device_generic_goldfish-opengl/shared, branch master Unnamed repository; edit this file 'description' to name the repository. Don't try to do robust buffer access checks on non-linked attrib indices 2020-09-17T03:33:43+00:00 Lingfeng Yang lfy@google.com 2020-09-17T01:55:01+00:00 62054e92be714b2f17aeea2f9347cfa0139f30c2 There's no danger from allowing out of bounds access there if the shader program does nothing with it. Don't upload client arrays in that case either. Change-Id: Ic356a75d28fcf880608f3d44ca123bba18913c65 
There's no danger from allowing out of bounds access there if the shader
program does nothing with it.

Don't upload client arrays in that case either.

Change-Id: Ic356a75d28fcf880608f3d44ca123bba18913c65
Merge "Move host validation to guest" 2020-09-10T19:55:47+00:00 Lingfeng Yang lfy@google.com 2020-09-10T19:55:47+00:00 e96843b734bcd0e2f5ac1a06703e438a43fc7086 






Move host validation to guest
2020-09-10T14:52:30+00:00

Lingfeng Yang
lfy@google.com

2020-09-04T02:11:02+00:00

ed069c5944c3066ea1bbb3db65e4791a96b717b5

Test: dEQP-GLES3 no failures when not using host-side glGetError

Change-Id: If602d7c784ea12c5ae47ce56a91669e14a9ac62f





Test: dEQP-GLES3 no failures when not using host-side glGetError

Change-Id: If602d7c784ea12c5ae47ce56a91669e14a9ac62f







memset all drm_virtgpu_map structs to 0 before using
2020-08-27T03:09:17+00:00

Lingfeng Yang
lfy@google.com

2020-08-27T00:03:15+00:00

85dfc611a7e29547287ce31119e31c17f55dd510

Change-Id: I627c64d4760ca465541f2d5fec69f3c88c4fc336





Change-Id: I627c64d4760ca465541f2d5fec69f3c88c4fc336







Don't glMapBufferRange on the host if possible
2020-08-19T22:13:07+00:00

Lingfeng Yang
lfy@google.com

2020-08-19T16:07:56+00:00

1ae08114d848824dd8bcc019b68352afa1c758a0

It is expensive to

- let glMapBufferRange to the host at all
- read from the GPU mapped pointer when on the host

The second item will introduce a stall similar to
glClientWaitSync/glFinish on the host GPU.

Thus, track when the buffer is possibly dirty and only read from it
during those times. At other times, mapped buffers that are only used
for write can be implemented in a completely feed-forward way.

Change-Id: Ife9393e4eb5238411e6eae0abb49272523a8d935





It is expensive to

- let glMapBufferRange to the host at all
- read from the GPU mapped pointer when on the host

The second item will introduce a stall similar to
glClientWaitSync/glFinish on the host GPU.

Thus, track when the buffer is possibly dirty and only read from it
during those times. At other times, mapped buffers that are only used
for write can be implemented in a completely feed-forward way.

Change-Id: Ife9393e4eb5238411e6eae0abb49272523a8d935







Add StateTrackingSupport that includes a class to track predicates
2020-08-19T22:13:02+00:00

Lingfeng Yang
lfy@google.com

2020-08-19T16:06:15+00:00

7d677c5a14d3170396d678f5daeb8b760d350807

The guest-side state tracker looks like it can be sped up by using
HybridComponentManager. One of the first uses of this is going to be a
generic "predicate" structure that will be used for tracking
existence/dirtiness of objects.

Change-Id: I1f4c52ba4b256a2a574196f64b0139891171c873





The guest-side state tracker looks like it can be sped up by using
HybridComponentManager. One of the first uses of this is going to be a
generic "predicate" structure that will be used for tracking
existence/dirtiness of objects.

Change-Id: I1f4c52ba4b256a2a574196f64b0139891171c873







Change GOLDFISH_VULKAN macro to GFXSTREAM macro
2020-08-19T22:12:33+00:00

Lingfeng Yang
lfy@google.com

2020-08-17T22:46:53+00:00

6f9871ba3a3683b3ce96f1e964c61fd69d3f834b

The macro's true purpose seems to be a general "modern"
specifier like "vulkan support exists" and
"this is an env where c++11 is available" etc.

Change-Id: I9fa61c80f71be1c6d521b4963d03b6192f14c85e





The macro's true purpose seems to be a general "modern"
specifier like "vulkan support exists" and
"this is an env where c++11 is available" etc.

Change-Id: I9fa61c80f71be1c6d521b4963d03b6192f14c85e







fuchsia: Replace FIDL C++ bindings with LLCPP.
2020-07-21T18:27:20+00:00

Yilong Li
liyl@google.com

2020-07-18T04:05:22+00:00

6dc9b5b39ac8df58e1a8077fece72f399605ea2f

Previously goldfish ICD uses FIDL C++ bindings which
has dependency on the default async dispatcher and
caused some binary dependency issues. This change
migrates all usages of C++ bindings to LLCPP to
eliminate that dependency.

Bug: fxbug.dev/56565
Bug: fxbug.dev/55650

TEST=gfx_unittests, escher_tests on Fuchsia

Change-Id: I0815e210b1630ce2ea9fd32222ddb4a32bbf8644





Previously goldfish ICD uses FIDL C++ bindings which
has dependency on the default async dispatcher and
caused some binary dependency issues. This change
migrates all usages of C++ bindings to LLCPP to
eliminate that dependency.

Bug: fxbug.dev/56565
Bug: fxbug.dev/55650

TEST=gfx_unittests, escher_tests on Fuchsia

Change-Id: I0815e210b1630ce2ea9fd32222ddb4a32bbf8644







Fix Fuchsia build.
2020-07-01T04:20:15+00:00

David Reveman
reveman@google.com

2020-07-01T03:29:14+00:00

744cad1c13ff7bc13faa9902b0cd3edfbd51a6f3

Change-Id: Ifad3cb5a3d18568b408b98f704fe653a7d0e5b1b





Change-Id: Ifad3cb5a3d18568b408b98f704fe653a7d0e5b1b







reland: virtio-gpu-asg transport
2020-06-29T21:54:18+00:00

Lingfeng Yang
lfy@google.com

2020-05-27T21:10:11+00:00

667e8ee06e889765f4c50b5cb55701f041023e39

Requires ro.boot.hardware.gltransport=virtio-gpu-asg or
ro.kernel.qemu.gltransport=virtio-gpu-asg

Bug: 160181061
Change-Id: Ia7865320ad7b3a9ec3f6971ead1cabe3fff36d27





Requires ro.boot.hardware.gltransport=virtio-gpu-asg or
ro.kernel.qemu.gltransport=virtio-gpu-asg

Bug: 160181061
Change-Id: Ia7865320ad7b3a9ec3f6971ead1cabe3fff36d27