OSDN Git Service

android-x86/device-generic-goldfish-opengl.git
7 years agoMerge "Fix jb build"
Yahan Zhou [Wed, 20 Jul 2016 23:26:20 +0000 (23:26 +0000)]
Merge "Fix jb build"

7 years agoFix jb build
Yahan Zhou [Wed, 20 Jul 2016 23:12:02 +0000 (16:12 -0700)]
Fix jb build

Change-Id: I31b9e3e44523819f54da362e688b2c5098ec27bd

7 years agoMerge "Fix jb arm boot"
Bo Hu [Wed, 20 Jul 2016 21:37:20 +0000 (21:37 +0000)]
Merge "Fix jb arm boot"

7 years agoFix jb arm boot
Yahan Zhou [Wed, 20 Jul 2016 21:31:33 +0000 (14:31 -0700)]
Fix jb arm boot

Change-Id: Ibf124424f703cc30ee7b5db7c3589b69728a4c32

7 years agoMerge "Misc fix in gralloc pipe"
Yahan Zhou [Tue, 19 Jul 2016 23:27:32 +0000 (23:27 +0000)]
Merge "Misc fix in gralloc pipe"

7 years agoMisc fix in gralloc pipe
Yahan Zhou [Mon, 18 Jul 2016 23:14:21 +0000 (16:14 -0700)]
Misc fix in gralloc pipe

Fix size and have a confirmation message in gralloc pipe. Also fix the
naming of puid.

Change-Id: I0b87b9e6bcf88172792705ac9643e150b227be96

7 years agoMerge "Create pre-process gralloc pipe on the guest"
Yahan Zhou [Mon, 18 Jul 2016 23:03:25 +0000 (23:03 +0000)]
Merge "Create pre-process gralloc pipe on the guest"

7 years agoMerge "Make goldfish-opengl compatible with master"
Yahan Zhou [Mon, 18 Jul 2016 23:03:08 +0000 (23:03 +0000)]
Merge "Make goldfish-opengl compatible with master"

7 years agoCreate pre-process gralloc pipe on the guest
Yahan Zhou [Mon, 11 Jul 2016 18:51:43 +0000 (11:51 -0700)]
Create pre-process gralloc pipe on the guest

https://buganizer.corp.google.com/issues/29457657

This patch create a pre-process pipe in gralloc. It tells the host which
process holds which color buffer, so that on process exit the host can
correctly reduce the reference counters of all gralloc color buffers.

Each process will get a 64bit identifier from the host, to help the host
identifying it.

It works with host patch:
https://android-review.googlesource.com/#/c/246823/

Change-Id: I8d9a512a1f151f4f29a3b318ccf17d62d8f46512

7 years agoMake goldfish-opengl compatible with master
Yahan Zhou [Mon, 18 Jul 2016 19:20:04 +0000 (12:20 -0700)]
Make goldfish-opengl compatible with master

Change-Id: I103424ff9e2bfdad8181850a216b7a0ee0290267

7 years ago[cherry-pick] Fix misc-macro-parentheses warnings in goldfish/opengl.
Chih-Hung Hsieh [Thu, 9 Jun 2016 21:02:30 +0000 (14:02 -0700)]
[cherry-pick] Fix misc-macro-parentheses warnings in goldfish/opengl.

Add parentheses around macro arguments used beside operators.
Use NOLINT to suppress warning or clang-tidy can put parentheses at
wrong places.

Bug: 28705665

cherry-pick from https://android-review.googlesource.com/#/c/237790/
Change-Id: I40653d3e617f84c07b308bb9c2869219f6563706
Test: build with clang-tidy

7 years ago[cherry-pick] Delete some useless files
Tina Zhang [Mon, 7 Sep 2015 02:47:07 +0000 (10:47 +0800)]
[cherry-pick] Delete some useless files

Currently, some files are not used by the guest part of
GL emulation any more.

cherry-pick from https://android-review.googlesource.com/#/c/169695/

Change-Id: I1fb87614aaba19be6863d6fb85e80937d0778384
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
7 years ago[cp] Build camera related goldfish HALs as brilloemulator for Brillo
Christopher Wiley [Fri, 12 Feb 2016 20:37:09 +0000 (12:37 -0800)]
[cp] Build camera related goldfish HALs as brilloemulator for Brillo

Brillo runs a reduced subset of the graphics stack and a different
system configuration from goldfish.  However, many of these HALs still
make sense for Brillo.  Build the HALs with a board name that will allow
the camera and graphics subsystems to pick them up.

Bug: 26936651
Test: brilloemulator is able to pick up and use these HALs to take
      fake pictures.

cherry-pick from:

https://android-review.googlesource.com/#/c/202700/

Change-Id: I79265ada33a5a97c6f670b79f9c239606465c9fc

7 years ago[cherry-pick] Refactor gralloc HAL to not repeat defintions
Christopher Wiley [Fri, 12 Feb 2016 20:32:07 +0000 (12:32 -0800)]
[cherry-pick] Refactor gralloc HAL to not repeat defintions

Bug: 26936651

cherry-pick from:

https://android-review.googlesource.com/#/c/202619/

Change-Id: If11a2af0496cdaf86e3bc87ded37fc8d93b7fa16

7 years ago[cherry-pick] GPU: Build gralloc.ranchu.so
Miroslav Tisma [Wed, 23 Sep 2015 10:54:56 +0000 (12:54 +0200)]
[cherry-pick] GPU: Build gralloc.ranchu.so

cherry-pick from:

https://android-review.googlesource.com/#/c/172935/

Change-Id: Ie449043b06b6fefa52bdbd68c243a1165ec95ee1

7 years agoMake it compatible with ics-mr1
Yahan Zhou [Fri, 1 Jul 2016 22:47:21 +0000 (15:47 -0700)]
Make it compatible with ics-mr1

