summaryrefslogtreecommitdiff
path: root/tests
AgeCommit message (Collapse)AuthorFilesLines
2016-03-06glx-shader-sharing: Report failure instead of crash on assert.Vinson Lee1-1/+3
Signed-off-by: Vinson Lee <vlee@freedesktop.org> Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
2016-03-05cl: clSetKernelArg test more args kindSerge Martin1-3/+27
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
2016-03-05cl: clSetKernelArg check for image supportSerge Martin1-14/+37
Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
2016-03-04arb_internalformat_query2: Added tests for {COLOR,DEPTH,STENCIL}_COMPONENTSAntia Puentes3-0/+283
Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for FILTER pnameAlejandro Piñeiro3-0/+245
From spec: " FILTER: The support for filter types other than NEAREST or NEAREST_MIPMAP_NEAREST for the resource is written to <params>. This indicates if sampling from such resources supports setting the MIN/MAG filters to LINEAR values. Possible values returned are FULL_SUPPORT, CAVEAT_SUPPORT, or NONE. If the resource or operation is not supported, NONE is returned." On the case of FILTER there are well known cases defined by the OpenGL spec where multi-texel filtering is not allowed: * Multi-sample textures ( GL_TEXTURE_2D_MULTISAMPLE, GL_TEXTURE_2D_MULTISAMPLE_ARRAY). * Any resource using a integer internalformat * Texture buffer objects So in addition to check that it returns NONE for not supported internalformat, we know that it should return NONE too for those cases. In other cases, it checks that the returned value is FULL_SUPPORT, CAVEAT_SUPPORT or NONE. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails: * Texture Buffer target is returning FULL_SUPPORT. * Multi-texture targets is returning FULL_SUPPORT. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: Added tests for image texturesAntia Puentes3-0/+218
Checks that: * TEXEL_SIZE * IMAGE_COMPATIBILITY_CLASS * IMAGE_PIXEL_FORMAT * IMAGE_PIXEL_TYPE return the values defined for them in Table 3.22 of the OpenGL 4.2 specification depending on the "Image Format" passed. For unsupported resources, this test checks that the returned value is equal to the 'unsupported' response defined by the extension specification. v2: check for unsupported values, so testing this pnames could be removed from generic-pnames Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails: * IMAGE_TEXEL_SIZE From spec: The size of a texel when the resource when used as an image texture is returned in <params>. This is the value from the /Size/ column in Table 3.22. If the resource is not supported for image textures, or if image textures are not supported, zero is returned. NVIDIA proprietary drivers returns 1, 2 or 4, a values not present on Table 3.22. * IMAGE_PIXEL_TYPE: From spec: The pixel type of the resource when used as an image texture is returned in <params>. This is the value from the /Pixel type/ column in Table 3.22. If the resource is not supported for image textures, or if image textures are not supported, NONE is returned. NVIDIA proprietary drivers returns in some cases GL_UNSIGNED_INT_8_8_8_8_REV, that although defined on core spec 4.2, it is not included on Table 3.22 * IMAGE_PIXEL_FORMAT: From spec: The pixel format of the resource when used as an image texture is returned in <params>. This is the value from the /Pixel format/ column in Table 3.22. If the resource is not supported for image textures, or if image textures are not supported, NONE is returned. NVIDIA proprietary drivers returns in some cases GL_R11F_G11F_B10F, that although defined on core spec 4.2, it is not included on Table 3.22 Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: port of query1 minmax test to query2Alejandro Piñeiro5-0/+402
It is basically the same, but testing with both GetInternalformativ and GetInternalformati64v, using the test_data struct defined at common.h Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for TEXTURE_COMPRESSED_BLOCK_<X>Alejandro Piñeiro3-0/+205
This test add a check for the following pnames: * TEXTURE_COMPRESSED_BLOCK_WIDTH * TEXTURE_COMPRESSED_BLOCK_HEIGHT * TEXTURE_COMPRESSED_BLOCK_SIZE On all those three, query2 spec says the following: "If the internal format is not compressed, or the resource is not supported, 0 is returned." We could have classified the existing internalformats on compressed/non-compressed (similar to color-format/non-color-format for COLOR_ENCONDING), but that seems pointless taking into account that we already have TEXTURE_COMPRESSED to query if a internalformat is compressed. So this test queries TEXTURE_COMPRESSED and INTERNALFORMAT_SUPPORTED, and if any of them is false, checks that the returned value for any of those three pnames is zero. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for COLOR_ENCODINGAlejandro Piñeiro3-0/+211
