- Also updates and adds tests for CommandBuffer since it is the last object with a Destroy name clash.
Bug: dawn:628
Change-Id: I028e0101a91a785aa90d2b656556d48fe0d6e736
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/68101
Commit-Queue: Loko Kung <lokokung@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
The changes should pass through the destroy changes such that when the device is destroyed, the respective destroy functionality currently existing in the backends should be called.
For buffers, destroy no longer causes validation errors since even error buffers may need to be destroyed in the case of mappedAtCreation.
Bug: dawn:628, dawn:1002
Change-Id: I42a475af5d67cc60f86d95ac53c2b377a9fd2e82
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/65863
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Loko Kung <lokokung@google.com>
Adds a generic SetDebugName utility for D3D12 and Vulkan. Passes down
debug label set by user for buffers and textures to be labeled by the
appropriate backend method.
Bug: dawn:840
Change-Id: I7158b537a6e37fdf733645e6830dc33833ee683e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/61588
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Brandon Jones (Intel) <brandon1.jones@intel.com>
Buffer allocations in Dawn may be padded. This padding
could be visible, depending on how the backend
implements robust vertex buffer access. This commit
updates buffer creation to clear all padding bytes
immediately after creation.
It is not counted as a lazy clear.
And, add a test which reads off the end of a padded
vertex buffer to check that the padding bytes are also
initialized.
Also: Update Metal buffers to always allocate enough
space for Tint's vertex pulling, and pass the full
allocated size to Tint. While writing the test in this
commit, a bug was found where Tint assumes there is at
least 4 bytes in the vertex pulling buffer. The WebGPU
API currently allows zero-sized bindings, so Dawn needs
to always allocate enough space for one element.
Also: Update Vulkan vertex/index buffers to allocate
at least one more byte than requested because
vkCmdSetVertexBuffers and vkCmdSetIndexBuffer disallow
the offset to be equal to the buffer size. We need
at least one more byte to make zero-sized bindings
as the end of the buffer valid.
Lastly: Update helper so that a T2B copy of a single
row that touches all bytes in a buffer updates the
buffer state to be fully initialized. This is a small
optimization that is necessary to write the test.
Bug: dawn:837
Change-Id: I829f4764509c4ec784c5aeaaf40b6dcbd4be6866
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/62161
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Buffers are padded to be at least 4 bytes. Dawn would
produce invalid barriers for buffers that were "zero"
bytes large. The size for the barrier must be either
non-zero or VK_WHOLE_SIZE. Since we always intend
to transition the entire buffer, simply use
VK_WHOLE_SIZE instead of GetSize() as the size.
Fixed: dawn:969
Change-Id: Ifed4e8f9164bafa3c44917074fcf214e72919c56
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/57940
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Austin Eng <enga@chromium.org>
This patch combines all the resource barriers added before each
dispatch() into one call to reduce the number of
vkCmdPipelineBarrier() in the Vulkan command buffer.
BUG=dawn:522
Change-Id: I1b5943e62eb0a09db96de12196fcabb3448e9e4d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28283
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
This patch fixes the issues on the initialization of buffers with
MapRead usage and mappedAtCreation == true.
1. The buffers with MapRead usage and mappedAtCreation == true can be
read from the CPU side just after the creation of the buffer,
however at that time the internal pending command buffers may not
be executed, thus causing the buffer is not cleared as is expected.
2. On D3D12 the buffer with MapRead and mappedAtCreation == true is
created on the READBACK heap, so all the data written in the CPU
side cannot be uploaded to the GPU memory. When the buffer is mapped
again all the original data written through the CPU pointer will be
overwritten by the data in the GPU memory (which means it is also
cleared to 0).
This patch fixes this issue by:
1. clearing the buffers with mappedAtCreation == true on the CPU side.
2. on D3D12 making the buffer with MapRead and mappedAtCreation == true
use the staging buffer instead of mapping itself.
Note that this change is only related to the code path with Toggle
"nonzero_clear_resources_on_creation_for_testing" enabled, currently
we don't plan to do the similar change when we enable Dawn wire.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I2b3d0840333e8d99759800ab9fc141d0a7cf2f8d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27220
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
MapAsync in dawn_native is fully implemented and only missing
a couple cleanups that can be done once MapRead/WriteAsync are
removed.
MapAsync in dawn_wire is left as a pure shim on top of
MapRead/WriteAsync and will be transitioned to its own commands
in follow-ups.
All MapRead/WriteAsync end2end and validation tests are duplicated
for MapAsync.
Bug: dawn:445
Change-Id: Ib1430b9257149917be19a84f13e0ddd2a8eccc32
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24260
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
This patch implements buffer lazy initialization before
CopyBufferToBuffer() behind the toggle LazyClearBufferOnFirstUse.
- If the source buffer is not initialized, it will be cleared to 0
before CopyBufferToBuffer().
