vulkan: Changes to 'refs/tags/vulkan-1.0.3.1+dfsg1-1'
Tag 'vulkan-1.0.3.1+dfsg1-1' created by Timo Aaltonen <tjaalton@debian.org> at 2016-03-10 12:29 +0000
Tagging upload of vulkan 1.0.3.1+dfsg1-1 to unstable.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=aQcm
-----END PGP SIGNATURE-----
Changes since the dawn of time:
Adam Jackson (5):
cmake: Don't check for installed glm, it's in-tree
cmake: Specify -std=c++11
nulldrv: Link with -Bsymbolic so vkGetDeviceProcAddress behaves
loader: Fix ICD and layer directory scan order on Linux
linux: Use standard GNU install paths
Andrzej Kotlowski (1):
loader: GL120, move critical section to cover loader_get_icd_and_device
Antoine Labour (5):
vkjson: initial commit
vkjson: add missing include
vkjson: handle uint64_t as hex strings
Make cJSON case-sensitive
Relicense vkjson from 3-BSD to MIT
BogDan Vatra (7):
Use primary_node when render_node is not available.
Remove autogenerated file from git.
Remove unused dependency.
Fix file permissions
Fix crash: pApplicationInfo is not mandatory
loader: MR111, Fix memleaks
Intel: MR112, Fix export vkDestroyPipelineCache
Cass Everitt (1):
check that cube array views are complete cubes
Chia-I Wu (760):
xglinfo: initial commit
add a terse xglDbg.h
xgl-generate: add a script for code generation
loader: add a generic ICD loader
xgl-generate: move _generate_icd_dispatch_table to Subcommand
xgl-generate: add is_dispatchable()
xgl-generate: improve constness
xglinfo: link against libXGL
cmake: set -std=c99 -Wall
xgl-generate: add pretty-dummy subcommand
main: silence warnings
cmake: find libdrm with pkg-config
xgl-generate: add _generate_icd_dispatch_entrypoints to Subcommand
loader: let cmake generate the source file
rename drivers/ to icd/
loader: ignore __icd{Load,Unload}
cmake: -Werror -Wextra -Wno...
common: add common code for all drivers
intel: use the common code
intel: remove xgl{Load,Unload}()
add and deal with -fvisibility=hidden everywhere
intel: add dummy dispatch tables
intel: properly set up the dispatch tables
loader: fix out-of-tree builds
loader: refactor loader_icd_scan()
intel: add a base header that every file can include
intel: add genhw
intel: add intel_gpu
intel: use intel_gpu for XGL_PHYSICAL_GPU
header: add XGL_DBG_OBJECT_TYPE
icd: assign each allocator a unique id
intel: add intel_{base,obj,state}
intel: add intel_winsys
intel: adapt intel_winsys for our needs
intel: add intel_dev and intel_queue
intel: use intel_dev for device creation and heap queries
xglinfo: fix device queue support
xglinfo: reindent
icd: fix driver reinit without alloc cb
intel: fix udev_device leaks
intel: fix init with maxGpus equal to 0
loader: short cut when init with maxGpus equal to 0
xglinfo: free all driver resources
intel: move intelGetGpuInfo and intelGetExtensionSupport
icd: add a note about thread safety
intel: implement all queue functions except intelQueueSubmit
intel: do not advertise pinning
intel: implement memory functions
intel: implement generic object functions
intel: split intel_queue to its own files
intel: let intel_queue_create return an error code
intel: implement fence functions
intel: comment on semaphores
intel: implement event functions
intel: add get_info() to intel_base
intel: make intel_base_dbg_create() more useful
intel: fix copying of create info
intel: fix intel_base_dbg creation
intel: add generic object creation function
intel: more comments on missing functions
intel: no need to include winsys.h in event.c
intel: add more intel_mem operations
intel: add align()
intel: implement query functions
intel: refactor query result processing
intel: hook up query functions for real
intel: implement intelGetFormatInfo()
intel: minor cleanup for intel_format_get_color_features
intel: add intel_dev_log()
intel: add more utility functions
intel: add tex layout code
intel: make tex layout code compilable
intel: implement image functions
intel: add layout code from Mesa
intel: make Mesa layout code compile
intel: update intel_img to use intel_layout
intel: add back size check against intel_max_resource_size
intel: intel_img separate stencil and aux support
intel: minor header cleanup
icd: add common utility functions
icd: add common format functions
intel: add some format helpers
intel: forgot to set create_info_size
intel: add intel_gpu as a param of format functions
intel: save some image states missing in intel_layout
intel: add minimal implementation of view functions
intel: add view state emission code from Mesa
intel: update intel_layout from Mesa
icd: add more utils for working with floats
intel: implement intelCreateSampler()
intel: implement descriptor set functions
intel: fix copying of dset create info
intel: switch on dbg type instead of struct type
intel: add minimal implementation of dynamic states
intel: sampler_init() instead of emit_sampler()
intel: fill in dynamic hardware states
intel: implement debug functions
intel: move cmd stubs to cmd.c
intel: implement command buffer create/begin/end/reset
intel: move stubs to the respective source files
intel: use intel_cmd instead of intel_bo for syncs
intel: implement intelQueueSubmit()
intel: fix queue creation
intel: minor cleanups
intel: rename dispatch_tables to dispatch
intel: use an accessor for the dispatch tables
intel: rename init_driver.c to intel.c
intel: avoid "emit" in function names
intel: allow cmd building to record an error
intel: remember intel_mem allocation size
intel: move private command buffer functions to cmd_priv.h
intel: add functions to ease command buffer IO
intel: export batch_buffer_reloc_count
intel: add command buffer relocs
intel: add command buffer bind points
intel: emit 3DSTATE_INDEX_BUFFER directly on GEN7.5+
intel: call intel_pipeline()
intel: emit 3DPRIMITIVE
intel: use width/height from intel_layout
intel: hide intel_cmd_reloc
intel: extend GEN_RENDER_CMD for different GENs
intel: emit 3DSTATE_DEPTH_BUFFER and etc.
