This patch fixes a crash issue in both D3D12 and Vulkan backends.
Previously on D3D12 and Vulkan before a compute pass we transitioned
the states of all the resources used in the pass, and before each
dispatch call we only checked if the states of the storage buffers,
read-only storage textures and write-only storage textures need to
be transitioned. This behavior causes two issues:
1. In a compute pass a buffer or texture can be used as both read-only
and writable usages in different dispatch calls (e.g. as storage
buffer in the first dispatch, and as the uniform buffer in the next
dispatch), while this is invalid state combination on D3D12 and isn't
allowed by D3D12 validation layer.
2. In the above case, the state of the buffer is not transitioned into
UNIFORM, which does not match the required state in the next dispatch.
This patch fixes this issue by transitioning all the states in the
current bind group before each dispatch() instead of the beginning
of the compute pass.
BUG=dawn:522
TEST=dawn_end2end_tests
Change-Id: Ibeb6c41dc493ee1068b43bd89ed5a15f2331ef75
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27942
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Also moves BindingNumber, BindGroupIndex, and BindingIndex to
IntegerTypes.h. Future TypedIntegers should be declared here.
Bug: dawn:442
Change-Id: I5ba8de3412fb48b7957b67e7c413a5097f8ec00f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27880
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Remove all usages of SPVC from the code and update the fuzzers. Some
of the include paths and deps came transitively from spvc, so needed
to update build rules.
This patch does NOT remove the flags related to spvc usage, they are
just no-ops as the moment. After this patch lands I will remove the
usage of those flags from the bots, then remove the flags.
BUG=dawn:521
Change-Id: I0d7c3e28f79354c78f00c48b6a383b823094a069
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27900
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
- Implement QuerySet on Vulkan backend.
- Enable end2end tests on Vulkan. The timestamp tests will be enabled
in following CL which implement timestamp query.
Bug: dawn:434
Change-Id: I7ee04380c5f6b5af561cc23e28637dcae70bc7b9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/26360
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Hao Li <hao.x.li@intel.com>
Format::aspect should be a mask so that it is easier to iterate over
and test if an aspect is present.
This CL also re-exports wgpu's EnumClassBitMask helpers in dawn_native.
It also adds an EnumMaskIterator which wraps BitSetIterator to allow
iterating over the enums in an enum mask.
Bug: dawn:439
Change-Id: I08180a45b27c6031e2f80b0fa1801716677fd813
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24682
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Austin Eng <enga@chromium.org>
GN recently added support for Apple frameworks to link, rather than
overloading the libs lists. This pulls .frameworks out of the libs
lists, so that GN can stop supporting .frameworks in libs in the
future.
Bug: chromium:1052560
Change-Id: I0b57a3c4c995887a57937317db30b67cf7a7f0f7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24281
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This CL adds per-stage and per-pipeline-layout limits according
to the WebGPU spec. It also slightly increases kMaxBindingsPerGroup
from 16 to 24 so that these limits can be effectively tested
without hitting kMaxBindingsPerGroup. kMaxBindingsPerGroup is not a
real WebGPU limit and will be removed in future patches.
Bug: dawn:443
Change-Id: I72be062cd31dea4ebd851f2d9f8274a77f286846
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24481
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
At the moment each instance extension needs special handling
in 6+ places when added to the Vulkan backend. This is very
error-prone and makes it difficult to do changes in how they
are extensions are handled.
This CL makes instance extensions linked with an enum class
and a bitset to know which are available (instead of individual
booleans). A table of known extensions with more information like
`versionPromoted` so that they can be handled programmatically.
Bug: dawn:457
Change-Id: I266deb730eb2b7f3ab0ee7ada1b06ff9748a60e4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22940
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
All the buffer backend files had basically the same implemenations
of MapRequestTracker and the tracker was owned by device backends.
This refactor puts MapRequestTracker into its own file
and has the tracker be owned by DeviceBase and BufferBase.
Bug: dawn:400
Change-Id: Id28422b575e9c04d4435d5f119e0ffe08c2d1ce8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/21760
Commit-Queue: Natasha Lee <natlee@microsoft.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Allows bindgroups that use the same samplers to share
a descriptor heap allocation. This is particularly important
for sampler heaps which incur expensive pipeline flushes
due to the smaller size requiring more frequent switches.
The device dolls out entries to a sampler heap allocation cache.
When the BindGroup is created, it does a lookup and refs the
allocation. This ensures the cache does not grow unbounded
or needlessly store unused entires.
This change is a follow-up of de-coupling heaps.
BUG=dawn:155
Change-Id: I3ab6f1bdb13a40905cb990cd7a2139e73da30303
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20783
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tint is behind a build flag, since Tint is not yet available in
chromium's third_party.
BUG=dawn:403
Change-Id: If5fa36b36216ef6965341b886c3bd70d27daf23a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/21300
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Allows bindgroups to be populated by heap type. Previously, failing to
populate one type of GPU descriptor heap required both GPU heaps
to be switched out. This resulted in extra copies and heap allocations
should only one heap type overflow.
This change also simplifies GPU descriptor heap management:
- Allocator no longer needs to operate on both heaps.
- Sub-allocation tracking can be moved into handles.
A follow-up change will remove duplicated sampler heap allocations.
