Commit Graph

154 Commits

Author SHA1 Message Date
steve-lunarg
265c0618b1 HLSL: allow implicit array sizing.
In HLSL array sizes need not be provided explicitly in all circumstances.
For example, this is valid (note no number between the [ ]):

  // no explicit array size
  uniform float g_array[] = { 1, 2, 3, 4, 5 };

This PR does not attempt to validate most invalid cases.

A new test is added to verify the resulting linker objects.
2016-09-27 14:28:26 -06:00
John Kessenich
10f7fc739c HLSL: Reverse what the driver is told about row/column majorness, matching the row-column reversal. 2016-09-25 20:26:03 -06:00
steve-lunarg
cf43e66125 Fix defects in uniform array flattening
Fix for two defects as follows:

- The IO mapping traverser was not setting inVisit, and would skip some AST nodes.
  Depending on the order of nodes, this could have prevented the binding from
  showing up in the generated SPIR-V.

- If a uniform array was flattened, each of the flattened scalars from the array
  is still a (now-scalar) uniform.  It was being converted to a temporary.
2016-09-22 15:58:06 -06:00
steve-lunarg
e0b9debda2 Flatten uniform arrays
This checkin adds a --flatten-uniform-arrays option which can break
uniform arrays of samplers, textures, or UBOs up into individual
scalars named (e.g) myarray[0], myarray[1], etc.  These appear as
individual linkage objects.

Code notes:

- shouldFlatten internally calls shouldFlattenIO, and shouldFlattenUniform,
  but is the only flattening query directly called.

- flattenVariable will handle structs or arrays (but not yet arrayed structs;
  this is tested an an error is generated).

- There's some error checking around unhandled situations.  E.g, flattening
  uniform arrays with initializer lists is not implemented.

- This piggybacks on as much of the existing mechanism for struct flattening
  as it can.  E.g, it uses the same flattenMap, and the same
  flattenAccess() method.

- handleAssign() has been generalized to cope with either structs or arrays.

- Extended test infrastructure to test flattening ability.
2016-09-22 08:47:48 -06:00
John Kessenich
6714bcc2ca HLSL: Fix result type of passing a flattened-aggregate to a function. 2016-09-21 17:50:12 -06:00
John Kessenich
a1e2d4952e HLSL: Move to correct parsing of annotations, improving all annotations and recent string grammar. 2016-09-20 13:22:58 -06:00
John Kessenich
2572b19e94 HLSL: Turn on reflection, with basic test file, to catch regressions.
This is not a claim that reflection is working right, only a way to
see it is occurring and catch any regression.
2016-09-19 23:12:48 -06:00
John Kessenich
6b71c400f8 HLSL: Remove extraneous built-in member decorations for IO structs used in non-IO situations. 2016-09-19 22:16:09 -06:00
John Kessenich
86f7138706 HLSL: Add string basic type and recognize string declaration grammar.
This includes the "< decl ; decl ; >" syntax which has its own namespace.
This functionality is not implemented, just silently accepted.
2016-09-19 20:29:45 -06:00
John Kessenich
eee9d536bc Track separate entry-point names and mangled names...
... and use each in the correct way at consumption sites.
This completes issue #513.
2016-09-19 18:09:30 -06:00
John Kessenich
6fccb3cd75 Non-functional: Sweep through the stack for consistent with "main" and entry point.
Partially addresses issue #513.
2016-09-19 16:01:41 -06:00
John Kessenich
142785f324 HLSL: Change the final syntax-error printf to go to the infoLog.
Fixes issue #510.
2016-09-19 14:56:55 -06:00
John Kessenich
28b28140bb HLSL: Fix assert: ensure flattened shadow is EvqTemporary. 2016-09-19 00:19:49 -06:00
John Kessenich
5159d4f1af HLSL: Intercept flatten aggregates passed to a function input, and copy member-by-member. 2016-09-19 00:06:19 -06:00
John Kessenich
f911500db8 HLSL: Non-functional; make flatten semantics be about aggregates, not just structures. 2016-09-18 23:36:39 -06:00
John Kessenich
fcea302dbc HLSL: Fix bug in previous checkin when non-flattened objects are not simple l-values. 2016-09-16 21:16:04 -06:00
John Kessenich
d2ce838a58 HLSL: Handle flatten for reads from flatten structs and parameter passing. 2016-09-16 20:24:14 -06:00
John Kessenich
34e7ee79bb HLSL: Improve setting and testing of interpolation qualifiers.
Notably, use of 'linear' on a non-input could mark it as an input.
2016-09-16 18:05:44 -06:00
John Kessenich
d21baed6bc HLSL: Flatten whole-struct assigns and returns when targeting flattened I/O structs. 2016-09-16 03:20:03 -06:00
John Kessenich
f8e494c18c HLSL: Flatten all input/output structs, regardless of stage.
This is needed because an output structure can contain embedded built-ins
(like SV_Position) which should not get locations assigned.
2016-09-16 01:52:14 -06:00
John Kessenich
7dc630f3da HLSL: Flatten a return struct from an entry point and assign locations after flattening.
Locations now get assigned in order, but skipping built-ins, which can be
done post flattening.
2016-09-16 01:44:43 -06:00
John Kessenich
7f702124ec HLSL: return correct error when HLSL parsing fails.
At least partially addresses issue #510.
2016-09-15 22:49:31 -06:00
John Kessenich
deb4940c17 HLSL: Register all entry-point in/out as part of the interface.
This makes the interface be invariant, whether or not individual
variables are used.
2016-09-12 11:55:47 -06:00
John Kessenich
cd0a78a0d9 HLSL: Flatten vertex input and fragment output structures.
Vulkan can't handle structures into the vertex stage or out
of the fragment stage.
2016-09-10 11:09:24 -06:00
John Kessenich
6295c27900 Merge pull request #505 from steve-lunarg/rowmajor-fix-2a
HLSL: alias HLSL matrix-row-column onto AST matrix-column-row
2016-09-09 14:00:27 -06:00
steve-lunarg
297ae211f1 WIP: HLSL: Treat HLSL rows as GLSL columns.
WIP: HLSL: EOpGenMul arg reversal
2016-09-09 12:02:42 -06:00
John Kessenich
d4032293ce HLSL: Report an error if SPIR-V for Vulkan wasn't selected. 2016-09-09 11:43:11 -06:00
John Kessenich
cfd7ce87cd HLSL: Support register(..., spaceN) for setting the descriptor set.
This was suggested in issue #454.
2016-09-05 16:03:45 -06:00
John Kessenich
e3218e270e HLSL: Accept layout(...) also as a post-decl. Issue #454. 2016-09-05 14:37:03 -06:00
John Kessenich
7735b94403 HLSL Non-Functional: Move to more robust capturing of postDecls into a qualifier.
This will prevent a possible future defect of thinking the type can be changed,
where there is a code path today that would drop that change.
2016-09-05 12:40:06 -06:00
John Kessenich
b804de605c HLSL: Track binding numbers to struct instances; fixes issue #496. 2016-09-05 12:19:18 -06:00
John Kessenich
7d01bd6f0b HLSL: Handle swizzles on vectors of size 1. Addresses issue #453. 2016-09-02 22:21:25 -06:00
John Kessenich
841db35bb3 HLSL: Fix issue #442, smear and truncate shape conversions for == and !=. 2016-09-02 21:12:23 -06:00
John Kessenich
07350f3382 HLSL: Handle "fake" entry points, by undoing their built-in variable declarations. 2016-09-02 20:24:07 -06:00
John Kessenich
9e079535a0 HLSL: Handle greater/less depth modes. Fixes issue #489. 2016-09-02 20:05:52 -06:00
John Kessenich
a305166ea4 HLSL: Error if funcion with return type doesn't return a value. 2016-09-02 19:13:36 -06:00
John Kessenich
1a4b775cd5 HLSL: Correct line numbers for function definitions. 2016-09-02 19:05:24 -06:00
John Kessenich
5e56423046 Front-ends: Remove now defunct afterEOF and related, use scanner's instead.
Code using atEndOfFile was dead, instead do something useful with
the scanners atEndOfInput().  This allows a better error message
for early termination of cascading errors.
2016-08-31 13:46:50 -06:00
John Kessenich
830b0cc98b HLSL: Start location numbering with the entry-point return value.
Also, increment location numbers by the size of the objects.
2016-08-29 18:10:47 -06:00
John Kessenich
a05d8b5604 HLSL: Remove recent change to put locations on SV_TARGET*.
This put locations on members of structures, which is not allowed
in either AST or SPIR-V.

