with the drivers or their compatibility with X.

Argh. Sorry. Told AMD about these continued failures.

I'm not sure what we should do. It seems a bad idea to downgrade to the ancient driver again. Can we suppress the other random test failures and temporarily stop running angle_end2end_tests on this machine? Or is there something more drastic we could do in this test harness to avoid creating OpenGL context configurations which are going to fail?

Yes, disabling end2end and investigating other failures sounds good.
I was going to attempt to reproduce the error in a few lines of code and
let AMG debug.
We could also try a fresh install of 16.04.3 to avoid potential X / old
driver / new driver conflicts.

Thanks Yuly.

AMD responded and it sounds like they already have reproduction of these X issues, and are starting to institute the same quality control for the Linux driver (including WebGL conformance runs and other tests) that's been present for their Windows driver for a while. So I think no more effort is needed here beyond suppressing the current failures on this bot.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a9a2d2a06b5b73525c319796727811d4007c4862

commit a9a2d2a06b5b73525c319796727811d4007c4862
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Fri Dec 22 01:48:37 2017

Disable angle_end2end_tests on Linux Release (AMD R7 240)

Temporarily due to problems with AMDGPU-PRO driver.

BUG= 786219 

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I809dd961e0d85eb496c7567f05151c0f3c50b547
Reviewed-on: https://chromium-review.googlesource.com/841042
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#525889}
[modify] https://crrev.com/a9a2d2a06b5b73525c319796727811d4007c4862/content/test/gpu/generate_buildbot_json.py
[modify] https://crrev.com/a9a2d2a06b5b73525c319796727811d4007c4862/testing/buildbot/chromium.gpu.fyi.json

Since I've disable angle_end2end_tests, the following fails on this bot consistently:

webgl_conformance_gl_passthrough_tests:
- WebglConformance_conformance_renderbuffers_depth_renderbuffer_initialization
  AssertionError: should be 0,0,255,255
  at (2, 0) expected: 0,0,255,255 was 255,255,102,102
  FAIL should be 0,0,255,255
  at (2, 0) expected: 0,0,255,255 was 255,255,102,102
  should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 51,51,153,153
  FAIL should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 51,51,153,153
  should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 0,0,0,0
  FAIL should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 0,0,0,0
  should be 0,0,255,255
  at (8, 8) expected: 0,0,255,255 was 0,0,0,0
  FAIL should be 0,0,255,255
  at (8, 8) expected: 0,0,255,255 was 0,0,0,0
  should be 0,0,255,255
  at (2, 0) expected: 0,0,255,255 was 255,255,102,102
  FAIL should be 0,0,255,255
  at (2, 0) expected: 0,0,255,255 was 255,255,102,102
  should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 0,0,0,0
  FAIL should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 0,0,0,0

- WebglConformance_conformance_renderbuffers_stencil_renderbuffer_initialization
  AssertionError: should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 0,0,0,0
  FAIL should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 0,0,0,0
  should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 153,170,102,51
  FAIL should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 153,170,102,51
  should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 153,170,102,51
  FAIL should be 0,0,255,255
  at (0, 0) expected: 0,0,255,255 was 153,170,102,51

gl_tests:
- TranslatorVariants/EXTBlendFuncExtendedDrawTest.ESSL1FragColor/0
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xCC' (204)
  actual
    Which is: '\x7F' (127)
 at 25, 75 channel 0
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 25, 75 channel 1
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 25, 75 channel 2
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: 'r' (114, 0x72)
  actual
    Which is: '\xFF' (255)
 at 25, 75 channel 3
../../gpu/command_buffer/tests/gl_ext_blend_func_extended_unittest.cc:186: Failure
Value of: GLTestHelper::CheckPixels(kWidth / 4, (3 * kHeight) / 4, 1, 1, 1, color, nullptr)
  Actual: false
Expected: true
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xCC' (204)
  actual
    Which is: '\x7F' (127)
 at 99, 0 channel 0
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 99, 0 channel 1
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 99, 0 channel 2
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: 'r' (114, 0x72)
  actual
    Which is: '\xFF' (255)
 at 99, 0 channel 3
../../gpu/command_buffer/tests/gl_ext_blend_func_extended_unittest.cc:188: Failure
Value of: GLTestHelper::CheckPixels(kWidth - 1, 0, 1, 1, 1, color, nullptr)
  Actual: false
Expected: true

