This is enough to get all of the sample apps running on a conformant ES 3.1 implementation, such as ANGLE/Vk or NVidia's OpenGL ES Linux driver.
Implements a new opengl::AdapterDiscoveryOptionsES subclass to specify its creation at adapter discovery time.
Adds a "-b opengles" command-line flag to the code samples.
Asserts on a call to glShaderStorageBlockBinding() on ES.
Works around missing indexed draw buffers support by asserting when a non-0 color attachment is specified.
Works around missing glClearTexSubImage() by asserting. :/
These will likely require front-end validation.
BUG=dawn:580
Change-Id: I4a4240ca695a22388c55073fd2aee0323cd4afc9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/31000
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Stephen White <senorblanco@chromium.org>
... before attempting to push instructions to `functions_.back()`
If this fires, we're in an invalid state anyway. At least make the explosion less head scratching.
Change-Id: I7b4a002043de4b55a12d9aba80a0393c630140c2
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33664
Commit-Queue: David Neto <dneto@google.com>
Reviewed-by: David Neto <dneto@google.com>
Migrate all uses to use this and the new `unique_type<T>()` and `types()` methods.
Remove the `type_mgr()` accessor. `TypeManager` is now an implementation detail of the module, allowing us to unify the allocation of types and nodes (if we so wish).
Fixes: tint:337
Bug: tint:307
Change-Id: I233fa9dc73d60515dd721f02ea7ba089ef7d374f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33667
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: David Neto <dneto@google.com>
First step to moving this to the `ast::Module`.
Also remove a bunch of redundant includes to `type_manager.h` as this is already included in `context.h`
Bug: tint:307
Bug: tint:337
Change-Id: Ic4baffa7b76ddefa29f56f758c25b1003ef40888
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33665
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: David Neto <dneto@google.com>
The Phi in the merge block was taking the value of the RHS
from the wrong basic block ID. Instead of taking it from
the first block of the expression for the RHS, take it from
the last block of the expression for the RHS.
Bug: tint:355
Change-Id: I1b79a1b107459fd420e39963ad7ab2e89bc4494f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33640
Commit-Queue: David Neto <dneto@google.com>
Auto-Submit: David Neto <dneto@google.com>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
There was a data-race between getting and using the pointer to
mLastSubmittedCommands in WaitForCommandsToBeScheduled and setting it
inside the scheduling handler.
Lock the mutex before calling waitUntilScheduled so that the object
doesn't get destroyed from underneath us.
Bug: chromium:1142024
Change-Id: Iadbf473530342de6d108d39285d723815452ac7c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/31260
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This change is a prerequisite to D3D pipeline caching.
This change introduces:
- Caching interface which enables the cache.
- Helper for backends to load/store blobs to be cached.
- Ability to cache HLSL shaders.
Bug:dawn:549
Change-Id: I2af759882d18b3f45dc63e49dcb6a3caa1be3485
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/32305
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
With the recent enabling of PartitionAlloc on Windows this test is
crashing because PartitionAlloc raises an exception on OOM instead of
returning nullptr.
Bug: dawn:579
Change-Id: I07577ef96272dc7ed26f6922c67ed09f1e5ca97d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33560
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Add `tint::writer::FloatToString()`:
Converts the float `f` to a string using fixed-point notation (not scientific).
The float will be printed with the full precision required to describe the float.
All trailing `0`s will be omitted after the last non-zero fractional number,
unless the fractional is zero, in which case the number will end with `.0`.
Use this for the wgsl, msl and hlsl backends.
Change-Id: If5701136579e4398c31c673942f30e8877e9f813
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33421
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: David Neto <dneto@google.com>
According to the http://www.eel.is/c++draft/temp.expl.spec:
An explicit specialization shall not use a storage-class-specifier
other than thread_local.
Clang doesn't claims about it, but GCC does.
An error example for GCC 8.4.0:
gen/third_party/dawn/src/dawn_wire/client/ApiObjects_autogen.h:25:5:
error: explicit template specialization cannot have a storage class
Bug: dawn:384
Change-Id: Iaf86722a943d19c9796a7f112885666ac88f20ca
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33480
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This patch adds another way to get DXC DLLs (dxcompiler.dll and
dxil.dll) by searching the default installation path of Windows x64
SDK (C:\Program Files (x86)\Windows Kits\10\\bin\10.0.[version].0\x64)
so that now Dawn can find these two DLLs when Windows SDK is correctly
installed at the default location.
This patch also forces enabling the "use_dxc" toggle when Dawn can
find DXC DLLs and the device is created with "ShaderFloat16" extension
enabled because currently the HLSL shaders with float16_t can only be
compiled with DXC DLLs.
BUG=dawn:402
Change-Id: I410195d9e079f54faebbca71fff77a71f489f08e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33180
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This CL updates the binary operator emission to use the `mul()` method
in the following cases:
- vector * matrix
- matrix * vector
- matrix * matrix
This is because the `*` operator works per-component in HLSL which does
not do the expected multiply.
Bug: tint:301
Change-Id: I0810522ac26fbbea323cf8a05a3ff6f2fb62117e
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33362
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Auto-Submit: dan sinclair <dsinclair@chromium.org>
For HLSL emission instead of:
`vector<float, N>` emit `floatN`
`vector<int, N>` emit `intN`
`vector<uint, N>` emit `uintN`
These are significantly shorter, more idiomatic, and is far easier to read.
Change-Id: Idef8cc550e0b49cc919087e281b72a7a0a0f11bc
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33424
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Handle wsgl parsing and spirv writing of:
textureSample(), textureSampleBias(), textureSampleLevel(),
textureSampleGrad(), textureSampleCompare()
Handle the different signature for array texture types.
Includes offset overloads.
Change-Id: I6802d97cd9a7083f12439b32725b9a4b666b8c63
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/32985
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: David Neto <dneto@google.com>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
The reason being that some tests called parse() twice, which will silently destruct the first parser.
Once the `Module` owns the AST nodes, the second call will end up deleting all the AST nodes. Tests would then perform use-after-free for the AST nodes belonging to the first parser / module.
There's no reason why the unique_ptr can't be returned, which is cleaner overall.
Bug: tint:335
Change-Id: I7ff2e9777a7ebeb76702f806294fe4c2c49bd7c9
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33241
Auto-Submit: Ben Clayton <bclayton@google.com>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Also moves the type determiner call out of the transformers into the
manager.
Cleans up the code to not have anything directly calling
Run() on the transformers other then the manager.
Bug: tint:308
Change-Id: I3343f2ba16dae6fb33f35e390ae4c797f2a05522
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33262
Auto-Submit: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
