| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Using void pointers for pointer arithmetic doesn't work. Fixes:
tests\util\piglit_ktx.c(333) : error C2036: 'void *' : unknown size
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=53013
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
|
|
Piglit would segfault if piglit_load_text_file() was accidentally passed
a directory name instead of a filename. Now return NULL if the file is
not regular or a symlink.
|
|
As with the previous commit, but VS.
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
This was only testing compile-time evaluation of sqrt(0).
It's more interesting to do render-time testing, which is done
in other tests.
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
Make sure that all the functions are available when they should be.
This is currently broken in Mesa.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52003
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
Otherwise, we'd see massive spam of WARN tests, due to the frequency
of the bug (about 2% of testcases on my system).
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
We've got a bunch of instances of this happening intermittently during
runs with gnome-shell. Because our tests are not generally able to
handle resizes, they would show up as spurious failures that are hard
to reproduce. Instead, provide a diagnostic message and report WARN.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
This is a lot like getactiveuniformblockname.
|
|
|
|
|
|
|
|
If any of the robustness flags are used, then the resulting context
must support the GL_ARB_robustness extension.
NVIDIA's closed-source driver passes this test. AMD's closed-source
driver has not been tested.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: James Jones <jajones@nvidia.com>
|
|
rejected
NVIDIA's closed-source driver passes this test. AMD's closed-source
driver has not been tested. The spec doesn't explicitly say what
error should be generated, but similar cases generate BadMatch.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: James Jones <jajones@nvidia.com>
|
|
rejected
NVIDIA's closed-source driver passes this test. AMD's closed-source
driver has not been tested.
v2: Based on discussions whtih James Jones (NVIDIA), expect
GLXBadProfileARB instead of BadMatch.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Cc: James Jones <jajones@nvidia.com>
|
|
The spec doesn't forbid indirect rendering with OpenGL ES 2.0.
There's no protocol defined, so it seems impossible that this could
ever work.
NVIDIA's closed-source driver fails this test because an unusable
indirect-rendering ES2 context is created.
AMD's closed-source driver has not been tested.
v2: The test will pass if the indirect-rendering context can actually
be used. For this we require that glGetShaderPrecisionFormat produce
a valid result.
v3: Fix the tests limits for mediump floats. Feedback from Chad.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: James Jones <jajones@nvidia.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
|
|
The profile mechanism is only meaningful for OpenGL 3.2 or later.
Verify that the profile setting is silently ignored when a version
earlier than 3.2 is requested.
NVIDIA's closed-source driver passes this test. AMD's closed-source
driver has not been tested.
v2: Try versions other than just 1.0. This catches a bug in recent
xserver code.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
References: http://lists.x.org/archives/xorg-devel/2012-July/032440.html
Cc: Chad Versace <chad.versace@linux.intel.com>
|
|
for OpenGL 3.2
NVIDIA's closed-source driver passes this test. AMD's closed-source
driver has not been tested. NVIDIA's closed-source driver supports
both core and compatibility profiles.
v2: Try to create a core profile both with and without the core
profile mask. This was suggested by Chad Versace.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
|
|
NVIDIA's closed-source driver passes this test. AMD's closed-source
driver has not been tested.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
|
|
v2: More clip tests and comments.
|
|
sys/mman.h is not available on Windows.
|
|
|
|
|
|
Tests the depth buffer only for coverage value equal to 0.0 and 1.0 because
depth buffer behaviour is undefined for intermediate coverage values.
This test found an issue with the current implemantation of msaa for i965
on mesa. Following patch on mesa makes this test to pass:
[PATCH] i965/msaa: Set KILL_ENABLE when GL_ALPHA_TO_COVERAGE enabled
commit 6c355cca9149e43850cf27f2d0821fab1e7a69f5
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
Computes expected depth values and probes the depth buffer only for coverage
value equal to 0.0 and 1.0 because depth buffer behaviour is undefined for
intermediate coverage values.
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
Lot of the functions defined in this test case are also shared by
draw-buffers-common.cpp. So this refactoring is done to avoid code
duplication.
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
Lot of the functions defined in this test case are also shared by
draw-buffers-common.cpp. So this refactoring is done to avoid code
duplication.
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
This test verifies that alpha values are not modified if
GL_SAMPLE_ALPHA_TO_ONE is enabled and the buffer is single-sampled.