Change-Id: I8661fa4cc93111dfdb5fe9fa73f8a47b2820f615

7 years agoMerge "Make opengl compatible with gingerbread"
Yahan Zhou [Thu, 30 Jun 2016 21:16:16 +0000 (21:16 +0000)]
Merge "Make opengl compatible with gingerbread"

7 years agoMerge "Revert "Workaround: disable checksum for a problem in glTexSubImage2D""
Yahan Zhou [Thu, 30 Jun 2016 21:09:10 +0000 (21:09 +0000)]
Merge "Revert "Workaround: disable checksum for a problem in glTexSubImage2D""

7 years agoMake opengl compatible with gingerbread
Yahan Zhou [Thu, 23 Jun 2016 20:58:09 +0000 (13:58 -0700)]
Make opengl compatible with gingerbread

This patch makes mnc opengl compatible with gingerbread. With this
patch, you can copy-paste the opengl folder to gingerbread and it will
just build.

Notice that in gingerbread, the GL folder is in
development/tools/emulator

Change-Id: I99f060edd517a7afd425fd2103cbceaa61f0fc82

7 years agoRevert "Workaround: disable checksum for a problem in glTexSubImage2D"
Yahan Zhou [Fri, 3 Jun 2016 00:24:03 +0000 (00:24 +0000)]
Revert "Workaround: disable checksum for a problem in glTexSubImage2D"

The checksum enable/disable should be control by feature control now.

This reverts commit 2a5a09abd50933d7cc4b1302769a19ffdf9d104b.

Change-Id: I8f33718f7b2112fd6abcfd51fadc336e9bd15a3a

7 years agoMake GL build script compatible with jb (and mr1.1, mr2)
Yahan Zhou [Mon, 20 Jun 2016 19:28:34 +0000 (12:28 -0700)]
Make GL build script compatible with jb (and mr1.1, mr2)

Notice that in jb, the GL encoder is located in sdk/emulator/opengl. In
jb-mr1.1, the encoder is in development/tools/emulator/opengl.

After this patch, it should just work if you copy the opengl directory
from mnc to jb.

Change-Id: I4b6ba6daeb64315ecf72b52b142acbbccdc4dbe0

7 years agoMerge "Revert "Temporarily rename makefile to avoid building duplicate opengl libs.""
Yahan Zhou [Tue, 28 Jun 2016 00:00:11 +0000 (00:00 +0000)]
Merge "Revert "Temporarily rename makefile to avoid building duplicate opengl libs.""

7 years agoRevert "Temporarily rename makefile to avoid building duplicate opengl libs."
Yahan Zhou [Mon, 27 Jun 2016 22:48:39 +0000 (22:48 +0000)]
Revert "Temporarily rename makefile to avoid building duplicate opengl libs."

This reverts commit 8b2652824b14889092d41d263762b9f3bdf6f86c.

Change-Id: I93ceaea56f8ef4ed90a73af5b04f7a1fed0083e0

7 years agoMerge "Fix 3Dmark crash - preserve more host connections"
Yahan Zhou [Mon, 27 Jun 2016 22:16:40 +0000 (22:16 +0000)]
Merge "Fix 3Dmark crash - preserve more host connections"

7 years agoFix 3Dmark crash - preserve more host connections
Lingfeng Yang [Fri, 24 Jun 2016 02:06:28 +0000 (19:06 -0700)]
Fix 3Dmark crash - preserve more host connections

The CL

https://googleplex-android-review.git.corp.google.com/#/c/1131769/

for killing host connections was too aggressive.

It incorrectly assumed that if eglInitialize was not called in the
current thread, it'd be treated as a gralloc-only connection.

It's perfectly possible to create a new thread,
not call eglInitialize, and call eglCreateContext instead,
which is within proper usage since after there is supposed
to be only one display per process.

This situation resulted in 3DMark crashing, because once the
connection is killed on a thread that is not actually gralloc-only,
all the state goes away and segfault is just a matter of time.

This CL makes it so that eglCreateContext / eglMakeCurrent
mark the connection as not gralloc only, preventing the
connection from being erroneously killed.

The behavior where no extra render threads are created
when opening recents is still preserved.

(Cherry-pick from ebff5363850076de2623b6f6a0ff035ea3e823d5)
Change-Id: I20a204aefca91e5ac8f6b097d9fcd799f6b02583

7 years agoTemporarily rename makefile to avoid building duplicate opengl libs.
Prathmesh Prabhu [Mon, 27 Jun 2016 20:30:13 +0000 (13:30 -0700)]
Temporarily rename makefile to avoid building duplicate opengl libs.

We've yet to drop the opengl code from the old location. Turns out that
simply including this new source location leads the Android build system
to build the lib from both the old and new location.

Renamed the Android.mk file here to avoid building from this location
until the migration is complete.

Change-Id: I66b5f4201f99b8a17c814918e5b42489d686d7c5

7 years agoMerge opengl/ from device/generic/goldfish to /
Surendra Vispute [Fri, 24 Jun 2016 00:24:14 +0000 (17:24 -0700)]
Merge opengl/ from device/generic/goldfish to /

7 years agoMerge "Fix glTexSubImage2D" into mnc-emu-dev
Yahan Zhou [Wed, 22 Jun 2016 23:13:17 +0000 (23:13 +0000)]
Merge "Fix glTexSubImage2D" into mnc-emu-dev

7 years agoFix glTexSubImage2D
Yahan Zhou [Wed, 22 Jun 2016 22:36:04 +0000 (15:36 -0700)]
Fix glTexSubImage2D

This is to port
https://googleplex-android-review.git.corp.google.com/#/c/1145109

into mnc

Change-Id: Ic53c1181ab1c69fb33c5758034c0c4ec131b010e

7 years agoFix build scripts for klp
Yahan Zhou [Wed, 22 Jun 2016 22:26:08 +0000 (15:26 -0700)]
Fix build scripts for klp

The previous build script makes klp fail to boot. This patch fixes it.

