platform/upstream/Vulkan-Tools.git
8 years agolayers: Add string array parameter checking
Dustin Graves [Tue, 8 Mar 2016 21:42:59 +0000 (14:42 -0700)]
layers: Add string array parameter checking

The param_checker layer had previously peformed a NULL
check on the pointer to a string array, but had not
performed a NULL check for the individual strings in the
array.  This adds NULL checks for the individual strings.

Change-Id: Id4527f670086187e8cd1f146027bfdfd1e134b7b

8 years agolayers: Remove redundant/unused param checker code
Dustin Graves [Tue, 8 Mar 2016 20:34:49 +0000 (13:34 -0700)]
layers: Remove redundant/unused param checker code

Removes param checker sType and NULL check code that is now
 auto-generated.  Also remove empty functions and conditional
statements.

Change-Id: I8be8aab0bf3a8fa60219e1066a7d3d6c9d26ec05

8 years agolayers: LX430, Modfiy object_tracker to handle multiple instances
Mark Lobodzinski [Tue, 8 Mar 2016 22:10:00 +0000 (15:10 -0700)]
layers: LX430, Modfiy object_tracker to handle multiple instances

Upon DestroyInstance, OT indicated that all objects in all maps were
not properly destroyed. Made this reporting work on a per-instance basis.

Change-Id: I9b4eae4364a24597cb449610a31bf9b872c45225

8 years agolayers: GH83 Fix updates of copies descriptor sets in draw_state
Tobin Ehlis [Tue, 8 Mar 2016 23:31:06 +0000 (16:31 -0700)]
layers: GH83 Fix updates of copies descriptor sets in draw_state

When a copied descriptor update is valid, need to update pUpdateStructs
for the destination set. There are still some issues with how copied
descriptors are handled in draw_state (noted in TODO on line 3049) but
this fix addresses specific issue in GH83.

8 years agolayers: Fix layer documentation so that it can be auto-validated
Tobin Ehlis [Tue, 8 Mar 2016 23:14:52 +0000 (16:14 -0700)]
layers: Fix layer documentation so that it can be auto-validated

Validator currently requires each table entry to be on a single line. Fixing a couple of multi-line entries.

8 years agolayers: Count attachments as the same if they have overlapping memory.
Michael Lentine [Mon, 29 Feb 2016 20:22:59 +0000 (14:22 -0600)]
layers: Count attachments as the same if they have overlapping memory.

8 years agoloader: Fix issue passing wrong instance if layer wraps instances
Jon Ashburn [Tue, 8 Mar 2016 22:21:30 +0000 (15:21 -0700)]
loader: Fix issue passing wrong instance if layer wraps instances

Change-Id: Iafd0f7dc3e4a911b165d3df1dadddf4d45fbef10

8 years agolayers: LX405 Avoid some checks when rastrization is disabled.
Karl Schultz [Tue, 8 Mar 2016 17:30:21 +0000 (10:30 -0700)]
layers: LX405 Avoid some checks when rastrization is disabled.

When rasterization is disabled, some pipeline state checks can
be skipped since they would not be used, including sample
counts and viewport/scissor counts.
Add rasterization state to graphics pipelines in the
test suite because it is required.

Change-Id: I88f76038c2acc0abd0565f1655ceeed128557f77

8 years agocodegen: LX424 Generate code for WSI-specific funs in GIPA.
Karl Schultz [Tue, 8 Mar 2016 22:14:11 +0000 (15:14 -0700)]
codegen: LX424 Generate code for WSI-specific funs in GIPA.

For the unique_objects and object_tracker layers, generate
system types (xcb, xlib, wayland, mir) in GetInstanceProcAddr.
The previous code would only generate code for one display
system, even if more than one was selected by the CMake config.
The code to return these proc addresses is conditionally compiled
with the corresponding USE_PLATFORM #ifdef, and the CMake config
can go ahead and set multiple USE_PLATFORM defines.
The result is that GIPA resolves proc address requests for these
functions and allows them to be hooked properly when the layers
are loaded.