intel: use intel_pipeline to emit 3DSTATE_VF and etc.
intel: move device dbg variables to intel_dev_dbg
intel: use XGL_GPU_SIZE for GPU sizes
intel: introduce intel_cmd_writer
intel: fix intel_cmd_end()
intel: remove unused generic intel_cmd_writer functions
intel: add a state buffer to intel_cmd
intel: improve batch/state writer helpers
intel: indicate cmd_len in intel_msaa_state
intel: emit intel_msaa_state into the batch buffer
inte: emit CC states
intel: allow blend and ds states to be NULL
intel: fix intel_viewport_state
intel: emit viewport states
intel: add INTEL_DEBUG env var to help debugging
intel: add INTEL_DEBUG=nohw to disable bo submission
intel: add intel_rmap for shader resource mapping
intel: create VS and FS resource mappings
intel: export the length of SURFACE_STATE
render_test: no descriptor set nor dynamic view mapping
intel: emit RTs and BINDING_TABLE_STATE
intel: use cmd_batch_reserve_reloc() when relocs are added
intel: add a command writer for kernel buffers
intel: refactor cmd_pipeline
intel: emit 3DSTATE_CLEAR_PARAMS
intel: emit 3DSTATE_DRAWING_RECTANGLE
intel: replace intel_queue_submit() by queue_submit_bo()
intel: add a queue bo to initialize the hw context
intel: intel_bo instead of intel_mem in intel_cmd_reloc
intel: add a note about cmd->reloc_count
intel: use absolute pos in cmd_writer_add_reloc()
intel: emit STATE_BASE_ADDRESS
intel: make PIPELINE_SELECT a queue state
intel: emit depth buffer workaround
intel: clean up queue init
intel: add a command buffer scratch bo
intel: rework command buffer workarounds
intel: update winsys
intel: update genhw
intel: declare intel_dev in layout.h
intel: clear intel_cmd_bind on reset
intel: record number of draws in intel_cmd_bind
intel: partial intelCmdPrepare*() to flush caches
intel: emit 3DSTATE_CONSTANT_*
tests: fix memory access with uninitialized values
render_tests: use XGL_NUM_FMT_UNORM
render_tests: write the result to TriangleTest.ppm
intel: need only one WA flag so far
intel: fix and clean up shader cache
icd: rename and prefix shader_il.h
intel: clean up intel_shader
render_tests: set BIL generator magic as a hack
intel: add some intel_ir info
intel: remove dummy pipeline_get_info()
intel: refactor intelCreateGraphicsPipeline()
intel: pCode is not a intel_shader
intel: remove some unused fields from intel_pipeline
intel: reject other GENS when INTEL_GEN_SPECIALIZED is defined
intel: export GT in intel_gpu
intel: emit 3DSTATE_URB_*
icd: remove generated icd-dispatch-table.h
intel: fix assertion failures on Haswell
intel: remove PCB and binding table pointer setup from PSO
intel: allow INTEL_DEBUG to override device ID
intel: remove unused members in intel_pipe_shader
intel: gen7_3DSTATE_GS is GEN7+ only
intel: give WAs some sanity
intel: document the commands in PSO
intel: set unused states to zero
intel: emit 3DSTATE_VERTEX_ELEMENT
intel: reserve space for PSO
intel: add sample_count to intel_msaa_state
intel: export more shader info
intel: emit 3DSTATE_WM and 3DSTATE_PS
intel: add 3DSTATE_VERTEX_ELEMENTS to INTEL_PSO_CMD_ENTRIES
intel: vs output count should be 2
intel: emit 3DSTATE_SF and 3DSTATE_SBE
intel: export viewport count in intel_viewport_state
intel: emit 3DSTATE_CLIP
intel: correct non-MSAA sample position
intel: honor image tiling request
intel: fix CLIP_VIEWPORT
intel: include obj.h in mem.h
intel: careful with the flags in cmd_batch_flush()
intel: implement intelCmd*Query()
intel: fix pipeline stats querying
render_tests: add functions to create query pools
render_tests: query and test XGL_PIPELINE_STATISTICS_DATA
intel: set urb_grf_start for our magic shaders
render_tests: no culling
render_tests: use built-in kernels
intel: fix max entries of 3DSTATE_URB_VS on Haswell GT2+
intel: limit INTEL_RELOC_GGTT to GEN6
intel: implement intelGetMultiGpuCompatibility()
intel: extend gen6_PIPE_CONTROL()
intel: add helpers for GEN6_PIPE_CONTROL_WRITE_*
intel: rename cmd_query.c to cmd_mi.c
intel: implement intelCmd*{Event,Timestamp}()
intel: remove more unused fields from intel_pipeline
add a top-level .gitignore
update copyright information
intel: remove unused pipeline defines
intel: rename intel_pipe_shader to intel_pipeline_shader
intel: rename pipeline_rmap.c to pipeline_shader.c
Revert "intel: rename pipeline_rmap.c to pipeline_shader.c"
Revert "intel: rename intel_pipe_shader to intel_pipeline_shader"
intel: rename pipeline builder to pipeline create info
intel: remove gpu from intel_pipeline_create_info
intel: cs is a part of XGL_COMPUTE_PIPELINE_CREATE_INFO
intel: move shader building/tearing to pipeline_shader.c
intel: make all rmap functions static
intel: clean up pipeline shader building a bit
intel: intel_pipeline should not have XGL_PIPELINE_SHADER
intel: rename intel_pipe_shader to intel_pipeline_shader
intel: shorten shader stage names
intel: rename rmap to pipeline rmap
intel: really rename pipeline_rmap.c to pipeline_shader.c
intel: remove urb_read_length and max_*_threads
intel: improve XGL_PIPELINE_IA_STATE_CREATE_INFO handling
intel: make cmd writers an array
intel: rename ptr_opaque to ptr
intel: rework cmd buffer management
intel: rework cmd helpers
icd: add u_uif
intel: improve command decoding
intel: upload viewport states separately
