Move the CPU descriptor allocators to the device and bucket them to
ensure only kMaxBindingsPerGroup exist rather than create them per BGL.
Also, renames NonShaderVisible => Staging.
BUG=dawn:155
Change-Id: If6dae368e7e2a2b349343bdf898041a049159038
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19001
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Prior to CMake 3.12 option() is a cache variable operation which meant
that the overriding of third_party project variables in
third_party/CMakeLists.txt needs to write to the cache.
Also remove a couple extra commas that were causing warnings.
Bug: None
Change-Id: I35efce70cf3e0cb923ebdbad654b2b9883b8f734
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19280
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
The basic change was to copy-paste targets in the new BUILD.gn files and
fixup paths / add includes. There's a couple more changes that had to be
done at the same time:
- Multiple files need to know if GLFW is supported so the variable was
moved to dawn_features.gni.
- The gtest_and_gmock target used to abstract between Dawn's copy of
GTest/GMock is only needed by tests and was moved in src/tests/BUILD.gn.
- A leftover dawn_end2end_tests target is left in the main BUILD.gn
file that is an exact copy of the on in src/tests/BUILD.gn. This is
because the GN path is hardcoded in Chromium's isolate_map.pyl that also
can't support GN groups. The only way to move a target I could figure
out was to duplicate it temporarily.
Bug: chromium:1064305
Change-Id: I96820e9d6510b8c9b9112c3e6cd8df2413f04287
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19201
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
When attempting to allocate more than Dawn's budget within a single
serial, all heaps in the LRU will be un-evictable because the last
submission serial is the same as the current serial. We can work
around this by instead using the LockHeap and UnlockHeap functions
instead of EnsureCanMakeResident when calling CreatePlacedResource.
Also added in some additional comments regarding the last submission serial.
Bug: dawn:193
Change-Id: Ie4ec7ed5350b0858ea817431fbf77df6ca8acd96
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18622
Commit-Queue: Brandon Jones <brandon1.jones@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Rafael Cintron <rafael.cintron@microsoft.com>
Destruction of the BindGroup needs to ensure that the BindGroupLayout is
destroyed after the BindGroup. This is done by using a custom deleter which
first creates an extra reference to the BGL before deleting the BindGroup.
Bug: dawn:355
Change-Id: I819bbce13473ee4738eaa304f6dac90e0501302a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19060
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Also adds generic support for generating a combinatorial value
as a const definition.
Bug: tint:3
Change-Id: Idae758d146264491679710967e48ea270436be91
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19107
Reviewed-by: dan sinclair <dsinclair@google.com>
Lots of little style nits needed to be fixed for this work.
BUG=tint:44
Change-Id: Ibb45d9e3f6795ee0c09f5eca994bb28e20979d97
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19221
Reviewed-by: dan sinclair <dsinclair@google.com>
When the libdawn_native target was renamed to dawn_native, the output
library name stayed the same (GN inserts a lib prefix if it isn't
present) but the @rpath annotation changed to be just dawn_native.dylib.
Fix this by adding the lib prefix in the rpath annotation. This requires
changing libdawn_proc to dawn_proc otherwise the rpath annotation would
be liblibdawn_proc.dylib.
Bug: dawn:380
Change-Id: Id8610a6318af3468dcc486ee8d3c035f0273fe0d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19200
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
The test was failing on Swiftshader, likely because the allocations for
the SSBO and the UBO where neighbors. The UBO test uses uvec4 instead of
uint so it needs 4 times less invocations to perform the test.
Bug: dawn:283
Change-Id: Id1952505763a0a3eb34718e000cb386e1faa4ade
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19082
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
The only test where this can matter at the moment is the OpArrayLength
test, so one of the buffers tested is repurposed to test wgpu::WholeSize
with an offset.
Bug: dawn:377
Change-Id: I43f22f614dff55e399246c46892a865ec77f13c4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19080
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
The viewport tests assume sub-pixels precision when the Vulkan
specification explicitly allows for no subpixel precision.