From spec: "COLOR_ENCODING: <skip> Possible values for color buffers are LINEAR or SRGB, for linear or sRGB-encoded color components, respectively. For non-color formats (such as depth or stencil), or for unsupported resources, the value NONE is returned." Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: add generic check for GET_TEXTURE_IMAGE_TYPEAlejandro Piñeiro1-0/+19
From spec: "GET_TEXTURE_IMAGE_TYPE: <skip> Possible values include any value that is legal to pass for the <type> parameter to GetTexImage, or NONE if the resource does not support this operation, or if GetTexImage is not supported." Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: add generic check for GET_TEXTURE_IMAGE_FORMATAlejandro Piñeiro1-0/+65
From query2 spec: "GET_TEXTURE_IMAGE_FORMAT: <skip> Possible values include any value that is legal to pass for the <format> parameter to GetTexImage, or NONE if the resource does not support this operation, or if GetTexImage is not supported." The possible list of values is the same that testing TEXTURE_IMAGE_FORMAT, but depending on the gl version, GL_STENCIL_INDEX is allowed too. This commits adds another list of possible values. We could have added a clone method that added extra values, but for now that would be an overkill. We could add more if we find more similar cases. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails. As with TEXTURE_IMAGE_FORMAT in some cases returns GL_R11F_G11F_B10F or GL_RGB9_E5, that are internalformats, not formats. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: add a generic check for TEXTURE_IMAGE_TYPEAlejandro Piñeiro1-0/+54
"TEXTURE_IMAGE_TYPE: <skip> Possible values include any value that is legal to pass for the <type> parameter to the Tex*Image*D commands, or NONE if the resource is not supported for this operation." Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: add a generic check for TEXTURE_IMAGE_FORMATAlejandro Piñeiro1-0/+48
From spec: "Possible values include any value that is legal to pass for the <format> parameter to the Tex*Image*D commands, or NONE if the resource is not supported for this operation." Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails: In some cases is returns GL_R11F_G11F_B10F or GL_RGB9_ES5, that are internalformats, not valid formats. Tested on AMD Radeon (TM) R9 380 Series: fails: It returns values different to NONE on cases where the target/internalformat are not supported (via INTERNALFORMAT_SUPPORTED). v2: rebased after changes in other commits Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: Check if the resource is supported in max-dimensionsAntia Puentes1-10/+51
The ARB_internalformat_query2 specification says about MAX_WIDTH, MAX_HEIGHT, MAX_DEPTH and MAX_COMBINED_DIMENSIONS <pnames>: "If the resource is unsupported, zero is returned." From the same specification: "In the following descriptions, the term /resource/ is used to generically refer to an object of the appropriate type that has been created with <internalformat> and <target>." We check if the /resource/ is supported by trying to create the object, if an error is raised, we consider the /resource/ as "unsupported". Before, we were only checking if the <internalformat> was supported independently of the <target>. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: after this commit, all the subtests on this test fails. Now that a stricter check for unsupported resources are in place, the test fails on NVIDIA because for some unsupported resources it returns a value different to zero. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: add MAX_COMBINED_DIMENSIONS subtest to max-dimensionsAlejandro Piñeiro3-0/+179
From spec: "MAX_COMBINED_DIMENSIONS: The maximum combined dimensions for the resource is returned in <params>. The combined dimensions is the product of the individual dimensions of the resource. For multisampled surfaces the number of samples is considered an additional dimension. Note that the value returned can be >= 2^32 and should be queried with the 64-bit query. <skip> If the resource is unsupported, zero is returned." In this case the returning value is a combination of the values of MAX_WIDTH, MAX_HEIGHT, MAX_DEPTH, SAMPLES and the number of faces, so the test can check that the returned value is correct in any case. Note that there are cases that the combined value is greater that 2^32, but the test is also testing the 32-bit query. The spec doesn't specify what the query should return in that case, so it is assumed that any value would be correct. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails for cube map related targets. From spec: "For cube map targets this is the maximum combined width, height and faces" But NVIDIA propietary drivers is just combining width and height. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: check returned max-dimensions values against ↵Alejandro Piñeiro1-6/+129
