Skip to content
Snippets Groups Projects
d3dbc.c 71.3 KiB
Newer Older
    {
        const char *semantic;
        bool output;
        enum vkd3d_shader_type shader_type;
        unsigned int major_version;
        D3DSHADER_PARAM_REGISTER_TYPE type;
        DWORD offset;
    }
    register_table[] =
    {
        {"color",       false, VKD3D_SHADER_TYPE_PIXEL, 1, D3DSPR_INPUT},
        {"texcoord",    false, VKD3D_SHADER_TYPE_PIXEL, 1, D3DSPR_TEXTURE},

        {"color",       true,  VKD3D_SHADER_TYPE_PIXEL, 2, D3DSPR_COLOROUT},
        {"depth",       true,  VKD3D_SHADER_TYPE_PIXEL, 2, D3DSPR_DEPTHOUT},
        {"sv_depth",    true,  VKD3D_SHADER_TYPE_PIXEL, 2, D3DSPR_DEPTHOUT},
        {"sv_target",   true,  VKD3D_SHADER_TYPE_PIXEL, 2, D3DSPR_COLOROUT},
        {"color",       false, VKD3D_SHADER_TYPE_PIXEL, 2, D3DSPR_INPUT},
        {"texcoord",    false, VKD3D_SHADER_TYPE_PIXEL, 2, D3DSPR_TEXTURE},

        {"color",       true,  VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_COLOROUT},
        {"depth",       true,  VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_DEPTHOUT},
        {"sv_depth",    true,  VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_DEPTHOUT},
        {"sv_target",   true,  VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_COLOROUT},
        {"sv_position", false, VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_MISCTYPE,    D3DSMO_POSITION},
        {"vface",       false, VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_MISCTYPE,    D3DSMO_FACE},
        {"vpos",        false, VKD3D_SHADER_TYPE_PIXEL, 3, D3DSPR_MISCTYPE,    D3DSMO_POSITION},

        {"color",       true,  VKD3D_SHADER_TYPE_VERTEX, 1, D3DSPR_ATTROUT},
        {"fog",         true,  VKD3D_SHADER_TYPE_VERTEX, 1, D3DSPR_RASTOUT,     D3DSRO_FOG},
        {"position",    true,  VKD3D_SHADER_TYPE_VERTEX, 1, D3DSPR_RASTOUT,     D3DSRO_POSITION},
        {"psize",       true,  VKD3D_SHADER_TYPE_VERTEX, 1, D3DSPR_RASTOUT,     D3DSRO_POINT_SIZE},
        {"sv_position", true,  VKD3D_SHADER_TYPE_VERTEX, 1, D3DSPR_RASTOUT,     D3DSRO_POSITION},
        {"texcoord",    true,  VKD3D_SHADER_TYPE_VERTEX, 1, D3DSPR_TEXCRDOUT},

        {"color",       true,  VKD3D_SHADER_TYPE_VERTEX, 2, D3DSPR_ATTROUT},
        {"fog",         true,  VKD3D_SHADER_TYPE_VERTEX, 2, D3DSPR_RASTOUT,     D3DSRO_FOG},
        {"position",    true,  VKD3D_SHADER_TYPE_VERTEX, 2, D3DSPR_RASTOUT,     D3DSRO_POSITION},
        {"psize",       true,  VKD3D_SHADER_TYPE_VERTEX, 2, D3DSPR_RASTOUT,     D3DSRO_POINT_SIZE},
        {"sv_position", true,  VKD3D_SHADER_TYPE_VERTEX, 2, D3DSPR_RASTOUT,     D3DSRO_POSITION},
        {"texcoord",    true,  VKD3D_SHADER_TYPE_VERTEX, 2, D3DSPR_TEXCRDOUT},
    };

    for (i = 0; i < ARRAY_SIZE(register_table); ++i)
    {
        if (!ascii_strcasecmp(semantic->name, register_table[i].semantic)
                && output == register_table[i].output
                && ctx->profile->type == register_table[i].shader_type
                && ctx->profile->major_version == register_table[i].major_version)
        {
            *type = register_table[i].type;
            if (register_table[i].type == D3DSPR_MISCTYPE || register_table[i].type == D3DSPR_RASTOUT)
                *reg = register_table[i].offset;
            else
                *reg = semantic->index;
            return true;
        }
    }

