Commit Graph
489 Commits
Author SHA1 Message Date
Igor Shaposhnik 5f8e54bbfa [glsl-out] Add stage postfix to generated global name (#1616) 2021-12-29 09:55:08 -05:00
Jay Oster 548ce66ed4 Fixes for fma function (#1580)
* [hlsl-out] Write `mad` intrinsic for `fma` function

- This should be enough because we only support f32 for now.
- Adds a new test for WGSL functions, in the spirit of operators.wgsl.
- Closes #1579

* Add FMA feature to glsl backend

- I think this is right. Just iterate all known expressions in all
  functions and entry points to locate any `fma` function call.
  Should not need to walk the statement DAG.

* Transform GLSL fma function into an airthmetic expression when necessary

* Add tests for GLSL fma function tranformation

* Remove the hazard comment from the webgl test input

* Add helper method for fma function support checks

* Address review comment
2021-12-29 09:54:33 -05:00
Dzmitry Malyshau c2328fea75 Release v0.8 2021-12-18 12:05:12 -05:00
Dzmitry Malyshau 8caa2bd87e Gather operations 2021-12-18 11:46:49 -05:00
Jim Blandy db98743a5a [glsl-out] Don't return early from snapshot tests. 2021-12-17 20:27:27 -05:00
Ashley 75ce45d4f3 spv-in: fix support for degrees and radians conversion 2021-12-17 11:56:47 -05:00
João Capucho e1552fd9d5 wgsl-in: Implement matrix construction from scalar 2021-12-17 10:39:31 -05:00
Dzmitry Malyshau 6df9656e96 spv-out: Force wrapping of push constant structs 2021-12-16 16:21:06 -05:00
João Capucho 9c75f3cdce glsl-in: Flatten entry point arguments
The IR doesn't allow having structs has bindings for entry point
input/output, so the glsl frontend must flatten them.

Glsl defines that the locations are sequential from the block base location.
2021-12-16 10:55:06 -05:00
João Capucho 4334f0e6a9 Support bitwise And on booleans 2021-12-15 23:53:23 -05:00
João Capucho c64d5eff50 Support bitwise Or on booleans 2021-12-15 23:53:23 -05:00
Dzmitry Malyshau 8ffd6ba929 Remove top_level property of structs 2021-12-15 09:33:33 -05:00
Jim Blandy 3867ef4f6c [msl-out] Correctly construct out-of-bounds values for all types. 2021-12-14 23:00:03 -05:00
Lain-dono e200e44ede [spv-out] Fixing incorrect pack/unpack (#1587)
* [spv-out] Fixing incorrect pack/unpack

* [spv-out] Fixing incorrect pack/unpack: make tests happy

* [spv-out] Fixing incorrect pack/unpack: remove unexpected newline
2021-12-11 21:29:31 -05:00
Jim Blandy f51f468a47 [msl-out] Implement index bounds check policies for non-texture accesses. 2021-12-06 15:53:41 -05:00
João Capucho 14719553a6 glsl-in: fix pre/postfix ops on swizzles 2021-12-05 22:16:28 -05:00
Dzmitry Malyshau 6fbd668bd3 Add missing GLSL tests 2021-12-02 13:11:23 -05:00
João Capucho 9fac7c48a6 glsl-in: Allow vector/matrix in vector constructor
The glsl spec defines that vector constructors flatten their arguments
and consume them in order, ignoring the remaining arguments if there are
more than needed
2021-11-29 11:06:18 -05:00
Igor Shaposhnik dba6beb941 [wgsl-in] Don't allow redefinition of module scope identifiers 2021-11-28 21:23:22 -05:00
Igor Shaposhnik 20b96ef47b Avoid function name redefinition in tests 2021-11-28 21:23:22 -05:00
Dzmitry Malyshau daa61b5fcb Fix shapshots to run without deserialize feature 2021-11-26 12:16:50 -05:00
Dzmitry Malyshau 74132ba947 Actually disable explicit types on most WGSL outputs 2021-11-26 11:36:46 -05:00
Dzmitry Malyshau f4ba74544a Refactor snapshot test arguments to avoid checking for deserialize feature too often.
We can safely assume that deserialize is enabled for all output tests.
Supporting other modes of operation has low utility.

This change also adds an option to skip the explicit types for WGSL output.
2021-11-26 11:36:46 -05:00
Dzmitry Malyshau ddbc1c4b1f WGSL: use underscore prefix like the other backends 2021-11-25 16:23:24 -05:00
Igor Shaposhnik 2107b20561 [wgsl-in] Don't allow keywords to be used as identifiers 2021-11-24 16:34:17 -05:00
Jim Blandy cc930180d9 Tighten checks on pointers to unsized values, and pointer arguments.
Pointers should not be `DATA`: they can never be stored in anything. (Function
arguments are not storage; they're like `let` bindings.)

Un-`SIZED` values may only live in the `Storage` storage class, so creating
pointers to them in other storage classes is meaningless.

The `ARGUMENT` flag should be set only on pointers in those storage classes that
are permitted to be passed to functions.

See comments in code for details.