GetInteger In addition to the previous checks, this checks that the returned values when supported are the same that the ones you receive calling GetIntegerv with equivalent pnames like GL_MAX_TEXTURE_SIZE, GL_MAX_3D_TEXTURE_SIZE, etc. All those are internal format-independent, meanwhile GetInternalformat allows to specify the internal format. So in theory there is the possibility of being different for some internal format. But in practice, this is not happening on any driver at this moment. Query2 spec mentions this case: "7) There some <pnames> which it makes no sense to be qualified by a per-format/target scope, how should we handle them? e.g. MAX_WIDTH and MAX_HEIGHT might be the same for all formats. e.g. properties like AUTO_GENERATE_MIPMAP and MANUAL_GENERATE_MIPMAP might depend only on the GL version. <skip> A) Just use this entry point as is, if there are no per-format or target differences, it is perfectly acceptable to have the implementation return the same information for all valid parameters. This does allow implementations to report caveats that may exist for some formats but not others, even though all formats/targets may be supported." So at this point, taking into account the current implementation, it makes sense to check against those values. Probably in the future this check needs to be removed. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass Tested on 4.5.13399 on AMD Radeon (TM) R9 380 Series: fails: * For MAX_WIDTH: * For 3d textures is returning MAX_TEXTURE_SIZE * For texture buffers is returning crap. Different values for each internalformat, including negative values. * For MAX_HEIGHT: * Ditto for 3d textures. * For RENDERBUFFER and TEXTURE_2D_MULTISAMPLE_ARRAY is returning 0 even for supported values. Anyway, take into account that even the basic dimension test (0 for unsupported) was failing on ATI proprietary drivers. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: max dimensions related checksAlejandro Piñeiro3-1/+357
Add a check for the following pnames (a subtest for each one) * MAX_WIDTH. From spec: "The maximum supported width for the resource is returned in <params>. For resources with only one-dimension, this one dimension is considered the width. If the resource is unsupported, zero is returned." It is only tested that returns zero if not supported. It makes sense to include it on this test because it is related to the other max dimension pnames. * MAX_HEIGHT. From spec: "The maximum supported height for the resource is returned in <params>. For resources with two or more dimensions, the second dimension is considered the height. If the resource does not have at least two dimensions, or if the resource is unsupported, zero is returned." So in addition to the usual zero-test if not supported, it is tested that if the target has less that two dimensions, it returns zero too, even if supported. * MAX_DEPTH: From spec: "The maximum supported depth for the resource is returned in <params>. For resources with three or more dimensions, the third dimension is considered the depth. If the resource does not have at least three dimensions, or if the resource is unsupported, zero is returned. So in addition to the usual zero-test if not supported, it is tested that if the target has less that three dimensions, it returns zero too, even if supported. * MAX_LAYERS: From spec: "The maximum supported number of layers for the resource is returned in <params>. For 1D array targets, the value returned is the same as the MAX_HEIGHT. For 2D and cube array targets, the value returned is the same as the MAX_DEPTH. If the resource does not support layers, or if the resource is unsupported, zero is returned." In addition to the usual zero-test if not supported, it is tested that the value returned is the same that the one returned by MAX_HEIGHT or MAX_DEPTH for the array texture targets. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: passes. Tested on 4.5.13399 on AMD Radeon (TM) R9 380 Series: the tests doesn't pass for this ATI proprietary drivers for any of the subtests. It returns a non-zero value for unsupported combinations. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: IMAGE_FORMAT_COMPATIBILITY_TYPE testAlejandro Piñeiro3-0/+174
