- 04 Aug, 2014 14 commits
-
-
Jamie Madill authored
By starting from a near-maximum size, and trying until we reach a a valid allocatable size, we can guarantee the test can run. Previously the allocation could fail under some systems and succeded in others, leading to flakiness. BUG=angle:716 Change-Id: If1690349f7028c4e6a88f20649fb255ea2dd2587 Reviewed-on: https://chromium-review.googlesource.com/210992Tested-by:
Jamie Madill <jmadill@chromium.org> Reviewed-by:
Geoff Lang <geofflang@chromium.org>
-
Geoff Lang authored
BUG=angle:658 Change-Id: I8134cde4d72796c51613594e15aded86e83c4af7 Reviewed-on: https://chromium-review.googlesource.com/206837Reviewed-by:
Nicolas Capens <capn@chromium.org> Reviewed-by:
-
Jamie Madill authored
In D3D11, the device doesn't allow overlapped buffer copies unless the D3D11_FEATURE_DATA_D3D11_OPTIONS::CopyWithOverlap feature is enabled. This isn't technically available until D3D11.1. Get around the restriction by using a different storage buffer as the source of the copy. Fixes funcional.buffer.copy.single_buffer.* in dEQP. BUG=angle:715 Change-Id: I03fea8a1c6180a29e07e05022e3454524e320803 Reviewed-on: https://chromium-review.googlesource.com/210902Tested-by:
-
Jamie Madill authored
This was causing a test failure in TwoElseIfRewriting. BUG=angle:716 Change-Id: I75e21ff586cdef10f0c35985a97fbb7c4783d2fe Reviewed-on: https://chromium-review.googlesource.com/210991Tested-by:
-
Jamie Madill authored
We were inserting excess newlines in several places. Change-Id: I746334d3696e9e3be585ab02074384ae63535ea5 Reviewed-on: https://chromium-review.googlesource.com/210823Tested-by:
-
Geoff Lang authored
BUG=angle:658 Change-Id: I8a4ce5f45054a1fa829a9647a94cf2bd0ba93bc0 Reviewed-on: https://chromium-review.googlesource.com/206836Reviewed-by:
-
Jamie Madill authored
We would trigger assertion failures in Debug mode, and fail to parse and translate correctly in Release. BUG=angle:711 Change-Id: Ibb7f33b288376617598578f48c7bbdbdec044279 Reviewed-on: https://chromium-review.googlesource.com/210822Tested-by:
-
Jamie Madill authored
This fixes some broken tests in dEQP: functional.shaders.conversions.matrix_to_matrix BUG=angle:712 Change-Id: I0538595f2913a2c1d4f0da901d65d608a6580d19 Reviewed-on: https://chromium-review.googlesource.com/210882Tested-by:
-
Jamie Madill authored
This fixes some broken tests in dEQP: functional.shaders.conversions.scalar_to_matrix BUG=angle:712 Change-Id: Ia01d720307ea6ca70da266b2085a878eaaab2412 Reviewed-on: https://chromium-review.googlesource.com/210881Tested-by:
-
Geoff Lang authored
Visual Studio 2010 express was the last version to not support 64 bit configurations. BUG=angle:714 Change-Id: Ie4383a437a98cecfc82702e766a460ade96062f8 Reviewed-on: https://chromium-review.googlesource.com/210824Reviewed-by:
-
Jamie Madill authored
The computation could copy more data than needed. Although benign it could cause performance degredation compared to the D3D9 back end. Change-Id: I3cf1dc79085c33d44040fd55153e63a4e5e63cb1 Reviewed-on: https://chromium-review.googlesource.com/210640Tested-by:
-
Geoff Lang authored
Removed all the separate query functions and simply expose the internal info structures. This reduces the number of std::map/std::set operations that were hidden behind the API. Moved the validation tables for ES3 format combinations and effective internal formats into validationES3.cpp so that formatutils.h only has generic GL format queries. BUG=angle:658 Change-Id: Ieb60d42b8eafcdb4f21dcbec130b39478ce5f7c5 Reviewed-on: https://chromium-review.googlesource.com/206835Reviewed-by:
-
Jamie Madill authored
BUG=angle:709 Change-Id: I93f92b916a0da26975cd459399cc2873cb4ab9f0 Reviewed-on: https://chromium-review.googlesource.com/210642Tested-by:
-
Jamie Madill authored
We were passing the arguments from the gl::Buffer stub to the implementation class, which was breaking the call. Also clean up a few comments. BUG=angle:709 Change-Id: Ib92a36d1c52d60d8929f567d0daf54e9d6a2d127 Reviewed-on: https://chromium-review.googlesource.com/210641Tested-by:
-
- 01 Aug, 2014 1 commit
-
-
Jamie Madill authored
Our build in GN differed slightly from our gyp build, causing the linker to complain about inconsistent linkage. Also remove some stray definitions of EGLAPI from our headers. Change-Id: I5d3a09ccb0cec528ead33944ce0996cceed32d86 Reviewed-on: https://chromium-review.googlesource.com/210821Tested-by:
-
- 31 Jul, 2014 7 commits
-
-
Jamie Madill authored
In a previous fix to a Clang build error, we had neglected to update the GN build. Broken in commit 07d49ef5. BUG=angle:707 Change-Id: Ic11aa907d591540e915bec60f3bafaea471689d5 Reviewed-on: https://chromium-review.googlesource.com/210652Tested-by:
Brett Wilson <brettw@chromium.org> Reviewed-by:
-
Jamie Madill authored
Each new entry point corresponds to one of the variable types: varyings, attributes, uniforms, output variables, and interface blocks. They return a pointer to the vector with all of the parsed variables, which then the app can copy to its own memory. Currently we do not support the staticUse field in the HLSL translator. BUG=angle:466 Change-Id: I7dc09e761ab070feef5360ad27740110c44853b3 Reviewed-on: https://chromium-review.googlesource.com/208750Tested-by:
Zhenyao Mo <zmo@chromium.org> Reviewed-by:
-
Jamie Madill authored
This allows the API to query for clear type introspection into the parsed GL types from the translator. The returned types are not expanded and have properly nested fields. This patch uses the types from ShaderVars.h to return GL type information. The app must include this header to get access to the types structs. BUG=angle:466 Change-Id: I28ad0d6f11a964804dd234ef0d00651f665d1ae3 Reviewed-on: https://chromium-review.googlesource.com/208751Tested-by:
-
Jamie Madill authored
This avoid unnecessary binary bloat from STL default constructors and assignment operators. BUG=angle:466,697 Change-Id: I679e524d3e3dbd0d3866fc30a5a01967d8769249 Reviewed-on: https://chromium-review.googlesource.com/208356Tested-by:
bratell at Opera <bratell@opera.com> Reviewed-by:
-
Jamie Madill authored
Most of these constructors are unused. Remove the unused ones, and change the usage of existing constructors to be consistent. BUG=angle:466,697 Change-Id: I455dd314036e1dfcb8c4690bab577b81dd0e060c Reviewed-on: https://chromium-review.googlesource.com/208355Tested-by:
-
Jamie Madill authored
This reverts commit 8c9038d6. Change-Id: I29c25121870e6f8d12144d235029f46fff90be7c Reviewed-on: https://chromium-review.googlesource.com/210651Tested-by:
-
Brandon Jones authored
BUG=angle:688 Change-Id: I43be738e0a42f7a7e1448e1d9300b30d4bc54125 Reviewed-on: https://chromium-review.googlesource.com/209343Reviewed-by:
Brandon Jones <bajones@chromium.org>
-
- 29 Jul, 2014 3 commits
-
-
Jamie Madill authored
We don't allow ReadPixels from multisampled targets in GL. Also update the assertions in Buffer11 to reflect our expectations. BUG=angle:511 Change-Id: Ia3d5f796f5405fbae1187f3d45b0080d7f24b6d9 Reviewed-on: https://chromium-review.googlesource.com/210053Tested-by:
-
Geoff Lang authored
BUG=angle:658 Change-Id: Iab7906309f6603d5364d891bb332af8c3bf457fe Reviewed-on: https://chromium-review.googlesource.com/206834Reviewed-by:
-
Geoff Lang authored
BUG=angle:658 Change-Id: Ic9fa53fb62d7eff62f07b68d7ddada461ecad859 Reviewed-on: https://chromium-review.googlesource.com/206832Tested-by:
-
- 28 Jul, 2014 1 commit
-
-
Nicolas Capens authored
D3D11 interprets an index value of 0xFFFF in a 16-bit index buffer as a primitive restart marker. This behavior can't be toggled off. http://msdn.microsoft.com/en-us/library/windows/desktop/bb205124(v=vs.85).aspx We work around it by converting to 32-bit indices. BUG=angle:708 Change-Id: Ibc92d6ba98e5f11a98d76cae14f90ca050a19964 Reviewed-on: https://chromium-review.googlesource.com/209883Tested-by:
-
- 25 Jul, 2014 4 commits
-
-
Jamie Madill authored