Change-Id: I091fdd29ef3f37d10d5f56c81e9ba17618214ebf

7 years agoMake build script compatible with klp and up
Yahan Zhou [Sat, 18 Jun 2016 00:40:14 +0000 (17:40 -0700)]
Make build script compatible with klp and up

It gets rid of c++11 and makes build script compatible with klp and up.

It also includes GL headers here, because headers in framework/native
might be different across API levels.

Change-Id: Ida425416a392cef878256965d69b54afe42b7263

7 years agoInitial empty repository
Bill Rassieur [Mon, 20 Jun 2016 22:30:41 +0000 (15:30 -0700)]
Initial empty repository

7 years agoMerge "Fix EGL cleanup code for CTS android.nativeopengl" into mnc-emu-dev
Yahan Zhou [Mon, 20 Jun 2016 19:58:05 +0000 (19:58 +0000)]
Merge "Fix EGL cleanup code for CTS android.nativeopengl" into mnc-emu-dev

7 years agoFix EGL cleanup code for CTS android.nativeopengl
Yahan Zhou [Fri, 10 Jun 2016 00:54:34 +0000 (17:54 -0700)]
Fix EGL cleanup code for CTS android.nativeopengl

In CTS, android.nativeopengl.EGLCleanupTest tries to repeatatively
create and destroy egl context for 1000 times to check for memory leaks.
This patch releases resources when destroying contexts.

It turns android.nativeopengl.EGLCleanupTest from failure to flaky,
because of two other issue:

(1) There is a 1 min timeout in the test:
cts/tools/tradefed-host/src/com/android/cts/tradefed/testtype/
WrappedGTest.java
Our emulator can barely finish it within 1 min.

(2) There is another bug that may trigger UI rendering events after the
test begun, which crashes randomly because this test releases rendering
resources. It needs to be fixed in another patch.

Change-Id: Id285f526149638c63eb78c167c7d2daef7bd2412

7 years agoMerge "Regenerate gl encoder from qemu" into mnc-emu-dev
Yahan Zhou [Mon, 20 Jun 2016 16:58:21 +0000 (16:58 +0000)]
Merge "Regenerate gl encoder from qemu" into mnc-emu-dev

7 years agoRegenerate gl encoder from qemu
Yahan Zhou [Fri, 17 Jun 2016 22:06:37 +0000 (15:06 -0700)]
Regenerate gl encoder from qemu

Regenerated from:
https://android-review.googlesource.com/#/c/237292/

Change-Id: I7972c4fddfb9765af39856ccd8aa29ac67558301

7 years agoDon't leave gralloc-only host connections hanging.
Lingfeng Yang [Mon, 13 Jun 2016 15:39:28 +0000 (08:39 -0700)]
Don't leave gralloc-only host connections hanging.

bug: 29318455

Memory leaks are becoming more of a problem, especially
with host machines that don't have much RAM and when
we use the emulator for long periods of time.

This CL fixes the problem detailed in the bug link,
which is that the number of render threads
(and thus thread-specific storage) skyrockets when
repeatedly pressing the Recents button while an app is active.

The thing is that there are lots of threads
that never use the EGL display, just gralloc, in order
to do gralloc things like update the screenshot display
of apps in recents.

These threads are not cleaned up because a thread pool
is often used to manage these sorts of tasks.
If a new or different thread suddenly wants to update
the recents screenshots through gralloc, a new connection
will be established.

This CL uses gralloc_unregister_buffer as a signal
for when a thread might be done with gralloc in general,
exiting the host connection immediately.

Of course, there may be other things that the thread wants
to do, but then it can just go ahead and create a new connection.

We also use the heuristic that we do not exit on gralloc_unregister_buffer
when the thread seems to be a bona fide rendering thread
(eglInitialize has been called).

There are still more leaks going on, this fixes just
one source of them.

After this CL, the render thread count should be held constant
when repeatedly starting apps and going to recents.

Change-Id: I5f57539429f149e3a938467e3b82fa911ab118c3

7 years agoUse correct format when reading color buffers
Bjoern Johansson [Mon, 13 Jun 2016 23:09:20 +0000 (16:09 -0700)]
Use correct format when reading color buffers

When reading back data from color buffers on the host make sure gralloc
uses the correct format and type parameters. Previously it was hardcoded
based on RGBA at 8 bits per component which would cause incorrect reads
for other pixel formats. This could cause QEMU pipe reads to call abort
because the reads went past a page boundary if the calculated size
exceeded the actual buffer size.

Change-Id: Ifa61f0a97d79577744b899b1bbb4120a0b9170eb

7 years agoMerge "Move stream->flush to non-generated code" into mnc-emu-dev
Yahan Zhou [Mon, 6 Jun 2016 19:42:17 +0000 (19:42 +0000)]
Merge "Move stream->flush to non-generated code" into mnc-emu-dev

7 years agoMove stream->flush to non-generated code
Yahan Zhou [Mon, 6 Jun 2016 19:30:39 +0000 (12:30 -0700)]
Move stream->flush to non-generated code

Move flush command out of auto-generated code.

