Using a designated initializer like this leaves padding bits, which form
part of the aliasing u64/f64 member of the union, uninitialised, but a
nir_const_value must always have the unused bits zeroed out. Thus, use
the nir_const_value_for_float helper instead like everywhere else which
will do a memset 0 for us first.
Without this, using the pan_blend shader in a build with validation
enabled fails with:
NIR validation failed after nir_lower_vars_to_ssa
...
vec4 32 ssa_58 = load_const (0x3f7cfcfd /* 0.988235 */, 0x3f7cfcfd /* 0.988235 */, 0x3f7cfcfd /* 0.988235 */, 0x3f800000 /* 1.000000 */)
error: memcmp(val, &cmp_val, sizeof(cmp_val)) == 0 (../src/compiler/nir/nir_validate.c:976)
Fixes:
1378c67bcf9e ("panfrost/blend: Inline blend constants")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20071>
(cherry picked from commit
750325730bb84981f68d78b205fb0cec1e165100)
"description": "panfrost/blend: Fix invalid const values leading to NIR validation errors",
"nominated": true,
"nomination_type": 1,
- "resolution": 0,
+ "resolution": 1,
"main_sha": null,
"because_sha": "1378c67bcf9e16aea98ddb8cb6bd37b9ed4e3a81"
},
float *floats = data;
const nir_const_value constants[4] = {
- { .f32 = floats[0] },
- { .f32 = floats[1] },
- { .f32 = floats[2] },
- { .f32 = floats[3] }
+ nir_const_value_for_float(floats[0], 32),
+ nir_const_value_for_float(floats[1], 32),
+ nir_const_value_for_float(floats[2], 32),
+ nir_const_value_for_float(floats[3], 32)
};
b->cursor = nir_after_instr(instr);