Change-Id: I2fa38847863f0a1187dd17d882ef9d94f546527f

8 years agoBump api_version to 1.0.5 in json files
Tony Barbour [Tue, 8 Mar 2016 21:50:03 +0000 (14:50 -0700)]
Bump api_version to 1.0.5 in json files

Change-Id: I71a03cd8043906074f6c3be98af0006e1aeaf5ee

8 years agolayers: Add func pointer NULL checks
Dustin Graves [Tue, 8 Mar 2016 00:52:14 +0000 (17:52 -0700)]
layers: Add func pointer NULL checks

Add code generated NULL checks for function pointers.  This enables
parameter checking for the VkAllocationCallbacks struct, with
param_checker additions for the vkDestroy functions that
receive VkAllocationCallbacks parameters.

Change-Id: I9910c52026aa584bb013ee5015b5cb04b2d6ddcd

8 years agolayers: GH41 Flag error if cmd buffer references destroyed framebuffer
Tobin Ehlis [Tue, 8 Mar 2016 18:19:34 +0000 (11:19 -0700)]
layers: GH41 Flag error if cmd buffer references destroyed framebuffer

Create framebuffer node and include set of cmd buffers that reference that FB.
At the time when RenderPass is bound to cmd buffer, or when secondary cmd buffer
is created with framebuffer inheritence, store a reference from FB to cmd buffer.

When FB destroyed, set any cmd buffers referencing it as INVALID and store FB ref.

At submit time, if CB is INVALID, check for any destroyed FBs and report them.

8 years agolayers: Fix android build of draw_state
Tobin Ehlis [Tue, 8 Mar 2016 17:51:34 +0000 (10:51 -0700)]
layers: Fix android build of draw_state

Need to static cast NULL to VkFence in FENCE_NODE constructor.

8 years agolayers: Fix android build from linux or Win32 system
Tobin Ehlis [Tue, 8 Mar 2016 17:39:10 +0000 (10:39 -0700)]
layers: Fix android build from linux or Win32 system

When we codegen layers for android build, cmd line is long which was causing us to not customize build based on "Android" option on cmd line.

Handling "Android" as special case currently so android can easily be built from Win32 or linux.
Need to clean this code up at some point to make it easier to build various targets from different dev system types.

8 years agoUse size_t types for output size_t.
Michael Lentine [Tue, 8 Mar 2016 16:34:05 +0000 (10:34 -0600)]
Use size_t types for output size_t.

8 years agobuild: Enable Xlib support to default ON on Linux builds
Jon Ashburn [Tue, 8 Mar 2016 17:04:06 +0000 (10:04 -0700)]
build: Enable Xlib support to default ON on Linux builds

Change-Id: I5a89e81caf31b6cc5a5f0668a97fb05e8e6462aa

8 years agoloader: gllvl #50, Add support for the WSI extension KHR_display
Jon Ashburn [Tue, 8 Mar 2016 16:30:30 +0000 (09:30 -0700)]
loader: gllvl #50, Add support for the WSI extension KHR_display

Change-Id: I8d5ecc7500beba3042a6b9147b2f3fd3055b3b1d

8 years agolayers: LX431, FENCE_NODE structures not initialized
Mark Lobodzinski [Mon, 7 Mar 2016 23:43:57 +0000 (16:43 -0700)]
layers: LX431, FENCE_NODE structures not initialized

The fenceMap structures are never initialized, resulting in indeterminate
behavior including bad draw_state validation messages from the smoke
test. Added a default constructor to FENCE_MAP.

Change-Id: I27e6b48e485bbb1e8774e2d5789e1e8d1b48020b

8 years agobuild: Fix CMake Xlib DisplayServer name.
Karl Schultz [Tue, 8 Mar 2016 15:07:13 +0000 (08:07 -0700)]
build: Fix CMake Xlib DisplayServer name.

Change-Id: Ia50157cd4f245971208a1baa28c823a62683dcd3

8 years agoloader: VulkanTools PR 13 Resolve ...XlibPresentationSupportKHR
Karl Schultz [Tue, 8 Mar 2016 14:55:27 +0000 (07:55 -0700)]
loader: VulkanTools PR 13 Resolve ...XlibPresentationSupportKHR

Change-Id: I1177e20b657ab00f41e9e4eb6f9ac997f7be459e

8 years agodemos: Change layer order in cube.c to put param_check sooner.
Karl Schultz [Tue, 8 Mar 2016 00:54:07 +0000 (17:54 -0700)]
demos: Change layer order in cube.c to put param_check sooner.