Change-Id: I3f3c8ca49344ea4c5357d7d6953364ce1015ea99

7 years agoMerge "[cherry-pick] Lose HAVE_ANDROID_OS in goldfish." into mnc-emu-dev
Yahan Zhou [Mon, 6 Jun 2016 17:51:45 +0000 (17:51 +0000)]
Merge "[cherry-pick] Lose HAVE_ANDROID_OS in goldfish." into mnc-emu-dev

7 years agoCheck vertex buffer size before glDraw*
Yahan Zhou [Thu, 2 Jun 2016 00:24:28 +0000 (17:24 -0700)]
Check vertex buffer size before glDraw*

This patch verifies if a vertex buffer is big enough before calling
glDraw*. It fixes crashes when trying to index outside of vertex buffer.

It should make it less crashy in glDrawElements and glDrawArrays:

https://code.google.com/p/android/issues/detail?id=207284

Change-Id: I36f0f568d96b576f0e0975435d2d7fd41cd9e732

7 years ago[cherry-pick] Lose HAVE_ANDROID_OS in goldfish.
Elliott Hughes [Wed, 12 Aug 2015 21:30:38 +0000 (14:30 -0700)]
[cherry-pick] Lose HAVE_ANDROID_OS in goldfish.

cherry pick from:

https://android-review.googlesource.com/#/c/166427

Change-Id: Id626aee8d65e0ded543ac12416c37f938d196676

7 years agoFix mnc build
Yahan Zhou [Wed, 1 Jun 2016 19:49:39 +0000 (12:49 -0700)]
Fix mnc build

Missing a bracket when cherry-pick.

Change-Id: I5202bf4dc5c64cb4663fdcadcbf6704272708b42

7 years ago[cherry-pick] Optimize glGetIntegerv to reduce the cost of translating
Lizhe Liu [Wed, 27 May 2015 03:13:23 +0000 (11:13 +0800)]
[cherry-pick] Optimize glGetIntegerv to reduce the cost of translating

Hardware GLES emulation in the Android platform is implemented
with a mix of components. Translator module is one of the most
important components in emulator. It translate the function
calls into calls to the appropriate desktop APIs and cost lots
of reference to complete it. So optimization must be beneficial
to us. GL_MAX_CUBE_MAP_TEXTURE_SIZE, GL_MAX_RENDERBUFFER_SIZE
and GL_MAX_TEXTURE_SIZE are the rough estimate value of the
largest texture which the OPENGL can handle. All of them are
constant value. So there is no need for us to translate them
from guest os to host os every time. We can cache the value
after first time and reuse it next time.

Cherry-pick from:

https://android-review.googlesource.com/#/c/151899

Change-Id: I1a0352c40b667bafd6ec1fb66a8060d7fe4a415d
Signed-off-by: Lizhe Liu <lizhe.liu@intel.com>
7 years agoMerge "[cherry-pick] opengl: GLEncoder: Fix potential array buffer misbinding" into...
Yahan Zhou [Wed, 1 Jun 2016 17:38:02 +0000 (17:38 +0000)]
Merge "[cherry-pick] opengl: GLEncoder: Fix potential array buffer misbinding" into mnc-emu-dev

7 years ago[cherry-pick] opengl: GLEncoder: Fix potential array buffer misbinding
Yu Ning [Wed, 19 Aug 2015 07:22:13 +0000 (15:22 +0800)]
[cherry-pick] opengl: GLEncoder: Fix potential array buffer misbinding

When encoding a glDrawElements() or glDrawArrays() call, the encoder
needs to figure out whether each array (e.g. vertex, color, etc.) uses
a bound buffer or not. It is possible that some arrays use bound buffers
while others immediate buffers. In this case, there is a chance that the
encoder fails to preserve the semantics of the input GLES calls,
resulting in misinterpretation by the translator that all arrays use
bound buffers.

Fix this by encoding an unbind operation before any glXXXPointer() call
that does not mean to use a bound buffer. For GLESv2, that is already
done:

 40a674e Fix emulator crashes on glDrawElements command

This CL does the same for GLESv1.

Cherry-pick from:

https://android-review.googlesource.com/#/c/167443/

Change-Id: Ifd17321b124a4f4b1037a43d5a54094541d367f1
Signed-off-by: Yu Ning <yu.ning@intel.com>
8 years agoDon't optimize out rcUpdateColorBuffer
Lingfeng Yang [Wed, 25 May 2016 15:31:06 +0000 (08:31 -0700)]
Don't optimize out rcUpdateColorBuffer

bug:28406160

If we want to use rcUpdateColorBuffer to
synchronize on the host in order to remove
out of order webcam frames, we need to call it
if (cb->hostHandle), not just when
the width and height are nonzero.

Otherwise, we get deadlocks.

Change-Id: Ib98774b935e8786e307dd38711a0f2f9ffea440e

8 years agoMerge "Save a few glBindBuffer calls" into mnc-emu-dev
Lingfeng Yang [Fri, 27 May 2016 00:32:31 +0000 (00:32 +0000)]
Merge "Save a few glBindBuffer calls" into mnc-emu-dev

8 years agoSave a few glBindBuffer calls
Lingfeng Yang [Thu, 26 May 2016 21:18:33 +0000 (14:18 -0700)]
Save a few glBindBuffer calls

When sending vertex attributes, we only need to
bind the context-wide GLClientState's VBO
if a particular vertex attrib state is enabled.

This CL makes it so we call glBindBuffer only if
the currently bound vbo changes, in many cases
completely eliminating extra glBindBuffer calls
in sendVertexAttributes. Each extra glBindBuffer
call takes a few microseconds.

Change-Id: I04de3be0114ec6df10e67bc801b6d089bf5096db

8 years agoRefactor GL extension getting part in checksum setup.
Lingfeng Yang [Mon, 23 May 2016 19:16:04 +0000 (12:16 -0700)]
Refactor GL extension getting part in checksum setup.

