Change-Id: Ide2e905f8bdfb44368ba7b75b2517f895796a85d

Change-Id: Icc5a10dac69c1cffd275ab0847e93388b751dc32

There is a conflict between libgbm and our custom libminigbm. To make it
work, run angle with environment variable `LD_PRELOAD=libminigbm.so`.

Change-Id: I45b33533983296dbad29d9c56c9dc2fab41b9178

Constantine Shablya authored Mar 22, 2022 and Antonio Caggiano committed Mar 23, 2022

not sure if there's anything missing still

Change-Id: I48d7f0c2e7d9d7dc8f4ab65d4acbc72c52b6fe84

Constantine Shablya authored Mar 19, 2022 and Antonio Caggiano committed Mar 21, 2022

Change-Id: If144a6306a8035289d0086b308823008f5ae3cc3

Constantine Shablya authored Mar 16, 2022 and Antonio Caggiano committed Mar 18, 2022

Change-Id: Idc1f57bd29b067e00a21d8d5375af2d01b0f6b3b

Constantine Shablya authored Mar 16, 2022 and Antonio Caggiano committed Mar 18, 2022

Change-Id: If9daedd3621ff1935da72c342e37e9710af33fdd

Constantine Shablya authored Mar 04, 2022 and Antonio Caggiano committed Mar 18, 2022

Change-Id: Ib585983dad81f957c2b109b40d723eaa5b1b8c28

Change-Id: I61f3b1d50ea8bb6a8c8636a31b7c111fb2f4d2b1

Add support for GBM platform by implementing a Display with no WSI
extension and no swapchain usage.

Change-Id: Ia9089131c5984dfdd926d2f85f1c218df1e84d9a

A DRM format may be converted into more than one Vulkan format,
therefore we try to create a dmabuf image with all candidate Vulkan
formats until we succeed.

Bug: angleproject:7065
Change-Id: I376d10a596ce81347c8ae21464c2c2db048f01d2

Improve GetPlatformDisplay by supporting EGL_PLATFORM_WAYLAND_EXT.

Bug: angleproject:6902
Change-Id: Ia4c342097f47b7c41778ac02517451f51a94d8be

Use switch constructs instead of if-else to allow reuse of code and
simplify addition of other platforms as new cases.

Bug: angleproject:6902
Change-Id: Ia941d0d7bc03538bbc8d96ebece60224994a6c66

Add a platform parameter to this function to possibly accept other
platforms besides EGL_PLATFORM_ANGLE_ANGLE.

Bug: angleproject:6902
Change-Id: I90bd9cf2e8142ca0e656dee4bf687953515b71a3

Bug: angleproject:6902
Change-Id: I6a27e3f610bd48dd31937255b287b11fe84c05ca

Improve ANGLE native display factory by supporting creation of a Wayland
native display.

Bug: angleproject:7065
Change-Id: I0882d8666b1d07ced0804b99d1998ca45f4ad73b

Implement EGL extensions image dmabuf import and import modifiers in
Vulkan Wayland Display.

Bug: angleproject:7065
Change-Id: I91535db3581cbe51f0eddc429e2752641c568148
Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Signed-off-by: Constantine Shablya <constantine.shablya@collabora.com>

Implement DisplayVkWayland and WindowSurfaceVkWayland. Get window size
from native window and check egl config is just empty.

Then add an EGL wayland test for testing rendering and buffers swapping.

Bug: angleproject:6902
Change-Id: I8204a5cc99f26330b74caba241bebf14c5650c2d

https://swiftshader.googlesource.com/SwiftShader.git/+log/c27e99245d42..d01dca1d18d0

2022-03-17 swiftshader.regress@gmail.com Regres: Update test lists @ b3b1a3fe
2022-03-16 capn@google.com Use slow legacy log2() implementation for Chromium
2022-03-16 capn@google.com Add exp/log optimization documentation
2022-03-16 capn@google.com Improve the precision of highp log2()
2022-03-16 capn@google.com Improve the precision and performance of highp exp2()
2022-03-16 capn@google.com Fix handling infinity for relaxed precision log2()
2022-03-16 srisser@google.com Add variable to control regres daily branch
2022-03-16 joshuaduong@google.com Fix integer overflow in vkAllocateMemory.
2022-03-16 capn@google.com Implement a relaxed precision log2()
2022-03-16 capn@google.com Implement a relaxed precision exp2()
2022-03-15 swiftshader.regress@gmail.com Regres: Update test lists @ c27e9924

If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/swiftshader-angle-autoroll
Please CC jmadill@google.com on the revert to ensure that a human
is aware of the problem.

To file a bug in SwiftShader: https://bugs.chromium.org/p/swiftshader/issues/entry
To file a bug in ANGLE: https://bugs.chromium.org/p/angleproject/issues/entry

To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md

Bug: None
Tbr: jmadill@google.com
Change-Id: I18059c70242f237a343729584a0d69f3786da7a8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3531933
Commit-Queue: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>
Bot-Commit: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>

This change exposes GL_KHR_blend_equation_advanced if
VK_EXT_blend_operation_advanced is supported.  No emulation is yet
implemented.

