Currently, if we access an array of non-scalar items we'll incorrectly
emit an OpVectorExtractDynamic which will fail as the result is not
scalar.
This CL updates the array accessor code such that if the base array is
an array of non-scalars we'll do load of the array and then access chain
into the loaded variable.
Bug: tint:267
Change-Id: Ia4d7052b57d8b31b835714b7b922c7859e3dce1f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29844
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Commit-Queue: David Neto <dneto@google.com>
Reviewed-by: David Neto <dneto@google.com>
The AST doesn't do detailed checking of numeric parameters.
That's deferred to the separate validation step.
Change-Id: I85aada7292fcc9757f9fd54e2d9a50ad1405ef9c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29940
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
BUG=tint:268
Change-Id: Ie90df82146f4853ef310e32219e5195d36a742db
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29960
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
This CL updates the WGSL parser to allow global variable decorations
to accept multiple blocks.
Bug: tint:240
Change-Id: Iaf8f794d285d87af6e2ff8c93e34d88331eedccb
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29840
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: Sarah Mashayekhi <sarahmashay@google.com>
The push_preamble method was dealing with multiple sections of the
SPIR-V binary layout. As we changed the way things write (like
extensions getting written later) the preamble section was ending up in
incorrect order.
This CL replaces push_preamble with push methods for each of the
sections at the start of the SPIR-V module which should fixup the
ordering issue.
Bug: tint:267
Change-Id: Ib73a66d0fdb2c67dd6e80582289dd18475fad9f9
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29841
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
This CL updates the WGSL parser to allow array decorations to accept
multiple blocks. The stride decoration on arrays was turned into a
proper decoration object instead of just storing the stride directly.
Bug: tint:240
Change-Id: I6cdc7400d8847e3e043b846ea5c9f86cb795cf86
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29780
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: Sarah Mashayekhi <sarahmashay@google.com>
Reviewed-by: David Neto <dneto@google.com>
This CL updates the struct member decoration parsing to allow multiple
blocks of decorations.
Bug: tint:240
Change-Id: I97293ef30333f63c33bbc6e728dba11abc020c7c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29280
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
BUG=tint:257
Change-Id: Iec0dca854dfa6380991c04544848c24f21496a93
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29524
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
This CL updates to match the spec change allowing multiple struct
decorations.
Bug: tint:240
Change-Id: Id859c6a331c67c46597fc3c70de06d6cc0f486ec
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29260
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
This CL adds support for retrieving the array length of a Runtime Array
in the SPIR-V backend.
Bug: tint:252
Change-Id: Ic13c4a99da5760738d57702c45f52c6a194a172d
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29220
Commit-Queue: David Neto <dneto@google.com>
Reviewed-by: David Neto <dneto@google.com>
This CL adds specialization to the HLSL backend. The specialization is
emulated using defined preprocessor macros.
Bug: tint:154
Change-Id: I73ab42360558967eee9a0da6bbe74d513fe4cc57
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29720
Commit-Queue: David Neto <dneto@google.com>
Reviewed-by: David Neto <dneto@google.com>
This CL adds the inspector into the architecture document.
Change-Id: I17f06e0ae29398a39b5082b0b4399c9c5e65769f
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29800
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
This is to match the upstream WebGPU spec.
Bug: dawn:445
Change-Id: I1a511ed9a2a04c7b95368ce724d69c128158f097
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29360
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This CL moves spirv-tools so that it will only build what is requested.
This means that the tool binaries will not be built as part of tint as
they aren't needed.
Change-Id: Idd89b45facbe835948c57fe6dda6068657a2f822
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29660
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
The upstream Tint issues have been fixed and Tint has rolled. This CL
re-enables the disabled test.
Bug: tint:263
Change-Id: Id8a73e198c15c74c55dc9b26603f4a3850d995b3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29680
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This is to match the upstream WebGPU spec.
Bug: dawn:445
Change-Id: I4246487247fdba8d90a119c1970d6d4df3235835
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29361
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This can be used to simplify some logic in Blink.
Bug: dawn:445
Change-Id: I9859c51bc95f564847035533426675188eb8ef99
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29362
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Currently if a constructor contains constructors we consider it const.
This falls down with the new type constructor syntax if the types don't
match. In the case they don't match we no longer consider the
constructor const as we'll generate OpBitcast and OpCopyObject
instructions which we need to build the composite from.
Bug: tint: 263
Change-Id: Ic85f58c8410e862a2ec30c7d93c9b87a61822f6d
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29523
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
All of the OpEntryPoint declarations must come before OpExecutionMode.
Currently if you have multiple fragment shaders we'll interleave the
OpEntryPoint and OpExeutionMode which will fail to validate.
Bug: tint:263
Change-Id: I7c925cf6b5345c03bfaf1aa15115caa1bdb9af4c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29522
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
This CL adds the code to emit function_constant attributes to the
generated MSL.