From spec: "IMAGE_FORMAT_COMPATIBILITY_TYPE: The matching criteria use for the resource when used as an image textures is returned in <params>. This is equivalent to calling GetTexParameter with <value> set to IMAGE_FORMAT_COMPATIBILITY_TYPE. Possible values are IMAGE_FORMAT_COMPATIBILITY_BY_SIZE or IMAGE_FORMAT_COMPATIBILITY_BY_CLASS. If the resource is not supported for image textures, or if image textures are not supported, NONE is returned." Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails for the same reasons it fails for INTERNALFORMAT_{X}_SIZE and INTERNALFORMAT_{X}_TYPE. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for INTERNALFORMAT_{X}_TYPE pnamesAlejandro Piñeiro3-0/+214
From spec: " INTERNALFORMAT_RED_TYPE INTERNALFORMAT_GREEN_TYPE INTERNALFORMAT_BLUE_TYPE INTERNALFORMAT_ALPHA_TYPE INTERNALFORMAT_DEPTH_TYPE INTERNALFORMAT_STENCIL_TYPE For uncompressed internal formats, queries for these values return the data type used to store the component. For compressed internal formats the types returned specify how components are interpreted after decompression. For textures this query returns the same information as querying GetTexLevelParameter{if}v for TEXTURE_*TYPE would return. Possible values return include, NONE, SIGNED_NORMALIZED, UNSIGNED_NORMALIZED, FLOAT, INT, UNSIGNED_INT, representing missing, signed normalized fixed point, unsigned normalized fixed point, floating-point, signed unnormalized integer and unsigned unnormalized integer components. NONE is returned for all component types if the format is unsupported." So this test calls GetInternalformat with INTERNALFORMAT_SUPPORTED: * If it is false, checks that the returned value is zero * If it is true, checks that all returned values are from that specific set of values * If it is true, checks that all the values are the same that GetTexLevelParameter Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails for the same reasons that for INTERNALFORMAT_{X}_SIZE tests. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for INTERNALFORMAT_{X}_SIZE pnamesAlejandro Piñeiro5-0/+438
From spec: " INTERNALFORMAT_RED_SIZE INTERNALFORMAT_GREEN_SIZE INTERNALFORMAT_BLUE_SIZE INTERNALFORMAT_ALPHA_SIZE INTERNALFORMAT_DEPTH_SIZE INTERNALFORMAT_STENCIL_SIZE INTERNALFORMAT_SHARED_SIZE For textures this query will return the same information as querying GetTexLevelParameter{if}v for TEXTURE_*_SIZE would return. If the internal format is unsupported, or if a particular component is not present in the format, 0 is written to <params>." So this test calls GetInternalformat with INTERNALFORMAT_SUPPORTED: * If it is false, it checks that the returned value is 0. * If it is true, for texture targets, checks that the returned value is the same that calling GetTexLevelParameter{if}v. In order to call GetTexLevelParameter, it requires to create a texture for the given internalformat/target combination. If that fails it is assumed that the current implementation doesn't support that combination, and it is tested against 0 too. Note that this means that the test would output some error messages even if the test passes. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: fails on the next cases: * Fails when trying to create a texture with some per-spec valid internalformat, like 1D/3D textures with GL_COMPRESSED_SIGNED_RED_RGTC1. Returns INVALID_ENUM, as if TexCompressedImage1D were called. * GL_TEXTURE_BUFFER has a limited ranged of internalformat supported (table 3.15 from spec 4.2). So for example, creating a texture with target GL_TEXTURE_BUFFER and internalformat GL_DEPTH_STENCIL would fail. On those cases the query should return 0, but it is not always the case. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for SAMPLES and NUM_SAMPLE_COUNTS pnamesAlejandro Piñeiro5-0/+397