Many of our tests rely on extensions and ES3 support. Until we have the ability to switch between Renderers for specific tests, the D3D11 Renderer should enable many failing tests to pass. Also fix a double-delete crash in the test cleanup routine. BUG=angle:702 Change-Id: I5e615ca1a35bd109d9f7bcbb5fa90b88e09157bd Reviewed-on: https://chromium-review.googlesource.com/209102Reviewed-by:
-
Brandon Jones authored
BUG=angle:688 Change-Id: I0cdc758000993f2d54240c69a84f5ae149306d9e Reviewed-on: https://chromium-review.googlesource.com/209332Reviewed-by:
-
Jamie Madill authored
Clang was complaining that we were casting a char* to a wchar_t*. BUG=angle:707 Change-Id: I063165e9add4e9b57d131181da30c05e5abbc1c2 Reviewed-on: https://chromium-review.googlesource.com/209618Tested-by:
-
Jamie Madill authored
We were not properly setting the masked channels for these APIs. Since we use a std::vector for tracking the render targets for a particular ClearBuffer call, we lose the indexing into the clear parameters. Fix this by storing the information in the std::vector along with the Render Target. BUG=angle:702 Change-Id: Ie3b1e870aa04054411c4f155682b86a340ec00cf Reviewed-on: https://chromium-review.googlesource.com/209103Tested-by:
-
- 24 Jul, 2014 7 commits
-
-
Geoff Lang authored
BUG=angle:700 Change-Id: I036901c397053a75677923304d7e1ed697c82aa2 Reviewed-on: https://chromium-review.googlesource.com/209570 Reviewed-by: <ehsan@mozilla.com> Reviewed-by:
-
Jamie Madill authored
MSVC's more permissing validator missed a dependent typename, and deleting a void pointer. BUG=angle:703,704 Change-Id: I227b94c9c1590973d7b11b0ed0bfda32f2a71cec Reviewed-on: https://chromium-review.googlesource.com/209613Tested-by:
-
Geoff Lang authored
MemoryBuffer has a similar interface to std::vector but returns a bool on resize to do error checking. BUG=angle:700 Change-Id: Ib201eeb91b07f5b7f970e153f5d1e110f9b2fa55 Reviewed-on: https://chromium-review.googlesource.com/209612Reviewed-by:
Nico Weber <thakis@chromium.org> Tested-by:
-
Brandon Jones authored
BUG=angle:688 Change-Id: Ie3e5ac2bedcffb816ef1537575f09afb3f5e7a35 Reviewed-on: https://chromium-review.googlesource.com/208381Reviewed-by:
-
Brett Wilson authored
Translator headers that are included by dependent code (example: ShaderLang.h) expect to be able to include files from "includes" (example: "KHR/khrplatform.h"). This automatically applies this include path to dependent targets. Change-Id: I6990f66132562f142a6cdf8b8c434d2dbaad1c98 Reviewed-on: https://chromium-review.googlesource.com/209452Reviewed-by:
-
Geoff Lang authored
Since filerability and renderability may depend on separate extensions, use functions to determine their support. Context now uses these functions to fill it's own TextureFormatCaps. BUG=angle:658 Change-Id: Ib4dc877ba7c24f84a3823fd5aa2e01a3b3621705 Reviewed-on: https://chromium-review.googlesource.com/206831Reviewed-by:
-
Jamie Madill authored
The code assumed an else-if would end in an else. We can also save a few allocations in the cases where there is no else. BUG=angle:699 Change-Id: I550857366775b4a34aea97e117ef732297d3f448 Reviewed-on: https://chromium-review.googlesource.com/208681Tested-by:
-
- 23 Jul, 2014 3 commits
-
-
Jamie Madill authored
The names of the tests were not properly capitalized, and the huge buffer test could fail unexpectedly when it was run after several prior tests which allocated large amounts of memory. Change-Id: Ied5d1b6f56354e748f0df9d52658d14de9a559ac Reviewed-on: https://chromium-review.googlesource.com/209611Tested-by:
-
Jamie Madill authored
We can compute the interface block member info entirely on the API side. This will allow us to get rid of some un-necessary code in the compiler. BUG=angle:466 Change-Id: I664ffc82de5f2723156e51f4e9ffc07e4de162aa Reviewed-on: https://chromium-review.googlesource.com/207781Reviewed-by:
-
Jamie Madill authored
With the previous cleanups, these fields aren't needed any longer. We can also clean up some unused methods and simplify existing code. BUG=angle:466 Change-Id: I96df8d152324bda5e6868b5eccdf52bdc09155e9 Reviewed-on: https://chromium-review.googlesource.com/207256Reviewed-by:
-