Bug: angleproject:3586
Change-Id: Ie7a442c22e028e3aa45ccef52db320e205664d49
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3487951


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>

Bug: angleproject:3586
Change-Id: I27086295e7455ce8848db1c6004917fea40202e8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3484785


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>

To avoid undefined behavior, D3D requires the
base mip level to have multiple-of-4 dimensions.
Add this validation check back in.

Bug: angleproject:7099
Change-Id: Id51f6cd4c745455062b864c0d291b9857e985e35
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3530499


Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Alexey Knyazev <lexa.knyazev@gmail.com>
Commit-Queue: Brandon Schade <b.schade@samsung.com>

This was fixed by 349636a0. Adding
a regression test which I didn't have ready at the time.

Bug: angleproject:7102
Change-Id: Icbb685c1ffad6725a278e5ed65f881f86c5688ef
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3527604


Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Commit-Queue: Charlie Lao <cclao@google.com>

ANGLE implements the EGL 1.5 specification, update the
README to reflect that.

Bug: angleproject:1944
Change-Id: Id61d21ebab05f81bbfcd2ea6ce6c137c8a564279
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3530487


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: mohan maiya <m.maiya@samsung.com>

This reverts commit 5c29d795.

Reason for revert: this is causing some test failures on chromium and blocking the angle-chromium auto roller job: https://chromium-review.googlesource.com/c/chromium/src/+/3529771/

Original change's description:
> Flush the texture staged updates when destroying context share group
>
> If we are using the extension EGL_ANGLE_display_texture_share_group,
> flush the texture staged updates upon destroying the context. With the
> extension enabled, the texture could still be alive when both context
> and its' EGL::ShareGroup are destroyed. If we have staged updates not
> yet flushed, the updates will keep the ShareGroupVk bufferpool occupied,
> causing an error upon ShareGroupVk::onDestroy().
>
> Bug: chromium:1299211
> Change-Id: I260de93c3a3099e023e31acbe017803e824459ad
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3495879


> Reviewed-by: Charlie Lao <cclao@google.com>
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Commit-Queue: Yuxin Hu <yuxinhu@google.com>

Bug: chromium:1299211
Change-Id: I214161d6a8aec834e1efc5fc9d2479e62e3bfae0
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3530505


Auto-Submit: Yuxin Hu <yuxinhu@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Fix capturing glGetSynciv with null lengths on Linux.

When calling the default WriteParamValueReplay handler with a
nullptr, ostream on Linux will write `(nil)`.

This fixes retracing all traces containing glGetSynciv calls with null
lengths.

Add a NullLength test to FenceSyncTests that used to fail in the
capture_replay_tests.

Test: capture_replay_tests.py --gtest_filter=FenceSyncTest.NullLength/*
Bug: angleproject:7071
Bug: angleproject:6808
Change-Id: I0a2bd4db744ce2a026cd932ba0f2f4747672fcc8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3526653


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Lubosz Sarnecki <lubosz.sarnecki@collabora.com>

Define macro and header dependency in BUILD.gn when Vulkan is enabled.

Bug: angleproject:7065
Change-Id: Ifbd9091425212689d4284b2bc46d3fe59267509c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3526654


Auto-Submit: Antonio Caggiano <antonio.caggiano@collabora.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Value being non-zero would unambiguously indicate that target is
Android, not sure if the actual value is really needed but it's probably
good to have.

Bug: angleproject:6854
Change-Id: I685e8e0ffbb666349801203e9d09b082135af17e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3529966


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Roman Lavrov <romanl@google.com>

--render-test-output-dir is supported by Android test runner so the file
is first written on the device then copied over to the host.

Bug: angleproject:6854
Change-Id: Ia56fd9c36ab68a495459efa3b1014e871a615565
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3526855


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Roman Lavrov <romanl@google.com>

https://chromium.googlesource.com/vulkan-deps.git/+log/a11411926c31..51988dcdccbf

Changed dependencies:
* spirv-headers: https://chromium.googlesource.com/external/github.com/KhronosGroup/SPIRV-Headers.git/+log/0e994ee9c4..4995a2f272
* vulkan-headers: https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-Headers.git/+log/2c45218b90..384881cc90
* vulkan-loader: https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-Loader.git/+log/a0c7be0505..ab207b0829
* vulkan-tools: https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-Tools.git/+log/795bad232d..99500cee1e
* vulkan-validation-layers: https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-ValidationLayers.git/+log/ca222d286f..6c8f1627a3

If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/vulkan-deps-angle-autoroll
Please CC jmadill@google.com on the revert to ensure that a human
is aware of the problem.

