|
eccfec93
|
2024-11-01T11:49:11
|
|
Metal: gl_ClipDistance fails validation
Failure is "Found child with two parents".
Result from reusing gl_ClipDistance in:
float gl_ClipDistance[8];
ClipDistance_0 = gl_ClipDistance[0];
...
Fix by creating new node for each symbol access.
Bug: angleproject:376718272
Change-Id: I7a866a10e389cc25d0cff4e6f18ae9c70ab376e0
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5979781
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
|
|
|
b03f0148
|
2024-11-01T10:26:04
|
|
Metal: interpolateAtOffset fails validation
Fix validation error Found function prototype with an invalid qualifier.
The offset parameter was created with "global" qualifier.
Bug: angleproject:376718268
Change-Id: I6f314a42fe1d35d54bb673cfa7d58eba526c9ea5
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5979778
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
2156cd6e
|
2024-10-31T14:22:31
|
|
Metal: Fix rewritten array variables clashes
It was possible to generate MSL name clashes by declaring arrays,
such as in:
mat3 a[1]
mat2 a_0;
Complex GLSL outputs need to be linearized into primitive MSL types.
This would happen for struct outputs, arrays and matrices. For
these new MSL variables, the translator needs to invent new names
that match, generated from both VS and FS.
The clashes were due to the encoding scheme:
For arrays, the MSL vertex output variable was named with _0 suffix for
the array and _0 for the matrix row 0. In the example above, this
would create a clash since also non-array mat would be linearized
row-wise.
For structs, field `a.b` would be encoded as `a_b`. By selecting
different a and b, clashes could be generated trivially.
It is redundant to encode named semantics in the out variable names. The
only needed element is the discriminator which associates VS variables
to FS variables. Currently this is done by using the user-provided
root field name.
Fix by encoding only the root variable name to the MSL variable name.
The rest of the field discriminators are just a running number.
Each GLSL name `a` is unique. MSL names get fixed suffix form `a_X`,
and thus all of them are unique too.
This is a continuation of following fixes that turned out to be
incomplete:
6fe8a399dd Metal: Fix rewritten out variables with underscores
27423bffff Metal: Generate names for rewritten inputs
Bug: angleproject:376417347
Change-Id: I407db373d201b3e321c6d8414bafdbecfd82cf19
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5979774
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
|
|
|
4397ff2f
|
2024-10-25T16:55:30
|
|
Metal: SeparateCompoundStructDeclarations fails validation
Consider GLSL:
struct S { int i; } s;
s=s;
SeparateCompoundStructDeclarations would rewrite this to:
struct S { int i; };
S s';
s=s;
The interm rewrite would rewrite the specification and declaration of s,
but not the use sites. The use sites would use the old type, and thus
something that was not in the tree anymore. This would fail the
validation.
This kind of bug was previously fixed for SeparateDeclarations
in commit 18fa02bebf901dd8501de3176f6052ae4ce984be.
Fix by adding the logic to SeparateDeclarations, as it is already
doing almost the exact task, separating `struct S { ..} a, b`.
The separation is tested in GLSLTests.StructInShader and various
other draw tests. These pass with MSL, but these would also fail
validation if that was enabled.
Bug: angleproject:375523825
Change-Id: I1697103d0ba47616dbd3159f36f9e71cb2831c4b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5964899
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
898a1c12
|
2024-10-24T13:09:36
|
|
Metal: Fix ToposortStructs validation == failures
ToposortStructs would convert mat, vec, struct == to ANGLE_equals()
calls. However, the functions called were not in AST. This would cause
"Found node calling previously undeclared function
<validateFunctionCall>" validation error and ASSERT.
Mat, vec equality calls would use prelude ANGLE_equals implementations.
The MSL emit already does the conversion from == to ANGLE_equals for
these builtins.
For user-defined structs the logic would be:
- collect struct decls, structs, create equality functions
- toposort
- for each sorted struct:
- insert struct decl
- insert equality function
Move the equality function creation after the toposort:
- collect struct decls, structs
- toposort
- for each sorted struct:
- insert struct decl
- create equality function
- insert equality function
This way the sort ensures that nested struct equality functions can
refer to previously introduced equality functions.
Bug: angleproject:375352601
Change-Id: I59efed98bca6d99b198abc2b5c7577cf5d1d5d83
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5961281
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
|
|
d193d51b
|
2024-06-17T22:46:08
|
|
Replace issue ids post migration to new issue tracker
This change replaces anglebug.com/NNNN links.