remove XGLConfig.h.in
xgl-generate: do not hardcode #include's
icd: add an API to scan Linux DRM subsystem
intel: use the common DRM subsystem scanner
intel: prepare for primary node support
intel: teach layout about scanout
intel: add intel_obj_bind_mem()
intel: set tiling before exporting
delete GitSHA1.c generation
tidy up the build system a bit
intel: remove intel_fence_get_status()
include: add preliminary XGL_WSI_X11 proposal
add stubs for XGL_WSI_X11
intel: add support for XGL_WSI_X11
intel: do not require render node
intel: check for xcb too
intel: emit samplers and image/memory views
intel: disable HiZ
intel: disable PCB
icd: add icd_format_get_channel_count
intel: fix untyped memory views on GEN6
include: update xgl.h with FF vertex fetch
intel: add support for FF vertex fetch
kmd: fix intel_bo_wait() on kernel 3.17
intel: fix for a semantic change in libdrm 2.4.57 and later
intel: emit SAMPLER_STATE for all stages
intel: emit BINDING_TABLE_STATE for all stages
intel: fix devid override
intel: fix 8x sample pattern
intel: handle XGL_LAST_MIP_OR_SLICE
intel: fix color/depth attachments size mismatch
intel: emit everything at draw time
intel: clean up cmd shader cache
icd: add icd_format_is_equal()
intel: allow a writer to be a reloc target
intel: define meta ops
intel: convert resource operations to meta ops
intel: implement cmd_draw_meta()
intel: fix an assertion in create info copying
intel: allow depthWriteEnable to be independently enabled
intel: use helper funtions in gen6_meta_depth_buffer()
intel: create intel_ds_state for depth/stencil clears
intel: disable PS for depth/stencil clears
intel: improve blob decoding
intel: fix clears with mipLevels > 1
intel: add a missing WA in the meta path
intel: always emit DEPTH_STENCIL_STATE for meta
intel: add INTEL_DEBUG=nocache
intel: make intel_ir opaque
intel: remove unused INTEL_EXT_COMPILE_GLSL
intel: use intel_gpu to set up the compiler context
render_tests: fix constant buffer setup
demos: add tri
demos/tri: allocate textures
demos/tri: allocate vertex buffer
demos/tri: allocate depth buffer
demos/tri: enable FF vertex fetch
intel/compiler: add brw_blorp_blit_eu from Mesa
intel/compiler: add intel_pipeline_shader_compile_meta()
intel: use the compiler to compile meta shaders
tests: allow image tiling to be specified
tests: add basic blit (copy, clear, and resolve) tests
demos/tri: fix the shaders
demos/tri: use xglCmdClear*
demos/tri: make it prettier
intel: add intel_disassemble_kernel()
wsi_x11: add xcb_window_t to xglWsiX11GetMSC()
intel: register a special XGE for each X window
demos: add a fence for presentation
update to the latest vertex fetch proposal
intel: define INTEL_MAX_*_COUNT for vertex fetch
intel: check presentable image format
build: remove icd-dispatch-table.h generation from loader
build: do not pollute source tree with icd-dispatch-table.h
build: fix in-tree build
intel: fix INTEL_DEBUG=nocache for the first draw
intel: DATA_TRANSFER may imply rendering/sampling
tests: add CopyMemoryHazard to XglBlitTest
intel: set INTEL_RELOC_WRITE for meta dst
intel: set INTEL_RELOC_WRITE for descriptor set entries
intel: define meta modes
intel: bump up max thread count for meta
intel: bump up VS URB entry count for meta
compiler: add intel_create_brw_context()
compiler: add support for Byte/DWord Scattered Read/Write
compiler: clean up intel_meta_compiler
compiler: add VS meta shaders for memory filling/copying
intel: use VS meta shaders for memory filling/copying
tests: add {Fill,Copy}MemoryLarge to XglBlitTest
compiler: allocate more temps in the meta compiler
compiler: add support for non-multisample image copying
compiler: add support for INTEL_DEV_META_FS_COPY_MEM_TO_IMG
Revert "blit_test: Remove incomplete test"
tests: add CopyMemoryToImageBasic to XglBlitTest
intel: update genhw headers
intel: fix copying to compressed images
intel: limit libdrm_intel VMA caches
intel: honor XGL_PIPELINE_CB_STATE
intel: detect xglEndCommandBuffer without xglBeginCommandBuffer
tests: merge common.[ch] back to xglinfo.c
tests: remove unused functions from xglinfo
tests: move xglinfo to demos
tests: fix build errors due to ImageMagick dependency
intel: honor XGL_CHANNEL_MAPPING on Gen7.5+
intel: revive intel_layout_get_slice_tile_offset()
intel: factor out meta compressed image adjustment
intel: fix meta for compressed array images
intel: add support for MRT
tests: fix depth image preparation
tests: remove unused ClearRenderBuffer()
tests: remove redundant m_render_target_format/m_cb_attachment_state
compiler: re-enable assert(execsize >= width) in validate_reg()
compiler: no unaligned memory copy on Gen6
tests: untar golden_images.tar.gz
tests: add MRT support
tests: add XglRenderTest.TriangleMRT
compiler: fix leaks in meta compilation
intel: fix a leak in intel_dset_destroy()
intel: fix a use-after-free in intel_dev_destroy()
intel: add helpers to enable parallel state emissions
intel: fix emit_samplers() when the cmd is near full
intel: fix bound checking for memory blit ops
tests: rewrite blit tests
intel, tests: fix -Wreorder warnings
intel, tests: fix -Wswitch warnings
intel, tests, glave: fix -Wunused-variable warnings
intel: fix a -Wunused-value warning