Change-Id: Ieb26ce5682a87821935a5b639e5242942d13ac21

8 years agoFix memory-handling issues in the OpenGL code.
Adam Buchbinder [Fri, 20 May 2016 22:25:59 +0000 (15:25 -0700)]
Fix memory-handling issues in the OpenGL code.

This fixes the folloiwng cppcheck reports:
[device/generic/goldfish/opengl/system/GLESv2_enc/GL2Encoder.cpp:789]:
  (error) Mismatching allocation and deallocation: str
[device/generic/goldfish/opengl/system/GLESv2_enc/GL2Encoder.cpp:795]:
  (error) Mismatching allocation and deallocation: str
[device/generic/goldfish/opengl/tests/gles_android_wrapper/egl.cpp:492]:
  (error) Mismatching allocation and deallocation: attrib

Change-Id: Iaf494eca1b927b2fa4b91f01ba340746b9da9bd3
(cherry picked from commit 3375e5e21450b78daf109225707d574976b7ec1c)

8 years agoMerge "Sends tex coord only when the texture is enabled in gles1" into mnc-emu-dev
Yahan Zhou [Tue, 17 May 2016 16:56:04 +0000 (16:56 +0000)]
Merge "Sends tex coord only when the texture is enabled in gles1" into mnc-emu-dev

8 years agoSends tex coord only when the texture is enabled in gles1
Yahan Zhou [Mon, 16 May 2016 23:23:22 +0000 (16:23 -0700)]
Sends tex coord only when the texture is enabled in gles1

Only send tex coord if the corresponding texture unit is enabled.

This is to prevent a crash in An3DBench:

https://b.corp.google.com/issues/9631994

Change-Id: I23ae030adba3a3fa05e00f63c96040b439b86baf

8 years agoCreate and read color buffer when expecting reads
Bjoern Johansson [Thu, 12 May 2016 17:38:27 +0000 (10:38 -0700)]
Create and read color buffer when expecting reads

Make sure color buffers are created when gralloc is allocating buffers
that indicate that software reads are going to happen. In this case
software will require the rendered data to be available from the host at
a later point and that means that rendering has to be stored in a color
buffer. Also make sure that when gralloc locks a buffer for reading it
will actually read the data from the host if the buffer is marked for
reading.

Change-Id: Ief4b7fe34401727cabec9fdd5c2b9812a6c49612

8 years agoMerge "Cleanup debugging code in a previous CL" into mnc-emu-dev
Yahan Zhou [Wed, 11 May 2016 23:04:02 +0000 (23:04 +0000)]
Merge "Cleanup debugging code in a previous CL" into mnc-emu-dev

8 years agoCleanup debugging code in a previous CL
Yahan Zhou [Wed, 11 May 2016 22:54:56 +0000 (15:54 -0700)]
Cleanup debugging code in a previous CL

Clean up my logs.

Change-Id: Ia447994af03358964bb39330824e8da8eeebf08b

8 years agoMerge "Add framebuffer checks to pass dEQP" into mnc-emu-dev
Yahan Zhou [Wed, 11 May 2016 22:34:26 +0000 (22:34 +0000)]
Merge "Add framebuffer checks to pass dEQP" into mnc-emu-dev

8 years agoAdd framebuffer checks to pass dEQP
Yahan Zhou [Wed, 11 May 2016 22:06:34 +0000 (15:06 -0700)]
Add framebuffer checks to pass dEQP

They are defined in GLES3 compatible context, but used in dEQP-GLES2
tests. So we need to enable them.

Should be used with this CL in the host:
https://android-review.googlesource.com/#/c/229251

dEQP-GLES2.functional.fbo.completeness.renderable failures 21->0

Change-Id: Ieacbbcc4afa688711c4d090971036505a74cc1ab

8 years agoMerge "Allow rgb10_a2 internal format in renderbuffer" into mnc-emu-dev
Yahan Zhou [Wed, 11 May 2016 17:54:59 +0000 (17:54 +0000)]
Merge "Allow rgb10_a2 internal format in renderbuffer" into mnc-emu-dev

8 years agoAllow rgb10_a2 internal format in renderbuffer
Yahan Zhou [Wed, 11 May 2016 17:45:31 +0000 (10:45 -0700)]
Allow rgb10_a2 internal format in renderbuffer

dEQP-GLES2.functional.fbo.completeness.renderable.renderbuffer failures
3->0.

Change-Id: If5b94f5b1edc182ec5ccf2bd3321306a826e757e

8 years agoMerge "eglCreateContext returns error code when taking wrong attributes" into mnc...
Yahan Zhou [Mon, 9 May 2016 16:48:23 +0000 (16:48 +0000)]
Merge "eglCreateContext returns error code when taking wrong attributes" into mnc-emu-dev

8 years agoeglCreateContext returns error code when taking wrong attributes
Yahan Zhou [Thu, 5 May 2016 01:10:46 +0000 (18:10 -0700)]
eglCreateContext returns error code when taking wrong attributes

CTS.android.nativeopengl.EGLCreateContextTest fail -> pass

Change-Id: Iec95f568b65cd60cd78cc5eb80ce3ff0d1ae1731

8 years agoMerge "Flush draw calls" into mnc-emu-dev
Lingfeng Yang [Wed, 4 May 2016 16:12:53 +0000 (16:12 +0000)]
Merge "Flush draw calls" into mnc-emu-dev

8 years agoFlush draw calls
Lingfeng Yang [Mon, 2 May 2016 17:47:00 +0000 (10:47 -0700)]
Flush draw calls

Increases Antutu Garden performance by 2-4 FPS,
depending on if host renderthread uses lock
(faster if no lock)

The idea is that if the draw calls are not flushed,
they may be called too late on the host,
resulting in lower performance:
the GPU is just sitting idly between pipe flushes
for more instructions.