Bug: None
Change-Id: I8ac3aec8d2a8a844b3d7b99fc0a6b2be8da31761
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5637912
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
3b650ffa
|
2024-03-19T09:53:06
|
|
Metal: Assert while using gl_VertexID as ivec
Metal [vertex_id] is uint while gl_VertexID is int. Replacing
gl_VertexID with gl_VertexIDMetal without rewrites of the expressions
would cause invalid expressions.
Fix by casting uint kgl_VertexIDMetal to int kgl_VertexID instead
of replacing variable during compile.
Bug: angleproject:8597
Change-Id: I76acdb2a0ab5982aa05181175925b3359068e901
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5376498
Reviewed-by: Alexey Knyazev <lexa.knyazev@gmail.com>
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
490ff869
|
2024-03-14T00:00:00
|
|
Always redeclare clip/cull distance built-ins
When clip or cull distance built-in arrays are implicitly
sized, various pruning passes may produce an inconsistent
AST state thus causing translation or linking failures.
Adjusted the dedicated validation pass to inject
an explicit declaration thus bypassing the issue.
Additional updates:
* Ensured that API clip distance state emulation is
applied when a variable is declared but not assigned
* Reverted previous clip/cull distance related changes to
TIntermBinary::hasSideEffects as they are redundant now
* Fixed failing AST validation for MSL varying emulation
* Aligned linking error messages with the specifications
* Updated tests and cleaned-up obsolete code
Fixed: angleproject:8591
Change-Id: Ic8cfaf37778b8532bbab32ab998d5350b85d67ef
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5372714
Reviewed-by: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
|
|
|
27423bff
|
2024-03-05T17:57:24
|
|
Metal: Generate names for rewritten inputs
When expanding multi-component fields to multiple single-component
fields, use AngleInternal namespace for the new names. The names are
generated with form "someField_0" where _0 is the component index.
If these are not created in AngleInternal, caller is able to create
a name clash by introducing single-component field "someField_0".
Fixes an assert where the vec4(a_) + vec4(a) would assert on size
mismatch because the variable lookup for "a_" would find a rewritten
variable for the expanded matrix row of "a".
Bug: angleproject:8558
Change-Id: I64b7a755d7d534543fdb0f4c43008dd5c63f4aad
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5323060
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kyle Piddington <kpiddington@apple.com>
|
|
|
2b1ef00a
|
2024-02-20T15:45:34
|
|
Metal: Fix validation for anonymous struct arrays
Fix validation failure for accessing an element of
array of anonymous struct.
SeparateCompoundStructDeclarations would create
new Structures to name unnamed structures.
The Structure instance of the array index accessor
node was incorrect, it was the old Structure.
Use same pattern as in
SeparateStructFromUniformDeclarations.
Bug: angleproject:8551
Change-Id: I04684e5ad99cc2b9038a0cc21c7eefcc6cf247df
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5310074
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
|
|
|
197beb4d
|
2024-02-13T15:47:32
|
|
Metal: Crash if for loop body is optimized away
Loops might have their bodies optimized away.
For case like `for(;;) if(false);` the constant pruning happens at
parse phase. The else branch (e.e. no else branch, nullptr) would be
selected as the body.
Some code treated the body as optional, some code treated it as
required. Define it as required, and remove all conditional code
related to the loop body.
Bug: angleproject:8532
Change-Id: Ic35f1bf78e63ceb2cee7b96ba99e788efc282e6f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5291554
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
1ab5d01d
|
2023-08-29T13:31:07
|
|
Metal: Fix dropped out arguments from functions with many args.
RewriteOutArgs has an early-exit if it spots a potentially aliased
arg. It's also responsible for marking out args as references,
which caused an issue in Google Earth.
Removing the early-exit fixes both issues.
Bug: chromium:1474736
Change-Id: Ib68dd3f3e2e0a1e773e4e09edcdfa3a4bdfc1ef2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4823006
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
280e79e1
|
2023-08-10T14:08:00
|
|
Rescope globals only used in one function to function-local.
We can demote global variables when they are only used in one function.
This has performance implications on Metal.
Bug: angleproject:8311
Change-Id: Id666e6b167be771c14768cd73efa61fdacd897d4
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4771215
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
02e7f967
|
2023-05-25T14:54:49
|
|
Translator: Remove the "variables" option
Variable collection is invariably enabled by the front-end as well as
other major users of ANGLE such as Firefox. All translator backends
except GLSL force-enable variable collection either way.