tests: fix -Wunused-but-set-variable warnings
intel: set -Wno-{unknown-pragmas,ignored-qualifiers} for glass files
cmake: add -Wall, -Wextra, and others to CMAKE_CXX_FLAGS
layers: fix -Wswitch warnings
layers: fix -Wunused-function warnings
layers: fix -Wunused-variable warnings
layers: fix -Wpointer-sign warnings
layers: do not override CMAKE_C_FLAGS
layers: fix parallel building
cmake: remove -Werror from CMAKE_C_FLAGS
icd: add u_ffs()
intel: compute max_threads in intel_pipeline_shader
intel: request scratch space from PSO
intel: program scratch space in command buffers
tests: improve blit check pattern
icd: add icd_format_get_raw_value()
intel: force image format when raw copying/clearing
tests: no format filtering in XglCmdCopy{,MemoryTo}ImageTest
compiler: add VS meta shaders for copying images to memories
intel: implement xglCmdCopyImageToMemory()
tests: add XglCmdCopyImageToMemoryTest.Basic
tests: allow tiled images in blit tests
tests: make XglCmdClearColorImageTest reusable for raw clears
tests: add XglCmdClearColorImageRawTest.Basic
intel: use memory copy for intelCmdCloneImageData()
tests: add XglCmdCloneImageDataTest.Basic
include: fix typedefs
remove unnecessary casting between XGL_CHAR/char
intel: fix intel_normal_dispatch_table
icd: make probe_syspath() static
xgl-generate.py: remove unused IcdDispatchTableSubcommand
xgl-generate.py: remove unused LayerFuncsSubcommand
xgl-generate.py: remove unused LayerDispatchSubcommand
xgl.py: add Extension
xgl.py: use lists instead of tuples
xgl.py: remove icd_dispatch_table
xgl-generate.py: fork _generate_dispatch_entrypoints()
tests: add a new XGL C++ binding
tests: use the new binding in blit tests
tests: remove unused debug.h
tests: remove unused displayengine.{cpp,h}
tests: remove base_render_test.cpp
tests: derive XglDevice from xgl_testing::Device
tests: remove dependencies on xglgpu.h
tests: remove unused xglgpu.{cpp,h}
tests: derive XglImage from xgl_testing::Image
tests: derive XglCommandBufferObj from xgl_testing::CmdBuffer
tests: store XGL_GPU_MEMORYs in XglMemoryRefManager
tests: derive XglConstantBufferObj from xgl_testing::GpuMemory
tests: fix resource types in XglShaderObj
tests: derive XglTextureObj from xgl_testing::Image
tests: derive XglSamplerObj from xgl_testing::Sampler
tests: derive XglDescriptorSetObj from xgl_testing::DescriptorSet
tests: derive XglShaderObj from xgl_testing::Shader
tests: derive XglPipelineObj from xgl_testing::Pipeline
tests: fold xglimage.{cpp,h} into xglrenderframework.{cpp,h}
tests: remove unused XglDevice::AllocAndBindGpuMemory()
tests: remove XglDevice::CreateImage()
tests: call {Init,Shutdown}Framework from SetUp/TearDwon
tests: skip XglDevice in init.cpp and image_tests.cpp
tests: fold xgldevice.{cpp,h} into xglrenderframework.{cpp,h}
tests: do not enable layers by default
demos/cube: set data size for xglGetImageSubresourceInfo()
layers: no explicit enum values for internal enums
xgl-layer-generate.py: change obj_type_mapping formatting
glave-generate.py: change remap types and list formating
icd: clean up alloc functions
icd: clean up logging functions
icd: remove icd.c
icd: stop generating icd-dispatch-entrypoints.c
intel: remove unused internal dispatch tables
icd: build a static library
intel: add dummy intelEnumerateLayers()
xgl-generate.py: add IcdDummyEntrypointsSubcommand
xgl-generate.py: add IcdGetProcAddrSubcommand
intel: generate xglGetProcAddr()
icd: no need to include xglLayer.h
xgl.py: add c_pretty_decl() to Proto
xgl-generate.py: clean up IcdDummyEntrypointsSubcommand
xgl-generate.py: rename LoaderSubcommand
xgl-generate.py: add DispatchTableOpsSubcommand
loader: use generated dispatch table ops
xgl.py: add names of objects
xgl.py: add the ability to parse xgl.h
xgl.py: update with parsed core extension
xgl.py: add more methods to Param and Proto
xgl.py: remove special-purpose functions
xgl-generate.py: simplify _is_dispatchable()
xgl-generate.py: simplify _does_function_create_object()
layers: simplify initLayerTable()
layers: simplify xglGetProcAddr() for Basic and Multi
xgl-generate.py: add generate_get_proc_addr_check()
xgl-generate.py: add LayerGetProcAddrHelperSubcommand()
layers: intercept xglCmdBindVertexData() in DrawState and MemTracker
layers: use generated layer_intercept_proc()
intel: work around raw/structured buffers
intel: pass XGL_PIPELINE_SHADER to intel_pipeline_shader_compile()
intel: move intel_pipeline_rmap creation to the compiler
intel: remove pipeline_*.[ch]
layers: xgl-helper.py depends on xgl.h
xgl.py: fix prototype paring
cmake: add find modules
cmake: use find_package() for XCB
cmake: use find_package() for libdrm
icd: clean up build rules
intel: clean up build rules
intel: separate out ossource.cpp
xgl.py: missing xgl prefix in XGL_LAYER_DISPATCH_TABLE
intel: fix undefined references
cmake: fix up gtest and set -fvisibility=hidden globally
intel: add -Wl,--exclude-libs,ALL
intel: add -Wmissing-declarations
intel: constify intel_base_magic
intel: constify intel_gpu_exts
update for buffer object changes
intel: fix DEPTH_STENCIL_STATE on Gen6
complete fixed-function vertex fetch support
intel: initialize raw buffer SURFACE_STATEs on view creation
demos, tests: use XGL_BUFFER_VIEW_RAW for constant buffers