Swiftshader seem to have only 4 bits of subpixel precision and fails
only this test surprisingly.
Bug: dawn:283
Change-Id: I77f6a6169a8fdff13448a612abdd4ab1dc42532e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19084
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Rendering +-INIFNITY with these formats results in a NaN when using
Swiftshader. Temporarily disable these tests while the Swiftshader issue
if being resolved.
Bug: dawn:283
Bug: swiftshader:147
Change-Id: I6e7c91ff72d0b4b6423175f5ab8586fdea42bb53
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19083
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
The build_overrides directory for the VVL will default to "" which
causes support for them to be skipped if the dependents don't set the
dawn_vulkan_validation_layers_dir variable.
Bug: chromium:1064305
Change-Id: I0a31f5d1d55982b5c953ce8ac6542d38829eedb5
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19085
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Zhenyao Mo <zmo@google.com>
This CL adds the source info into the type determiner errors so they
will include line and column number if available. If the line number is
0 then it's considered the source info is missing and it is not emitted.
Add line and colummn to type determiner errors
Change-Id: I18764a71db80082fd31c8509c5e9b193800f1d95
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19105
Reviewed-by: David Neto <dneto@google.com>
This CL converts a few more std::vector<std::unique_ptr<foo>>'s to
FooList aliases.
Change-Id: If0dd672f11beaeabd9e89e82903c2174b11be6e1
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19103
Reviewed-by: David Neto <dneto@google.com>
If a non-const variable in a function has a kNone storage class we
update it to kFunction. If there is a storage class other then kFunction
we emit an error.
Bug: tint:5
Change-Id: If45eb91bd0a0095e625eb1d0e1d1e361c784e35d
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19102
Reviewed-by: David Neto <dneto@google.com>
This CL adds the relational add expression to the spirv writer.
Bug: tint:5
Change-Id: Ideed225c63e60f53a019b79803809f8e973299c6
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18605
Reviewed-by: David Neto <dneto@google.com>
Skia uses more warnings than Dawn, enable in Dawn directly so that
rolls of Dawn into Skia don't introduce warnings. These warnings
seem useful anyway.
Bug: chromium:1064305
Change-Id: I13dc776af84151131584a95caeee2cd21ae80fea
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18964
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
The server didn't take intercept the destroy() call which meant the
buffer could be unmapped by dawn_native without the status updated in
ServerBuffer. This caused crash when a subsequent UpdateMappedData
command was handled and tried to write into the mapped buffer.
The client needs to also track destroy() otherwise it could sent an
UpdateMappedData to a destroyed buffer which is a fatal error.
Tests are added that cover the client-server interaction for this, but
the pattern that the following is unfortunately not tested directly
against the wire server:
- CreateBufferMapped
- Destroy
- UpdateMappedData
Bug: chromium:1068466
Change-Id: If5185d4a8a81cd5f6bb41c9888a18c44c14b2de4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18961
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
I am about to add a new entry to the enum being used in this switch on
the spvc side. Temporarily adding in a default here, so that can roll
correctly, then this function can be updated to use the new entry.
BUG=dawn:367
Change-Id: I9345f651e2e23604599a1b25fcd2993127062695
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19040
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Tests FunctionEmitter directly.
This is mostly refactoring to be able to selectively run parts of the
parsing flow, and to access relevant internal data.
Bug: tint:3
Change-Id: Ic2b166a2e9623a7e30e6769806088d12e78dcf45
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18704
Reviewed-by: dan sinclair <dsinclair@google.com>
Change-Id: Icab0d32f199d4b0f95e2c1ebcd9ae9ea9ec7b61c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19020
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Add a dawn_has_build override similarly to angle_has_build that makes
Dawn not depend on Chromium //build if it can avoid it. (this was
previously used for mac_sdk.gni).
Only load the Vulkan Validation Layers' gni if it is enable via
dawn_enable_vulkan_validation_layers.
Fix a GN build warning where the source of dawn_components would be
built twice: for example libdawn_native.MetalBackend.o would be built
once for libdawn_native static and once for libdawn_native shared. Fix
this by making libdawn_native as a static library libdawn_native_static.
