wine/vkd3d
0
0
Fork 0
mirror of https://gitlab.winehq.org/wine/vkd3d.git synced 2025-04-13 05:43:18 -07:00
Code Issues Packages Projects Releases Wiki Activity

vkd3d-shader/hlsl: Divert written uniform derefs to temps before copy-propagation passes.

Browse Source 
The following pixel shader currently triggers an infinite loop during
copy propagation, which is fixed by this commit:

    sampler s;
    Texture2D t1, t2;

    float4 main() : sv_target
    {
        Texture2D t = t1;
        t1 = t2;
        t2 = t;
        return t1.Sample(s, float2(0, 0)) + t2.Sample(s, float2(0, 0));
    }

The infinite loop occurs because copy_propagation_transform_object_load()
replaces t1 in the resource_load(t1, ...) instruction
with t2, t1, t2, ... repeatedly.
This commit is contained in:
Shaun Ren
2025-03-21 17:39:42 -04:00
committed by Henri Verbeet
parent 3ead8d532b
commit b1d8915440
Notes: Henri Verbeet 2025-04-02 19:43:31 +02:00 
Approved-by: Francisco Casas (@fcasas)
Approved-by: Elizabeth Figura (@zfigura)
Approved-by: Henri Verbeet (@hverbeet)
Merge-Request: https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/1436
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
libs/vkd3d-shader/hlsl_codegen.c
View File

@@ -12763,6 +12763,9 @@ static void process_entry_function(struct hlsl_ctx *ctx,
hlsl_transform_ir(ctx, lower_resource_load_bias, body, NULL); hlsl_transform_ir(ctx, lower_resource_load_bias, body, NULL);
} }
compute_liveness(ctx, entry_func);
transform_derefs(ctx, divert_written_uniform_derefs_to_temp, &entry_func->body);
loop_unrolling_execute(ctx, body); loop_unrolling_execute(ctx, body);
hlsl_run_const_passes(ctx, body); hlsl_run_const_passes(ctx, body);
@@ -12773,9 +12776,6 @@ static void process_entry_function(struct hlsl_ctx *ctx,
lower_ir(ctx, lower_casts_to_bool, body); lower_ir(ctx, lower_casts_to_bool, body);
lower_ir(ctx, lower_int_dot, body); lower_ir(ctx, lower_int_dot, body);
compute_liveness(ctx, entry_func);
transform_derefs(ctx, divert_written_uniform_derefs_to_temp, &entry_func->body);
if (hlsl_version_lt(ctx, 4, 0)) if (hlsl_version_lt(ctx, 4, 0))
hlsl_transform_ir(ctx, lower_separate_samples, body, NULL); hlsl_transform_ir(ctx, lower_separate_samples, body, NULL);