For those pnames, it verify that the defined on the spec is fulfilled: NUM_SAMPLE_COUNTS <skip> If <internalformat> is not color-renderable, depth-renderable, or stencil-renderable (as defined in section 4.4.4), or if <target> does not support multiple samples (ie other than TEXTURE_2D_MULTISAMPLE, TEXTURE_2D_MULTISAMPLE_ARRAY, or RENDERBUFFER), 0 is returned. SAMPLES <skip> If <internalformat> is not color-renderable, depth-renderable, or stencil-renderable (as defined in section 4.4.4), or if <target> does not support multiple samples (ie other than TEXTURE_2D_MULTISAMPLE, TEXTURE_2D_MULTISAMPLE_ARRAY, or RENDERBUFFER), <params> is not modified. This test is really similar to api-errors test at arb_internalformat_query, with the difference that on that case an error was defined, and here was "softed" to return some specific values. It also covers query1 overrun test. Those two tests fails on the following proprietary drivers: * 4.5.13399 on AMD Radeon (TM) R9 380 Series * 4.5.0 NVIDIA 352.55 on GeForce GTX 950/PCIe/SSE2 For NUM_SAMPLES_COUNTS, they report a value different of 0 (5 on the NVIDIA, 4 on ATI) for target/internalformat that should return 0 (example: GL_TEXTURE_1D+GL_COMPRESSED_RGB). It is worth to note that for those cases, INTERNALFORMAT_SUPPORTED returns TRUE. It is not clear if they provide multi-sample capabilities for those targets. That seems to suggest that the paragraph was C&P from query1, and it is being too restrictive, when the purpose of the query2 itself is "a) provide a mechanism for implementations to declare support *above* the minimum required by the specification". For the SAMPLES case, it changes the first element of <params>, for cases where it should remain unmodified. Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: test for the most generic pname checksAlejandro Piñeiro5-1/+812
The spec includes conditions for all the pnames. On most cases, it is generic like this: "Possible values returned are <set>. If the resource is not supported, or if the operation is not supported, NONE is returned." So this test checks that for those pnames: * If it is not supported (using INTERNALFORMAT_SUPPORTED), the returned value is zero. * If it is supported, and there is available a list of possible values, the returned value is among one of those values. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass. v2: IMAGE_TEXEL_SIZE, IMAGE_COMPATIBILITY_CLASS, IMAGE_PIXEL_TYPE, IMAGE_PIXEL_FORMAT tested on a specific test (Antía) Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query2: query2 api-error checkAlejandro Piñeiro6-0/+464
Similar to the equivalent for arb_internalformat_query, but in addition to test when a invalid pname/target combination should return INVALID_ENUM, it also checks that a valid combination returns NO_ERROR, and testing both GetInternalformativ and GetInternalformati64v. The rationale of this is that is really likely that the implementation of arb_internal_format_query2 would reuse a lot of bits of arb_internal_format_query, so we want to be sure that a combination that was invalid with arb_internal_format_query is not considered invalid by arb_internal_format_query2. Tested on NVIDIA GeForce GTX 950 - NVIDIA 352.55: pass Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04arb_internalformat_query: remove references to arb_internalformat_query2Alejandro Piñeiro2-32/+6
arb_internalformat_query tests was taking into account query2, in order to avoid testing some pnames/internalformats valid on query2. But taking into account how deeply arb_internalformat_query2 extended and changed the behaviour of GetInternalformativ, it is really more clear to isolate both. With this commit arb_internalformat_query tests will test when only arb_internalformat_query is present, and arb_internalformat_query2 will test when both are present (as arb_internalformat_query2 has arb_internalformat_query as a requirement). This patch makes this one obsolete: http://lists.freedesktop.org/archives/piglit/2015-October/017746.html Acked-by: Dave Airlie <airlied@redhat.com>
2016-03-04shaders: add test for recursive variable indexing into arraysIago Toral Quiroga1-0/+45
v2: - Move to tests/spec/glsl-1.30 (Timothy) - Fix code style and rename some variables (Iago) Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
2016-03-04arb_enhanced_layouts: xfb_stride layout qualifier compile testsTimothy Arceri8-0/+200
2016-03-04arb_enhanced_layouts: xfb_buffer layout qualifier compile testsTimothy Arceri17-0/+496
2016-03-04arb_enhanced_layouts: xfb_offset layout qualifier compile testsTimothy Arceri16-0/+392
2016-03-03cl: add image attribute testZoltan Gilian1-0/+31
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: add image reading testsZoltan Gilian2-0/+151
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: add image writing testsZoltan Gilian1-0/+81
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: do not allow mismatching types for in-out argsZoltan Gilian1-3/+19
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: reword arg_out value errorZoltan Gilian1-1/+1
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: add image and sampler argument support to program-testerZoltan Gilian1-32/+574
Only 2D images are supported yet. Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: rename buffer_arg to mem_argZoltan Gilian1-47/+47