Fixes #1513.
2021-11-18 17:39:56 -05:00
Jim Blandy 633f1b3377 [wgsl-in] Don't be confused by pointers to globals.
The WGSL front end knows that globals in the `Handle` storage class do not
produce references, but it attempts to manage this in two places which trip over
each other. As a consequence, referring to a `let` binding holding a pointer to
a global flips the variable's type from WGSL `ptr` to WGSL reference, with
bewildering consequences.
2021-11-18 17:39:56 -05:00
Jim Blandy 6a5b33fa24 Move BoundsCheckPolicy/Policies into proc, from back. (#1537) 2021-11-17 19:15:59 -08:00
Igor Shaposhnik 30a0a37af0 Update dependencies 2021-11-16 23:46:54 -05:00
Jim Blandy ce18eba695 Check that stores are permitted by the pointer's storage access.
Fixes #1533.
2021-11-16 23:40:56 -05:00
Boris-Chengbiao Zhou 1dcde48d09 [valid] Check that switches have a default case (#1529)
From the WGSL spec: "Each switch statement must have exactly one default clause."
2021-11-15 00:07:55 +00:00
Jay Oster 66c3499df8 [spv-out] Fix scalar-times-matrix operations (#1524) 2021-11-14 19:07:38 -05:00
Jim Blandy e69a70bfb7 [spv-in] Permit pointers to runtime arrays only in StorageBuffer. (#1522)
* [spv-in] Change shadow.spv test input to use StorageBuffer.

The ecosystem around Naga will generally not be able to use Vulkan adapters that
don't support the SPV_KHR_storage_buffer_storage_class (which was incorporated
into SPIR-V 1.3), so we can assume it is present.

Changing the test not to use runtime-sized arrays in the Uniform storage class
will allow Naga to tighten up some validation checks.

* [spv-in] Permit pointers to runtime arrays only in StorageBuffer.

Fixes #1519.
2021-11-10 16:58:57 -05:00
João Capucho d168f15d81 [glsl-in] Allow constructors from non-scalar to scalar (#1520) 2021-11-10 12:15:43 -05:00
João Capucho c3c4ab7548 [glsl-in] Add test for matrix from matrix constructor 2021-11-09 15:15:11 +00:00
João Capucho 7234d19d00 [glsl-in] Implement matrix to matrix constructor 2021-11-09 15:15:11 +00:00
Dzmitry Malyshau 8bc3aa824b Refactor namer scheme, fix trailing digits (#1510) 2021-11-09 09:24:41 -05:00
Dzmitry Malyshau 28c45321e5 hlsl: respect array stride in storage buffers (#1507) 2021-11-03 22:44:46 -04:00
Jim Blandy c1b378842e Make tests properly exercise Some("") global names. (#1503)
The GLSL empty-global-name.frag test doesn't suffice because the GLSL front end
doesn't produce the same IR as the SPIR-V included in the bug report. As far as
I know, only a genuine SPIR-V input test can produce a global whose name is
`Some("")`.

Include the SPIR-V assembly source.
2021-11-01 14:07:11 -04:00
Igor Shaposhnik d1cfdd15a0 Add test for last falltrough switch case 2021-10-27 11:12:35 +01:00
Igor Shaposhnik 00bbbed90a Handle empty variable names in namer (#1484)
* Handle empty variable names in namer

* Add glsl-in test with empty global name
2021-10-26 17:46:27 -04:00
João Capucho 63dbd38edc Make default a switch case (#1477)
* Make default a switch case

Previously the default case of a switch statement was encoded as a block
in the statement but the wgsl spec defines it in such a way that the
default case ordering matters.

* [spv-out] Support for the new switch IR

* [dot-out] Use different labels for default cases
2021-10-26 13:31:54 -04:00
Alex Es ec001c3ead Use span information in analyzer errors (proof of concept) (#1470)
* Proof-of-concept for adding spans to validation errors.

Still missing: actually printing the damn stuff.

* Emit errors from analyzer in the CLI.

TODO: tests, I guess!

* Simplification refactoring: avoid avoiding allocations so vehemently.

* Mask helper traits with `as _`.

* Fix block iterator throwing up when span feature is disabled.

* Nest use statements.

* Add basic docs.

* Axe AddSpanResult.
2021-10-24 22:47:03 -04:00
Dzmitry Malyshau fd5085277a Switch the constant in bevy-pbr to avoid display incosistencies 2021-10-23 23:39:49 -04:00
João Capucho c3a0d71f8b [spv-out] Write function arguments names (#1472) 2021-10-23 23:20:12 -04:00
João Capucho 70ad084ad4 [spv-in]: Cull unused builtins inside structs (#1469)
Some compilers like shaderc introduce a full gl_PerVertex struct, this
includes gl_ClipDistance. Normally this isn't a problem since most
drivers optimize it away, but naga zero inits globals if they weren't
previously initialized. This causes gl_ClipDistance to be initialized to
zero which can be really bad for performance.
2021-10-19 22:55:20 -04:00
Jim Blandy ed4cc8e4d8 Use Cow to reduce allocation in proc::Namer. (#1465)
Minimize allocation in the namer. Heap-allocate string only when needed to
provide an owned key for a hash table, or to hold synthesized text. Try to reuse
allocations.

Many unnecessary trailing `_` characters are removed from test output. These
were all superfluous; separators are still inserted where necessary.
2021-10-16 17:56:59 +00:00
Dzmitry Malyshau ee450c1ee4 Fix float-bool casts in MSL, SPV, and HLSL backends (#1459) 2021-10-12 11:42:20 -04:00
Igor Shaposhnik 943235cd5e [glsl-out] Convert modulo operator on float to SPIR-V OpFRem equivalent function (#1452) 2021-10-07 16:59:39 -04:00
Jim Blandy 0e3fbc8166 Replace pointer-access.spv snapshot test with WGSL source. (#1450)
The original pointer access test used SPIR-V for its input because WGSL didn't
have a working pointer indirection operator at the time. Now that it does, we
can just write this test in WGSL directly.

Fixes #1432.
2021-10-06 16:30:36 -04:00