Also make the order match the standard meta layer.

Change-Id: Id250b9fe89b9d85d9e8ab6b803474f2c449fc6b1

8 years agoglslang / spirv-tools: get latest versions
GregF [Mon, 7 Mar 2016 23:53:42 +0000 (16:53 -0700)]
glslang / spirv-tools: get latest versions

8 years agodemos: Remove allocation callback code from tri demo
Tony Barbour [Mon, 7 Mar 2016 22:48:06 +0000 (15:48 -0700)]
demos: Remove allocation callback code from tri demo

From LoaderAndValidationLayers issue #84:
demos/tri.c implements VkAllocationCallbacks with aligned_alloc/_aligned_malloc, free/_aligned_free, and realloc.

I belive that is broken in many ways:

realloc does not return a pointer with the necessary alignment.
realloc(_aligned_malloc(...)) crashes.
_aligned_free(realloc(...)) crashes.
aligned_alloc requires that size must be a multiple of alignment, while Vulkan provides no such guarantee to these callbacks.
On POSIX-based implementations, aligned_alloc requires that alignment must be a multiple of sizeof(void *) as well as a power of two, while Vulkan merely requires it to be a power of two.
aligned_alloc(0), realloc(p, 0) might return non-NULL, while I think Vulkan requires NULL.
_aligned_malloc(0, a) is documented to invoke the invalid parameter handler (although I'm not sure it actually does in practice).
Implementing a valid allocator is quite non-trivial, so it may be best to just delete the allocator entirely from the demo.

(See also KhronosGroup/Vulkan-Docs#21 and KhronosGroup/Vulkan-Docs#103 for related discussions.)

Change-Id: I3ef3ba19c64127bd65a7dbe7b0b54de42e68ebb0

8 years agolayers: GHLVL88, Fix ValidateBarriers in draw_state
Mark Lobodzinski [Mon, 7 Mar 2016 23:11:47 +0000 (16:11 -0700)]
layers: GHLVL88, Fix ValidateBarriers in draw_state

Apps using QUEUE_FAMILY_IGNORED might get bad validation errors.

Change-Id: I3c8b3723701b8b24e2e9cd81a20f7192bcef15a7

8 years agolayers: GHLVL87, Fix cut/paste VerifyImageLayout call in draw_state
Mark Lobodzinski [Mon, 7 Mar 2016 22:45:49 +0000 (15:45 -0700)]
layers: GHLVL87, Fix cut/paste VerifyImageLayout call in draw_state

Change-Id: I4c34930338805ac964f0a4c70c2356977c1adabd

8 years agolayers: LX402, Invalid CmdBindPipeline validation error
Mark Lobodzinski [Mon, 7 Mar 2016 19:23:52 +0000 (12:23 -0700)]
layers: LX402, Invalid CmdBindPipeline validation error

The draw_state layer was emitting an error if this API was
called outside a renderPass, which is indeed valid.

Change-Id: I712b381e7ea9d9e66e091ba7325ab8d884f56edb

8 years agoheader: update to version 1.0.5
Jon Ashburn [Mon, 7 Mar 2016 20:01:35 +0000 (13:01 -0700)]
header: update to version 1.0.5

Change-Id: I2c7fdb2349d50d06b0a875e4da0c9ab50d2249de

8 years agomisc: Update to header version 1.0.4
Jon Ashburn [Mon, 7 Mar 2016 18:07:19 +0000 (11:07 -0700)]
misc: Update to header version 1.0.4

Change-Id: I61657dc28c7850075b787413955a6c880447a03d

8 years agolayers: Add param checks for CreateInstance/Device
Dustin Graves [Thu, 3 Mar 2016 21:17:08 +0000 (14:17 -0700)]
layers: Add param checks for CreateInstance/Device

Add parameter checks for vkCreateDevice and vkCreateInstance.
Includes a local fix for GitHub Vulkan-Docs issue #104 to
prevent false positives from being generated for the
pCreateInfo->pEnabledFeatures == NULL case.