xgl: move to new resource binding model
xgl: fix constness of pointers
intel: refactor cmd_batch_begin()
intel: BINDING_TABLE_STATEs are in surface state bo
intel: allocate a separate surface state bo
intel: honor 64KB limit of BINDING_TABLE_STATEs
intel: do not emit 3DSTATE_SBE on Gen6
intel: set up dummy drawing rect when unused in meta
intel: double PCB alloc on GT3
intel: use the same PCB alloc setup for normal and meta paths
intel: set up mappable and total aperture sizes
intel: always emit "post" WAs
intel: add cmd_fail() to signal command buffer errors
intel: do not submit invalid command buffers
intel: add intel_winsys_read_reset_stats()
intel: add INTEL_DEBUG=hang
icd: add U_ASSERT_ONLY
intel: fix -Wunused-(but-set-)variable warnings
intel: fix -Wmaybe-uninitialized warnings
intel: fix mulitple descriptor sets/regions
intel: fix stencil refs
intel: fix stencil refs for stencil clears
intel: fix the param type of img_type_to_view_type()
intel: fix a compile error with clang
cmake: support clang as the compiler
icd: add likely()/unlikely()
intel: rework xglQueueSubmit()
intel: rework command buffer waiting
intel: silence a compiler warning
loader: do not set RTLD_DEEPBIND
intel: rename intel_framebuffer to intel_fb
intel: rename intel_rp_from_obj() to intel_render_pass_from_obj()
intel: move xglCmd{Begin,End}RenderPass to cmd_pipeline.c
intel: do not cast away constness in intel_cmd_begin()
intel: mostly cosmetic changes to fb.[ch]
Revert "intel: Be sure to set width & height for depth-only"
intel: minor improvements to fb.c
intel: rename cmd_prepare.c to cmd_barrier.c
intel: format API functions as declared in xgl.h
intel: ICD_EXPORT all API functions
intel: remove intel_format_is_stencil()
intel: add intel_format_has_depth()
intel: avoid unnecessary stencil and hiz relocs
intel: make sense of gen6_meta_DEPTH_STENCIL_STATE()
intel: add cmd_meta_ds_op()
intel: enable HiZ on Gen7+
demos/cube: use a single layout
intel: remove libdrm dependency from libintelcompiler.a
intel: import stripped-down libdrm-2.4.59-23-g1f73578
intel: add and link with libintelkmd.a
cmake: remove unused FindDRM.cmake
intel: add intel_bo_map_async()
intel: use intel_bo_map_async() for xglMapMemory()
intel: use unlimited VMA cache
update BUILD.md for imported libdrm
intel: fix waiting of present fences
intel: move xglEnumerateLayers() to gpu.c
intel: rename intel_gpa.c to gpa.c
intel: rename intel.c to instance.c
intel: allow GPU re-enumeration
intel: an XGL_INSTANCE is not an XGL_OBJECT
intel: add support for per-instance GPU enumeration
intel: add intel_handle
intel: enable L3 cache
demos/cube: use a single layout
intel: make intel_base_create() take a intel_handle
icd: add generic icd_instance
intel: make intel_instance own an icd_instance
icd: add support for per-instance DRM enumeration
intel: add intel_{alloc,free,logv,log}()
intel: use intel_alloc()/intel_free() when possible
intel: use intel_alloc()/intel_free() everywhere
nulldrv: remove icd-alloc.h dependency
icd: remove unused icd-alloc.[ch]
icd: remove stale function declarations
nulldrv: fix gcc errors/warnings
nulldrv: build it on Linux
intel: update genhw headers
intel: add intel_pipeline_init_default_sample_patterns()
intel: determine 3DSTATE_MULTISAMPLE from intel_fb
intel: emit FB-dependent states conditionally
intel: emit 3DSTATE_PUSH_CONSTANT_ALLOC_x in intel_cmd_begin()
intel: update kmd
intel: add support for xglPinSystemMemory()
intel: call intel_bo_set_tiling() on presentable images
intel: update layout code from ilo
intel: honor programPointSize
intel: honor pointOrigin
intel: honor depthMode
intel: fix culling for XGL_COORDINATE_ORIGIN_LOWER_LEFT
intel: convert pixels to blocks in meta
intel: remove an invalid assert
intel: fix XGL_TOPOLOGY_POINT_LIST
tests: fix environment tear-down without X11 setup
intel: overhaul the internal WSI interface
intel: xglQueueWaitIdle() should wait for presents
intel: replace intel_wsi_x11_window by intel_wsi_swap_chain
intel: move code around in wsi_x11.c
intel: clean up wsi_x11_dri3_pixmap_from_buffer()
intel: clean up wsi_x11_create()
intel: restore original drm_intel_gem_bo_map()
intel: update to libdrm-2.4.60-30-gd20314d
intel: add xf86drmMode.[ch] back to our stripped-down libdrm
intel: add x11_display_scan() to scan displays
demos: remove non-__linux__ paths
demos: set EXTERNAL_SPV from CMakeLists.txt
demo: remove XCB_NVIDIA paths
xgl-generate.py: add WinDefFileSubcommand
xgl-generate.py: add and use helpers from loader.h
xgl-generate.py: rewrite LoaderEntrypointsSubcommand
demos/tri: fix a leak
icd: remove unused icd-log.[ch]
binding: disable layers and tools
binding: rename descriptor region to descriptor pool
binding: update XGL_DESCRIPTOR_SET_LAYOUT_CREATE_INFO
binding: rename xglClearDescriptorPool()
binding: remove XGL_DESCRIPTOR_TYPE_RAW_BUFFER*
binding: introduce XGL_DESCRIPTOR_SET_LAYOUT_CHAIN
binding: update for pImmutableSamplers
binding: update for xglCmdBindDescriptorSets()
intel: plug a serious command buffer leak
intel: save layout chain in intel_pipeline
vulkan: Remove VkDescriptorSetLayoutChain from vkCmdBindDescriptorSets
loader: add a missing VKAPI
demos/tri: remove outdated code
update vk_wsi_lunarg.h ...