Change-Id: I78e278fc9b30b4a135b48f56d912b6aba8aadea7

8 years agoMerge "FBO completeness check in glDrawElements." into mnc-emu-dev
Evan Birenbaum [Thu, 28 Apr 2016 21:07:02 +0000 (21:07 +0000)]
Merge "FBO completeness check in glDrawElements." into mnc-emu-dev

8 years agoFBO completeness check in glDrawElements.
Evan Birenbaum [Thu, 28 Apr 2016 19:47:57 +0000 (12:47 -0700)]
FBO completeness check in glDrawElements.

dEQP-GLES2.functional.negative_api.vertex_array.*:
3 Fail -> 0 Fail

The host framebuffer (not the guest state) is not complete in these
tests, so a call to the host glCheckFramebufferStatus is needed to
confirm that. The cached state is unreliable, since things may have
changed since the last guest call to glCheckFramebufferStatus.

Change-Id: I2cf026129c871df68b328949e9c79b3f5df590eb

8 years agoMerge "glActiveTexture should set GL_INVALID_ENUM when taking wrong input" into mnc...
Yahan Zhou [Thu, 28 Apr 2016 18:19:12 +0000 (18:19 +0000)]
Merge "glActiveTexture should set GL_INVALID_ENUM when taking wrong input" into mnc-emu-dev

8 years agoMerge "Stricter FBO completeness checks" into mnc-emu-dev
Lingfeng Yang [Mon, 25 Apr 2016 16:39:43 +0000 (16:39 +0000)]
Merge "Stricter FBO completeness checks" into mnc-emu-dev

8 years agoMerge "Validate presence of framebuffer attachment's object" into mnc-emu-dev
Lingfeng Yang [Mon, 25 Apr 2016 16:39:36 +0000 (16:39 +0000)]
Merge "Validate presence of framebuffer attachment's object" into mnc-emu-dev

8 years agoSupport glDrawElements with GL_UNSIGNED_INT again
Yahan Zhou [Wed, 20 Apr 2016 23:02:10 +0000 (16:02 -0700)]
Support glDrawElements with GL_UNSIGNED_INT again

This is defined in GLES2 extension.

Change-Id: Id6ac22bb207a88ebb13282c443c0bc990f25cdf3

8 years agoWorkaround: disable checksum for a problem in glTexSubImage2D
Yahan Zhou [Tue, 12 Apr 2016 17:02:38 +0000 (10:02 -0700)]
Workaround: disable checksum for a problem in glTexSubImage2D

This is a temporary workaround for a problem in glTexSubImage2D. It
disables the GL checksum to avoid crashes. Please revert this CL when
the root cause is solved.

Change-Id: I8401449e42da87f580f119982a1bcb1b6b9a4c8b
(cherry picked from commit c1a6ab7b5ce1e23cf65a8e936860bcea879b6eef)

8 years agoStricter FBO completeness checks
Lingfeng Yang [Tue, 12 Apr 2016 16:29:11 +0000 (09:29 -0700)]
Stricter FBO completeness checks

dEQP-GLES2.functional.fbo.completeness.*:
39 Fail -> 0 Fail

Based on similar logic in Swiftshader's libGLESv2/libGLESv2.cpp

Also cache the result of glCheckFramebufferStatus,
if already called from somewhere else.

Change-Id: I3fb42dd6902cfce4043d08fd459a7069ef491f7e

8 years agoValidate presence of framebuffer attachment's object
Lingfeng Yang [Fri, 8 Apr 2016 21:42:34 +0000 (14:42 -0700)]
Validate presence of framebuffer attachment's object

dEQP-GLES2.functional.negative_api.state.get_framebuffer_attachment_parameteriv
Fail -> Pass
dEQP-GLES2.functional.fbo.api.attachment_query_empty_fbo
Fail -> Pass

Track texture and renderbuffer object associations to
framebuffer attachment points.

Change-Id: Iff646bc57ef0202105768123cfa99384f37b01a3

8 years agoMerge "Sanity checks in vertex array functions." into mnc-emu-dev
Evan Birenbaum [Fri, 15 Apr 2016 22:03:51 +0000 (22:03 +0000)]
Merge "Sanity checks in vertex array functions." into mnc-emu-dev

8 years agoSanity checks in vertex array functions.
Evan Birenbaum [Wed, 13 Apr 2016 19:53:09 +0000 (12:53 -0700)]
Sanity checks in vertex array functions.

dEQP-GLES2.functional.negative_api.vertex_array.*:
6 Fail -> 3 Fail.
dEQP-GLES2.functional.rasterization.*:
1 Fail -> 0 Fail.

The remaining vertex array failures will be fixed using some of
Frank's framebuffer status code.

Change-Id: Id8623029180beb6e792a3250763f08c4a03c7499

8 years agoSend glDrawArrays even if there is no vertex array
Lingfeng Yang [Fri, 15 Apr 2016 19:23:55 +0000 (12:23 -0700)]
Send glDrawArrays even if there is no vertex array

The spec says that if there are no arrays,
vertex attributes are to be used next.