Change-Id: I1f3642e7e18c8bfb32b0996b6e2e57a5f9d8e011

8 years agolayers: Add NULL/sType checks for struct members
Dustin Graves [Thu, 3 Mar 2016 01:23:29 +0000 (18:23 -0700)]
layers: Add NULL/sType checks for struct members

Add param_checker support for validating struct members.  Messages
are logged for the following conditions, based on conditions
specified in vk.xml:
 - A pointer is NULL and is not marked as optional in the XML
 - An array is NULL and is not marked as optional in the XML, unless
   its count is 0
 - An array count is 0 and is not marked as optional in the XML
 - A structure's sType value does not match the value specified
   in the XML

Addresses GL105, GL109, GH82

Change-Id: I7063fe2582b30fdfc0006fe945a0f9c84a2aa66a

8 years agolayers: Handle indirection in param_checker
Dustin Graves [Mon, 29 Feb 2016 20:35:07 +0000 (13:35 -0700)]
layers: Handle indirection in param_checker

Code was being generated for param_checker that looked like:
  validate_pointer(pAllocInfo);
  validate_array(pAllocInfo->count, array);
Where validate_array could dereference a NULL pointer.  The code
generator was updated to insert NULL checks for this case:
  validate_pointer(pAllocInfo);
  if (pAllocInfo != NULL) {
      validate_array(pAllocInfo->count, array);
  }

Change-Id: I1eda66530dbff2e95d604a20d122179be172bedf

8 years agolayers: Add vkFreeCommandBuffers param check
Dustin Graves [Fri, 26 Feb 2016 18:53:28 +0000 (11:53 -0700)]
layers: Add vkFreeCommandBuffers param check

Add NULL pointer checks for vkFreeCommandBuffers.

Change-Id: I7bef6a9d55fda6607b483c873485156b8c8148a1

8 years agolayers: LX413, Fix dynamicOffsets crash in draw_state
Mark Lobodzinski [Mon, 7 Mar 2016 17:39:59 +0000 (10:39 -0700)]
layers: LX413, Fix dynamicOffsets crash in draw_state

Properly handle CmdBindDescriptorSets being called multiple times
consecutively. DynamicOffsets being overwritten on subsequent calls,
resulting in a crash when the array was indexed out-of-bounds.

8 years agoFix assert for layouts.
Michael Lentine [Fri, 4 Mar 2016 00:30:28 +0000 (18:30 -0600)]
Fix assert for layouts.

Currently we assume that the layout must have been set for a subresource.
While this is true it may be set globally instead of on that specific
subresource directly causing the old assert to throw.

8 years agolayers: GL156 Ignore pQueueFamilyIndices if exclusive sharing mode is requested
Dominik Witczak [Wed, 2 Mar 2016 16:24:56 +0000 (17:24 +0100)]
layers: GL156 Ignore pQueueFamilyIndices if exclusive sharing mode is requested

8 years agoRegester event on creation and delete on deletion.
Michael Lentine [Fri, 4 Mar 2016 23:53:37 +0000 (17:53 -0600)]
Regester event on creation and delete on deletion.

8 years agolayers: Validate subpass index in CreateGraphicsPipelines
Chris Forbes [Thu, 3 Mar 2016 22:52:03 +0000 (11:52 +1300)]
layers: Validate subpass index in CreateGraphicsPipelines

Not validating this upfront meant that users got really weird errors
about mismatches between their FS outputs and color attachments. Now
they should hit the real problem first.

Fixes GH #75

V2: Run through git-clang-format.

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: Remove vestigial bits of old derivatives API from DS
Chris Forbes [Thu, 3 Mar 2016 22:19:52 +0000 (11:19 +1300)]
layers: Remove vestigial bits of old derivatives API from DS

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: Validate pipeline derivative requirements in DrawState
Chris Forbes [Thu, 3 Mar 2016 21:40:09 +0000 (10:40 +1300)]
layers: Validate pipeline derivative requirements in DrawState

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agoMerge pull request #80 from lentinem/mlentine-buildfix
tobine [Sat, 5 Mar 2016 01:33:45 +0000 (18:33 -0700)]
Merge pull request #80 from lentinem/mlentine-buildfix

Remove unused var.

8 years agoRemove unused var.
Michael Lentine [Fri, 4 Mar 2016 23:11:00 +0000 (17:11 -0600)]
Remove unused var.

8 years agolayers: LX423 Set string var with handle value before using.
Karl Schultz [Fri, 4 Mar 2016 22:20:27 +0000 (15:20 -0700)]
layers: LX423 Set string var with handle value before using.