    return false;
}

bool hlsl_sm1_usage_from_semantic(const struct hlsl_semantic *semantic, D3DDECLUSAGE *usage, uint32_t *usage_idx)
{
    static const struct
    {
        const char *name;
        D3DDECLUSAGE usage;
    }
    semantics[] =
    {
        {"binormal",        D3DDECLUSAGE_BINORMAL},
        {"blendindices",    D3DDECLUSAGE_BLENDINDICES},
        {"blendweight",     D3DDECLUSAGE_BLENDWEIGHT},
        {"color",           D3DDECLUSAGE_COLOR},
        {"depth",           D3DDECLUSAGE_DEPTH},
        {"fog",             D3DDECLUSAGE_FOG},
        {"normal",          D3DDECLUSAGE_NORMAL},
        {"position",        D3DDECLUSAGE_POSITION},
        {"positiont",       D3DDECLUSAGE_POSITIONT},
        {"psize",           D3DDECLUSAGE_PSIZE},
        {"sample",          D3DDECLUSAGE_SAMPLE},
        {"sv_depth",        D3DDECLUSAGE_DEPTH},
        {"sv_position",     D3DDECLUSAGE_POSITION},
        {"sv_target",       D3DDECLUSAGE_COLOR},
        {"tangent",         D3DDECLUSAGE_TANGENT},
        {"tessfactor",      D3DDECLUSAGE_TESSFACTOR},
        {"texcoord",        D3DDECLUSAGE_TEXCOORD},
    };

    unsigned int i;

    for (i = 0; i < ARRAY_SIZE(semantics); ++i)
    {
        if (!ascii_strcasecmp(semantic->name, semantics[i].name))
        {
            *usage = semantics[i].usage;
            *usage_idx = semantic->index;
            return true;
        }
    }

    return false;
}

static uint32_t sm1_version(enum vkd3d_shader_type type, unsigned int major, unsigned int minor)
{
    if (type == VKD3D_SHADER_TYPE_VERTEX)
        return D3DVS_VERSION(major, minor);
    else
        return D3DPS_VERSION(major, minor);
}

static D3DXPARAMETER_CLASS sm1_class(const struct hlsl_type *type)
{
    switch (type->class)
    {
        case HLSL_CLASS_ARRAY:
            return sm1_class(type->e.array.type);
        case HLSL_CLASS_MATRIX:
            assert(type->modifiers & HLSL_MODIFIERS_MAJORITY_MASK);
            if (type->modifiers & HLSL_MODIFIER_COLUMN_MAJOR)
                return D3DXPC_MATRIX_COLUMNS;
            else
                return D3DXPC_MATRIX_ROWS;
        case HLSL_CLASS_OBJECT:
            return D3DXPC_OBJECT;
        case HLSL_CLASS_SCALAR:
            return D3DXPC_SCALAR;
        case HLSL_CLASS_STRUCT:
            return D3DXPC_STRUCT;
        case HLSL_CLASS_VECTOR:
            return D3DXPC_VECTOR;
        default:
            ERR("Invalid class %#x.\n", type->class);
            vkd3d_unreachable();
    }
}

static D3DXPARAMETER_TYPE sm1_base_type(const struct hlsl_type *type)
{
    switch (type->base_type)
    {
        case HLSL_TYPE_BOOL:
            return D3DXPT_BOOL;
        case HLSL_TYPE_FLOAT:
        case HLSL_TYPE_HALF:
            return D3DXPT_FLOAT;
        case HLSL_TYPE_INT:
        case HLSL_TYPE_UINT:
            return D3DXPT_INT;
        case HLSL_TYPE_PIXELSHADER:
            return D3DXPT_PIXELSHADER;
        case HLSL_TYPE_SAMPLER:
            switch (type->sampler_dim)
            {
                case HLSL_SAMPLER_DIM_1D:
                    return D3DXPT_SAMPLER1D;
                case HLSL_SAMPLER_DIM_2D:
                    return D3DXPT_SAMPLER2D;
                case HLSL_SAMPLER_DIM_3D:
                    return D3DXPT_SAMPLER3D;
                case HLSL_SAMPLER_DIM_CUBE:
                    return D3DXPT_SAMPLERCUBE;
                case HLSL_SAMPLER_DIM_GENERIC:
                    return D3DXPT_SAMPLER;
                default:
                    ERR("Invalid dimension %#x.\n", type->sampler_dim);
                    vkd3d_unreachable();
            }
            break;
        case HLSL_TYPE_STRING:
            return D3DXPT_STRING;
        case HLSL_TYPE_TEXTURE:
            switch (type->sampler_dim)
            {
                case HLSL_SAMPLER_DIM_1D:
                    return D3DXPT_TEXTURE1D;
                case HLSL_SAMPLER_DIM_2D:
                    return D3DXPT_TEXTURE2D;
                case HLSL_SAMPLER_DIM_3D:
                    return D3DXPT_TEXTURE3D;
                case HLSL_SAMPLER_DIM_CUBE:
                    return D3DXPT_TEXTURECUBE;
                case HLSL_SAMPLER_DIM_GENERIC:
                    return D3DXPT_TEXTURE;
                default:
                    ERR("Invalid dimension %#x.\n", type->sampler_dim);
                    vkd3d_unreachable();
            }
            break;
        case HLSL_TYPE_VERTEXSHADER:
            return D3DXPT_VERTEXSHADER;
        case HLSL_TYPE_VOID:
            return D3DXPT_VOID;
        default:
            vkd3d_unreachable();
    }
}

