This CL does the following:
* Adds a "transient attachment" bit to texture usage. This bit
specifies that the created texture will be used only during
this render pass.
* Adds a TransientAttachments Feature that gates the usage of
transient attachments.
* Adds support for transient attachments on Metal, where they're
used to create textures as memoryless.
* Adds validation tests and an E2T test of the feature.
A followup CL will add support in Vulkan.
Bug: dawn:1695
Change-Id: I3c7322dd1e4bee113062aae2e0494d292ee8cbc3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/129080
Commit-Queue: Colin Blundell <blundell@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This reverts commit 8cc6205bf7.
Reason for revert: Graphite actually reuses the bind groups between
draw calls using different pipelines and this change prevents it
from happening.
Original change's description:
> Disable frontend cache when implicit device sync is on.
>
> Normal behavior of ApiObjectBase's APIRelease() which only locks the
> device when last ref dropped is not thread safe if the object is cached
> as raw pointers by the device. Example of cached objects: bind group
> layout, pipeline, sampler, shader module.
>
> The following scenario could happen:
> - thread A:
> - shaderModuleA.APIRealease()
> - shaderModuleA.refCount.Decrement() == true (ref count has reached zero)
> - going to call shaderModuleA.LockAndDeleteThis().
> - thread B:
> - device.CreateShaderModule().
> - lock()
> - device.GetOrCreateShaderModule()
> - shaderModuleA is in the cache, so return it.
> - unlock()
> - thread A:
> - starting to call shaderModuleA.LockAndDeleteThis()
> - lock()
> - erase shaderModuleA from the cache.
> - delete shaderModuleA.
> - unlock()
>
> This CL disables caching when ImplicitDeviceSynchronization is turned on
> until we find a better solution.
>
> Bug: dawn:1769
> Change-Id: Ideb2a717ece0a40e18bd1c2bef00817262bd25da
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/127900
> Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
> Reviewed-by: Austin Eng <enga@chromium.org>
> Kokoro: Kokoro <noreply+kokoro@google.com>
TBR=cwallez@chromium.org,enga@chromium.org,noreply+kokoro@google.com,dawn-scoped@luci-project-accounts.iam.gserviceaccount.com,lehoangquyen@chromium.org
Change-Id: Ib13bba8005402d06963865fae919388a91e718f0
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: dawn:1769
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/128440
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
Normal behavior of ApiObjectBase's APIRelease() which only locks the
device when last ref dropped is not thread safe if the object is cached
as raw pointers by the device. Example of cached objects: bind group
layout, pipeline, sampler, shader module.
The following scenario could happen:
- thread A:
- shaderModuleA.APIRealease()
- shaderModuleA.refCount.Decrement() == true (ref count has reached zero)
- going to call shaderModuleA.LockAndDeleteThis().
- thread B:
- device.CreateShaderModule().
- lock()
- device.GetOrCreateShaderModule()
- shaderModuleA is in the cache, so return it.
- unlock()
- thread A:
- starting to call shaderModuleA.LockAndDeleteThis()
- lock()
- erase shaderModuleA from the cache.
- delete shaderModuleA.
- unlock()
This CL disables caching when ImplicitDeviceSynchronization is turned on
until we find a better solution.
Bug: dawn:1769
Change-Id: Ideb2a717ece0a40e18bd1c2bef00817262bd25da
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/127900
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
1. Auto-generate mutex locking code in DawnNative's ProcTable's
functions. Using a mutex owned by the related Device.
- Unless the function/class has "no autolock" attribute (new) in
dawn.json. In which cases, no locking code will be auto-generated.
- Currently Instance, Adapter, Surface, Encoder classes and
Reference/Release method have "no autolock".
2. Added Feature::ImplicitDeviceSynchronization to enable Device's
mutex.
- If this feature is disabled, lock/unlock Device's mutex is no-op.
Auto-generated locking code will have no effect. This is the default.
- This approach is used instead of generating two sets of ProcTable
because it's cleaner and the internal code doesn't need to care about
whether there is a mutex being locked or not. Furthermore, if there
were two sets of ProcTable, and user used dawnProcSetProcs() to set
global ProcTable, it would affect all other modules using different
Devices. Even though those modules don't need thread safety.
Bug: dawn:1662
Change-Id: I75f0d28959f333318e4159916b259131581f79f5
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/119940
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This CL refactor the logic adapter creating device toggles set when
creating device and the way device holding its toggles. This CL also
introduce the concept "toggle stage", currently "device stage" only but
in future will add "instance stage" and "adapter stage" for instance and
adapter toggles. No changes on Dawn API.
More details:
1. Introduce `TogglesState` objects that represent the complete toggles
state of a device (and will used for instance and adapter in future).
2. When creating a device, adapter set up a TogglesState object for it
in `AdapterBase::CreateDeviceInternal` and
`Adapter::SetupBackendDeviceToggles`, no other place would change
the device's toggles state. This change simplify the logic.
3. Introduce the `ToggleStage` enum for every toggle and `TogglesState`
object. Currently we only have `Device` toggle stage, but in future
will have `Instance` and `Adapter` for instance and adapter toggles.
Bug: dawn:1495
Change-Id: Ifafac6a6a075b5b9a733159574ae5b6d4f3ebde9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/118030
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
Add scaffolding for structured Dawn wire fuzzer.
This CL contains a basic fuzzer for Dawn wire server
that shows some simple design ideas:
1) A basic protobuf spec that is generated using dawn.json
2) conversion from protobuf message to a dawn wire server
command.
This is not the complete implementation and serves as a
foundation for the fuzzer so that subsequent CLs will be
easier to review.
Bug: chromium:1374747
Change-Id: Ife1642dda13d01d3308bdd5fe56cf85978399fd3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/109406
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Brendon Tiszka <tiszka@chromium.org>
These were not marked as optional previously because it would mean that
C++ methods would look like void F(a = null, b, c) which is invalid in C++
because default arguments must be contiguous and at the end of the
signature.
This commit adds a special case when optional=true where no_default=true
still marks the argument as optional, but doesn't produce a C++ default
argument.
Fixed: dawn:1502
Change-Id: I1b648d37edb94b8412109fc0a06a91bbbd31b8c4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/97002
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This helps output in the headers information about which structure can
be used to extend which. In the future it could also be used to generate
helpers that validate that the chain for a root structure contains only
allowed extension structs.
Fixed: dawn:1486
Change-Id: I6134332d477503e242b3bec9f8e9bedeeb352351
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/96000
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
The current presubmit has the filter inverted so it would only attempt
to match the filtered files. The file name also has to be converted to
`LocalPath` otherwise it's attempting to compare a python object to a
string and always fails to match.
Bug: dawn:1339
Change-Id: Ie7712dee60f6b9df2cb78c9feab11769f7ea1f02
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87080
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Auto-Submit: Dan Sinclair <dsinclair@chromium.org>
This has already been moved once, but with more consideration, there's far less fuzzer-related stuff to put in the same root directory here.
Bug: dawn:1275
Change-Id: Ic4bb556d9a9f104293cca3316cae84b66a6362b6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/79104
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
