encounter/dawn-cmake
1
0
Fork 0
mirror of https://github.com/encounter/dawn-cmake.git synced 2025-12-21 02:39:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

HLSL: work around FXC failures when dynamically indexing matrices

Browse Source 
This fixes errors like "error X3500: array reference cannot be used as
an l-value; not natively addressable". Note that FXC treats matrices
like arrays. We still get this error for dynamically indexed arrays in
structs.

Also improved HLSL assign tests, and add missing ones for vector
indexing.

Manually removed 20 e2e skip hlsl SKIP files that are now passing with
this change.

Bug: tint:1333
Change-Id: If23881a667857a4d4ec6881e72666af0a666ef10
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/71982
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
This commit is contained in:
Antonio Maiorano
2021-12-08 23:03:33 +00:00
committed by Tint LUCI CQ
parent 3703522d41
commit a8c202b8e7
65 changed files with 1296 additions and 2594 deletions
Download Patch File Download Diff File Expand all files Collapse all files

102
test/vk-gl-cts/graphicsfuzz/cov-unused-matrix-copy-inside-loop/0-opt.spvasm.expected.hlsl
View File

@@ -1,102 +0,0 @@
SKIP: FAILED
void set_float4(inout float4 vec, int idx, float val) {
vec = (idx.xxxx == int4(0, 1, 2, 3)) ? val.xxxx : vec;
}
cbuffer cbuffer_x_6 : register(b1, space0) {
uint4 x_6[4];
};
cbuffer cbuffer_x_10 : register(b0, space0) {
uint4 x_10[1];
};
static float4 x_GLF_color = float4(0.0f, 0.0f, 0.0f, 0.0f);
void main_1() {
float4x4 m0 = float4x4(0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
int c = 0;
float4x4 m1 = float4x4(0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
const int x_40 = asint(x_6[1].x);
const float x_41 = float(x_40);
m0 = float4x4(float4(x_41, 0.0f, 0.0f, 0.0f), float4(0.0f, x_41, 0.0f, 0.0f), float4(0.0f, 0.0f, x_41, 0.0f), float4(0.0f, 0.0f, 0.0f, x_41));
const int x_48 = asint(x_6[2].x);
c = x_48;
[loop] while (true) {
const int x_53 = c;
const uint scalar_offset = ((16u * uint(0))) / 4;
const int x_55 = asint(x_6[scalar_offset / 4][scalar_offset % 4]);
if ((x_53 < x_55)) {
} else {
break;
}
m1 = m0;
const int x_59 = c;
const int x_61 = asint(x_6[3].x);
const int x_64 = asint(x_6[2].x);
const uint scalar_offset_1 = ((16u * uint(0))) / 4;
const float x_66 = asfloat(x_10[scalar_offset_1 / 4][scalar_offset_1 % 4]);
set_float4(m1[(x_59 % x_61)], x_64, x_66);
const int x_68 = c;
const int x_70 = asint(x_6[3].x);
const int x_73 = asint(x_6[2].x);
const uint scalar_offset_2 = ((16u * uint(0))) / 4;
const float x_75 = asfloat(x_10[scalar_offset_2 / 4][scalar_offset_2 % 4]);
set_float4(m0[(x_68 % x_70)], x_73, x_75);
{
c = (c + 1);
}
}
const float4x4 x_79 = m0;
const int x_81 = asint(x_6[1].x);
const int x_84 = asint(x_6[2].x);
const int x_87 = asint(x_6[1].x);
const int x_90 = asint(x_6[1].x);
const int x_93 = asint(x_6[1].x);
const int x_96 = asint(x_6[2].x);
const int x_99 = asint(x_6[1].x);
const int x_102 = asint(x_6[1].x);
const int x_105 = asint(x_6[1].x);
const int x_108 = asint(x_6[2].x);
const int x_111 = asint(x_6[1].x);
const int x_114 = asint(x_6[1].x);
const int x_117 = asint(x_6[1].x);
const int x_120 = asint(x_6[2].x);
const int x_123 = asint(x_6[1].x);
const int x_126 = asint(x_6[1].x);
const float4x4 x_132 = float4x4(float4(float(x_81), float(x_84), float(x_87), float(x_90)), float4(float(x_93), float(x_96), float(x_99), float(x_102)), float4(float(x_105), float(x_108), float(x_111), float(x_114)), float4(float(x_117), float(x_120), float(x_123), float(x_126)));
if ((((all((x_79[0u] == x_132[0u])) & all((x_79[1u] == x_132[1u]))) & all((x_79[2u] == x_132[2u]))) & all((x_79[3u] == x_132[3u])))) {
const int x_156 = asint(x_6[2].x);
const int x_159 = asint(x_6[1].x);
const int x_162 = asint(x_6[1].x);
const int x_165 = asint(x_6[2].x);
x_GLF_color = float4(float(x_156), float(x_159), float(x_162), float(x_165));
} else {
const int x_169 = asint(x_6[1].x);
const float x_170 = float(x_169);
x_GLF_color = float4(x_170, x_170, x_170, x_170);
}
return;
}
struct main_out {
float4 x_GLF_color_1;
};
struct tint_symbol {
float4 x_GLF_color_1 : SV_Target0;
};
main_out main_inner() {
main_1();
const main_out tint_symbol_3 = {x_GLF_color};
return tint_symbol_3;
}
tint_symbol main() {
const main_out inner_result = main_inner();
tint_symbol wrapper_result = (tint_symbol)0;
wrapper_result.x_GLF_color_1 = inner_result.x_GLF_color_1;
return wrapper_result;
}
C:\src\tint\test\Shader@0x0000013C243CAFF0(36,20-30): warning X3556: integer modulus may be much slower, try using uints if possible.
C:\src\tint\test\Shader@0x0000013C243CAFF0(22,10-21): error X3531: can't unroll loops marked with loop attribute