static void write_sm1_type(struct vkd3d_bytecode_buffer *buffer, struct hlsl_type *type, unsigned int ctab_start)
{
    const struct hlsl_type *array_type = hlsl_get_multiarray_element_type(type);
    unsigned int array_size = hlsl_get_multiarray_size(type);
1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684
    unsigned int field_count = 0;
    size_t fields_offset = 0;
    size_t i;

    if (type->bytecode_offset)
        return;

    if (array_type->class == HLSL_CLASS_STRUCT)
    {
        field_count = array_type->e.record.field_count;

        for (i = 0; i < field_count; ++i)
        {
            struct hlsl_struct_field *field = &array_type->e.record.fields[i];

            field->name_bytecode_offset = put_string(buffer, field->name);
            write_sm1_type(buffer, field->type, ctab_start);
        }

        fields_offset = bytecode_align(buffer) - ctab_start;

        for (i = 0; i < field_count; ++i)
        {
            struct hlsl_struct_field *field = &array_type->e.record.fields[i];

            put_u32(buffer, field->name_bytecode_offset - ctab_start);
            put_u32(buffer, field->type->bytecode_offset - ctab_start);
        }
    }

    type->bytecode_offset = put_u32(buffer, vkd3d_make_u32(sm1_class(type), sm1_base_type(array_type)));
    put_u32(buffer, vkd3d_make_u32(type->dimy, type->dimx));
    put_u32(buffer, vkd3d_make_u32(array_size, field_count));
    put_u32(buffer, fields_offset);
}

static void sm1_sort_extern(struct list *sorted, struct hlsl_ir_var *to_sort)
{
    struct hlsl_ir_var *var;

    list_remove(&to_sort->extern_entry);

    LIST_FOR_EACH_ENTRY(var, sorted, struct hlsl_ir_var, extern_entry)
    {
        if (strcmp(to_sort->name, var->name) < 0)
        {
            list_add_before(&var->extern_entry, &to_sort->extern_entry);
            return;
        }
    }

    list_add_tail(sorted, &to_sort->extern_entry);
}

static void sm1_sort_externs(struct hlsl_ctx *ctx)
{
    struct list sorted = LIST_INIT(sorted);
    struct hlsl_ir_var *var, *next;

    LIST_FOR_EACH_ENTRY_SAFE(var, next, &ctx->extern_vars, struct hlsl_ir_var, extern_entry)
    {
        if (var->is_uniform)
            sm1_sort_extern(&sorted, var);
    }
    list_move_tail(&ctx->extern_vars, &sorted);
}

static void write_sm1_uniforms(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        struct hlsl_ir_function_decl *entry_func)
{
    size_t ctab_offset, ctab_start, ctab_end, vars_start, size_offset, creator_offset, offset;
    unsigned int uniform_count = 0;
    struct hlsl_ir_var *var;

    LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry)
    {
        enum hlsl_regset regset = hlsl_type_get_regset(var->data_type);

        if (!var->semantic.name && var->regs[regset].allocated)
        {
            ++uniform_count;

            if (var->is_param && var->is_uniform)
            {
                struct vkd3d_string_buffer *name;

                if (!(name = hlsl_get_string_buffer(ctx)))
                {
                    buffer->status = VKD3D_ERROR_OUT_OF_MEMORY;
                    return;
                }
                vkd3d_string_buffer_printf(name, "$%s", var->name);
                vkd3d_free((char *)var->name);
                var->name = hlsl_strdup(ctx, name->buffer);
                hlsl_release_string_buffer(ctx, name);
            }
        }
    }

    sm1_sort_externs(ctx);

    size_offset = put_u32(buffer, 0);
    ctab_offset = put_u32(buffer, VKD3D_MAKE_TAG('C','T','A','B'));

    ctab_start = put_u32(buffer, sizeof(D3DXSHADER_CONSTANTTABLE));
    creator_offset = put_u32(buffer, 0);
    put_u32(buffer, sm1_version(ctx->profile->type, ctx->profile->major_version, ctx->profile->minor_version));
    put_u32(buffer, uniform_count);
    put_u32(buffer, sizeof(D3DXSHADER_CONSTANTTABLE)); /* offset of constants */
    put_u32(buffer, 0); /* FIXME: flags */
    put_u32(buffer, 0); /* FIXME: target string */

    vars_start = bytecode_align(buffer);

    LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry)
    {
        enum hlsl_regset regset = hlsl_type_get_regset(var->data_type);

        if (!var->semantic.name && var->regs[regset].allocated)
        {
            put_u32(buffer, 0); /* name */
            if (var->data_type->class == HLSL_CLASS_OBJECT
                    && (var->data_type->base_type == HLSL_TYPE_SAMPLER
                    || var->data_type->base_type == HLSL_TYPE_TEXTURE))
            {
                assert(regset == HLSL_REGSET_SAMPLERS);
                put_u32(buffer, vkd3d_make_u32(D3DXRS_SAMPLER, var->regs[regset].id));
                put_u32(buffer, 1);
            }
            else
            {
                assert(regset == HLSL_REGSET_NUMERIC);
                put_u32(buffer, vkd3d_make_u32(D3DXRS_FLOAT4, var->regs[regset].id));
                put_u32(buffer, var->data_type->reg_size[regset] / 4);
            }
            put_u32(buffer, 0); /* type */
            put_u32(buffer, 0); /* FIXME: default value */
        }
    }

    uniform_count = 0;

    LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry)
    {
        enum hlsl_regset regset = hlsl_type_get_regset(var->data_type);

        if (!var->semantic.name && var->regs[regset].allocated)
        {
            size_t var_offset = vars_start + (uniform_count * 5 * sizeof(uint32_t));
            size_t name_offset;

            name_offset = put_string(buffer, var->name);
            set_u32(buffer, var_offset, name_offset - ctab_start);

            write_sm1_type(buffer, var->data_type, ctab_start);
            set_u32(buffer, var_offset + 3 * sizeof(uint32_t), var->data_type->bytecode_offset - ctab_start);
            ++uniform_count;
        }
    }

    offset = put_string(buffer, vkd3d_shader_get_version(NULL, NULL));
    set_u32(buffer, creator_offset, offset - ctab_start);

    ctab_end = bytecode_align(buffer);
    set_u32(buffer, size_offset, vkd3d_make_u32(D3DSIO_COMMENT, (ctab_end - ctab_offset) / sizeof(uint32_t)));
}

static uint32_t sm1_encode_register_type(D3DSHADER_PARAM_REGISTER_TYPE type)
{
    return ((type << D3DSP_REGTYPE_SHIFT) & D3DSP_REGTYPE_MASK)
            | ((type << D3DSP_REGTYPE_SHIFT2) & D3DSP_REGTYPE_MASK2);
}

struct sm1_instruction
{
    D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode;

    struct sm1_dst_register
    {
        D3DSHADER_PARAM_REGISTER_TYPE type;
        D3DSHADER_PARAM_DSTMOD_TYPE mod;
        unsigned int writemask;
        uint32_t reg;
    } dst;

    struct sm1_src_register
    {
        D3DSHADER_PARAM_REGISTER_TYPE type;
        D3DSHADER_PARAM_SRCMOD_TYPE mod;
        unsigned int swizzle;
        uint32_t reg;
    } srcs[3];
    unsigned int src_count;

    unsigned int has_dst;
};

static void write_sm1_dst_register(struct vkd3d_bytecode_buffer *buffer, const struct sm1_dst_register *reg)
{
    assert(reg->writemask);
    put_u32(buffer, (1u << 31) | sm1_encode_register_type(reg->type) | reg->mod | (reg->writemask << 16) | reg->reg);
}

static void write_sm1_src_register(struct vkd3d_bytecode_buffer *buffer,
        const struct sm1_src_register *reg)
{
    put_u32(buffer, (1u << 31) | sm1_encode_register_type(reg->type) | reg->mod | (reg->swizzle << 16) | reg->reg);
}

static void write_sm1_instruction(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct sm1_instruction *instr)
{
    uint32_t token = instr->opcode;
    unsigned int i;

    if (ctx->profile->major_version > 1)
        token |= (instr->has_dst + instr->src_count) << D3DSI_INSTLENGTH_SHIFT;
    put_u32(buffer, token);

    if (instr->has_dst)
        write_sm1_dst_register(buffer, &instr->dst);

    for (i = 0; i < instr->src_count; ++i)
        write_sm1_src_register(buffer, &instr->srcs[i]);
};

static void sm1_map_src_swizzle(struct sm1_src_register *src, unsigned int map_writemask)
{
    src->swizzle = hlsl_map_swizzle(src->swizzle, map_writemask);
}

static void write_sm1_dp2add(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_reg *dst, const struct hlsl_reg *src1, const struct hlsl_reg *src2,
        const struct hlsl_reg *src3)
{
    struct sm1_instruction instr =
    {
        .opcode = D3DSIO_DP2ADD,

