- 24 Mar, 2014 1 commit
-
-
Jamie Madill authored
It got broken in r2423. We were referring to field-name as struct-name. This patch fixes the regression and also improves the error message by adding the field-name. BUG=angle:459 R=kbr@chromium.org Review URL: https://codereview.appspot.com/12891043 Change-Id: I9ac9a8dafcad1857f7d39e6d494e457d1fc50f70 Reviewed-on: https://chromium-review.googlesource.com/190530Reviewed-by:
Shannon Woods <shannonwoods@chromium.org> Reviewed-by:
Nicolas Capens <nicolascapens@chromium.org> Tested-by:
Jamie Madill <jmadill@chromium.org>
-
- 18 Mar, 2014 7 commits
-
-
Geoff Lang authored
This reverts commit 4251b758. Change-Id: I95afebcaf91ee38a121df20f403eb06f19b2510a Reviewed-on: https://chromium-review.googlesource.com/190560Reviewed-by:
Geoff Lang <geofflang@chromium.org>
-
Shannon Woods authored
Would cause UNREACHABLE to be triggered if target was invalid, and INVALID_VALUE to be returned instead of INVALID_ENUM if it were the only erroneous argument. Change-Id: I2606e77379fa2832a50b3a299e6474a4b2f68afa Reviewed-on: https://chromium-review.googlesource.com/189701Reviewed-by:
-
Geoff Lang authored
BUG=angle:493 Change-Id: Ibeae41da2b83ab63d237912602ba5d2aad0bbec2 Reviewed-on: https://chromium-review.googlesource.com/185036Tested-by:
-
Jamie Madill authored
In D3D11, map-discard is reserved for DYNAMIC usage buffer, which are resources that live on the GPU. Staging buffers don't have the notion of map-discard, and using them with discard gives a D3D11 error. This patch fixes crashes in dEQP buffer.map.write.invalidate. BUG=angle:587 Change-Id: Ie225e72d4e226c69e73826c9fb67bbb940bbc466 Reviewed-on: https://chromium-review.googlesource.com/190072Reviewed-by:
-
Jamie Madill authored
We were incorrectly setting the signed parameter to true in the vertex format table. This fixes a dEQP test in vertex_arrays.output_types. BUG=angle:586 Change-Id: I84dcdee89a95369cc8e41d451f96dbb8daa37b6d Reviewed-on: https://chromium-review.googlesource.com/190056Reviewed-by:
-
Jamie Madill authored
The dEQP varying struct linkage tests were triggering asserts when we tried to query component info of struct types. BUG=angle:580 Change-Id: I61e33573c5577a327c58fec7a07ae7718690ac42 Reviewed-on: https://chromium-review.googlesource.com/189194Reviewed-by:
-
Jamie Madill authored
We can't assume we have a simple mapping between input attribs and shader attribs. The app specifies matrix input attribs as several vector inputs, while the shader uses a single matrix type. BUG=angle:576 Change-Id: I4e3cdde4a14a5f06b86052dfe0325113cbe8e599 Reviewed-on: https://chromium-review.googlesource.com/189193Reviewed-by:
-
- 17 Mar, 2014 5 commits
-
-
Shannon Woods authored
bug=angle:582 Also performs format checks for non-subimage copies, and permits unsized format textures as destination for the copy. Change-Id: Id647bb968416fbd35e1c70e5da21c81367e08f07 Reviewed-on: https://chromium-review.googlesource.com/189431Reviewed-by:
-
Scott Graham authored
This is the initial draft of a spec that extends EGL_EXT_platform_base to allow choosing a particular ANGLE D3D backend. Change-Id: I1d80fa8b3c4a59fe998f6d15aa0f393bdb9d26c1 Reviewed-on: https://chromium-review.googlesource.com/184975Reviewed-by:
Scott Graham <scottmg@chromium.org> (cherry picked from commit dc18dc30) Reviewed-on: https://chromium-review.googlesource.com/190026Reviewed-by:
-
Geoff Lang authored
D3D10 cards perform poorly when the pixel shader outputs to multiple render targets that are not bound. BUG=325581 Change-Id: I13f6f98861ffeb77cd21f7401ded4425e4ea7841 Reviewed-on: https://chromium-review.googlesource.com/185400Reviewed-by:
-
Geoff Lang authored
Cherry-pick of bd79517a. Change-Id: I87d790c14d68789524df76f63a61fc30a8792e4e Reviewed-on: https://chromium-review.googlesource.com/190350Reviewed-by:
-
Geoff Lang authored