dEQP-GLES2.functional.draw.random.50
Fail -> Pass

Change-Id: I81ab0bcac1c483436f8930212159a5079fe7b725

8 years agoValidation for glGetVertexAttribPointerv
Lingfeng Yang [Fri, 8 Apr 2016 19:46:42 +0000 (12:46 -0700)]
Validation for glGetVertexAttribPointerv

dEQP-GLES2.functional.negative_api.state.get_vertex_attrib_pointerv
Fail -> Pass

Based on Intel's OpenGL validation for goldfish opengl,
introduced in nyc-dev

Change-Id: If67192368935f21de20c0f531036db027b867b37

8 years agoValidation for glGetVertexAttrib i/f v
Lingfeng Yang [Fri, 8 Apr 2016 19:38:35 +0000 (12:38 -0700)]
Validation for glGetVertexAttrib i/f v

dEQP-GLES2.functional.negative_api.state.get_vertex_attrib i/f v:
Fail -> Pass

Based on Intel's OpenGL validation for goldfish opengl,
introduced in nyc-dev

Change-Id: I9b539024e88a5afb7c78f8543e7c9ffff4f837e0

8 years agoValidation for glGetUniform i/f v
Lingfeng Yang [Fri, 8 Apr 2016 19:21:13 +0000 (12:21 -0700)]
Validation for glGetUniform i/f v

dEQP-GLES2.functional.negative_api.state.get_string:
Fail -> Pass

Based on Intel's OpenGL validation for goldfish opengl,
introduced in nyc-dev

Change-Id: I9c3b820a5fa6b3b3e4b5e8590cca590cf9ef86d2

8 years agoValidation for glGetString
Lingfeng Yang [Fri, 8 Apr 2016 19:13:45 +0000 (12:13 -0700)]
Validation for glGetString

dEQP-GLES2.functional.negative_api.state.get_string:
Fail -> Pass

Based on Intel's OpenGL validation for goldfish opengl,
introduced in nyc-dev

Change-Id: I74c480cf38a01e9c13f35ee3f58168415f7b6ab3

8 years agoMore gracefulness on dequeueBuffer_* fail on flip
Lingfeng Yang [Mon, 4 Apr 2016 16:26:19 +0000 (09:26 -0700)]
More gracefulness on dequeueBuffer_* fail on flip

BUG: https://code.google.com/p/android/issues/detail?id=204176

Features like Instant Run will do things like abandon
the current set of buffers SurfaceFlinger gave to the UI,
which makes functions like eglSwapBuffers fail,
since they are dequeueing a buffer that is considered 'invalid'
(as it is before the instant run).

If we set the EGL error to EGL_BAD_ALLOC, this interacts badly
with libs/hwui/renderthread/EGLManager.cpp, causing
a LOG_ALWAYS_FATAL(...) to be called, which aborts the current process.

This CL sets the EGL error to EGL_BAD_SURFACE instead,
which allows hwui to recover.

Change-Id: I7cb5bf3b3834b3696981746f2319804c1adf6a25

8 years agoMerge "Fix a typo in GLES encoder headers" into mnc-emu-dev
Yahan Zhou [Mon, 21 Mar 2016 23:17:53 +0000 (23:17 +0000)]
Merge "Fix a typo in GLES encoder headers" into mnc-emu-dev

8 years agoFix a typo in GLES encoder headers
Yahan Zhou [Mon, 21 Mar 2016 23:11:38 +0000 (16:11 -0700)]
Fix a typo in GLES encoder headers

Unimportant but let's fix it.

Change-Id: I360ba0dc0b676238adaeb5d4360c10a9ec8c2371

8 years agoMerge "Exit when QemuPipe is corrupted by GL" into mnc-emu-dev
Yahan Zhou [Fri, 18 Mar 2016 21:25:44 +0000 (21:25 +0000)]
Merge "Exit when QemuPipe is corrupted by GL" into mnc-emu-dev

8 years agoChecksum for GLES messages from host to guest
Yahan Zhou [Wed, 16 Mar 2016 19:41:08 +0000 (12:41 -0700)]
Checksum for GLES messages from host to guest

Add checksum for GLES messages from host to guest. This patch contains
generated code (gl_enc.cpp, gl2_enc.cpp, renderControl_enc.cpp) and
shared code with the host (ChecksumCalculator.cpp,
ChecksumCalculator.h).

Change-Id: I577a34cae5208d3f11a165f861a69fa0971cf864

8 years agoChecksum for GL pipe communication
Yahan Zhou [Thu, 10 Mar 2016 19:45:02 +0000 (11:45 -0800)]
Checksum for GL pipe communication

This is the guest side implementation for cl go/oag/c/207292.

It initializes checksum whenever a renderControl is created. Each
HostConnection owns a checksum structure.

The list contains mostly auto-generated code by emugen.

Change-Id: I5f7bb9ec8a0cea82f3687f8a15839081651fd841

8 years agoExit when QemuPipe is corrupted by GL
Yahan Zhou [Tue, 8 Mar 2016 22:02:22 +0000 (14:02 -0800)]
Exit when QemuPipe is corrupted by GL

Previously when QemuPipe is corrupted, it prints an error message, and
the application will continue to use it, which leads to random behaviour
in GL on the host.

This patch crashes the APP and pops up the crash dialog when the pipe is
(going to be) corrupted. It would happen if the user asks GL to write /
read a NULL pointer (i.e. passing NULL to glGenBuffers). It might also
happen for other unknown reasons.

Change-Id: I303772029bb72a845fa0d4c79b0d38dba1fca2c7

8 years agoCache glGetIntegerv(GL_MAX_VERTEX_ATTRIBS, int*).
Lingfeng Yang [Wed, 2 Mar 2016 05:27:49 +0000 (21:27 -0800)]
Cache glGetIntegerv(GL_MAX_VERTEX_ATTRIBS, int*).

BUG: https://buganizer.corp.google.com/issues/27418143

If we don't, the chatter over the pipe will
kill framerate.

I believe this is OK for the emulator,
as the parameter is dependent on only the
GPU hardware.

Change-Id: I9bd74de6317316902bddf54b915ef509625cee52

8 years agoglActiveTexture should set GL_INVALID_ENUM when taking wrong input
Yahan Zhou [Tue, 1 Mar 2016 20:03:48 +0000 (12:03 -0800)]
glActiveTexture should set GL_INVALID_ENUM when taking wrong input

The GL spec says it generates GL_INVALID_ENUM if taking a wrong input.
Currently it generates GL_INVALID_OPERATION which is not the expected
error code.