This change removes this compile option and enables variable collection
unconditionally.
The flag itself remains in ShCompileOptions until references to it are
removed from Chromium.
Bug: chromium:1447314
Change-Id: I4d3b30c1bfbd345c5ad269abc62c0a6a59de2f56
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4568524
Auto-Submit: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
6f80f0f0
|
2022-08-06T02:29:19
|
|
Translator: Clean up the compile flag passing interface
Historically, compile flags were sent to the translator as a bitmask.
Recently, we were getting close to running out of bits. Additionally,
direct-to-metal work had started to introduce constants to be passed to
the translator, which were misplaced in ShBuiltInResources and Caps.
Recent work on Pixel Local Storage adds even more constants, aggravating
the situation.
In this change, the interface to passing compile flags is reworked. A
struct is passed (instead of a bitmask) that has one bit for each flag.
This can be indefinitely extended. Additionally, the constants needed
by metal and PLS are also placed in this struct. In turn, the backends
can set these options directly, and don't have to hack them into Caps to
further get hacked into ShBuiltInResources.
Bug: angleproject:7559
Change-Id: If93f1e1b8818ad3a0ac708ab04ab93b4b397d114
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3812562
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Commit-Queue: Amirali Abdolrashidi <abdolrashidi@google.com>
Auto-Submit: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
da3db87e
|
2021-07-06T14:00:58
|
|
Upstream latest changes to Metal backend from Apple to 7/1/2021
This CL merges in the ANGLE changes between these two WebKit commits:
https://git.webkit.org/?p=WebKit.git;a=commit;h=8648b353ab1d7730438c2e08319e1a4d64982c31
https://git.webkit.org/?p=WebKit.git;a=commit;h=166e4924a52971d6a32ad48247a439b16c00e062
Include provoking vertex buffer out of bounds fix
from https://bugs.webkit.org/show_bug.cgi?id=230107
Fix bad merge of resetting of dirty bits, breaking
DepthStencilFormatsTest.DepthTextureRender test and perhaps others.
Disable GL_APPLE_clip_distance when the direct-to-Metal compiler is
active. It can not yet handle the gl_ClipDistance array.
Disable use of rectangular textures for IOSurfaces. Metal can bind
IOSurfaces to 2D textures, and this was passing all tests in the
SPIR-V Metal backend. Introducing rectangular textures breaks the
SPIR-V Metal backend, and the tests currently fail on the
direct-to-Metal backend.
Fix several bugs with ProvokingVertex, which was causing
both the SpirV and Direct backends to incorrectly draw
indices.
(https://bugs.webkit.org/show_bug.cgi?id=230107)
Skip the following tests on the Metal backend which is still failing
RobustResourceInitTestES3.BlitDepthStencilAfterClearBuffer
GLSLTest_ES3.GLVertexIDIntegerTextureDrawArrays/ES3_Metal
With these changes, angle_end2end_tests again runs to completion.
Bug: angleproject:6395
Change-Id: I3cc58f531426a95fc8f177a4ad87f56c1855a546
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3167010
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kyle Piddington <kpiddington@apple.com>
|
|
d33a2222
|
2021-04-26T16:56:15
|
|
Upstream Apple's direct-to-Metal backend: compile libANGLE.
This change is meant to merge the metal backend changes from Apple's
direct-to-Metal backend. Taken from Kyle Piddington's CL:
https://chromium-review.googlesource.com/c/angle/angle/+/2857366/
The goal of this CL is to merge the metal backend code in a state
that compiles, but not to switch the Metal backend over to using
the direct-to-metal backend yet.
Bug: angleproject:5505
Bug: angleproject:6127
Change-Id: If6783e06e0086b3a1dd25c6f53caca5cfc96cb86
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2950067
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
|
|
|
d7aa0130
|
2021-04-26T16:56:15
|
|
Upstream Apple's direct-to-Metal backend: compile translator.
This change is meant to merge the translator changes from Apple's
direct-to-Metal backend. Taken from Kyle Piddington's CL:
https://chromium-review.googlesource.com/c/angle/angle/+/2857366/
The goal of this CL is to merge the translator code in a state that
compiles, but not to switch the Metal backend over to use this
translator backend yet.
Bug: angleproject:5505
Change-Id: I68a6354604498cd5fd1eb96c13fc56f3b38f2bd0
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2897536
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
|