- TranslatorVariants/EXTBlendFuncExtendedDrawTest.ESSL1FragColor/1
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xCC' (204)
  actual
    Which is: '\x7F' (127)
 at 25, 75 channel 0
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 25, 75 channel 1
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 25, 75 channel 2
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: 'r' (114, 0x72)
  actual
    Which is: '\xFF' (255)
 at 25, 75 channel 3
../../gpu/command_buffer/tests/gl_ext_blend_func_extended_unittest.cc:186: Failure
Value of: GLTestHelper::CheckPixels(kWidth / 4, (3 * kHeight) / 4, 1, 1, 1, color, nullptr)
  Actual: false
Expected: true
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xCC' (204)
  actual
    Which is: '\x7F' (127)
 at 99, 0 channel 0
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 99, 0 channel 1
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: '\xFF' (255)
  actual
    Which is: '\x7F' (127)
 at 99, 0 channel 2
../../gpu/command_buffer/tests/gl_test_utils.cc:183: Failure
Expected equality of these values:
  expected
    Which is: 'r' (114, 0x72)
  actual
    Which is: '\xFF' (255)
 at 99, 0 channel 3
../../gpu/command_buffer/tests/gl_ext_blend_func_extended_unittest.cc:188: Failure
Value of: GLTestHelper::CheckPixels(kWidth - 1, 0, 1, 1, 1, color, nullptr)
  Actual: false
Expected: true

gl_unittests:
- GLImageNativePixmapTest.GLTexture2DToDmabuf
[3802:3802:1221/205206.867124:7662394531:ERROR:gl_surface_egl.cc(464)] No suitable EGL configs found.
[3802:3802:1221/205206.867169:7662394573:ERROR:gl_context_egl.cc(81)] eglGetConfigAttrib failed with error EGL_BAD_CONFIG
[3802:3802:1221/205206.867190:7662394593:FATAL:scoped_refptr.h(208)] Check failed: ptr_. 
#0 0x0000004f4c6c base::debug::StackTrace::StackTrace()
#1 0x0000004fb20c logging::LogMessage::~LogMessage()
#2 0x000000394a38 gl::GLSurfaceEGL::InitializeOneOffCommon()
#3 0x0000003941c5 gl::GLSurfaceEGL::InitializeOneOff()
#4 0x00000056220b gl::init::InitializeGLOneOffPlatform()
#5 0x00000055de1a gl::init::InitializeGLOneOffImplementation()
#6 0x0000004ee87a gl::GLSurfaceTestSupport::InitializeOneOffImplementation()
#7 0x0000002d9062 gl::(anonymous namespace)::GLImageNativePixmapTest::SetUp()
#8 0x0000003351b1 testing::Test::Run()
#9 0x000000335a20 testing::TestInfo::Run()
#10 0x000000335f07 testing::TestCase::Run()
#11 0x00000033c467 testing::internal::UnitTestImpl::RunAllTests()
#12 0x00000033c0b7 testing::UnitTest::Run()
#13 0x000000596502 base::TestSuite::Run()
#14 0x0000005aaa87 base::(anonymous namespace)::LaunchUnitTestsInternal()
#15 0x0000005aa8fb base::LaunchUnitTests()
#16 0x0000002daf52 main
#17 0x7f03b1e35830 __libc_start_main
#18 0x0000002b402a _start

- GLSurfaceEGLTest.SurfaceFormatTest
[3800:3800:1221/205206.862869:7662390274:ERROR:gl_surface_egl.cc(464)] No suitable EGL configs found.
[3800:3800:1221/205206.862906:7662390309:ERROR:gl_context_egl.cc(81)] eglGetConfigAttrib failed with error EGL_BAD_CONFIG
[3800:3800:1221/205206.862920:7662390322:FATAL:scoped_refptr.h(208)] Check failed: ptr_. 
#0 0x0000004f4c6c base::debug::StackTrace::StackTrace()
#1 0x0000004fb20c logging::LogMessage::~LogMessage()
#2 0x000000394a38 gl::GLSurfaceEGL::InitializeOneOffCommon()
#3 0x0000003941c5 gl::GLSurfaceEGL::InitializeOneOff()
#4 0x00000056220b gl::init::InitializeGLOneOffPlatform()
#5 0x00000055de1a gl::init::InitializeGLOneOffImplementation()
#6 0x0000004ee87a gl::GLSurfaceTestSupport::InitializeOneOffImplementation()
#7 0x0000002d4420 gl::(anonymous namespace)::GLSurfaceEGLTest_SurfaceFormatTest_Test::TestBody()
#8 0x000000335246 testing::Test::Run()
#9 0x000000335a20 testing::TestInfo::Run()
#10 0x000000335f07 testing::TestCase::Run()
#11 0x00000033c467 testing::internal::UnitTestImpl::RunAllTests()
#12 0x00000033c0b7 testing::UnitTest::Run()
#13 0x000000596502 base::TestSuite::Run()
#14 0x0000005aaa87 base::(anonymous namespace)::LaunchUnitTestsInternal()
#15 0x0000005aa8fb base::LaunchUnitTests()
#16 0x0000002daf52 main
#17 0x7fcdb1914830 __libc_start_main
#18 0x0000002b402a _start