Change-Id: I3313ad49130651d0358dbd3bdeb67b7e2d4fd5d9

8 years agoGL encoder null pointer check before write
Yahan Zhou [Fri, 19 Feb 2016 21:59:34 +0000 (13:59 -0800)]
GL encoder null pointer check before write

Previously when encoding some functions, if the user gives us a null
pointer, the encoder will write a header with some data length while not
actually writing any data. This patch fix it by writing 0 into the data
length field for all those functions.

The changed functions include:
glBufferSubData
glCompressedSubTexImage2D
glTexSubImage3DOES
glCompressedTexImage3DOES
glCompressedTexSubImage3DOES

Other functions do not directly write user's pointer, so they don't need
to be updated.

Since this is auto-generated code, there would be another patch in
emu-master-dev to update the code generator.

Change-Id: I05fca078dfcf01e20fa757f1050820b21ca080b7

8 years agoFix crash when forgot to call glVertexAttribPointer
Yahan Zhou [Thu, 11 Feb 2016 00:10:46 +0000 (16:10 -0800)]
Fix crash when forgot to call glVertexAttribPointer

Calling glDraw* will crash the emulatorif one forgot to call
glVertexAttribPointer first. This patch fixes it.

There are two situations: (1) user forgot to enable any vertex
attribute arrays, or (2) user enabled a vertex attribute array but
forgot to bind any data to it. This patch fix both situations. After the
patch, in (1) user will get an error message, in (2) user will get an
error message as well as GL error code GL_INVALID_OPERATION. These are
similar behaviors as running it in swift shader or on a real device.

(Running it on a real device with (1) will give nothing and with (2)
will crash the app.)

Change-Id: Id045c0c81a489a7c6ee4b6efc91f5ab148432f35

8 years agoSupport glDrawElements with GL_UNSIGNED_INT in the encoder
Yahan Zhou [Wed, 10 Feb 2016 01:42:33 +0000 (17:42 -0800)]
Support glDrawElements with GL_UNSIGNED_INT in the encoder

Currently glDrawElements with type GL_UNSIGNED_INT is not supported by
our encoder and decoder. The emulator will crash if someone uses it.
This patch add the support in the encoder side.

The related issue is here: b.android.com/200338
The patch on the decoder:
https://android-review.googlesource.com/#/c/202101

Change-Id: I9a6cde9983d34f39cfcc8c66a7ccf021dd5963eb

8 years agoMerge "Fix the GLES version number in the emulator encoder" into mnc-emu-dev
Yahan Zhou [Wed, 3 Feb 2016 22:03:43 +0000 (22:03 +0000)]
Merge "Fix the GLES version number in the emulator encoder" into mnc-emu-dev

8 years agoFix the GLES version number in the emulator encoder
Yahan Zhou [Fri, 29 Jan 2016 20:12:57 +0000 (12:12 -0800)]
Fix the GLES version number in the emulator encoder

The native code, emulator encoder and emulator decoder each keeps its
own GLES version number. Previously they are inconsistent when
EGL_CONTEXT_CLIENT_VERSION=3. In this patch, because our emulator
currently doesn't support GLES3, we return EGL_NO_CONTEXT with an
error code EGL_BAD_CONFIG when the user tries to use GLES 3.

In future when we impelment GLES3, please remmeber to change the
version number here, as well as those in native code and the emulator
decoder.

Release Note:
Report error code when setting up GLES3

Change-Id: I1d1083310d28bb57f42625daaed710711495a3ed

8 years agoClean up minor opengl errors/warnings.
Lingfeng Yang [Sat, 30 Jan 2016 00:02:25 +0000 (16:02 -0800)]
Clean up minor opengl errors/warnings.

There are always disconcerting messages printed to logcat
whenever developing apps on the emulator.

glUtilsParamSize and other functions
often fail with "unknow param 0x00..."

These are missing GLenums and types that for
some reason are not in glUtils.cpp
in OpenglCodecCommon.

There also has been a long-standing warning
that eglSurfaceAttrib is not implemented.

It seems we don't need to implement that to get things done,
so we will give a no-op implementation that suppresses
the warning.

Change-Id: I1b672cedf1672fc09cfe6a832f908e1496cc7a46

8 years agoFix pedantic cast error (EGL).
Nicolas Capens [Thu, 10 Dec 2015 22:45:35 +0000 (17:45 -0500)]
Fix pedantic cast error (EGL).

Change-Id: I6afa419f6e2519c3293452f273a0cebac4c00b6f

8 years agoFix pedantic cast error (GLES1).
Nicolas Capens [Thu, 10 Dec 2015 20:10:25 +0000 (15:10 -0500)]
Fix pedantic cast error (GLES1).

Change-Id: I74c2756cbf9190a1e2b9c30546bf60233e4ed543

8 years agoFix pedantic cast error.
Nicolas Capens [Thu, 10 Dec 2015 18:47:02 +0000 (13:47 -0500)]
Fix pedantic cast error.

Change-Id: I1d5bfbc17f4ed36774b8713af14d1c7a7177d3e5

8 years agoAdvertise EGL_KHR_gl_texture_2d_image
Nicolas Capens [Fri, 20 Nov 2015 20:10:09 +0000 (15:10 -0500)]
Advertise EGL_KHR_gl_texture_2d_image

EGL_KHR_gl_texture_2d_image must be supported unconditionally.

Bug 24517776

Change-Id: Ib1e4fc549ceee2291b2313dc46ae7c9c1159e4c9