BUG=dawn:155
Change-Id: I1960cf3a8bc3d86d3e8b2775da3d0c92125bcf82
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19887
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This is a reland of 96c4019214
It includes a fix to add a dummy descriptor count if the
VkDescriptorPool would be empty, and adds a test that a bind group
with an empty bind group layout may be created and used.
Original change's description:
> Slab-allocate VkDescriptorSets
>
> This introduces a slab allocator for VkDescriptorSets which creates
> a VkDescriptorPool pre-allocated with multiple VkDescriptorSets per
> BindGroupLayout. In the future, we can deduplicate pools that have
> the same, or roughly the same, descriptor counts.
>
> This CL also removes the old DescriptorSetService and moves most of
> the functionality onto the DescriptorSetAllocator itself to keep
> the tracking logic in one place.
>
> Bug: dawn:340
> Change-Id: I785b17f4353fb3d40c9ccc33746600d6794efe7c
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19320
> Reviewed-by: Austin Eng <enga@chromium.org>
> Commit-Queue: Austin Eng <enga@chromium.org>
Bug: dawn:340
Change-Id: Iabb744f110d0cab442bb857b31c87ba46bf0ad7a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20321
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This reverts commit 96c4019214.
Reason for revert: Breaks the roll, WebGPU CTS hits the
ASSERT(totalDescriptorCount > 0)
Original change's description:
> Slab-allocate VkDescriptorSets
>
> This introduces a slab allocator for VkDescriptorSets which creates
> a VkDescriptorPool pre-allocated with multiple VkDescriptorSets per
> BindGroupLayout. In the future, we can deduplicate pools that have
> the same, or roughly the same, descriptor counts.
>
> This CL also removes the old DescriptorSetService and moves most of
> the functionality onto the DescriptorSetAllocator itself to keep
> the tracking logic in one place.
>
> Bug: dawn:340
> Change-Id: I785b17f4353fb3d40c9ccc33746600d6794efe7c
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19320
> Reviewed-by: Austin Eng <enga@chromium.org>
> Commit-Queue: Austin Eng <enga@chromium.org>
TBR=cwallez@chromium.org,kainino@chromium.org,enga@chromium.org,bryan.bernhart@intel.com
Change-Id: Icb58485f1080eab79b24fbcd834a89fc6206b80e
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: dawn:340
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20280
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Uses the staging descriptor allocator to enable
sub-allocation for RTV/DSVs.
This change also simplifies CPU descriptor heap
management for render-passes:
- Allocating slot-by-slot removes extra attachment
pass.
- No need for the slower direct allocation allocator.
- Move RP creation of view handles into RP builder.
BUG=dawn:155
Change-Id: I508492a2e56a897bf8c85f9a45cd13f62fa0a2ef
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20042
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
This introduces a slab allocator for VkDescriptorSets which creates
a VkDescriptorPool pre-allocated with multiple VkDescriptorSets per
BindGroupLayout. In the future, we can deduplicate pools that have
the same, or roughly the same, descriptor counts.
This CL also removes the old DescriptorSetService and moves most of
the functionality onto the DescriptorSetAllocator itself to keep
the tracking logic in one place.
Bug: dawn:340
Change-Id: I785b17f4353fb3d40c9ccc33746600d6794efe7c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19320
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
- Move RefCounted to common (from dawn_native) so that we can use
it from additional places.
- Use EXPECT_ macros instead of ASSERT_ in RefCounted tests for
improved logging on failures.
- Add a missing test for Ref::Detach.
- Plug memory leak in RaceOnReferenceRelease
Change-Id: Iaa7b11b5a6fa146e3c322143279a21a4ac027547
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19903
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Rafael Cintron <rafael.cintron@microsoft.com>
On macOS we can't rely on having the Vulkan loader installed in the
system. So we add the Vulkan loader as an optional dependency of Dawn
and use it on macOS when building Dawn in standalone with Vulkan
support.
Usage of building our own loader might broaden if the loader gains
features that are useful on other OSes. For example the ability to pass
in the "root ICD" entrypoint to the loader so we can have both
Swiftshader and the system driver at the same time.
Bug: dawn:388
Change-Id: I7ade4961cce0463c66846ad17aebf95224f1afcc
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19723
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This is required for the call to the IsWindow function that's in
Surface.cpp.
Bug: chromium:1064305
Change-Id: I0c4a6ed17ba4aada994262d393de142de3f26144
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19282
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
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>
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>
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>
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>
This helps take advantage of the GN feature where when a directory is
used as a target name, like //foo/bar/baz, the //foo/bar/baz:baz target
is used automatically.
Bug: chromium:1064305
Change-Id: I2e2d9f308fda1b811482026962ab0770ac45e988
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18862
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This will help external projects that want to use Dawn, like Skia, refer
to GN targets without GN discovering Dawn's main BUILD.gn file that
causes all the tests and third_party dependencies to be discovered.
This CL just splits off chunks of BUILD.gn into separate file, adds
necessary includes and fixes up GN paths. It also introduces temporary
groups for targets that are used in Chromium so that the paths can be
fixed in a 3-way patch.
Bug: chromium:1064305
Change-Id: Ib4b73bd8d3121ef67d4ecee2e54ec158875f2117
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18861
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Jiawei Shao
Austin Eng
Ryan Harrison
Hao Li
Bryan Bernhart
Sylvain Defresne
Corentin Wallez
Brandon Jones
Natasha Lee
Rafael Cintron
Zhenyao Mo