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
This test varifies that alpha values are not changed with
GL_SAMPLE_ALPHA_TO_ONE enabled if GL_MULTISAMPLE is disabled.
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
When draw buffer zero is bound to an integer format buffer,
GL_SAMPLE_ALPHA_TO_COVERAGE should have no effect. This should work
properly even if there are other draw buffers that are not in integer
format.
Note: Test passes on Sandybridge with following patch applied on mesa.
Patch is under review on mesa-dev mailing list:
[PATCH] i965/msaa: Disable alpha-to-{coverage, one} when drawbuffer zero
is in integer format
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
When draw buffer zero is bound to an integer format buffer,`
GL_SAMPLE_ALPHA_TO_ONE should have no effect. This should work properly
even if there are other draw buffers that are not in integer format.
Note: Test passes on Sandybridge with following patch applied on mesa.
Patch is under review on mesa-dev mailing list:
[PATCH] i965/msaa: Disable alpha-to-{coverage, one} when drawbuffer zero
is in integer format
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
When rendering to multiple draw buffers, the alpha value used by
GL_SAMPLE_ALPHA_TO_COVERAGE should come from draw buffer zero, but it should
have an effect on all the draw buffers.
Reviewed-by: Paul Berry <stereotype441@gmail.com>
|
|
When rendering to multiple draw buffers, GL_SAMPLE_TO_ONE should modify
the alpha values sent to each draw buffer.
OpenGL 3.3 specification's section 4.1.3 is little unclear about the exact
behavior in above mentioned case. So, this test assumes the behavior shown
by NVIDIA's proprietary drivers.
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
Functions defined in these files are utilized by following test cases:
draw-buffers-alpha-to-one
int-draw-buffers-alpha-to-one
draw-buffers-alpha-to-coverage
int-draw-buffers-alpha-to-coverage
V2: Removed unnecessary conditional checks
Switched to shared visualize_image function
Updated few comments
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
visualize_image function is utilized by multiple msaa test cases.
Reviewed-by: Paul Berry<stereotype441@gmail.com>
|
|
|
|
From the GL 3.0 spec, section 4.3.3, in the documentation for
CopyPixels():
"An INVALID_OPERATION error will be generated if the object bound
to READ_FRAMEBUFFER_BINDING is framebuffer complete and the value
of SAMPLE_BUFFERS is greater than zero."
The same applies to CopyTexImage...() and CopyTexSubImage...()
functions, since they are defined in terms of CopyPixels().
Previously we were checking for an INVALID_FRAMEBUFFER_OPERATION error
in these cases.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
This test reproduces a problem in Mesa's i965 driver as of 7/12/12
(commit 1a06e84): when no color buffer is present, but a multisampled
depth or stencil buffer is present, Mesa fails to set up the rendering
pipeline for multisampling.
The test includes variants for testing both the stencil buffer and the
depth buffer. The depth buffer test may be run in a "computed" mode,
in which case the depth values are output by the fragment shader (this
is an important corner case for i965).
Verified on nVidia's proprietary binary driver for Linux.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
|
|
This patch modifies the common code for MSAA tests, so that when
testing depth buffer behaviour, we can choose to either output the
depth value from the vertex shader (using gl_Position) or from the
fragment shader (using gl_FragDepth).
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
|
|
This patch modifies the common code for MSAA tests, so that we can
easily configure framebuffers that lack a color, depth, and/or stencil
buffer.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
|
|
The GL spec mandates that signed normalized color formats be clamped
to [0, 1] when in GL_FIXED_ONLY mode, despite the fact that they can
represent values in the range [-1, 1]. So, when testing these
formats, we use a color clamping mode of GL_FALSE.
Note: it would have been easier to simply use a color clamping mode of
GL_FALSE for all formats, however since part of the purpose of the
test is to verify that unnecessary clipping does not occur, we want to
give as many opportunities as possible for the implementation to
incorrectly clip.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Previously, the MSAA tests for floating-point color buffers only
exercised color values in the range [0, 1]. This patch extends the
range to [-10, 10], in order to verify that floating-point color
values are not inadvertently clamped.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
where execfile() calls in python do not automatically declare variables as globals.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
|
|
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
|
|
platform where the os seperator is not '\'.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
|
|
test name will always use '/'.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
|