Change-Id: Ib8fb72282435eb475d7a11dae0d5a9977572fd57
Bug: chromium:1064305
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18963
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This CL completes a few of the DISABLED tests.
Change-Id: I38806fb9381240d6e3da700f1fb2dac5f899ca6d
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/19000
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Match the more common usage for the expression type.
Bug: tint:37
Change-Id: Ia5d48a0444742ec4e304ea1036e499b3d7cad682
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18981
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
This CL adds tests for break, continue and return without attached
expressions.
Bug: tint:5
Change-Id: I8c4221e787af95f5597db3ea4e42b38962223bf4
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18980
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
This CL adds type determination for the unary op expression.
Bug: tint:5
Change-Id: I5b9c0c80bb48527f1f26febb2310f9640e5f7849
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18850
Reviewed-by: David Neto <dneto@google.com>
This CL adds the type determination for the unary method expression.
Bug: tint:5
Change-Id: I9f94a79b9715cf74e37c74eb1a612ca84b3c241f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18849
Reviewed-by: David Neto <dneto@google.com>
This Cl adds the type determination for unary derivative expressions.
Bug: tint:5
Change-Id: I44ad22733ce6b94e4b6d77b432024bacc110e8af
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18848
Reviewed-by: David Neto <dneto@google.com>
This CL adds the type determination for each of the relation types in
the relational expression.
Bug: tint:5
Change-Id: I15e8dae2f90cc4a0f720692f5addb944b26811ec
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18847
Reviewed-by: David Neto <dneto@google.com>
This Cl adds the member accessor type determination for both structures
and vector swizzles.
Bug: tint:5
Change-Id: I1172db29d8cbed2d9e0ae228ebc3a818d4930b7f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18846
Reviewed-by: David Neto <dneto@google.com>
This CL adds the type determination for cast expressions.
Bug: tint:5
Change-Id: I35ae28e3b70c554cd48c6abcd78d95c2ba7211fa
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18845
Reviewed-by: David Neto <dneto@google.com>
This CL adds the type determination for call expressions.
Bug: tint:5
Change-Id: Ibe08f90ec3905dd1e2169f6e69d1d74943720819
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/18844
Reviewed-by: David Neto <dneto@google.com>
What was previously the Device's loss status is now a state that also
contains the "being created" state. Its transitions are entirely
handled in the frontend which enforces somewhat uniform lifecycles
between backends.
The backend devices' ShutDownImpl() function is now guaranteed to be
called only during the destructor, which leads to further simplification.
Previously Destroy() could also be called when the device was first
lost. This require complications because, for example, a WGPUBuffer
could still exist, and would want to call some resource allocator
service after the call to Destroy(). Now destruction of the device's
backing API objects is deferred to the destructor. (that's ok as long
as the application can't submit any more work).
WaitForCompletion is now guaranteed to be called before ShutDownImpl() iff
the call to DeviceBase::Initialize was succesful and the backing device
not lost. The idea is that after DeviceBase::Initialize, the GPU can
have some work enqueued and we need to wait for it to complete before
deleting backing API objects. In the future we might also have backend
be reentrant, using WebGPU itself to implement parts of the backend.
Reentrant calls would only be allowed after DeviceBase::Initialize.
Also the DynamicUploader that was special-cased in all backends is
now handled entirely by the frontend.
Bug: dawn:373
Change-Id: I985417d67727ea3bc11849c999c5ef0e02403223
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18801
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
To do perfect forwarding of GN targets, the proper method is to make a
group with public_deps. For example in the following:
- Config C
- Target T with a public_deps including C
- Group G that proxies B
- Target T2 that depends on G
If G uses deps = [ T ] then T2 won't see C, whereas if G uses
public_deps = [ T ] then T2 will dep on T, which will make it dep on C.
Bug: chromium:1064305
Change-Id: Iae236150c119b1a4003b957dcacf42e7759a936c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18965
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