To file a bug in ANGLE: https://bugs.chromium.org/p/angleproject/issues/entry

To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md

Bug: None
Tbr: jmadill@google.com
Change-Id: Ie1322074d40a6eb1e0f5914ea5eb5b5d5e096728
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3528360
Commit-Queue: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>
Bot-Commit: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>

The position value in a frame refers to its relative position to
its parent. We do not want to copy that to its new child layer.

Test: when using ANGLE (with metal or swiftshader backend) with
emulator frame, it should render the emulator at the right
position.
Bug: angleproject:7097
Change-Id: I94f78e1253d7ee016bf7ab512d908f75250ce9c7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3523622


Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Auto-Submit: Yahan Zhou <yahan@google.com>
Reviewed-by: Yahan Zhou <yahan@google.com>
Commit-Queue: Kenneth Russell <kbr@chromium.org>

The previous fix crrev.com/c/3513553 has run into corner case that
requires more follow up change crrev.com/c/3522565. But with that, there
is report that now we are hitting assertion in
handleDirtyGraphicsIndexBuffer(). This becomes a bit fragile This new
fix relies on the DIRTY_BIT_INDEX_BUFFER dirty bit and should be more
reliable as long as the dirty bit is set properly (if not, then we have
other bug that it won't even send down vulkan command to bind the
correct element buffer). We could further optimize the code path and
create a fast path for most common usages in the future.

Bug: chromium:1299261
Change-Id: Ifa8f86d431798c9ca4c128ed71a3e9e0a3537ccb
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3526021


Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Charlie Lao <cclao@google.com>

If we are using the extension EGL_ANGLE_display_texture_share_group,
flush the texture staged updates upon destroying the context. With the
extension enabled, the texture could still be alive when both context
and its' EGL::ShareGroup are destroyed. If we have staged updates not
yet flushed, the updates will keep the ShareGroupVk bufferpool occupied,
causing an error upon ShareGroupVk::onDestroy().

Bug: chromium:1299211
Change-Id: I260de93c3a3099e023e31acbe017803e824459ad
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3495879


Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Yuxin Hu <yuxinhu@google.com>

Remove the ValidCompressedDimension() check for non-WebGL
contexts of compressed formats. This is because compressed
texture images can have dimensions that are neither multiples
of four nor small values like one or two.

Test: gtest_filter=*DXT1CompressedTextureTest.NonBlockSizesMipLevels*
Bug: angleproject:7099
Change-Id: Ic86c15dd3187ffab0665f6db21434fffcbe04fef
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3526017


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Brandon Schade <b.schade@samsung.com>

If vertex attribute is enabled and buffer is bound, but buffer size is
0, we should not crash. This CL skips mapImpl and data copy all together
if size is 0 to avoid crash when calling mapImpl while buffer is
invalid.

This CL also added a test for this.

Bug: chromium:1296467
Change-Id: I79af348f133e1d3b4427f044e370652d0875dc91
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3516700


Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Charlie Lao <cclao@google.com>

In CaptureVertexArrayState when a non-default VAO is bound and no
GL_ARRAY_BUFFER is bound, VertexAttribPointer calls with a non-null
offset are invalid and will produce:
"Client data cannot be used with a non-default vertex array object."

Add a test that leaves a vertex array with an invalid vertex
attribute pointer and used to produce an invalid trace with MEC.

Test: angle_end2end_tests --gtest_filter="VertexAttributeTestES3.InvalidAttribPointer/*"
Bug: angleproject:7042
Bug: angleproject:7098
Change-Id: Iaf91fe7f191c44df79db47bb5455e5df922f2ed5
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3475353


Reviewed-by: Cody Northrop <cnorthrop@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Lubosz Sarnecki <lubosz.sarnecki@collabora.com>

Add functions to perform a lossy conversion from VkFormat to DrmFourCC
and to convert a VkFormat to one or more DrmFourCC formats. Compilation
only happens when Vulkan is actually enabled.

Bug: angleproject:7065
Change-Id: I97c4585b9032e1ff4f773c3c427541bf7f6117fe
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3525754


Auto-Submit: Antonio Caggiano <antonio.caggiano@collabora.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Bug: angleproject:7096
Change-Id: I63ce1773282ac79ee9c0656b32715182d2e89d92
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3525814
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This reverts commit daef9577.

Reason for revert: Breaking roll into Skia due to missing Vk headers.

Original change's description:
> Vulkan: VkFormat/DrmFourCC
>
> Add functions to perform a lossy conversion from VkFormat to DrmFourCC
> and to convert a VkFormat to one or more DrmFourCC formats.
>
> Bug: angleproject:7065
> Change-Id: Iecccea6a89fbf8adeb50c3cb0c737ad3aedf7c62
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3495123


> Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
> Auto-Submit: Antonio Caggiano <antonio.caggiano@collabora.com>
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Commit-Queue: Jamie Madill <jmadill@chromium.org>

Bug: angleproject:7065
Change-Id: I7cbad4277929717c211f70435cd6e462f11e4bda
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3525813
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Bug: angleproject:3402
Change-Id: I90f17307efb98934b1faf44f2efbdbcfa4db9e0c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3524342


Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>