Bug: tint:153
Change-Id: I0e4f50257fde7e8db8f53e15bc9f460ebfc809ee
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29520
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Add some missing doxygen comments
Change-Id: Ic5d6684d90f5b2fe9ed3914bbb7c4058effc040c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29500
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
BUG=tint:257
Change-Id: Iaf03bfaeb622b7315d65e46eccfe90244bced339
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29420
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
This is a reland of b04a92f01be0e7dc2faf91e0f6f974f990e8b2f2
with the deletion of a duplicate exported function in dawn_wire that
was causing a compilation failure on Windows.
Original change's description:
> Add a per-thread proc table using thread local storage
>
> In situations where both dawn_wire and dawn_native are used on separate
> threads (Chrome with --single-process or --in-process-gpu), it's
> desirable to have a per-thread proc table so that the WebGPU C++ API can
> still be used. This eliminates classes of bugs with manual
> reference/release errors.
>
> This also changes many of the GetProcs functions to return const
> references to the static proc tables known at compile time, instead of a
> copy.
>
> Bug: none
> Change-Id: I8775bb715b312dd9476a1903fbd797d4b1302614
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29240
> Reviewed-by: Stephen White <senorblanco@chromium.org>
> Reviewed-by: Corentin Wallez <cwallez@chromium.org>
> Commit-Queue: Austin Eng <enga@chromium.org>
Bug: none
Change-Id: Id90e5372132cd93a2f8631c8185d0e71b01bc1af
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29443
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
This CL suppresses some Doxygen warnings for code doxygen doesn't
process correctly and adds the missing return for GenerateEntryPoint
method.
Change-Id: If97443a7177caa51c1054de83fb0711692a7ab22
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29461
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
This CL moves the visibility=hidden directive from the top level applied
flags to libtint specifically. This fixes up an issue between
SPIRV-Tools and the sample app when using std::rfind which ends up with
different visibility.
Change-Id: Ib06949b3755db66027d3656d3d6ce2504bfda81e
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29460
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
The OSX machines in "luci.flex.try" pool that run Dawn tryjobs
have been upgraded from OSX 10.14 to OSX 10.15. Per [1],
the preferred Xcode version on OSX 10.15 is '12a7209' and most
tryjobs (from other projects) install that version. But when
a Dawn tryjob runs, it rolls Xcode back to '11b52'. Reinstalling Xcode
back and forth like that takes time (~20 min), it caused some tryjobs
in the "infra" project to timeout.
Unhardcode Xcode version in Dawn configs. It should allow the recipe
to pick the "best" version ('12a7209' in this case).
[1]: https://source.chromium.org/chromium/chromium/tools/depot_tools/+/master:recipes/recipe_modules/osx_sdk/api.py;l=21?q=11b52&ss=chromiumR=cwallez@chromium.org, tandrii@chromium.org
Change-Id: I8499a763679b9e9270854bf66c2a21f1e6354594
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29440
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This reverts commit b04a92f01be0e7dc2faf91e0f6f974f990e8b2f2.
Reason for revert: Build failed on win-archive-dbg; see:
https://chromium-review.googlesource.com/c/chromium/src/+/2450792
Original change's description:
> Add a per-thread proc table using thread local storage
>
> In situations where both dawn_wire and dawn_native are used on separate
> threads (Chrome with --single-process or --in-process-gpu), it's
> desirable to have a per-thread proc table so that the WebGPU C++ API can
> still be used. This eliminates classes of bugs with manual
> reference/release errors.
>
> This also changes many of the GetProcs functions to return const
> references to the static proc tables known at compile time, instead of a
> copy.
>
> Bug: none
> Change-Id: I8775bb715b312dd9476a1903fbd797d4b1302614
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29240
> Reviewed-by: Stephen White <senorblanco@chromium.org>
> Reviewed-by: Corentin Wallez <cwallez@chromium.org>
> Commit-Queue: Austin Eng <enga@chromium.org>
TBR=cwallez@chromium.org,senorblanco@chromium.org,enga@chromium.org
Change-Id: I4587b457b7b2dd5d3c7457065bf4e77b95af59d7
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: none
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29442
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
This CL makes the AST type format name match the WGSL name so the
overloaded operator<< can be used in the WGSL writer. This removes a
bunch of duplication of WGSL type format name strings.
Bug: tint:230
Change-Id: I14aaefd21ced267ceed31f21faba3bd291275f0d
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29402
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
In situations where both dawn_wire and dawn_native are used on separate
threads (Chrome with --single-process or --in-process-gpu), it's
desirable to have a per-thread proc table so that the WebGPU C++ API can
still be used. This eliminates classes of bugs with manual
reference/release errors.
This also changes many of the GetProcs functions to return const
references to the static proc tables known at compile time, instead of a
copy.
Bug: none
Change-Id: I8775bb715b312dd9476a1903fbd797d4b1302614
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29240
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
This Cl adds countOneBits and reverseBits intrinsic support to the MSL
and HLSL backends.
Bug: tint:250
Change-Id: Idb572da32e9c10767a27326cacf58aafef44a77a
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/29401
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
