Age | Commit message (Collapse) | Author | Files | Lines |
|
The draw-sync test (from 2011) was never placed in all.py.
Signed-off-by: Albert Freeman <albertwdfreeman@gmail.com>
|
|
This covers more cases.
|
|
The rules changed from GLSL ES 1.00 to GLSL ES 3.00, so the two tests
expect opposite results.
NOTE: Mesa currently fails
tests/spec/glsl-es-3.00/compiler/constant-sequence.vert and the related
GLES3 conformance test.
v2: s/sequnece/sequence/ noticed by Matt.
v3: Move tests into const-initializer subdirectories. Add tests with
more complex use of sequence operator.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com> [v2]
Cc: Tapani Pälli <tapani.palli@intel.com>
Cc: Mark Janes <mark.a.janes@intel.com>
Cc: Marta Lofstedt <marta.lofstedt@intel.com>
|
|
Trivial.
|
|
The second parameter to glBindTexture should accept a texture
handle, not the address of one.
Cc: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
There are two problems with the way this test reports subtests:
First, it only reports subtests when they fail. This has all kinds of
problems. It's not at all what the summary code expects, the
summary code expects that subtests are always present, which means that
they will show up under disabled/enabled pages instead of
fixes/regressions, which in turn, will confuse developers.
Second, it tries to report lots of information as the subtest name. This
causes the JSON parser in python (JSON is used by the tests to
communicate with the runner) to choke and die.
This patch makes a couple of changes to correct this.
First, it reports the subtest result always, whether pass or fail.
Second, it simplifies the names of the subtests, and passes the other
information to stdout, which the runner will scoop up and record for a
developer to go look at.
v2: - Fix some formatting that wasn't right (indents with 4 spaces not 8
space tabs)
- Report all subtests in both pass and fail cases
- Fix some minor things as reported by Brian Paul
- simplify some control flow
- add function to report each quadrant as a separate subtest with
nice names
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
|
|
This reverts commit e68f387df54767c177fcbf9e2f0b44a98525eb2a.
This test is wrong. Varyings as GL_PROGRAM_INPUT should be in the
resource list only when using SSO, not in case of full shader programs
with multiple stages. I will write a separate test for that case.
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
|
|
The tests for the integer/long minimums do not properly
check that the value is stored in the correct type.
E.g. (-2147483648) actually gets parsed as a long by the preprocessor,
and therefore INT_MIN is actually stored as long and can't be automatically
vectorized.
By subtracting a vector with value of 0 from the given defined *_MIN
and then grabbing the first element of the resulting vector, we can
make sure that the values are actually stored in the correct type.
v2: Remove char/short modifications.
Test *_MAX vectorization as well as _MIN for int/long
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
Reported-By: Moritz Pflanzer <moritz.pflanzer14@imperial.ac.uk>
Signed-off-by: Aaron Watry <awatry@gmail.com>
|
|
To silence warning: "/*" within comment.
|
|
To silence warning: implicit declaration of function ‘dsa_create_program’
|
|
Replaced by new piglit gl-2.0-two-sided-stencil test.
Reviewed-by: Brian Paul <brianp@vmware.com>
|
|
This test replaces the original glean tstencil2.cpp test.
Reviewed-by: Brian Paul <brianp@vmware.com>
|
|
Check that glCreateTextures rejects target=0, per the spec. Also, Mesa
internally uses target=0 for a special state (texture gen'd but not bound)
so it's good to test.
|
|
Previously the test was trying to convert a sample number to the range
[0,1] by dividing by 10.0. This doesn't work when testing 16x MSAA
because it ends up with values greater than 1.0 which get clamped.
This patch makes it divide by miplevels so that it will work whatever
the number of samples is.
Cc: Chris Forbes <chrisf@ijw.co.nz>
Cc: Marek Olšák <marek.olsak@amd.com>
|
|
V3: rebase and test double types
V2: renamed variables from 'inner' that should have been 'outer'
|
|
These tests check that 2D texture arrays work for the 5x5 and
12x12 block sizes.
v2. add to all.py (Ilia).
create test for gles2.
v3. upgrade required GLES version to 3.1 (for 2d array support).
v4. Move CONFIG block to the top of the file (Chad).