        .dst.type = D3DSPR_TEMP,
        .dst.writemask = dst->writemask,
        .dst.reg = dst->id,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(src1->writemask),
        .srcs[0].reg = src1->id,
        .srcs[1].type = D3DSPR_TEMP,
        .srcs[1].swizzle = hlsl_swizzle_from_writemask(src2->writemask),
        .srcs[1].reg = src2->id,
        .srcs[2].type = D3DSPR_TEMP,
        .srcs[2].swizzle = hlsl_swizzle_from_writemask(src3->writemask),
        .srcs[2].reg = src3->id,
        .src_count = 3,
    };

    write_sm1_instruction(ctx, buffer, &instr);
}

static void write_sm1_binary_op(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode, const struct hlsl_reg *dst,
        const struct hlsl_reg *src1, const struct hlsl_reg *src2)
{
    struct sm1_instruction instr =
    {
        .opcode = opcode,

        .dst.type = D3DSPR_TEMP,
        .dst.writemask = dst->writemask,
        .dst.reg = dst->id,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(src1->writemask),
        .srcs[0].reg = src1->id,
        .srcs[1].type = D3DSPR_TEMP,
        .srcs[1].swizzle = hlsl_swizzle_from_writemask(src2->writemask),
        .srcs[1].reg = src2->id,
        .src_count = 2,
    };

    sm1_map_src_swizzle(&instr.srcs[0], instr.dst.writemask);
    sm1_map_src_swizzle(&instr.srcs[1], instr.dst.writemask);
    write_sm1_instruction(ctx, buffer, &instr);
}

static void write_sm1_binary_op_dot(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode, const struct hlsl_reg *dst,
        const struct hlsl_reg *src1, const struct hlsl_reg *src2)
{
    struct sm1_instruction instr =
    {
        .opcode = opcode,

        .dst.type = D3DSPR_TEMP,
        .dst.writemask = dst->writemask,
        .dst.reg = dst->id,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(src1->writemask),
        .srcs[0].reg = src1->id,
        .srcs[1].type = D3DSPR_TEMP,
        .srcs[1].swizzle = hlsl_swizzle_from_writemask(src2->writemask),
        .srcs[1].reg = src2->id,
        .src_count = 2,
    };

    write_sm1_instruction(ctx, buffer, &instr);
}

static void write_sm1_unary_op(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode, const struct hlsl_reg *dst,
        const struct hlsl_reg *src, D3DSHADER_PARAM_SRCMOD_TYPE src_mod, D3DSHADER_PARAM_DSTMOD_TYPE dst_mod)
{
    struct sm1_instruction instr =
    {
        .opcode = opcode,

        .dst.type = D3DSPR_TEMP,
        .dst.mod = dst_mod,
        .dst.writemask = dst->writemask,
        .dst.reg = dst->id,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(src->writemask),
        .srcs[0].reg = src->id,
        .srcs[0].mod = src_mod,
        .src_count = 1,
    };

    sm1_map_src_swizzle(&instr.srcs[0], instr.dst.writemask);
    write_sm1_instruction(ctx, buffer, &instr);
}

static void write_sm1_constant_defs(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer)
{
    unsigned int i, x;

    for (i = 0; i < ctx->constant_defs.count; ++i)
    {
        uint32_t token = D3DSIO_DEF;
        const struct sm1_dst_register reg =
        {
            .type = D3DSPR_CONST,
            .writemask = VKD3DSP_WRITEMASK_ALL,
            .reg = i,
        };

        if (ctx->profile->major_version > 1)
            token |= 5 << D3DSI_INSTLENGTH_SHIFT;
        put_u32(buffer, token);

        write_sm1_dst_register(buffer, &reg);
        for (x = 0; x < 4; ++x)
            put_f32(buffer, ctx->constant_defs.values[i].f[x]);
    }
}

static void write_sm1_semantic_dcl(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_ir_var *var, bool output)
{
    struct sm1_dst_register reg = {0};
    uint32_t token, usage_idx;
    D3DDECLUSAGE usage;
    bool ret;

    if (hlsl_sm1_register_from_semantic(ctx, &var->semantic, output, &reg.type, &reg.reg))
    {
        usage = 0;
        usage_idx = 0;
    }
    else
    {
        ret = hlsl_sm1_usage_from_semantic(&var->semantic, &usage, &usage_idx);
        assert(ret);
        reg.type = output ? D3DSPR_OUTPUT : D3DSPR_INPUT;
        reg.reg = var->regs[HLSL_REGSET_NUMERIC].id;
    }