Going to disable them, too.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e08ee6def83ce5f5a7c15b74a70d73dfbf372149

commit e08ee6def83ce5f5a7c15b74a70d73dfbf372149
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Fri Dec 22 23:36:06 2017

Suppress EXTBlendFuncExtendedDrawTest.ESSL1FragColor on Linux AMD

Likely due to AMDGPU-PRO driver bug.

BUG= 786219 
TBR=kbr@chromium.org,zmo@chromium.org

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I920e71fb110ec0ec53e96fe3ec2efa6259e74871
Reviewed-on: https://chromium-review.googlesource.com/843168
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#526091}
[modify] https://crrev.com/e08ee6def83ce5f5a7c15b74a70d73dfbf372149/gpu/command_buffer/tests/gl_ext_blend_func_extended_unittest.cc

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/149f0c18b2abf9aa2757deb68e71c42d7b404e10

commit 149f0c18b2abf9aa2757deb68e71c42d7b404e10
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Sat Dec 23 00:25:10 2017

Suppress couple WebGL CTS on Linux AMD passthrough

conformance/renderbuffers/depth-renderbuffer-initialization.html
conformance/renderbuffers/stencil-renderbuffer-initialization.html

BUG= 794339 , 786219 
TBR=kbr@chromium.org

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I50b57532022b9a6456a0288a5326887d5a3fb523
Reviewed-on: https://chromium-review.googlesource.com/843512
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#526105}
[modify] https://crrev.com/149f0c18b2abf9aa2757deb68e71c42d7b404e10/content/test/gpu/gpu_tests/webgl_conformance_expectations.py

Actually, I'm reluctant about disabling gl_unittests.
I don't see an obvious way to disable them only on AMD, and I wouldn't like to disable them on all Linux.

Ken, could you own handling gl_unittests crashes?

From the stack, I think this happens in tests that call
GLSurfaceTestSupport::InitializeOneOffImplementation(GLImplementation::kGLImplementationEGLGLES2, true);

The crash I think is in
    scoped_refptr<GLContext> context = InitializeGLContext(new GLContextEGL(nullptr), surface.get(), GLContextAttribs());
called from GLSurfaceEGL::InitializeOneOffCommon() where it tries to check if GL_OES_surfaceless_context is supported.

Ken, looks like this was off your radar.
Assigning this to you to decide what to do with the failing gl_unittests.
Perhaps disable them alltogether on this bot?

Actually, I went ahead and uploaded https://chromium-review.googlesource.com/c/chromium/src/+/855048, so you just need to +1 it if you like this approach.

Thanks Yuly for putting that together and sorry for not looking into it until this point. Pretty swamped here. It seems OK to me to disable these tests just on this bot until AMD ships a driver that is more robust.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/256e4d847565b3512850efc93961034f2dc0669f

commit 256e4d847565b3512850efc93961034f2dc0669f
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Tue Jan 09 06:39:08 2018

Disable gl_unittests on Linux Release (AMD R7 240)

Temporarily due to problems with AMDGPU-PRO driver.

BUG= 786219 

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: Ibe65165cbf305bda7067733cfecee78cab3c0288
Reviewed-on: https://chromium-review.googlesource.com/855048
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#527916}
[modify] https://crrev.com/256e4d847565b3512850efc93961034f2dc0669f/content/test/gpu/generate_buildbot_json.py
[modify] https://crrev.com/256e4d847565b3512850efc93961034f2dc0669f/testing/buildbot/chromium.gpu.fyi.json

Bot is green as of https://ci.chromium.org/buildbot/chromium.gpu.fyi/Linux%20Release%20%28AMD%20R7%20240%29/3890

Not sure if you want to close this, or keep for tracking AMD driver fixes.

Thanks Peter for your help upgrading the machine and driver, and Yuly for getting the bot green. Let's close this and file follow-on bugs for the remaining issues.