This was caught by asserts in the debug build.
2016-08-29 16:49:39 -06:00
John Kessenich
81d4714908 Merge branch 'HLSL_Semantic_Mapping' of https://github.com/dankbaker/glslang into dankbaker-HLSL_Semantic_Mapping 2016-08-29 16:07:29 -06:00
Dan Baker
6f220c0fd1 HLSL: Setting SV_DEPTHGREATEREQUAL and SV_DEPTHLESSEQUAL to EbvFragDepth for now 2016-08-29 15:56:55 -04:00
John Kessenich
6a70eb7161 HLSL: Emulate write-to-output on return-from-entry-point, for return value.
This fixes issue #487 and #480.
It also correctly handles output parameters from the entry point.
2016-08-28 20:13:07 -06:00
John Kessenich
81cd764b5f Non-functional: Add some missing const, related to signature selection. 2016-08-26 14:01:43 -06:00
John Kessenich
e3f2c8f98a HLSL: Include shape-changing conversions in overloaded signature selection.
This also enables vecN -> vec1 shape conversions for all places doing shape
conversions.

For signature selection, makes shape changes worse than any other comparison
when deciding what conversions are better than others.
2016-08-25 23:57:39 -06:00
Dan Baker
26aa8a4b16 HLSL: Format updates and some minor adjustments to SV_ handling 2016-08-25 17:13:25 -04:00
John Kessenich
90dd70f752 HLSL: Allow arbitrary baseType -> baseType conversion of calling arguments.
This also puts a stake in the ground as to which is better when selection
from multiple signatures.
2016-08-25 10:51:29 -06:00
Dan Baker
deec03cfca First stab at system value interpretation 2016-08-25 12:00:25 -04:00
steve-lunarg
36e87d0871 HLSL: add precise keyword 2016-08-25 08:48:54 -06:00
John Kessenich
fcc0aa3b64 HLSL: Switch to generic selector, but using GLSL #version 400 rules.
Next step is to modify for HLSL rules.
2016-08-24 18:34:43 -06:00