Giovanni Mascellani (@giomasce) commented about libs/vkd3d-shader/tpf.c:
break; } }
- else if (register_type == VKD3D_SM4_RT_IMMCONSTBUFFER)
- {
if (param->idx_count != 1)
{
WARN("Unexpected idx count %u.\n", param->idx_count);
vkd3d_shader_parser_error(&priv->p, VKD3D_SHADER_ERROR_TPF_INVALID_REGISTER_INDEX_COUNT,
"Invalid index count %u for immediate const buffer register; expected count 1.", param->idx_count);
}
param->idx[1] = param->idx[0];
param->idx[0].offset = 0;
param->idx[0].rel_addr = NULL;
param->idx_count = 2;
This breaks the `d3d-asm` disassembly for TPF. Relatedly, at some point we might want to start having two different dialects for the disassembly: one that closely mimics native disassembly for D3DBC and TPF and another one that retains information that is relevant for developing VSIR, including showing the indices "hidden" from the standard disassembly and possibly showing indices in a more consistent fashion.