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-03-03cl: add image and sampler utility functionsZoltan Gilian2-0/+265
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
2016-02-29getteximage-formats: use new fbo_use_test_set() functionBrian Paul1-1/+6
So that we can pass arguments to specify a format set name and an option such as init-by-rendering. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94322 Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
2016-02-29fbo-formats: add new fbo_use_test_set() functionBrian Paul1-0/+18
As an alternative to fbo_formats_init() which takes a single string. fbo_formats_init() doesn't work for tests which may take both a format set name and some other arguments. Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
2016-02-29fbo-formats: refactor fbo_lookup_test_set() / fbo_formats_init()Brian Paul1-2/+9
Move the error message and exit() into the later. Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
2016-02-27arb_uniform_buffer_object: add new test exercising an array in a UBOBrian Paul3-0/+212
This test puts the pos, size, rotation and color info for four objects in an array which is stored in a UBO. For each drawing command, index into the array to get the object parameters. Signed-off-by: Brian Paul <brianp@vmware.com>
2016-02-27getteximage-formats: pass argc, not 1, to fbo_formats_init() callBrian Paul1-1/+1
Now we can specify the format group on the command line and have it actually work. Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
2016-02-27gl-1.0-dlist-bitmaps: add code to exercise deleting listsBrian Paul1-3/+43
Signed-off-by: Brian Paul <brianp@vmware.com>
2016-02-26arb_compute_shader: Add test for Mesa bug 93840Jordan Justen1-0/+39
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93840 Signed-off-by: Jordan Justen <jordan.l.justen@intel.com> Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
2016-02-25arb_gpu_shader_fp64: Explicit conversion from double to floatAndres Gomez7-33/+83
Double and float derived types also need explicit conversions. Signed-off-by: Andres Gomez <agomez@igalia.com> Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
2016-02-23CL: Add tests to make sure that ilogb(0.0 or nan) returns FP_ILOGB0|FP_ILOGBNANAaron Watry1-0/+31
Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Jan Vesely <jan.vesely@rutgers.edu>
2016-02-20compute: add an execution test for shared atomicsIlia Mirkin1-0/+36
This adds a very simple test of atomicAdd on a shared variable. Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> Tested-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Tested-by: Jordan Justen <jordan.l.justen@intel.com> Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
2016-02-18readpixels-oob: Add a glReadPixels out-of-bounds testNanley Chery3-0/+135
Test that glReadPixels for an area which reaches out of bounds behaves like a glReadPixels into a sub-rectangle for the valid area. This behavior reduces the number of corner cases associated with this function. v2 (Brian Paul): - Rename test and change path - Use larger window dimensions and compute big buffer dims at run-time - Use bool instead of GLboolean - Remove extra function arguments - Free allocated data - Exercise more clipping code v3: - Replace extra &= with = (Brian Paul) - printf only when an error is triggered Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=92193 Signed-off-by: Nanley Chery <nanley.g.chery@intel.com> Reviewed-by: Brian Paul <brianp@vmware.com>
2016-02-18arb_sample_shading: ensure number of samples is at least oneIlia Mirkin1-1/+1
From the ARB_sample_shading spec: "gl_NumSamples is the total number of samples in the framebuffer, or one if rendering to a non-multisample framebuffer" Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu> Reviewed-by: Neil Roberts <neil@linux.intel.com>
2016-02-16arb_compute_shader: Add test for dispatches with a zero sizeJordan Justen3-0/+84
The ARB_compute_shader spec says: "If the work group count in any dimension is zero, no work groups are dispatched." Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94100 Signed-off-by: Jordan Justen <jordan.l.justen@intel.com> Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
2016-02-16arb_compute_shader: Support ID tests with work group sizes of 0Jordan Justen1-3/+3
The ARB_compute_shader spec says: "If the work group count in any dimension is zero, no work groups are dispatched." Signed-off-by: Jordan Justen <jordan.l.justen@intel.com> Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-27 22:35:14 +0000