- If the destination buffer is not initialized and the copy doesn't
overwrite the whole buffer, it will be cleared to 0 before
CopyBufferToBuffer(), otherwise the buffer shouldn't be cleared.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I3d0512c6376a1ed8928e86f8e56fefebc16910fa
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24360
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
In the validation tests, we relied on Queue.Submit(0, nullptr) to
resolve mapping operations. This is fragile so we replace it with a
FlushMappingOperations() function that uses device.Tick() instead.
This allows removing the mapSerial argument from
Buffer::MapRead/WriteAsyncImpl (which was the actual goal of this CL).
Bug: dawn:445
Change-Id: Id98822287370c371bebb83afb8e290e17f3c1b55
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24381
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This required changing DeviceBase::CreateBufferImpl to return
ResultOrError<Ref<BufferBase>>
Bug: chromium:1103154
Change-Id: I1a5811d293333b6ef29c988a08f2f1f84ac65702
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24500
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This CL:
- Adds mappedAtCreation to dawn.json
- Changes dawn_native to implement CreateBufferMapped in terms of
mappedAtCreation.
- Duplicates all the CreateBufferMappedTests to mappedAtCreation tests
(both validation and end2end).
- Implements dawn_wire's mappedAtCreation in terms of
CreateBufferMapped. The reversal in dawn_wire will be done in a
follow-up CL.
Bug: dawn:445
Change-Id: I70b9fa729b1402524a6b993c3f288987eb65c6c4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24083
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
GetMappedRange never produces errors and instead returns nullptr when it
is disallowed. When in a correct state, should return a valid pointer as
much as possible, even if the buffer is an error or if the device is
lost.
Adds tests for error buffers and device loss, and modify existing tests
to not expect a device error.
Also removes some dead code in the Vulkan backend and adds a fix for
missing deallocation of VkMemory on device shutdown.
Bug: dawn:445
Change-Id: Ia844ee3493cdaf75083424743dd194fa94faf591
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24160
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
This patch implements buffer lazy initialization before writeBuffer():
if the buffer is not initialized and writeBuffer() doesn't cover the
whole buffer, the buffer will be cleared to 0, otherwise the buffer
shouldn't be cleared.
This patch also introduces a toggle LazyClearBufferOnFirstUse for the
development of buffer lazy initialization: before buffer lazy
initialization being completely supported, all the related code will
only be enabled behind this toggle to prevent the buffers with valid
content being unexpectedly cleared.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I99a2aa98ca4b9b21d69c6b32080afb525e2c4ad3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/24041
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This patch enables nonzero_clear_resources_on_creation_for_testing
toggle on buffer on Vulkan backends as a preparation of supporting
buffer lazy-initialization in Dawn.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I7619d02ae898e30fd15438d35437802a09e959cc
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22981
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
This patch combines barriers in a render/compute pass into one
function call.
Previously, we need to dispatch barrier(s) for each buffer/texture
in a pass. So we may need quite a lot function calls to deliver
barriers in a pass in real web applications. One example is that
we did see that too many function calls to deliver barriers in
Aquarium (WebGPU porting) contributed to CPU usage and bottleneck.
Bug: dawn:441
Change-Id: Ibe44967fefd2e1e6e64df4587146c4fb7fbe8e73
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22700
Commit-Queue: Yunchao He <yunchao.he@intel.com>
Reviewed-by: Austin Eng <enga@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>
RefCounted (and derived) destructors should be protected on the class
to ensure the objects can ONLY be destructed by calling Release. This
avoids errors cause by destroying objects out from under code which
has an active reference count.
Unfortunately, many of the 'base' classes must continue having public
destructors because they are used as "blueprint" objects created on
the stack.
Added final on most-derived classes.
Ideas for future improvement:
- Change "base" objects to have protected destructors but create new
blueprint objects that privately derive from base objects. This
limits the blueprint object's usefulness to only be a blueprint.
- Modify createX methods to return Ref<Object> instead of Object*
Change-Id: I6f3b3b178118d135c4342cb912e982a3873d71af
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18780
Commit-Queue: Rafael Cintron <rafael.cintron@microsoft.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This removes the duplication of the memory allocators in preparation for
using sub-allocation in the Vulkan backend too.
Also renames ResourceMemory to ResourceHeap and MemoryResourceAllocator
to ResourceMemoryAllocator, and fixes a number of unused includes.
BUG=dawn:27
Change-Id: I1a9e7d41e5efafa5192bda1d89dc06455fa2af40
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/12660
Reviewed-by: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This makes all the regular WebGPU object creation handle errors properly
in the Vulkan backend instead of ASSERTing no Vulkan error is raised.
Static Create functions are added to all these types so that the details
of how the initialization is done is private, and it isn't possible to
construct an object but forget to initialize it.
