Skia command buffer bot failing floating point texture test on recent command buffer builds |
||||
Issue descriptionSometime yesterday our command buffer bot (which syncs ToT Chrome) started failing a simple floating point texture test. These can be seen here: https://chromium-swarm.appspot.com/task?id=349dea9a554fca10&refresh=10 The failures look like this: FAILURE: ../../../tests/FloatingPointTextureTest.cpp:60 [Command Buffer]: fpTexture->readPixels(0, 0, DEV_W, DEV_H, desc.fConfig, readBuffer.begin(), 0) FAILURE: ../../../tests/FloatingPointTextureTest.cpp:62 [Command Buffer]: 0 == memcmp(readBuffer.begin(), controlPixelData.begin(), readBuffer.bytes()) FAILURE: ../../../tests/FloatingPointTextureTest.cpp:60 [Command Buffer]: fpTexture->readPixels(0, 0, DEV_W, DEV_H, desc.fConfig, readBuffer.begin(), 0) FAILURE: ../../../tests/FloatingPointTextureTest.cpp:62 [Command Buffer]: 0 == memcmp(readBuffer.begin(), controlPixelData.begin(), readBuffer.bytes()) There are some GL errors in the log as well. However, those are also present prior to the floating point texture failures. https://skia.org/dev/chrome/commandbuffer has instructions on how to run the command buffer through Skia's dm test program. Using flags -"-match Floating --config commandbuffer" should reproduce this.
,
Feb 28 2017
It looks like this started happening sometime in this range: https://chromium.googlesource.com/chromium/src/+log/9c42082..220ff33
,
Mar 7 2017
cc'ing more people based on hand waving.
,
Mar 18 2017
Suspecting this change in the range. commit 8507215795207d4868c9ad379957819c2f15b881 author ccameron <ccameron@chromium.org> Fri Feb 24 19:53:23 2017 Add GL_EXT_color_buffer_half_float support Only expose for GLES2 contexts. For desktop support, add simliar checks to GL_EXT_color_buffer_float. Change a number of instances of accidentally referring to the extension as "color_buffer_float" instead of "color_buffer_half_float". BUG= 687439 Review-Url: https://codereview.chromium.org/2713553005 Cr-Commit-Position: refs/heads/master@{#452909}
,
Mar 20 2017
Taking a look at this. There was a similar sRGB failure that I also need to look at.
,
May 1 2017
The following revision refers to this bug: https://skia.googlesource.com/skia/+/bb3e7ffa3e6ba647b89d54ea32082b82e0421906 commit bb3e7ffa3e6ba647b89d54ea32082b82e0421906 Author: Ben Wagner <benjaminwagner@google.com> Date: Mon May 01 14:37:49 2017 Omit HalfFloatAlphaTextureTest on CommandBuffer bot. Bug: chromium:697030 No-Try: true Change-Id: Ib56c88c384e6b363aefe91c7027fa587d0b000b5 Reviewed-on: https://skia-review.googlesource.com/14648 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Ben Wagner <benjaminwagner@google.com> [modify] https://crrev.com/bb3e7ffa3e6ba647b89d54ea32082b82e0421906/infra/bots/recipes/test.expected/Test-Mac-Clang-MacMini6.2-GPU-HD4000-x86_64-Debug-CommandBuffer.json [modify] https://crrev.com/bb3e7ffa3e6ba647b89d54ea32082b82e0421906/infra/bots/recipes/test.py
,
Jul 26 2017
|
||||
►
Sign in to add a comment |
||||
Comment 1 by bsalo...@google.com
, Feb 28 2017Labels: OS-Mac