This CL removes all internal usage of the Context object. It is still
accepted as a parameter until we update Dawn, but all usage is removed.
The namer has been removed from the SPIR-V backend with this change and
the emitted names reverted to their non-modified version.
Change-Id: Ie6c550fab1807b558182cd7188ab6450a627f154
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/34740
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Auto-Submit: dan sinclair <dsinclair@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
We're seeing some chrome bots fail unittests in ways that suspiciously
look like dynamic casts are doing Wrong Things.
The ClassID::Of() logic depends on the linker folding away duplicate
compilation unit definitions based on ODR rules. If we were to somehow
end up with different definitions, then we'd have two or more different
ClassIDs for the same T type - leading to issues similar to what we're
seeing.
I'm not entirely sure why/how this could happen - and we've so far been
entirely unable to locally reproduce - but it _might_ have something to
do with the goma cache.
In an attempt to work around this, move the static symbol definition out
of a header-local-static and into the .cc file for each of the types.
Change-Id: If914d3045b9dac6fbe8824dac71153a768cfceb9
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/34563
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
If there is an issue with the shader, i.e. validation failing, the
former ASSERT_EQs will kill the thread that is running the lambda.
Since the main thread is waiting for a state change, if this assert
kills the thread, the test will hang forever.
This change allows the test to complete and fail correctly in the case
that the test condition fails.
Bug: tint:578
Change-Id: I6f5cc636193bf01c2d5e5d77a883fab7ec591c3e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34700
Auto-Submit: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
In WGSL, textureLoad translates to an OpImageFetch without the
combined sampler. In OpenGL, we need to use SPIRV-Cross to insert
a dummy nearest filtering sampler and bind that in the backend.
Bug: dawn:585
Change-Id: I92ae6ad35263d3720e59fa93688ca914a9495a81
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34401
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
When handling OpCompositeExtract, the SPIRV reader uses a
stack-allocated U32 type instead of one heap allocated by the
type manager. This causes type determination later to dereference
a garbage address.
Change-Id: I7d60b6dbf8310e53565d7db47eac4dd92b1bbfa0
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/34684
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
We've been using |blah| in various places to markup code, however this is not doxygen markup. So instead:
* If the code links to a parameter, use `blah`.
* If the code links to a member field, use #blah.
* If the code links to a method use blah().
* If the code is somewhere unlinkable use `blah`.
Change-Id: Idac748a4c2531b5bae77e1a335e3d3ef6fab48b6
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33787
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
This CL enable CopyTextureForBrowser to accept options. The first
supported option is flipY, which can be implemented through scale and
offset uniforms.
BUG=dawn:465
Change-Id: Ia90153ee63a50e0e40beb1c13c63764d19a0b809
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34402
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This is mostly the set of tests which are enabled on Desktop GL,
with the exception of tests which use glTextureView() (unsupported
on GLES).
BUG=dawn:580
Change-Id: I1dbefa394298f7dd31d65e920f1f4efeaebe90ec
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34520
Commit-Queue: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Deep-clones all `Node`s and `Type`s into a new module.
Instead of writing a million standalone tests that'll only ever test the
existing fields of each type, I've opted to write the tests using
wgsl<->ast<->wgsl conversion. This means the tests require the enabling
of TINT_BUILD_WGSL_READER and TINT_BUILD_WGSL_WRITER, but I believe this
is much easier to maintain.
I'm aware there are probably gaps in the tests, and that even full
coverage is likely to rapidly rot, so I've also added
fuzzers/tint_ast_clone_fuzzer.cc - a fuzzer based test that ensures that
all AST modules can be cloned with identical reproduction.
I've run this across 100 cores of a 3990x for 4 hours, fixing the
single issue it detected.
Note: Expressions do not currently clone their `TypeManager` determined
types. This is for two reasons:
(a) This initial CL is mahoosive enough.
(b) I'm uncertain whether we actually want to clone this info, or to
re-run the `TypeDeterminer` after each AST transform. Maybe it should
be optional. Time will tell.
Fixed: tint:307
Change-Id: Id90fab06aaa740c805d12b66f3f11d1f452c6805
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/33300
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
OpenGL ES doesn't have the glShaderStorageBlockBinding() call, so we
modify the binding decorations set in the shader instead. This requires
plumbing through some more pipeline state at shader translation time.
BUG=dawn:584
Change-Id: Ib7fdb6a7ad1eff1a99d44d55e9d923214affe702
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34400
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Adds in a toggle to control using Tint or SPIRV Cross for reflection.
The current implementation only gets name and stage information via Tint,
backfilling the rest from SPIRV Cross.
Future CLs will fill in the rest of the entry point metadata.
Bug: dawn:578
Change-Id: Iee16b9dbb9a17d4aaa68e5e52e4aa91ee0a62624
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33500
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
CopyTextureForBrowser end2end tests had been skipped on Vulkan/Metal
backends due to crbug.com/tint/63.
Because this issue has been fixed, enable CopyTextureForBrowserTests on
Vulkan/Metal backends.
However, D3D12 backend with validation layer still stucks. Raise issue
crbug.com/dawn/592 and keep tests skip.
BUG=dawn:465
Change-Id: I819b3608d07ea7467b33263004e9e19158e73eff
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34440
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
In previous internal shader, we draw a just fit-in rectangle to do the
copy. But there is arithmetic precision issues in border pixels when
copy from a large texture.
ANGLE handle the same issue by drawing a large triangle. Dawn adopte the
same idea here
BUG=dawn:465
Change-Id: I2366e28b1e96e7a33116a170023a5138d8c9f770
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33900
Reviewed-by: Shaobo Yan <shaobo.yan@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Shaobo Yan <shaobo.yan@intel.com>
They already exist in a `ast::type` namespace, so `ast::type::BlahType` is just stuttering.
This is more important now that Is<> and As<> use the full type name.
Change-Id: I7c661fe58cdc33ba7e9a95c82c996a799786661f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/34321
Reviewed-by: dan sinclair <dsinclair@chromium.org>