Change-Id: I50e7a2778bd14ff4e0b8fc3f2fcb201faefce548

8 years agoadd generated files for smoke demo.
Karl Schultz [Fri, 4 Mar 2016 20:27:05 +0000 (13:27 -0700)]
add generated files for smoke demo.

Change-Id: I7169419113c74fc7e1e16c827f6553cf86abe989

8 years agovalidation: Promote validation warnings to errors
Tony Barbour [Wed, 2 Mar 2016 22:12:01 +0000 (15:12 -0700)]
validation: Promote validation warnings to errors

Change-Id: Ib527e646aabbd0aecbadf9071ca9e271a8729735

8 years agolayers: LX401, Handle VK_WHOLE_SIZE in flush/invalidateMappedMemoryRanges
Mark Lobodzinski [Thu, 3 Mar 2016 21:50:52 +0000 (14:50 -0700)]
layers: LX401, Handle VK_WHOLE_SIZE in flush/invalidateMappedMemoryRanges

Mem_tracker was ignoring this setting and emitting invalid error messages.

Change-Id: Id26a5e0bdf3f5c907356a384b9febbae1e885c27

8 years agolayers: LVLGH47 DS Include swapchain images in search
Karl Schultz [Fri, 4 Mar 2016 15:59:07 +0000 (08:59 -0700)]
layers: LVLGH47 DS Include swapchain images in search

Change-Id: I08941af0153c123d5f552bae8dc02023dc4f9965

8 years agoloader: gllavl#46 add check for device extensions
Jeremy Hayes [Thu, 3 Mar 2016 23:03:03 +0000 (16:03 -0700)]
loader: gllavl#46 add check for device extensions

Make GetInstanceProcAddr check layers for device extensions.

Change-Id: Ib23c5d42d11be415e9a6acd3b8427d0e474087a5

8 years agoMerge pull request #59 from chrisforbes/ds-more-fixes
tobine [Fri, 4 Mar 2016 13:59:59 +0000 (06:59 -0700)]
Merge pull request #59 from chrisforbes/ds-more-fixes

Miscellaneous fixes for DrawState layer

8 years agowinrtinstaller: sign uninstaller
David Pinedo [Fri, 4 Mar 2016 04:25:45 +0000 (21:25 -0700)]
winrtinstaller: sign uninstaller

8 years agoRemove generated files from source control
Tony Barbour [Thu, 3 Mar 2016 22:32:58 +0000 (15:32 -0700)]
Remove generated files from source control

8 years agolayers: GHPR43, Add tracking of total number of levels and layers for images
Chris Forbes [Fri, 26 Feb 2016 05:04:30 +0000 (18:04 +1300)]
layers: GHPR43, Add tracking of total number of levels and layers for images

8 years agobuild: LVLPR65, Fix android extensions support in unique_objects
Michael Lentine [Wed, 2 Mar 2016 23:28:55 +0000 (17:28 -0600)]
build: LVLPR65, Fix android extensions support in unique_objects

8 years agoMerge pull request #42 from chrisforbes/push-constants
tobine [Thu, 3 Mar 2016 20:00:01 +0000 (13:00 -0700)]
Merge pull request #42 from chrisforbes/push-constants

Validation of push constants

8 years agolayers: validate push constant ranges accessed by shader
Chris Forbes [Fri, 26 Feb 2016 03:55:34 +0000 (16:55 +1300)]
layers: validate push constant ranges accessed by shader

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: add more shader checker error codes
Chris Forbes [Fri, 26 Feb 2016 03:55:12 +0000 (16:55 +1300)]
layers: add more shader checker error codes

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: move get_shader_stage_id up a bit
Chris Forbes [Fri, 26 Feb 2016 03:54:29 +0000 (16:54 +1300)]
layers: move get_shader_stage_id up a bit

We're about to need this in another place too

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: extract helper for getting to block/struct type
Chris Forbes [Fri, 26 Feb 2016 03:53:51 +0000 (16:53 +1300)]
layers: extract helper for getting to block/struct type

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: Add push constant storage class to stringify support
Chris Forbes [Fri, 26 Feb 2016 01:52:55 +0000 (14:52 +1300)]
layers: Add push constant storage class to stringify support

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: Remove nonsense `sinterface` param from
Chris Forbes [Fri, 26 Feb 2016 01:51:25 +0000 (14:51 +1300)]
layers: Remove nonsense `sinterface` param from
collect_interface_by_descriptor_slot

This was only used to produce a misleading error message.