BUG=dawn:19
Change-Id: I362b2d66b74dd7799ffbf69d732bc58caa97950b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/11861
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Refactor existing memory allocators by using a common
memory type and handle.
BUG=dawn:27
Change-Id: Ieed4fa30a0bd8fedfb3a3c580920805f40b56fae
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10680
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
This is to match the naming convention of WebGPU's WebIDL and webgpu.h
BUG=dawn:22
Change-Id: Ia91c5a018403e6a72eb0311b5f1a072d102282a2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10461
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Adds a structure that contains wait / signal semaphores and the current command buffer. Will allow us to have a list of pending semaphores after recording, either to consume or do something smarter with in the future.
Bug: chromium:976495
Change-Id: Ib61455039bd97ac8b0ff701af2b694cc8794226d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9600
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This uses an intermediate staging buffer to copy data into the buffer.
Bug: dawn:7
Change-Id: I3bda19a8450ef0eddc5b4382ce1b9120f074b917
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/7500
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This is the first command to return a struct. This patch also
updates the code generator to support structure return values.
Bug: dawn:7
Change-Id: Ie8acec895c0ec88429672138ffc900032fbbc447
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/4780
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
This reverts commit 9bf529ec9421dcd6a27b9d07fbe3edf6bea598d3.
Reason for revert:
Fixed test failure by submitting basic render pass to clear out texture
before running the tests.
The test was failing previously because the texture pixel color was not
cleared before running the tests, causing unexpected
pixel colors to be compared. Creating a basic render pass clears
the texture, but since the first test fails on submit expectedly,
the pixel is never cleared.
Bug: dawn:46
Change-Id: Ic190c2d8d6af3f9d8def3370b92c6974a82a0096
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/5500
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Natasha Lee <natlee@microsoft.com>
This reverts commit b6a80b321e8fa0b79d9a947656ea0ad649ed5a3c.
Reason for revert: dawn_end2end_tests are failing on the Chromium GPU FYI bots. Example here: https://ci.chromium.org/p/chromium/builders/ci/Win10%20FYI%20Release%20%28NVIDIA%29/4226
Original change's description:
> Destroy backend implementation for Buffers
>
> Destroy can be used to free the GPU memory associated with resources
> without waiting for javascript garbage collection to occur.
> The buffer is validated at submission to the queue.
> So any buffer that has been destroyed before submission, will then
> invalidate the submit and result in an error.
>
> Bug: dawn:46
> Change-Id: I40df56ce97baef01deea7552d7a6d40b558fc985
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/5320
> Reviewed-by: Kai Ninomiya <kainino@chromium.org>
> Commit-Queue: Natasha Lee <natlee@microsoft.com>
TBR=cwallez@chromium.org,kainino@chromium.org,enga@chromium.org,rafael.cintron@microsoft.com,natlee@microsoft.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: dawn:46
Change-Id: Iadf37a8a6675c744207ec7daaa3fd2fde7da3714
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/5480
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Destroy can be used to free the GPU memory associated with resources
without waiting for javascript garbage collection to occur.
The buffer is validated at submission to the queue.
So any buffer that has been destroyed before submission, will then
invalidate the submit and result in an error.
Bug: dawn:46
Change-Id: I40df56ce97baef01deea7552d7a6d40b558fc985
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/5320
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Natasha Lee <natlee@microsoft.com>
Manages a single persistently mapped GPU heap which is sub-allocated
inside of ring-buffer for uploads. To handle larger buffers without additional
unused heaps, ring buffers are created on-demand.
BUG=dawn:28
TEST=dawn_unittests
Change-Id: If4d3e717186895b1409502c1dea5ab751a4776b2
Reviewed-on: https://dawn-review.googlesource.com/c/4440
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
This is part of making buffer mapping match WebGPU
Bug: dawn:7
Change-Id: Ia97c872e398112aef0f93c98618c8a7f3ff0c19a
Reviewed-on: https://dawn-review.googlesource.com/c/4580
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Manages a single persistently mapped GPU heap which is sub-allocated
inside of ring-buffer for uploads. To handle larger buffers without additional
unused heaps, ring buffers are created on-demand.
BUG=dawn:28
TEST=dawn_unittests
Change-Id: Ic2a5df3142fc24fa772b9a85b38248eea8c7e003
Reviewed-on: https://dawn-review.googlesource.com/c/4260
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Manages a single persistently mapped GPU heap which is sub-allocated
inside of ring-buffer for uploads. To handle larger buffers without additional
unused heaps, ring buffers are created on-demand.
BUG=dawn:28
TEST=dawn_unittests
Change-Id: Ifc5a1b06baf8633f1e133245ac1ee76275431cc5
Reviewed-on: https://dawn-review.googlesource.com/c/3160
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
We still keep a dummy BufferBuilder object around so that it can be used
for the builder error callback tests of the wire.
Change-Id: If0c502bb8b62ee3ed61815e34e9b6ee6c03a65ef