switch to VK_WSI_LunarG
intel: fix wrong use of xcb_request_check()
demos/tri: fix facing and enable culling
demos/cube: use GL conventions
intel: only VK_COORDINATE_ORIGIN_UPPER_LEFT is supported
intel: support only VK_DEPTH_MODE_ZERO_TO_ONE
v98: rename count to arraySize in VkDescriptorSetLayoutBinding
v98: remove vk{Begin,End}DescriptorPoolUpdate()
v98: replace vkUpdateDescriptors() by vkUpdateDescriptorSets()
update vulkan.h for multi-level command buffers (V121)
intel: add support for vkCmdExecuteCommands()
demos: remove an extraneous vkCmdBeginRenderPass()
demos: destroy VkRenderPass and VkFramebuffer with VkDevice
demos: fix color attachment store op
intel: get depth/stencil format from VkRenderPassCreateInfo
intel: get depth/stencil layout from VkRenderPassCreateInfo
intel: replace intel_{rt,ds}_view with intel_att_view
intel: add intel_render_pass_{attachment,subpass}
update vulkan.h for multi-pass command buffers (V131, #14075)
tests/binding: add Handle and NonDispHandle
test/binding: inherit PhysicalGpu from Handle
test/binding: inherit Device from Handle
test/binding: inherit Queue from Handle
test/binding: inherit GpuMemory from NonDispHandle
test/binding: inherit Fence from NonDispHandle
test/binding: inherit Semaphore from NonDispHandle
test/binding: inherit Event from NonDispHandle
test/binding: inherit QueryPool from NonDispHandle
test/binding: inherit views from NonDispHandle
test/binding: inherit Shader* from NonDispHandle
test/binding: inherit PipelineLayout from NonDispHandle
test/binding: inherit Pipeline from NonDispHandle
test/binding: inherit Sampler from NonDispHandle
test/binding: inherit Descriptor* from NonDispHandle
test/binding: inherit Dynamic* from NonDispHandle
test/binding: inherit CmdBuffer from Handle
test/binding: inherit Buffer and Image from NonDispHandle
intel: silence -Wmissing-declarations warnings
intel: fix -Wdiscarded-qualifiers warnings
vulkan.py: fix vulkan.h parsing
vktrace_generate.py: support array parameters
vulkan.py: use generated "core" API
bug 14844: Query Pool issues (WIP)
bug 14365: Rename slopeScaledDepthBias to depthBiasSlopeScaled (WIP)
bug 14861: rasterization review issues (WIP)
bug 14855: vkGetPipelineCacheSize is unique (consistency issue) (WIP)
bug 14462: endianness of code for shader modules (WIP)
bug 14725: Remove or replace VK_SPARSE_MEMORY_BIND_REPLICATE_64KIB_BLOCK_BIT (WIP)
bug 14822: timestampFrequency underspecified (WIP)
v184: bug 14719: vkCmdCopyBufferToImage alignment requirements (WIP)
bug 14868: Replace VkTimestampType with VkPipelineStageFlagBits (WIP)
support and use VkSubmitInfo semaphore signaling and waiting
MR 461: Documented updated vkQueueSubmit and remove wait + signal semaphore operations on queues. (WIP)
bug 14864: Remove redundant VkPipelineShaderStageCreateInfo.stage field? (WIP)
v185: bump API minor version
bug 14362: VK_NULL_HANDLE does not work in C (WIP)
bug 14876: Can timestamps be written inside a renderpass? (WIP)
bug 14567: What is VK_QUEUE_EXTENDED_BIT? (WIP)
v186: bump API minor version
bug 14857: Adding extensibility to VkSubmitInfo (WIP)
bug 14630: Issues with regard to primitive restart index (WIP)
v187: bump API minor version
bug 14905: Reserve VkStructureType values for loader/layer bootstrapping (WIP)
bug 14909: Define "length" as it relates to push constants be in units of bytes not words (WIP)
bug 14892: 'count' parameters and members named inconsistently in many cases (WIP)
bug 12921: Memory callback (WIP)
bug 14914: Should strictLines be a feature or a limit (WIP)
v188: bump API minor version
bug 14543: rename VkTex* enums to VkTexture* similarly for VK_TEX_* to VK_TEXTURE_* (WIP)
v189: bump API minor version
bug 14848: Verifying Sparse Behavior (WIP)
vktrace_generate.py: fix vkCreateShader tracing
bug 14641: VkXXXCreateFlags are not consistently used for the various VkXXXCreateInfo (WIP)
v190: bump API minor version
v191: bug 14574: Capability bit for stores from non-fragment/compute shaders (WIP)
bug 14918 part1: consistency/cleanup ideas (from jbolz/pdaniell/dkoch) (WIP)
v192: bump API minor version
bug 14778: VK_IMAGE_VIEW_CREATE_READ_ONLY_DEPTH_BIT underspecified (WIP)
bug 14892 part 2: rename more "count" (WIP)
bug 14918 part 2: some simple renames (WIP)
v193: bump API minor version
MR 542: Remaining items from the cleanup bug (WIP)
v194: bump API minor version
v195: bug 14932: Pipeline Cache clarifications needed (WIP)
bug 14921 part 1: sed renames
v196: bug 14921 part 2: Expanding abbreviations for consistency (WIP)
bug 14570: Which formats are supported for image loads and stores? (WIP)
bug 14936: Sparse questions/issues (WIP)
bug 14922: Consistent memory properties (WIP)
v197: bump API minor version
bug 14940: Capitalization of Abbreviations/Acryonyms (WIP)
bug 14968: Limitations of transfer operations on DMA queues (WIP)
bug 14921 part 3: typos
bug 14918 part 3: sed renames
bug 14918 part 4: more cleanups (WIP)
v198: bump API minor version
bug 14965: Merge VK_MEMORY_INPUT/OUTPUT flags to better express write-after-wite hazards (WIP)
bug 15000: aspect/aspectMask inconsistencies (WIP)
v199: bug 15018: Name all anonymous structs and enums (WIP)
bug 14887: Documenting Valid Behavior in the Spec (WIP)
bug 12921 part 2: Memory callback (WIP)
bug 14608: VkShaderModule object lifetime (WIP)
v200: bump API minor version
bug 14987: Document dependency of Vulkan API spec on SPIR-V spec (WIP)
v201: bump API minor version
bug 14936 part 2: Sparse questions/issues (WIP)
bug 15022: Need a way to specify dummy descriptors for unused bindings (WIP)
bug 14537: vkGetPhysicalDeviceImageFormatProperties doesn't express all limitations (WIP)
bug 14861 part 2: rasterization review issues (WIP)
bug 15003: vkCmdPipelineBarrier should be allowed on a DMA queue (WIP)
bug 15040: Add device limit for maximum number of samplers (and other objects?) (WIP)
v202: bump API minor version
bug 15045: vkGetPhysicalDeviceImageFormatProperties for unsupported formats (WIP)
v203: bump API minor version
bug 15068: Use hardfp calling convention on Android for 32-bit ARM targets
bug 15050: Add device limit for max number of input attachments
v204: bump API minor version
v205: bump API minor version
bug 15038: Need a way to express the total limit of resources per stage
bug 15004: Non-coherent memory, when is flush and invalidate required
bug 15085: queue creation naming issues
bug 15086: maxDrawIndirectInstanceCount?
