transform/shader_io: Generate a wrapper function

This is a major reworking of this transform. The old transform code
was getting unwieldy, with part of the complication coming from the
handling of multiple return statements. By generating a wrapper
function instead, we can avoid a lot of this complexity.

The original entry point function is stripped of all shader IO
attributes (as well as `stage` and `workgroup_size`), but the body is
left unmodified. A new entry point wrapper function is introduced
which calls the original function, packing/unpacking the shader inputs
as necessary, and propagates the result to the corresponding shader
outputs.

The new code has been refactored to use a state object with the
different parts of the transform split into separate functions, which
makes it much more manageable.

Fixed: tint:1076
Bug: tint:920
Change-Id: I3490a0ea7a3509a4e198ce730e476516649d8d96
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/60521
Auto-Submit: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
This commit is contained in:
James Price
2021-08-04 22:15:28 +00:00
committed by Tint LUCI CQ
parent 3e92e9f8ba
commit a5d73ce965
3866 changed files with 49323 additions and 26508 deletions

View File

@@ -13,9 +13,7 @@ struct tint_symbol_2 {
uint local_invocation_index : SV_GroupIndex;
};
[numthreads(1, 1, 1)]
void f(tint_symbol_2 tint_symbol_1) {
const uint local_invocation_index = tint_symbol_1.local_invocation_index;
void f_inner(uint local_invocation_index) {
{
for(uint idx = local_invocation_index; (idx < 64u); idx = (idx + 1u)) {
const uint i = idx;
@@ -24,5 +22,10 @@ void f(tint_symbol_2 tint_symbol_1) {
}
GroupMemoryBarrierWithGroupSync();
result.Store(0u, asuint(s.data[asint(ubo[0].x)]));
}
[numthreads(1, 1, 1)]
void f(tint_symbol_2 tint_symbol_1) {
f_inner(tint_symbol_1.local_invocation_index);
return;
}

View File

@@ -14,14 +14,18 @@ struct Result {
/* 0x0000 */ int out;
};
kernel void f(uint local_invocation_index [[thread_index_in_threadgroup]], constant UBO& ubo [[buffer(0)]], device Result& result [[buffer(1)]]) {
threadgroup S tint_symbol_1;
void f_inner(constant UBO& ubo, device Result& result, uint local_invocation_index, threadgroup S* const tint_symbol) {
for(uint idx = local_invocation_index; (idx < 64u); idx = (idx + 1u)) {
uint const i = idx;
tint_symbol_1.data.arr[i] = int();
(*(tint_symbol)).data.arr[i] = int();
}
threadgroup_barrier(mem_flags::mem_threadgroup);
result.out = tint_symbol_1.data.arr[ubo.dynamic_idx];
result.out = (*(tint_symbol)).data.arr[ubo.dynamic_idx];
}
kernel void f(uint local_invocation_index [[thread_index_in_threadgroup]], constant UBO& ubo [[buffer(0)]], device Result& result [[buffer(1)]]) {
threadgroup S tint_symbol_1;
f_inner(ubo, result, local_invocation_index, &(tint_symbol_1));
return;
}

View File

@@ -13,9 +13,7 @@ struct tint_symbol_2 {
uint local_invocation_index : SV_GroupIndex;
};
[numthreads(1, 1, 1)]
void f(tint_symbol_2 tint_symbol_1) {
const uint local_invocation_index = tint_symbol_1.local_invocation_index;
void f_inner(uint local_invocation_index) {
{
for(uint idx = local_invocation_index; (idx < 64u); idx = (idx + 1u)) {
const uint i = idx;
@@ -25,5 +23,10 @@ void f(tint_symbol_2 tint_symbol_1) {
GroupMemoryBarrierWithGroupSync();
s.data[asint(ubo[0].x)] = 1;
result.Store(0u, asuint(s.data[3]));
}
[numthreads(1, 1, 1)]
void f(tint_symbol_2 tint_symbol_1) {
f_inner(tint_symbol_1.local_invocation_index);
return;
}

View File

@@ -14,15 +14,19 @@ struct Result {
/* 0x0000 */ int out;
};
kernel void f(uint local_invocation_index [[thread_index_in_threadgroup]], constant UBO& ubo [[buffer(0)]], device Result& result [[buffer(1)]]) {
threadgroup S tint_symbol_1;
void f_inner(constant UBO& ubo, device Result& result, uint local_invocation_index, threadgroup S* const tint_symbol) {
for(uint idx = local_invocation_index; (idx < 64u); idx = (idx + 1u)) {
uint const i = idx;
tint_symbol_1.data.arr[i] = int();
(*(tint_symbol)).data.arr[i] = int();
}
threadgroup_barrier(mem_flags::mem_threadgroup);
tint_symbol_1.data.arr[ubo.dynamic_idx] = 1;
result.out = tint_symbol_1.data.arr[3];
(*(tint_symbol)).data.arr[ubo.dynamic_idx] = 1;
result.out = (*(tint_symbol)).data.arr[3];
}
kernel void f(uint local_invocation_index [[thread_index_in_threadgroup]], constant UBO& ubo [[buffer(0)]], device Result& result [[buffer(1)]]) {
threadgroup S tint_symbol_1;
f_inner(ubo, result, local_invocation_index, &(tint_symbol_1));
return;
}