Since a DirectBuffer11 may return different buffers from getBuffer depending on the required buffer type and if the internal buffers are dirty or not, the current serial is not reliable. Use the actual buffer object so that there can be no missmatch. BUG=350581 Change-Id: I6ea1616273346117a17646edb68e6518e584f2b1 Reviewed-on: https://chromium-review.googlesource.com/189435Reviewed-by:
-
- 14 Mar, 2014 4 commits
-
-
Jamie Madill authored
We should use the cross-platform os.path.join() to support platforms other than Windows. BUG=angle:577 Change-Id: I9d96fc1b13731a57ffcd97469632f0acedb8e5c7 Reviewed-on: https://chromium-review.googlesource.com/188879Reviewed-by:
Zhenyao Mo <zmo@chromium.org> Reviewed-by:
-
Geoff Lang authored
Change-Id: I135c4457fdc95f2d331e2924104a30f90a40bd1a Reviewed-on: https://chromium-review.googlesource.com/189456Reviewed-by:
-
Jamie Madill authored
Move ReadPixels error generation out of the implementation and into the API level. BUG=angle:571 Change-Id: I0b32294f359fedd13d1af2c95baf37a3e5ac1d5b Reviewed-on: https://chromium-review.googlesource.com/188014Reviewed-by:
-
Jamie Madill authored
GetBufferPointerv, GetBufferParameteri64v, and new values for GetBufferParameteriv. BUG=angle:565 Change-Id: I179a3193987b2bff1b0628e47f6a0d627541b0cf Reviewed-on: https://chromium-review.googlesource.com/188203Reviewed-by:
-
- 12 Mar, 2014 1 commit
-
-
Nicolas Capens authored
BUG=348205 Change-Id: I9be17832b1a35d2995725ab147977992c2732274 Reviewed-on: https://chromium-review.googlesource.com/188711Tested-by:
-
- 11 Mar, 2014 2 commits
-
-
Jamie Madill authored
Change-Id: I0a385ff823ef492b4fced89a734d8053fdb0ffd0 Reviewed-on: https://chromium-review.googlesource.com/189474Reviewed-by:
-
Zhenyao Mo authored
We use hashing to map all variables/strcuture field names, etc, so we no longer need this option. Checked with Firefox and WebKit, they no longer use this option either. Time to remove it. Change-Id: Ie3e79b91a05258b04af419a9c42b2fd1b00e67c4 Reviewed-on: https://chromium-review.googlesource.com/189236Reviewed-by:
Kenneth Russell <kbr@chromium.org> Reviewed-by:
-
- 10 Mar, 2014 1 commit
-
-
Shannon Woods authored
Change-Id: I2fe1d02cdf59409cb2d8a68b407bbd01ab7d8c0a Reviewed-on: https://chromium-review.googlesource.com/188954Reviewed-by:
-
- 07 Mar, 2014 5 commits
-
-
Zhenyao Mo authored
1) Before this workaround is hardwired on mac, now we move it behind a compil 2) Fix the issue where "break" inside the loop isn't handled while unrolled. BUG=338474 TEST=webgl conformance test sampler-array-using-loop-index.html Change-Id: I4996a42c2dea39a8a5af772c256f8e3cb383f59a Reviewed-on: https://chromium-review.googlesource.com/188079Reviewed-by: -
Geoff Lang authored
BUG=angle:575 Change-Id: Ibad77fda193c850c29483aa56d0d2275c131429a Reviewed-on: https://chromium-review.googlesource.com/189163Tested-by:
-
Geoff Lang authored
Instead of opening and closing the debug log file with each message, keep it open permanently and flush it with each message, allowing the OS to flush to disk as needed. With debug logging enabled, performance is improved by 100x! Also, dynamically size the output buffer to the max message size. When running all the samples, the largest debug message was 318 characters so an initial size of 512 (instead of 32768) seems reasonable. BUG=angle:575 Change-Id: If4a1387431fc2d32a6b2d1f9ca1f2590473feb86 Reviewed-on: https://chromium-review.googlesource.com/188713Reviewed-by:
-
Jamie Madill authored
BUG=angle:565 Change-Id: I14cee4eec032d4b6b8d9479da4234b81e4ec6bec Reviewed-on: https://chromium-review.googlesource.com/188202Reviewed-by:
-
Jamie Madill authored
A query helper method in Context and a validation helper method reduce code redundancy quite a bit. BUG=angle:565 Change-Id: I96b9ecc4614e7e31240e31594223e8b9636d22c2 Reviewed-on: https://chromium-review.googlesource.com/188201Reviewed-by:
-
- 05 Mar, 2014 3 commits
-
-
Geoff Lang authored