intel: make sure intel_cmd_dset_data is allocated
bug 12998 part 1: append _BLOCK to compressed formats
bug 12998 part 2: append _PACK* to packed formats that change conventions
bug 12998 part 3: append _PACK* to remaining packed formats
bug 12998 part 4: drop explicit enum values temporarily
bug 12998 part 5: reorder formats
bug 12998 part 6: add missing formats to icd and demos
bug 12998 part 7: drop unwanted USCALED formats
bug 12998 part 8: add new formats
bug 12998 part 9: add back explicit enum values
MR632: Valid usage xml global changes
MR640: Alloc -> Allocate naming fixes
v207: bump API minor version
bug 15089: Does "Across Command Buffer Boundaries" apply to second-level command buffers?
v208: bump API minor version
MR671: Rename VK_STRUCTURE_TYPE_LAYER_* to ..LOADER_*
bug 15120: rename vkCmdPushConstants "values" to "pValues"
v209: bump API minor version
Chris Forbes (116):
shader_checker: initial plumbing
shader_checker: capture spir-v for every shader at vkCreateShader time
shader_checker: collect interface definitions from shaders
shader_checker: add hook for xglCreateGraphicsPipeline
shader_checker: collect pipeline stages in vkCreateGraphicsPipeline
shader_checker: add validation of interface between vs and fs
shader_checker: validate vertex attribs against vs inputs
shader_checker: validate fs outputs against cb
shader_checker: first pass at typechecks
shader_checker: add plumbing for debug
shader_checker: convert printf to debug output
layers: remove assumption that there's a glslang source tree alongside
intel: Check viewport/scissor count against proper limit
intel: Fill out the rest of VkPhysicalDeviceProperties
shader_checker: fix fs output/cb attachment walking after assert fixes
shader_checker: add helper to walk a type tree and return the basic type
shader_checker: add type checking between VF and VS
shader_checker: check types between FS output and CB
shader_checker: Add support for delivering layer messages via callback
shader_checker: current object & locking
layer_validation_tests: Enable ShaderChecker layer
tests: allow run_all_tests to be run from other directory
tests: ensure Object::bound is initialized
tests: Add ScopedUseSpv helper to remove some duplicated boilerplate
tests: add test for vs output not consumed
tests: add test for fs input not written by previous stage
tests: add test for type mismatch between vs and fs
shader_checker: Only continue with pipeline config if no errors
shader_checker: validate VS inputs against VI even when no VI
tests: Allow failures in pipeline creation
tests: Add test for a VI attrib not consumed by the VS
tests: Add test for a VS input not provided by a VI attrib
tests: Add test for VI attrib type not matching VS input
tests: Add test for FS broadcast with non-float attachment
tests: Add test for CB attachment not written by FS
tests: Add test for FS output with no matching CB attachment
tests: Add test for FS output type mismatch with CB attachment
shader_checker: remove stray debug output from typechecker implementation
intel: fix (benign) use of uninitialized stack variable
tests: Fixup conditional compilation around shader_checker
Fix a bunch of mismatched new[]/delete.
shader_checker: Extract pipeline validation into own function
shader_checker: Also validate vkCreateGraphicsPipelineDerivative
tests: Add test to ensure ShaderChecker doesnt interfere with non-spirv
shader_checker: Add typechecker support for extra array level
shader_checker: Support pipelines with more stages, and be tolerant
shader_checker: fix invalid iterator dereference (LunarXchange #60)
shader_checker: Check for duplicate VI binding descriptions.
tests: Add test for render target clear via load op
intel: implement color clear load ops
tests: Dont try to begin or end null renderpasses
cube: Use render pass load ops for color clear
vulkan.h: Remove vkGetFormatInfo, add features/limits (#12827, v125)
intel: Add shader module object support
tests: only destroy instance if creation succeeded
vulkan.h: Update features/limits structures (#12827, v125)
vulkan.h: Split attachment and image clears (#13914, v126)
intel: Add support for depth/stencil clears via render pass load op
cube: Use render pass load op for depth clear
tri: Use render pass load ops for color and depth clears
intel: Implement attachment clears
vulkan.h: v129 -- rework clear colors, remove raw clear [#14020]
intel: Fill out some device limits
layers: Update ShaderChecker to use new logging mechanism
layers: Update ShaderChecker for 14084
layers: Fixup and reenable fs output validation in ShaderChecker
vulkan.h: API changes for #13744 - sharing across queues
mem_tracker: Provide ctor for embedded union (and name it so we can)
layers: Adjust SC to validate bulk pipeline creation
layers: Shadow descriptor set layout in ShaderChecker
layers: Collect descriptor usage from SPIRV images
layers: Require pipeline layout to contain all referenced descriptors
icd: clear correct miplevel of depth/stencil attachment
icd: do depth/stencil attachment clears together
icd: fixup renderpass load ops for less waste and correct stencil clears
layers: Make passing a non-SPIRV shader to CreateShaderModule an error
tests: Remove nonsense ScopedUseGlsl use in test that requires GLSL
tests: remove remaining ScopedUseGlsl use and support
intel: Copy dynamic state use flags into the pipeline state
intel: Fix application of PSO-defined scissor state
layers/SC: Delete disabled FS broadcast checking code
tests: remove comment about FS broadcast rules
layers: Use spirv.hpp from colocated glslang tree
layers/SC: fix a bogus leftover comment in descriptor validation
tests: Add a test for matrix vertex attrib types with ShaderChecker
layers/SC: consider types larger than vec4 when collecting interfaces
layers/SC: LX187: compute size of array and matrix types
layers/SC: Fix location counting for stages which take arrayed input.