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: Add push constant validation to draw_state
Tobin Ehlis [Wed, 17 Feb 2016 17:35:18 +0000 (10:35 -0700)]
layers: Add push constant validation to draw_state

Validate that push constants don't exceed maxPushConstantSize at create or update time.
Also validate that size at creation time is >0 and a multiple of 4.

8 years agowinrtinstaller: Don't create a menu item for each version of RT installed
David Pinedo [Thu, 3 Mar 2016 19:36:24 +0000 (12:36 -0700)]
winrtinstaller: Don't create a menu item for each version of RT installed

8 years agowinrtinsaller: change comment
David Pinedo [Thu, 3 Mar 2016 18:57:41 +0000 (11:57 -0700)]
winrtinsaller: change comment

8 years agomisc: Remove lunarg_debug_marker extension
Jon Ashburn [Thu, 3 Mar 2016 19:03:58 +0000 (12:03 -0700)]
misc: Remove lunarg_debug_marker extension

This device extensions was not completely supported in any layer. It had partial
support (intercept) in draw_state  and device_limits. This extension is being
changed and has a formal proposal in Khronos as debug_marker_ext. Removing
the existing extension since it is deprecated; once the debug_marker_ext gets
registered by Khronos, layers might want to start using the ObjectTag/ObjectName
commands to supplement error report messages in debug_report_ext. Since
debug_report_ext might get changed as part of Khronos review, don't add it yet.

8 years agolayers: Allow use of QUEUE_FAMILY_IGNORED for BufferMemoryBarriers
Michael Lentine [Wed, 2 Mar 2016 20:35:44 +0000 (14:35 -0600)]
layers: Allow use of QUEUE_FAMILY_IGNORED for BufferMemoryBarriers

This fixed a bug in drawstate that produced invalid error messages
running apps that used FAMILY_IGNORED.

8 years agolayers: add missing error check
Jeremy Hayes [Thu, 3 Mar 2016 00:27:58 +0000 (17:27 -0700)]
layers: add missing error check

Change-Id: I18e36f8007d6555a152ac4e0632e2d38a03352be

8 years agoloader: Update documentation to remove editing comments.
Jon Ashburn [Thu, 3 Mar 2016 00:26:31 +0000 (17:26 -0700)]
loader: Update documentation to remove editing comments.
Also better describe layer requirements for layers which creat new dispatchable
objects.

8 years agolayers: fix imageLayout bug
Jeremy Hayes [Wed, 2 Mar 2016 23:48:17 +0000 (16:48 -0700)]
layers: fix imageLayout bug

8 years agolayers: Fix false positive on initialLayout in image layer
Tony Barbour [Wed, 2 Mar 2016 23:22:40 +0000 (16:22 -0700)]
layers: Fix false positive on initialLayout in image layer

8 years agoloader: ghlvl 33, ghlvl 34 Fix aliasing of VkPhysicalDevice
Jon Ashburn [Wed, 2 Mar 2016 02:51:07 +0000 (19:51 -0700)]
loader: ghlvl 33, ghlvl 34 Fix aliasing of VkPhysicalDevice

This fixes problems with not selecting the proper PhysicalDevice within
the loader if multiple PhysicalDevices are enumerated by the loader.
Fix involves wrapping the VkPhysicalDevice object in both trampoline code
and terminator code.

Also get rid of the ICD device_extension_cache. It is now always queried from
ICD whenever needed.

8 years agoLX #410 - Use ${PYTHON_EXECUTABLE} for PYTHON_CMD
Tony Barbour [Tue, 1 Mar 2016 23:40:39 +0000 (16:40 -0700)]
LX #410 - Use ${PYTHON_EXECUTABLE} for PYTHON_CMD

8 years agosmoke: use $<TARGET_FILE> instead of LOCATION prop
Tony Barbour [Tue, 1 Mar 2016 23:08:41 +0000 (16:08 -0700)]
smoke: use $<TARGET_FILE> instead of LOCATION prop

8 years agolayers: DS: populate features correctly
Chris Forbes [Wed, 2 Mar 2016 04:21:08 +0000 (17:21 +1300)]
layers: DS: populate features correctly

- Remove the redundant first fetching of the device features. This was
  always clobbered.
- Treat pEnabledFeatures==NULL as disabling all features, as per
  30.1 of the spec.