BUG=angle:495 Change-Id: I2d9fbf9c245bc519b8c5a724ca3912aaa7a23d97 Reviewed-on: https://chromium-review.googlesource.com/185034Tested-by:
-
Jamie Madill authored
GPU conversion means we should be using a default of 1 as for int types instead of the MAXINT value for normalized types. Fixes the WebGL test for vertex attribs with offsets. BUG=angle:550 Change-Id: I1aa1b1bae2282e8eda96366adeec492ffe1d5728 Reviewed-on: https://chromium-review.googlesource.com/188433Reviewed-by:
-
Jamie Madill authored
Since we never, and never will use a default value of zero for the alpha channel, we can use a value of zero to indicate no widening. This has the benefit of not needing a default value where it does not apply. BUG=angle:550 Change-Id: Ifdfedcb16039aa1d7f8ee644b65c6b4d89f9a384 Reviewed-on: https://chromium-review.googlesource.com/188432Reviewed-by:
-
- 04 Mar, 2014 1 commit
-
-
Nicolas Capens authored
BUG=angle:525 Change-Id: I3c3ec2cecebf9e745f0c02a132433e3076a6fdea Reviewed-on: https://chromium-review.googlesource.com/187534Tested-by:
-
- 03 Mar, 2014 1 commit
-
-
Geoff Lang authored
Skip over files that begin with '.' since they are often hidden temporary files. BUG=angle:553 Change-Id: I74f8412e6959c49bfca7f0ae063af42027e12549 Reviewed-on: https://chromium-review.googlesource.com/188035Reviewed-by:
-
- 27 Feb, 2014 1 commit
-
-
Jamie Madill authored
Because libGLESv2 needs a static translator to link against internal types like Attribute and Varying, as well as using some other internal methods, we can for now disable the component build for the translator. In the future we can support a proper component build with exports across library boundaries. BUG=angle:568 Change-Id: Ie3efc0a29342d02491e91952c3c2398568f83576 Reviewed-on: https://chromium-review.googlesource.com/187699Reviewed-by:
-
- 26 Feb, 2014 5 commits
-
-
Jamie Madill authored
The previous fix would allow the calling app to trigger an ASSERT if it did not first call ShFinalize. Chrome would do this in some cases, causing a test failure. BUG=344043 Change-Id: I33fa0c4668dac8750ca949febc7f02cc7d3d95e7 Reviewed-on: https://chromium-review.googlesource.com/187849Reviewed-by:
-
Nicolas Capens authored
BUG=346489 Change-Id: I28f5751580729a4d4d77fa6fdee0b4a6628a05de Reviewed-on: https://chromium-review.googlesource.com/188030Tested-by:
-
Jamie Madill authored
Because of a local static, we could get stuck and never re-initialize the translator after calling ShFinalize. BUG=344043 Change-Id: Ib4d18da886eac28f3e1bc959a7eadf5b34d9410f Reviewed-on: https://chromium-review.googlesource.com/186975Reviewed-by:
-
Jamie Madill authored
BUG=angle:565 Change-Id: I4f416981bc63098f0364653ff279b3f54eeb756e Reviewed-on: https://chromium-review.googlesource.com/186979Reviewed-by:
-
Jamie Madill authored
Mostly these calls pass through to BufferStorage. BUG=angle:565 Change-Id: I38ab7d4f49bfad940773428f5bb330de55d3f573 Reviewed-on: https://chromium-review.googlesource.com/187080Reviewed-by:
-
- 25 Feb, 2014 3 commits
-
-
Geoff Lang authored
BUG=angle:570 Change-Id: I6654c60aab69861771b6917081baaecbded71bab Reviewed-on: https://chromium-review.googlesource.com/187840Tested-by:
-
Jamie Madill authored
Chromium directly includes our .gypi files in its own standalone tests, hence we need to keep files where it looks for includes. Previously we had removed the gypi files with the new project structures. BUG=angle:568 Change-Id: I2cf87bdd3d9635ca8e58b991b9a3e55e00261851 Reviewed-on: https://chromium-review.googlesource.com/187650Reviewed-by:
-
Jamie Madill authored
Using two calls to vsnprintf allows us to correctly determine maximum buffer size as well as fixing garbage characters from a missing NULL terminator. Bug report from roland@lagoa.com. BUG=angle:561 Change-Id: I4eb156342d9106d40783a3cf57c0c1d1889390e2 Reviewed-on: https://chromium-review.googlesource.com/186988Reviewed-by:
-