    token = D3DSIO_DCL;
    if (ctx->profile->major_version > 1)
        token |= 2 << D3DSI_INSTLENGTH_SHIFT;
    put_u32(buffer, token);

    token = (1u << 31);
    token |= usage << D3DSP_DCL_USAGE_SHIFT;
    token |= usage_idx << D3DSP_DCL_USAGEINDEX_SHIFT;
    put_u32(buffer, token);

    reg.writemask = (1 << var->data_type->dimx) - 1;
    write_sm1_dst_register(buffer, &reg);
}

static void write_sm1_semantic_dcls(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer)
{
    bool write_in = false, write_out = false;
    struct hlsl_ir_var *var;

    if (ctx->profile->type == VKD3D_SHADER_TYPE_PIXEL && ctx->profile->major_version >= 2)
        write_in = true;
    else if (ctx->profile->type == VKD3D_SHADER_TYPE_VERTEX && ctx->profile->major_version == 3)
        write_in = write_out = true;
    else if (ctx->profile->type == VKD3D_SHADER_TYPE_VERTEX && ctx->profile->major_version < 3)
        write_in = true;

    LIST_FOR_EACH_ENTRY(var, &ctx->extern_vars, struct hlsl_ir_var, extern_entry)
    {
        if (write_in && var->is_input_semantic)
            write_sm1_semantic_dcl(ctx, buffer, var, false);
        if (write_out && var->is_output_semantic)
            write_sm1_semantic_dcl(ctx, buffer, var, true);
    }
}

static void write_sm1_constant(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_ir_node *instr)
{
    const struct hlsl_ir_constant *constant = hlsl_ir_constant(instr);
    struct sm1_instruction sm1_instr =
    {
        .opcode = D3DSIO_MOV,

        .dst.type = D3DSPR_TEMP,
        .dst.reg = instr->reg.id,
        .dst.writemask = instr->reg.writemask,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_CONST,
        .srcs[0].reg = constant->reg.id,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(constant->reg.writemask),
        .src_count = 1,
    };

    assert(instr->reg.allocated);
    assert(constant->reg.allocated);
    sm1_map_src_swizzle(&sm1_instr.srcs[0], sm1_instr.dst.writemask);
    write_sm1_instruction(ctx, buffer, &sm1_instr);
}

static void write_sm1_per_component_unary_op(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_ir_node *instr, D3DSHADER_INSTRUCTION_OPCODE_TYPE opcode)
{
    struct hlsl_ir_expr *expr = hlsl_ir_expr(instr);
    struct hlsl_ir_node *arg1 = expr->operands[0].node;
    unsigned int i;

    for (i = 0; i < instr->data_type->dimx; ++i)
    {
        struct hlsl_reg src = arg1->reg, dst = instr->reg;

        src.writemask = hlsl_combine_writemasks(src.writemask, 1u << i);
        dst.writemask = hlsl_combine_writemasks(dst.writemask, 1u << i);
        write_sm1_unary_op(ctx, buffer, opcode, &dst, &src, 0, 0);
    }
}

static void write_sm1_expr(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, const struct hlsl_ir_node *instr)
{
    struct hlsl_ir_expr *expr = hlsl_ir_expr(instr);
    struct hlsl_ir_node *arg1 = expr->operands[0].node;
    struct hlsl_ir_node *arg2 = expr->operands[1].node;
    struct hlsl_ir_node *arg3 = expr->operands[2].node;

    assert(instr->reg.allocated);

    if (instr->data_type->base_type != HLSL_TYPE_FLOAT)
    {
        /* These need to be lowered. */
        hlsl_fixme(ctx, &instr->loc, "SM1 non-float expression.");
        return;
    }

    switch (expr->op)
    {
        case HLSL_OP1_ABS:
            write_sm1_unary_op(ctx, buffer, D3DSIO_ABS, &instr->reg, &arg1->reg, 0, 0);
            break;

        case HLSL_OP1_DSX:
            write_sm1_unary_op(ctx, buffer, D3DSIO_DSX, &instr->reg, &arg1->reg, 0, 0);
            break;

        case HLSL_OP1_DSY:
            write_sm1_unary_op(ctx, buffer, D3DSIO_DSY, &instr->reg, &arg1->reg, 0, 0);
            break;

        case HLSL_OP1_EXP2:
            write_sm1_per_component_unary_op(ctx, buffer, instr, D3DSIO_EXP);
            break;

        case HLSL_OP1_LOG2:
            write_sm1_per_component_unary_op(ctx, buffer, instr, D3DSIO_LOG);
            break;