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: gllavl#78 validate base pipeline
Jeremy Hayes [Wed, 2 Mar 2016 00:51:54 +0000 (17:51 -0700)]
layers: gllavl#78 validate base pipeline

Change-Id: I08313295ac545652aca2060744b55274ed3cd4ca

8 years agolayers: gllavl#77 validate layout during clear
Jeremy Hayes [Tue, 1 Mar 2016 23:51:02 +0000 (16:51 -0700)]
layers: gllavl#77 validate layout during clear

Change-Id: I427cd1fcc6ecf115dbaf521387a909f117c13d07

8 years agolayers: gllavl#76 validate initialLayout
Jeremy Hayes [Tue, 1 Mar 2016 21:01:45 +0000 (14:01 -0700)]
layers: gllavl#76 validate initialLayout

Change-Id: Iede6dcf3bc83a5663b80b62c3239ef54bc74278a

8 years agolayers: Separate some variables
Chris Forbes [Tue, 1 Mar 2016 20:45:13 +0000 (09:45 +1300)]
layers: Separate some variables

Don't abuse maxInputAttachmentCount in the second loop which has nothing
to do with input attachments.

Signed-off-by: Chris Forbes <chrisforbes@google.com>
8 years agolayers: Update QueueFamilyIndex validation
Mark Lobodzinski [Tue, 1 Mar 2016 18:05:45 +0000 (11:05 -0700)]
layers: Update QueueFamilyIndex validation

Add handling of VK_QUEUE_FAMILY_IGNORED and allow VkImageMemoryBarrier types
inside render passes. Also plumbed functionName into ValidateBarriers.

8 years agoloader: refactor EnumerateDevice*Properties terminate/trampoline code
Jon Ashburn [Mon, 29 Feb 2016 20:00:51 +0000 (13:00 -0700)]
loader: refactor EnumerateDevice*Properties terminate/trampoline code

Should be no functional change.
Restructure so the terminator functions don't contain trampoline code.

8 years agolayers: GH LVL#49, Updated layer docs/README for new layers
Mark Lobodzinski [Mon, 29 Feb 2016 18:49:51 +0000 (11:49 -0700)]
layers: GH LVL#49, Updated layer docs/README for new layers

8 years agolayers: GL107, Validate queueFamilyIndices for CreateSwapchain
Mark Lobodzinski [Thu, 25 Feb 2016 20:56:57 +0000 (13:56 -0700)]
layers: GL107, Validate queueFamilyIndices for CreateSwapchain

8 years agolayers: GL207, Validate queue_family_index
Mark Lobodzinski [Thu, 25 Feb 2016 22:09:52 +0000 (15:09 -0700)]
layers: GL207, Validate queue_family_index

Validates index is less than number of queue families for these APIs:
CmdWaitEvents, CmdPipelineBarrier, CreateBuffer, and CreateImage.
Also:  Updated layer details doc with additional error code.
       Added layer validation test for CreateBuffer

Conflicts:
layers/draw_state.h
layers/vk_validation_layer_details.md

8 years agolayers: LX322, Track semaphore state acroos QueueBindSparse calls
Mark Lobodzinski [Thu, 25 Feb 2016 17:54:31 +0000 (10:54 -0700)]
layers: LX322, Track semaphore state acroos QueueBindSparse calls

8 years agobuild: GH PR17, Remove unused displayserver ENV settings
Mark Lobodzinski [Mon, 29 Feb 2016 15:35:38 +0000 (08:35 -0700)]
build: GH PR17, Remove unused displayserver ENV settings

8 years agobuild: GH PR17, Add Android WSI info to Windows version of codegen script
Cody Northrop [Fri, 26 Feb 2016 23:49:33 +0000 (16:49 -0700)]
build: GH PR17, Add Android WSI info to Windows version of codegen script

8 years agolayers: GH PR17, Add support for building for Android on Windows
Cody Northrop [Wed, 24 Feb 2016 19:28:41 +0000 (12:28 -0700)]
layers: GH PR17, Add support for building for Android on Windows

A better fix would be to stop using sys.platform when making
decisions, but this is quick.

