If the function has no return statements, we need to do this to
capture the potential non-uniformity coming from nested function
calls.
This also removes the need to add explicit edges for discard
statements, so remove them.
Bug: tint:880
Change-Id: I88b1132faf35a6d36460ef353912f77a15f8abaa
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89861
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Chromium tracks the devices which live on the wire so it can
automatically call tick on devices that have pending work. This used
to be done by querying an (id, generation) pair and checking if it
resolves to a non-null device.
This CL adds a new way to query directly using the device, since a
refactor in Chrome will change creation such that the id and generation
is not known when a new device is requested.
Also fixes a bug for device callbacks where the required callback
userdata wasn't fully populated for devices created with
requestAdapter. Update a test to check for this as well.
Bug: chromium:1315260
Change-Id: I7468edc3e77bade191e1e9f3eaadebbf4441d88a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89520
Reviewed-by: Loko Kung <lokokung@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Austin Eng <enga@chromium.org>
When producing an error from the uniformity analysis, add notes to
show the underlying reason for the uniformity requirement.
For function calls that are required-to-be-uniform, show the innermost
builtin call that has the requirement.
For function parameters that are required-to-be-uniform, recurse into
that function to show where its requirement comes from.
Add some new tests to specifically test the error messages.
Bug: tint:880
Change-Id: Ib166fdeceaffb156a3afc50ebc5a4ad0860dc002
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89722
Reviewed-by: Ben Clayton <bclayton@google.com>
Move the control flow graphs to `FunctionInfo` so that they are not
destroyed when we finish processing the function.
This will make it easier to produce more detailed diagnostics for
uniformity issues, which will be coming in subsequent CLs.
There is no functional change in this CL.
Bug: tint:880
Change-Id: I1322fb54b16bd1c660799a62435fbdcd7fb39cb2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89822
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
This implements the uniformity analysis as currently described in the
WGSL specification. Uniformity issues are presented as warnings, and
will be switched to errors in a future release.
A follow-up patch will improve the error messages, which currently
just show the point at which a uniformity was detected.
In a future release, once we have obtained initial feedback from
users, uniformity issues will become errors.
Bug: tint:880
Change-Id: I7d0b3080932c786c5d50b55720fec6d19f00d356
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88368
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Alan Baker <alanbaker@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
There is currently a race condition in the metal backend with the
updating of `mCompletedSerial`. It is currently possible for the
`addCompletedHandler` to set to one value and then have the
`CheckAndUpdateCompletedSerial` call immediately set it back to
a lower value. This can then cause hangs as the serial never
correctly updates again.
This was happening on `dawn.node` when running a large number of
CTS test cases all at the same time.
Change-Id: I28fc58ab2b3737ca8039559718e539ce819e88bb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89780
Auto-Submit: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Otherwise, calling `Sem().Get()` on an empty loop continuing block
will return nullptr, which causes issues when inspecting the semantic
information of ast::BlockStatement nodes generically.
Change-Id: Ib3665b750c96eda02355fa879cf6300b8d69293a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89721
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
This is done via a new extension, which in the future could also be
used by shader authors as an escape hatch while we are still refining
the analysis.
The transform is run by the sanitizers for all of the non-WGSL
backends.
Bug: tint:880
Change-Id: Ibe90d7437d34c741a91eda65dff6d21d8469b9c7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88464
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Unsuffixed float literals are currently treated as f32,
but will shortly become AbstractFloat. To keep tests behaving
identically to how they are currently, change all float literals
to explicitly use the f32 '_f' suffix.
Bug: tint:1504
Change-Id: I2a00725ee1b34a6efbe15ac4ba438c00c4416dd8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89402
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Generate different tokens for:
• 'f' suffixed float literals
• unsuffixed integer literals
'f' and unsuffixed are currently both treated as f32 by the resolver,
but this is the first step to supporting abstract floats.
Bug: tint:1504
Change-Id: Id3b1fe420b6eb8901f88d6a5de06ef4f54aa3edf
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89031
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
• Declare all the unary ops in the intrinsics.def file.
• Reimplement the bulk of Resolver::UnaryOp() with the IntrinsicTable.
This will simplify maintenance of the operators, and will greatly
simplify the [AbstractInt -> i32|u32] [AbstractFloat -> f32|f16] logic.
Bug: tint:1504
Change-Id: Ifc646d086fc93cfbe3f3f861b8c447178664c1f7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89028
Reviewed-by: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
• Declare all the binary ops in the intrinsics.def file.
• Reimplement Resolver::BinaryOpType() with the IntrinsicTable.
This will simplify maintenance of the operators, and will greatly
simplify the [AbstractInt -> i32|u32] [AbstractFloat -> f32|f16] logic.
Bug: tint:1504
Change-Id: Ie028602e05b59916c3f2168c92f200f10e402b96
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89027
Reviewed-by: James Price <jrprice@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Adapt the builtin parsing and resolving to also support operators.
Will be used to generate intrinsic table entries for operators.
This will simplify maintenance of the operators, and will greatly
simplify the [AbstractInt -> i32|u32] [AbstractFloat -> f32|f16] logic.
Bug: tint:1504
Change-Id: Id75735ea24e501877418812185796f3fba88a521
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89026
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
• Rename 'builtin-gen' back to 'intrinsic-gen', as 'intrinsics' now
include both builtins and operators.
• Move the intrinsic definitions, and IntrinsicTable to the resolver
package, where it belongs.
Bug: tint:1504
Change-Id: I5ad5c285c1e360a224ee1235c293ccd55eb2693d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89025
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Tint has already implemented the enable directive for using WGSL
extension in the future, and using a WGSL extension that is not allowed
for the device should result in a shader creation error.
In this patch a WGSL extension allow list is added in DeviceBase, and
a validation is added in shader module base initialization to make sure
all extensions used in the WGSL program are in the allow list. This
patch also rename the `ValidateShaderModuleDescriptor` to
`ValidateAndParseShaderModule`, which is more descriptive for what it
actually does.
Bug: tint:1472
Change-Id: I4b039a3e37c25159b4fc6cfa37488aa817004ab2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88241
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This change ensures that when errors are raised from WebGPU, including
lost context errors, they can include any associated Vulkan validation
layer messages in the message text if backend validation is enabled.
This will allow these messages to be surfaced in places like when
running CTS tests on browser build infrastructure.
Because of how Vulkan messages are raised at the instance level, some
routing of the messages to the appropriate device needed to be added.
This is based on appending message IDs to object labels and then parsing
them out of the message when it comes in.
This change also makes Vulkan validation messages cause a device loss,
which is inline with the behavior of the D3D12 backend.
Bug: dawn:1396
Change-Id: I2b309ba5475a338e7e99d4b4daaa7c3d5de61011
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88460
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Brandon Jones <bajones@chromium.org>
Unsuffixed integer literals are currently treated as i32,
but will shortly become AbstractInteger. To keep tests behaving
identically to how they are currently, change all test literals
to using either 'i' or 'u' suffixes.
Bug: tint:1504
Change-Id: Ic373d18ce1c718a16b6905568aec89da3641d36b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88845
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