114
test/vk-gl-cts/graphicsfuzz/cov-unused-matrix-copy-inside-loop/0-opt.wgsl.expected.hlsl
View File

@@ -1,114 +0,0 @@
SKIP: FAILED
void set_float4(inout float4 vec, int idx, float val) {
vec = (idx.xxxx == int4(0, 1, 2, 3)) ? val.xxxx : vec;
}
cbuffer cbuffer_x_6 : register(b1, space0) {
uint4 x_6[4];
};
cbuffer cbuffer_x_10 : register(b0, space0) {
uint4 x_10[1];
};
static float4 x_GLF_color = float4(0.0f, 0.0f, 0.0f, 0.0f);
void main_1() {
float4x4 m0 = float4x4(0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
int c = 0;
float4x4 m1 = float4x4(0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
const int x_40 = asint(x_6[1].x);
const float x_41 = float(x_40);
m0 = float4x4(float4(x_41, 0.0f, 0.0f, 0.0f), float4(0.0f, x_41, 0.0f, 0.0f), float4(0.0f, 0.0f, x_41, 0.0f), float4(0.0f, 0.0f, 0.0f, x_41));
const int x_48 = asint(x_6[2].x);
c = x_48;
[loop] while (true) {
const int x_53 = c;
const uint scalar_offset = ((16u * uint(0))) / 4;
const int x_55 = asint(x_6[scalar_offset / 4][scalar_offset % 4]);
if ((x_53 < x_55)) {
} else {
break;
}
m1 = m0;
const int x_59 = c;
const int x_61 = asint(x_6[3].x);
const int x_64 = asint(x_6[2].x);
const uint scalar_offset_1 = ((16u * uint(0))) / 4;
const float x_66 = asfloat(x_10[scalar_offset_1 / 4][scalar_offset_1 % 4]);
set_float4(m1[(x_59 % x_61)], x_64, x_66);
const int x_68 = c;
const int x_70 = asint(x_6[3].x);
const int x_73 = asint(x_6[2].x);
const uint scalar_offset_2 = ((16u * uint(0))) / 4;
const float x_75 = asfloat(x_10[scalar_offset_2 / 4][scalar_offset_2 % 4]);
set_float4(m0[(x_68 % x_70)], x_73, x_75);
{
c = (c + 1);
}
}
const float4x4 x_79 = m0;
const int x_81 = asint(x_6[1].x);
const int x_84 = asint(x_6[2].x);
const int x_87 = asint(x_6[1].x);
const int x_90 = asint(x_6[1].x);
const int x_93 = asint(x_6[1].x);
const int x_96 = asint(x_6[2].x);
const int x_99 = asint(x_6[1].x);
const int x_102 = asint(x_6[1].x);
const int x_105 = asint(x_6[1].x);
const int x_108 = asint(x_6[2].x);
const int x_111 = asint(x_6[1].x);
const int x_114 = asint(x_6[1].x);
const int x_117 = asint(x_6[1].x);
const int x_120 = asint(x_6[2].x);
const int x_123 = asint(x_6[1].x);
const int x_126 = asint(x_6[1].x);
const float4x4 x_132 = float4x4(float4(float(x_81), float(x_84), float(x_87), float(x_90)), float4(float(x_93), float(x_96), float(x_99), float(x_102)), float4(float(x_105), float(x_108), float(x_111), float(x_114)), float4(float(x_117), float(x_120), float(x_123), float(x_126)));
bool tint_tmp_2 = all((x_79[0u] == x_132[0u]));
if (tint_tmp_2) {
tint_tmp_2 = all((x_79[1u] == x_132[1u]));
}
bool tint_tmp_1 = (tint_tmp_2);
if (tint_tmp_1) {
tint_tmp_1 = all((x_79[2u] == x_132[2u]));
}
bool tint_tmp = (tint_tmp_1);
if (tint_tmp) {
tint_tmp = all((x_79[3u] == x_132[3u]));
}
if ((tint_tmp)) {
const int x_156 = asint(x_6[2].x);
const int x_159 = asint(x_6[1].x);
const int x_162 = asint(x_6[1].x);
const int x_165 = asint(x_6[2].x);
x_GLF_color = float4(float(x_156), float(x_159), float(x_162), float(x_165));
} else {
const int x_169 = asint(x_6[1].x);
const float x_170 = float(x_169);
x_GLF_color = float4(x_170, x_170, x_170, x_170);
}
return;
}
struct main_out {
float4 x_GLF_color_1;
};
struct tint_symbol {
float4 x_GLF_color_1 : SV_Target0;
};
main_out main_inner() {
main_1();
const main_out tint_symbol_3 = {x_GLF_color};
return tint_symbol_3;
}
tint_symbol main() {
const main_out inner_result = main_inner();
tint_symbol wrapper_result = (tint_symbol)0;
wrapper_result.x_GLF_color_1 = inner_result.x_GLF_color_1;
return wrapper_result;
}
C:\src\tint\test\Shader@0x000001ABB4809E40(36,20-30): warning X3556: integer modulus may be much slower, try using uints if possible.
C:\src\tint\test\Shader@0x000001ABB4809E40(22,10-21): error X3531: can't unroll loops marked with loop attribute