8 years agocmake: GH PR17, Improve error detection for displayserver selection
Mark Lobodzinski [Fri, 26 Feb 2016 01:14:56 +0000 (18:14 -0700)]
cmake: GH PR17, Improve error detection for displayserver selection

8 years agoloader: GH PR17, Add windows support for new displayserver types
Mark Lobodzinski [Wed, 24 Feb 2016 22:04:11 +0000 (15:04 -0700)]
loader: GH PR17, Add windows support for new displayserver types

8 years agocmake: GH PR17, Fix vulkan.py code generation script for Windows builds
Mark Lobodzinski [Fri, 26 Feb 2016 21:29:48 +0000 (14:29 -0700)]
cmake: GH PR17, Fix vulkan.py code generation script for Windows builds

8 years agocmake: GH PR17, Add DisplayServer variables for the WSI interfaces
Mun, Gwan-gyeong [Mon, 22 Feb 2016 11:32:17 +0000 (20:32 +0900)]
cmake: GH PR17, Add DisplayServer variables for the WSI interfaces

8 years agoBuild: GH PR17, Add Android WSI info to android codegen script
Mun, Gwan-gyeong [Mon, 22 Feb 2016 11:29:23 +0000 (20:29 +0900)]
Build: GH PR17, Add Android WSI info to android codegen script

Conflicts:
buildAndroid/android-generate.sh

8 years agocmake: GH PR17, Add WSI options to loader codegen script
Mun, Gwan-gyeong [Mon, 22 Feb 2016 11:27:31 +0000 (20:27 +0900)]
cmake: GH PR17, Add WSI options to loader codegen script

8 years agolayers: GH PR17, Add WSI argument passing to vk-generate.py
Mun, Gwan-gyeong [Mon, 22 Feb 2016 11:23:59 +0000 (20:23 +0900)]
layers: GH PR17, Add WSI argument passing to vk-generate.py

8 years agocmake: GH PR17, Pass WSI arg to layer generating scripts
Mun, Gwan-gyeong [Mon, 22 Feb 2016 02:02:58 +0000 (11:02 +0900)]
cmake: GH PR17, Pass WSI arg to layer generating scripts

8 years agolayers: GH PR17, Add Wayland, Xlib, Android & Mir extensions to codegen scripts
Mun, Gwan-gyeong [Mon, 22 Feb 2016 00:43:09 +0000 (09:43 +0900)]
layers: GH PR17, Add Wayland, Xlib, Android & Mir extensions to codegen scripts

Conflicts:
vulkan.py

8 years agoloader: GH PR17, Fix typo in comment
Mun, Gwan-gyeong [Mon, 22 Feb 2016 00:33:58 +0000 (09:33 +0900)]
loader: GH PR17, Fix typo in comment

8 years agodocs: Add CONTRIBUTING.md
Karl Schultz [Fri, 26 Feb 2016 22:42:57 +0000 (15:42 -0700)]
docs: Add CONTRIBUTING.md

8 years agodemos: Add Hologram snapshot as Smoke test/demo
Tony Barbour [Thu, 25 Feb 2016 22:44:10 +0000 (15:44 -0700)]
demos: Add Hologram snapshot as Smoke test/demo

8 years agoloader: Better delineate trampoline and terminator functions
Jon Ashburn [Fri, 26 Feb 2016 20:14:27 +0000 (13:14 -0700)]
loader: Better delineate trampoline and terminator functions

No functional changes!

Trampoline functions  have the Vulkan API function name and are in trampoline.c.
Terminator functions have terminator_<Vulkan API name> (omitting the "vk").
Instance extension trampoline and terminator functions are in together in the
same file for the extension.

8 years agolayers: Fix compliation error in draw_state
Jon Ashburn [Fri, 26 Feb 2016 17:55:14 +0000 (10:55 -0700)]
layers: Fix compliation error in draw_state

8 years agoandroid: Drop layer validation tests for now
Cody Northrop [Wed, 24 Feb 2016 21:44:39 +0000 (14:44 -0700)]
android: Drop layer validation tests for now