        case HLSL_OP1_NEG:
            write_sm1_unary_op(ctx, buffer, D3DSIO_MOV, &instr->reg, &arg1->reg, D3DSPSM_NEG, 0);
            break;

        case HLSL_OP1_SAT:
            write_sm1_unary_op(ctx, buffer, D3DSIO_MOV, &instr->reg, &arg1->reg, 0, D3DSPDM_SATURATE);
            break;

        case HLSL_OP1_RCP:
            write_sm1_per_component_unary_op(ctx, buffer, instr, D3DSIO_RCP);
            break;

        case HLSL_OP1_RSQ:
            write_sm1_per_component_unary_op(ctx, buffer, instr, D3DSIO_RSQ);
            break;

        case HLSL_OP2_ADD:
            write_sm1_binary_op(ctx, buffer, D3DSIO_ADD, &instr->reg, &arg1->reg, &arg2->reg);
            break;

        case HLSL_OP2_MAX:
            write_sm1_binary_op(ctx, buffer, D3DSIO_MAX, &instr->reg, &arg1->reg, &arg2->reg);
            break;

        case HLSL_OP2_MIN:
            write_sm1_binary_op(ctx, buffer, D3DSIO_MIN, &instr->reg, &arg1->reg, &arg2->reg);
            break;

        case HLSL_OP2_MUL:
            write_sm1_binary_op(ctx, buffer, D3DSIO_MUL, &instr->reg, &arg1->reg, &arg2->reg);
            break;

        case HLSL_OP1_FRACT:
            write_sm1_unary_op(ctx, buffer, D3DSIO_FRC, &instr->reg, &arg1->reg, D3DSPSM_NONE, 0);
            break;

        case HLSL_OP2_DOT:
            switch (arg1->data_type->dimx)
            {
                case 4:
                    write_sm1_binary_op_dot(ctx, buffer, D3DSIO_DP4, &instr->reg, &arg1->reg, &arg2->reg);
                    break;

                case 3:
                    write_sm1_binary_op_dot(ctx, buffer, D3DSIO_DP3, &instr->reg, &arg1->reg, &arg2->reg);
                    break;

                default:
                    vkd3d_unreachable();
            }
            break;

        case HLSL_OP3_DP2ADD:
            write_sm1_dp2add(ctx, buffer, &instr->reg, &arg1->reg, &arg2->reg, &arg3->reg);
            break;

        default:
            hlsl_fixme(ctx, &instr->loc, "SM1 \"%s\" expression.", debug_hlsl_expr_op(expr->op));
            break;
    }
}

static void write_sm1_load(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer, const struct hlsl_ir_node *instr)
{
    const struct hlsl_ir_load *load = hlsl_ir_load(instr);
    const struct hlsl_reg reg = hlsl_reg_from_deref(ctx, &load->src);
    struct sm1_instruction sm1_instr =
    {
        .opcode = D3DSIO_MOV,

        .dst.type = D3DSPR_TEMP,
        .dst.reg = instr->reg.id,
        .dst.writemask = instr->reg.writemask,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].reg = reg.id,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(reg.writemask),
        .src_count = 1,
    };

    assert(instr->reg.allocated);

    if (load->src.var->is_uniform)
    {
        assert(reg.allocated);
        sm1_instr.srcs[0].type = D3DSPR_CONST;
    }
    else if (load->src.var->is_input_semantic)
    {
        if (!hlsl_sm1_register_from_semantic(ctx, &load->src.var->semantic,
                false, &sm1_instr.srcs[0].type, &sm1_instr.srcs[0].reg))
        {
            assert(reg.allocated);
            sm1_instr.srcs[0].type = D3DSPR_INPUT;
            sm1_instr.srcs[0].reg = reg.id;
        }
        else
            sm1_instr.srcs[0].swizzle = hlsl_swizzle_from_writemask((1 << load->src.var->data_type->dimx) - 1);
    }

    sm1_map_src_swizzle(&sm1_instr.srcs[0], sm1_instr.dst.writemask);
    write_sm1_instruction(ctx, buffer, &sm1_instr);
}

static void write_sm1_store(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_ir_node *instr)
{
    const struct hlsl_ir_store *store = hlsl_ir_store(instr);
    const struct hlsl_ir_node *rhs = store->rhs.node;
    const struct hlsl_reg reg = hlsl_reg_from_deref(ctx, &store->lhs);
    struct sm1_instruction sm1_instr =
    {
        .opcode = D3DSIO_MOV,

        .dst.type = D3DSPR_TEMP,
        .dst.reg = reg.id,
        .dst.writemask = hlsl_combine_writemasks(reg.writemask, store->writemask),
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].reg = rhs->reg.id,
        .srcs[0].swizzle = hlsl_swizzle_from_writemask(rhs->reg.writemask),
        .src_count = 1,
    };

    if (store->lhs.var->data_type->class == HLSL_CLASS_MATRIX)
    {
        FIXME("Matrix writemasks need to be lowered.\n");
        return;
    }

    if (store->lhs.var->is_output_semantic)
    {
        if (!hlsl_sm1_register_from_semantic(ctx, &store->lhs.var->semantic,
                true, &sm1_instr.dst.type, &sm1_instr.dst.reg))
        {
            assert(reg.allocated);
            sm1_instr.dst.type = D3DSPR_OUTPUT;
            sm1_instr.dst.reg = reg.id;
        }
        else
            sm1_instr.dst.writemask = (1u << store->lhs.var->data_type->dimx) - 1;
    }
    else
        assert(reg.allocated);

    sm1_map_src_swizzle(&sm1_instr.srcs[0], sm1_instr.dst.writemask);
    write_sm1_instruction(ctx, buffer, &sm1_instr);
}

static void write_sm1_swizzle(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_ir_node *instr)
{
    const struct hlsl_ir_swizzle *swizzle = hlsl_ir_swizzle(instr);
    const struct hlsl_ir_node *val = swizzle->val.node;
    struct sm1_instruction sm1_instr =
    {
        .opcode = D3DSIO_MOV,

        .dst.type = D3DSPR_TEMP,
        .dst.reg = instr->reg.id,
        .dst.writemask = instr->reg.writemask,
        .has_dst = 1,

        .srcs[0].type = D3DSPR_TEMP,
        .srcs[0].reg = val->reg.id,
        .srcs[0].swizzle = hlsl_combine_swizzles(hlsl_swizzle_from_writemask(val->reg.writemask),
                swizzle->swizzle, instr->data_type->dimx),
        .src_count = 1,
    };

    assert(instr->reg.allocated);
    assert(val->reg.allocated);
    sm1_map_src_swizzle(&sm1_instr.srcs[0], sm1_instr.dst.writemask);
    write_sm1_instruction(ctx, buffer, &sm1_instr);
}

static void write_sm1_instructions(struct hlsl_ctx *ctx, struct vkd3d_bytecode_buffer *buffer,
        const struct hlsl_ir_function_decl *entry_func)
{
    const struct hlsl_ir_node *instr;

    LIST_FOR_EACH_ENTRY(instr, &entry_func->body.instrs, struct hlsl_ir_node, entry)
    {
        if (instr->data_type)
        {
            if (instr->data_type->class == HLSL_CLASS_MATRIX)
            {
                /* These need to be lowered. */
                hlsl_fixme(ctx, &instr->loc, "SM1 matrix expression.");
                continue;
            }
            else if (instr->data_type->class == HLSL_CLASS_OBJECT)
            {
                hlsl_fixme(ctx, &instr->loc, "Object copy.");
                break;
            }

            assert(instr->data_type->class == HLSL_CLASS_SCALAR || instr->data_type->class == HLSL_CLASS_VECTOR);
        }

        switch (instr->type)
        {
            case HLSL_IR_CALL:
                vkd3d_unreachable();

            case HLSL_IR_CONSTANT:
                write_sm1_constant(ctx, buffer, instr);
                break;

            case HLSL_IR_EXPR:
                write_sm1_expr(ctx, buffer, instr);
                break;

            case HLSL_IR_LOAD:
                write_sm1_load(ctx, buffer, instr);
                break;

            case HLSL_IR_STORE:
                write_sm1_store(ctx, buffer, instr);
                break;

            case HLSL_IR_SWIZZLE:
                write_sm1_swizzle(ctx, buffer, instr);
                break;

            default:
                hlsl_fixme(ctx, &instr->loc, "Instruction type %s.", hlsl_node_type_to_string(instr->type));
        }
    }
}

int hlsl_sm1_write(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry_func, struct vkd3d_shader_code *out)
{
    struct vkd3d_bytecode_buffer buffer = {0};
    int ret;

    put_u32(&buffer, sm1_version(ctx->profile->type, ctx->profile->major_version, ctx->profile->minor_version));

    write_sm1_uniforms(ctx, &buffer, entry_func);

    write_sm1_constant_defs(ctx, &buffer);
    write_sm1_semantic_dcls(ctx, &buffer);
    write_sm1_instructions(ctx, &buffer, entry_func);

    put_u32(&buffer, D3DSIO_END);

    if (!(ret = buffer.status))
    {
        out->code = buffer.data;
        out->size = buffer.size;
    }
    return ret;
}