Age | Commit message (Collapse) | Author | Files | Lines |
|
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
|
|
You can't put #extension directly in C source so we need to make a magic
qonos macro to allow declaring extensions.
|
|
|
|
The requirement of the Vulkan API is that the semaphore signal/wait
operations come in nicely interleaved pairs. There is no requirement
that the semaphore must be idle when you submit it, however. Let's be a
bit tougher on the implementation.
|
|
|
|
|
|
|
|
|
|
Most tests don't perform a transition from the UNDEFINED layout. Make this
happen automatically by changing the default attachment parameters.
|
|
|
|
There isn't a single type index that works for all mmapable things.
Instead, we need to actually look at the requirements.
|
|
|
|
|
|
|
|
|
|
|
|
Vulkan doesn't allow you to use features you haven't enabled. The easy
solution is to just enable everything.
|
|
|
|
We were getting the ranges wrong.
|
|
This test demonstrates a bug in Mesa f4821daed1ec185d16f9ee3cc0951d306ce6e2b9.
When arrayed depth buffers are cleared, only the first layer returns the clear
value.
|
|
|
|
Fixes the following validation error on func.first (which allocates two sets),
DS(ERROR): object: 0x1 type: 22 location: 1509 msgCode: 31: Unable to allocate
2 descriptorSets from pool 0x1. This pool only has 1 descriptorSets
remaining.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
|
|
|
|
|
|
Clients are required to set a scissor and this test wasn't.
|
|
|
|
The layout doesn't mention the push constants, but the test
goes ahead and uses push constants.
|
|
It's generated by miptree_gen.py
|
|
|
|
|
|
|
|
|
|
This test demonstrates a bug in mesa 29f53d793781b67a92bb95fe66d7d38adc5488bb ,
in which Anvil fails an ISL assertion that the maximum buffer range is less
than a certain size.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
Commit 17685f385bad0f7e79772e2e5990e2e2572b6897 changed the uniform buffer
test's reference image although the test logic would have generated the
same image if it didn't have the bugs that commit introduced. Restore the
original image.
Since storage test's output is equivalent to that of the uniform test,
remove its custom image and reference the uniform buffer test's image.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
The uniform buffer test is fixed by updating the right DescriptorSet binding.
The storage test failed, at least, because of the following bugs:
* It assigned uint array indices float color values
* It provided 1 dynamic offset instead of 3
* It did not specify the dynamic variant of
VkDescriptorSetLayoutBinding::descriptorType
This patch removes these bugs by simply making the storage buffer test
parallel to that of the uniform buffer test.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
The vertex shader has only one vertex input attribute.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
This is required by the Vulkan Specification and silences the
validation layers.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
The buffer has two usages in each test variant. Create it with usage flags equal
to the union of both usage sets. This field is required to be non-zero.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
This test isolates a bug in mesa 0f7a6ea5e7b95cfe10dd5c176858ca078b36a197 ,
where the driver segfaults when binding partly defined dynamic descriptor
sets. This bug was discovered through func.desc.dynamic.storage-buffer.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
Because the variable use_dynamic_offsets is true in this block,
the resource descriptor to be updated is actually a dynamic
uniform buffer.
Fixes:
* stress.lots-of-surface-state.vs.dynamic
* stress.lots-of-surface-state.fs.dynamic
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
I used the Vulkan Validation layers to determine the areas of
non-conformance.
Changes:
* Provide required fields:
- VkFramebufferCreateInfo::renderPass
- VkSubmitInfo::sType
- VkDeviceQueueCreateInfo::sType
- VkDeviceQueueCreateInfo::pQueuePriorities
* Assign the right bit to VkDescriptorPoolCreateInfo::flags for the
freeing which occurs during cleanup.
* Use the correct VkDescriptorPoolCreateInfo::poolSizeCount value.
This fixes a bug in DescriptorPool creation in which DescriptorTypes
after STORAGE_TEXEL_BUFFER were not being added to the pool.
* Set VkGraphicsPipelineCreateInfo::pDynamicState to NULL if
VkPipelineDynamicStateCreateInfo::dynamicStateCount is zero
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
|
|
This causes tests that attempt to create linear depth buffers on
Gen7+ (unsupported), to be skipped.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
|
|
|
|
This allows us to easily reproduce the dota2 flicker bug
|
|
|
|
This test needs big descriptor sets. Use a custom pool.
|
|
|
|
|
|
The reference image for func.depthstencil.basic-depth.clear-0.5.op-greater-equal
was wrong. It was produced by an old Mesa in which depth clears did not
work yet. That Mesa "cleared" all depth buffers to 0.0 regardless of the
actual clear value.
|