Acked-by: Chad Versace <chad.versace@intel.com>
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
These tests check that the GL error returned by valid and invalid API calls
are as defined by the spec.
v2. add test to all.py (Ilia).
update copyright date (Dylan).
numerous changes (Chad).
v3. cube map array support is in OpenGL ES 3.2 (not 3.0).
Reviewed-by: Chad Versace <chad.versace@intel.com>
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
These tests run through every ASTC configuration, comparing the
render of a compressed texture against a render of the decompressed
version of that compressed texture. The compressed and decompressed
texture was generated with a reference codec.
v2. numerous changes (Chad).
v3. upgrade required GLES version to 2.
Reviewed-by: Chad Versace <chad.versace@intel.com>
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
ASTC formats are added to the list of formats that should not be returned by the
COMPRESSED_TEXTURE_FORMATS query.
v2. add astc test to all.py (Chad).
Reviewed-by: Chad Versace <chad.versace@intel.com>
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
This function compares two rectangles for equality. This is useful
to compare the rendering of individual miplevels in a miptree while
avoiding too much resource consumption.
v2: insert spaces in addition operation (Chad).
Reviewed-by: Chad Versace <chad.versace@intel.com>
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
Without glBindTexture(), the test was unintentionally modifying the default
texture object. Make it use the generated texture instead.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
|
|
Use the piglit utility function for determining format block dimensions
instead of using a test-specific function. This requires adding some formats
not originally in the piglit-util function.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
|
|
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
|
|
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
in more cases.
The subtest with at most the maximum number of atomic counter buffers
can be run whenever the combined limit is at least one more than the
limit for the FS stage. The subtest exceeding the maximum number of
combined atomic counter buffers can be run whenever the sum of the
limits for the VS and FS stages is greater than the combined limit.
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
|
|
offsets in the VS of the combined test.
It is illegal for different atomic counter uniform declarations to
overlap, even across stages, use different offsets in the VS to avoid
that.
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
|
|
literal.
This subtest was probably only tested on nVidia and this happened to
work by luck.
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
|
|
max-counters.
This isn't used anywhere except for the max-counters test so we can
move the definition into the C file of that test and make it static.
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
|
|
Check for GL_INVALID_VALUE, not GL_INVALID_OPERATION. The extension spec
isn't clear on this, but GL_INVALID_VALUE is typically generated for
out-of-range int/uint values.
That's what NVIDIA does. Plus, that's what the other bind-texture-unit
piglit test checks for at line 91. Let's at least be consistent.
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
|
|
After looking at CTS I realised we have to accept all types
for get uniform for double conversion, and we should do double
conversion on the non-double types.
I assume this is due to it being the only interface to check
if the location has a value or something.
This adds getting non-double values with GetUniformdv, and
getting double values with float/int/uint interfaces.
notes: currently mesa fails this.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
|
|
CTS tests this a lot better than piglit, but this is a good place
to start, I've got a bunch of fixes queued for mesa to address
some of these problems.
This adds 3 tests:
simple double vs->tcs->tes->fs
simple dvec2 vs->tcs->tes->fs
simple double[2] vs->tcs->tes->fs
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
This tests vertex shaders work with shader subroutine.
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
|
|
The command is:
ssbo <size>
where <size> is the size in bytes to allocate for the SSBO buffer.
This command only sets of a buffer for SSBO index 0.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
|
|
|
|
|
|
v2: (Jan Vesely comments)
Remove extra whitespace
Single flag test cases
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
|
|
v2: Fix typo in comment noticed by Matt. When interpolating from the
north pole to the south pole of the sphere, correctly interpolate from 0
to pi instead of 0 to 2pi. See the "NOTE" in the code.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
The test sets the filter modes to GL_NEAREST, so levels other than zero
are never used.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Why was gluOrtho2D ever even a function?!?
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
I verified correctness by calling gluProject inside project and
comparing the results. My initial implementation forgot to do the
perspective divide, so this helped catch that bug.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
I think this was a holdover from having originated in glean.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Thanked-by: Kristian Høgsberg <krh@bitplanet.net>
|