layers: MR128, Revert "layers: Work around Windows release build crash in cube"
layers: MR128, Fix mismatched new[]/delete in validate_pipeline_shaders
layers: MR128, Don't complain about declarations without Builtin or Location
layers: MR130, Handle VK_WHOLE_SIZE properly in mem_tracker:initializeAndTrackMemory
nulldrv: Initialize memory types mask
nulldrv: Claim to support any WSI surface.
nulldrv: Expose some basic formats as presentable
nulldrv: Expose some present modes
nulldrv: Claim that our supported image formats can be used as color attachments
nulldrv: Add minimal support for pipeline cache objects.
nulldrv: Add more proper support for ShaderModule objects
nulldrv: Initialize image format properties
layers: Fix mishandling of VI vs VS validation (Gitlab#69)
layers: Add minimal actual analysis of interface blocks to DrawState
layers: MR140, Remove some commented merge remnants
layers: MR140, Pass whole shader_module to build_type_def_index
layers: MR140, Add SPIRV instruction stream iterator helper
layers: MR140, Add get_type_def helper in DrawState
nulldrv: fill out physical device memory info
tests: Only check memory types that actually exist
nulldrv: Set memory type mask for buffers.
layers: Only set up tracking for new descriptors if allocation succeeded.
layers: MR159, Rename get_type_def, type_def_index
layers: MR159, Collect offsets of instructions producing constants
layers: MR159/LX309/LX310/LX304, Fix handling of OpTypeArray size in DrawState
layers: MR159, Add helper to get the value of an assumed integral constant
tests: MR159, Add a test for an array size mismatch between VS and FS
layers: MR191, Some cmd buffer tracking structures are being left uninitialized
Cody Northrop (247):
compiler: Bring over glsl compiler from alchemist-10.2.6+steamos-LunarG-1.1
compiler: Bring over select required files from alchemist-10.2.6+steamos-LunarG-1.1
compiler: Import Mesa compiler components
compiler: enable SSO for standalone shader compile and link
compiler: Turn on enough to invoke front end during XGL execution
compiler: Get enough turned on to generate ISA
compiler: Progagate compile results to xgl, don't use them yet
compiler: Turn off SIMD16 for bring up
compiler: Disable standalone build for now
compiler: C linkage for icd alloc and free
compiler: Start using fs compile results
compiler: Blue triangle
compiler: Start adding new files for VS
compiler: Start using VS backend results
compiler: Turn on URB handling and interstage I/O
compiler: Print resulting data structs
compiler: Turn off push constants for fragment shader
compiler: Turn on program limit initialization
compiler: Swap FS and VS ordering in pipeline compiler interface
compiler: Don't use samplers for uniform loads yet
compiler: Print out more push/pull param info
triangle: Hook up a uniform read in fragment shader
compiler: Start dumping vertex shader data
compiler: Use pipeline shader rather than info shader for state update
compiler: Changes in the direction of textures and samplers
render_test: Start turning on sampler and texture setup
render_test: Populate the texture more correctly, render purple triangle
Undo changes to driver that mirror what Olv cooked up
render_test: refactor image/sampler setup, render textured triangle
render_test: Restore use of %
triangle: Hook up a uniform read in fragment shader
render_tests: Turn on FS with two uniforms
render_test: Render a triangle using compiler and vertex fetch
render_test: Turn all tests back on, easier to debug with just one
compiler: Turn off push constants for vec4 shaders
compiler: Propagate VS binding table size
render_test: Start turning on simplified VS uniform test
compiler: Turn off uniform loads via sampler for vec4 for now
render_test: Clean up VS uniform test, draw rotated triangle
compiler: Clean up front end changes a bit
compiler: Remove some (not nearly all) dead files from frontend
compiler: Clean up backend some more
compiler: Turn off all the compiler spew by default
compiler_render_tests: Fork the render_tests file for quick compiler test turn on
compiler_render_tests: Turn on the compiler_render_tests build
compiler_render_tests: Turn on textured triangle test, with just t0 and s0
compiler: Print out link errors from CreateShader
compiler: Bump up some const values to allow UBO
render_test: Bring over constant buffer stride fix from base_render_test
compiler_render_tests: Turn on VS and FS uniform block tests
compiler_render_tests: Clean up single texture test a bit
compiler: Set some better defaults for gl_context constants
compiler: Add clean copy of uniforms.c
compiler: Remove most of the contents of uniforms.c and add to build
compiler: Allow sampler unit bindings to propagate in backend
compiler: Allow sampler index to be looked up, although bindings not present yet
compiler_render_tests: Add a simple VS texturing test
compiler: Create an API enum for XGL, add checks where appropriate
compiler: Propagate buffer bindings in glassy translator
compiler: If XGL is being used, assign buffer bindings directly from the shader
compiler_render_tests: Add test that pulls from uniform buffer based on binding
compiler: Propagate sampler bindings through glassy translator
compiler: Allow texture unit bindings to also set sampler index
compiler_render_tests: Turn on test that uses sampler bindings and multiple textures
compiler: Use metadata qualifier to determine IO status
Add LunarG copyright to files with substantative changes
Clean up: convert compiler_render_tests to all use BIL, remove redundant tests
test_framework: Print glslang log on parse or link failure
Reply to: