diff options
Diffstat (limited to 'py/emitx64.c')
| -rw-r--r-- | py/emitx64.c | 530 |
1 files changed, 388 insertions, 142 deletions
diff --git a/py/emitx64.c b/py/emitx64.c index 5af5614385..5b2b6f0c56 100644 --- a/py/emitx64.c +++ b/py/emitx64.c @@ -1,15 +1,20 @@ -/* This code is equivalent to emitx64.c but pre-allocates stack - * space and uses mov instead of push/pop instructions to access - * the temporary stack. It runs in similar time, but uses 3*n - * more bytes, where n is number of push/pop instructions. - * - * This code is preferred because it keeps the stack aligned on a - * 16 byte boundary. - * - * Improvements: - * Doesn't call stub functions, does all the work inline. - * Has optimisations for loading i64s to stack. - */ +// Essentially normal Python has 1 type: Python objects +// Viper has more than 1 type, and is just a more complicated (a superset of) Python. +// If you declare everything in Viper as a Python object (ie omit type decls) then +// it should in principle be exactly the same as Python native. +// Having types means having more opcodes, like binary_op_nat_nat, binary_op_nat_obj etc. +// In practice we won't have a VM but rather do this in asm which is actually very minimal. + +// Because it breaks strict Python equivalence it should be a completely separate +// decorator. It breaks equivalence because overflow on integers wraps around. +// It shouldn't break equivalence if you don't use the new types, but since the +// type decls might be used in normal Python for other reasons, it's probably safest, +// cleanest and clearest to make it a separate decorator. + +// Actually, it does break equivalence because integers default to native integers, +// not Python objects. + +// for x in l[0:8]: can be compiled into a native loop if l has pointer type #include <unistd.h> #include <stdlib.h> @@ -38,30 +43,49 @@ typedef enum { NEED_TO_PUSH_I64, } need_to_push_t; +typedef enum { + VTYPE_UNBOUND, + VTYPE_PYOBJ, + VTYPE_BOOL, + VTYPE_INT, + VTYPE_PTR, + VTYPE_PTR_NONE, + VTYPE_BUILTIN_V_INT, +} vtype_kind_t; + struct _emit_t { int pass; + + bool do_viper_types; + int all_vtype_alloc; + vtype_kind_t *all_vtype; + vtype_kind_t *local_vtype; + vtype_kind_t *stack_vtype; int stack_start; int stack_size; + bool last_emit_was_return_value; + need_to_push_t need_to_push; - int last_r64; - int64_t last_i64; + vtype_kind_t last_vtype; + int last_reg; + int64_t last_imm; scope_t *scope; asm_x64_t *as; - bool do_native_types; }; emit_t *emit_x64_new(uint max_num_labels) { emit_t *emit = m_new(emit_t, 1); + emit->do_viper_types = false; + emit->all_vtype = NULL; emit->as = asm_x64_new(max_num_labels); - emit->do_native_types = false; return emit; } -static void emit_x64_set_native_types(emit_t *emit, bool do_native_types) { - emit->do_native_types = do_native_types; +static void emit_x64_set_viper_types(emit_t *emit, bool do_viper_types) { + emit->do_viper_types = do_viper_types; } static void emit_x64_start_pass(emit_t *emit, pass_kind_t pass, scope_t *scope) { @@ -72,6 +96,24 @@ static void emit_x64_start_pass(emit_t *emit, pass_kind_t pass, scope_t *scope) emit->need_to_push = NEED_TO_PUSH_NOTHING; emit->scope = scope; + if (emit->all_vtype == NULL) { + emit->all_vtype_alloc = scope->num_locals + scope->stack_size + 100; // XXX don't know stack size on entry, should be maximum over all scopes + emit->all_vtype = m_new(vtype_kind_t, emit->all_vtype_alloc); + emit->local_vtype = emit->all_vtype; + emit->stack_vtype = emit->all_vtype + scope->num_locals; + } + + if (emit->do_viper_types) { + for (int i = 0; i < emit->all_vtype_alloc; i++) { + emit->all_vtype[i] = VTYPE_UNBOUND; + } + // TODO set types of arguments based on type signature + } else { + for (int i = 0; i < emit->all_vtype_alloc; i++) { + emit->all_vtype[i] = VTYPE_PYOBJ; + } + } + asm_x64_start_pass(emit->as, pass); // entry to function @@ -144,12 +186,14 @@ static void stack_settle(emit_t *emit) { break; case NEED_TO_PUSH_R64: - asm_x64_mov_r64_to_local(emit->as, emit->last_r64, emit->stack_start + emit->stack_size); + emit->stack_vtype[emit->stack_size] = emit->last_vtype; + asm_x64_mov_r64_to_local(emit->as, emit->last_reg, emit->stack_start + emit->stack_size); adjust_stack(emit, 1); break; case NEED_TO_PUSH_I64: - asm_x64_mov_i64_to_r64_optimised(emit->as, emit->last_i64, REG_RAX); + emit->stack_vtype[emit->stack_size] = emit->last_vtype; + asm_x64_mov_i64_to_r64_optimised(emit->as, emit->last_imm, REG_RAX); asm_x64_mov_r64_to_local(emit->as, REG_RAX, emit->stack_start + emit->stack_size); adjust_stack(emit, 1); break; @@ -167,36 +211,57 @@ static void emit_pre(emit_t *emit) { emit_pre_raw(emit, 0); } -static void emit_pre_pop_r64(emit_t *emit, int r64) { +static vtype_kind_t peek_vtype(emit_t *emit) { switch (emit->need_to_push) { case NEED_TO_PUSH_NOTHING: - asm_x64_mov_local_to_r64(emit->as, emit->stack_start + emit->stack_size - 1, r64); + return emit->stack_vtype[emit->stack_size - 1]; + + case NEED_TO_PUSH_R64: + case NEED_TO_PUSH_I64: + return emit->last_vtype; + + default: + assert(0); + return VTYPE_UNBOUND; + } +} + +static void emit_pre_pop_reg(emit_t *emit, vtype_kind_t *vtype, int reg) { + switch (emit->need_to_push) { + case NEED_TO_PUSH_NOTHING: + *vtype = emit->stack_vtype[emit->stack_size - 1]; + asm_x64_mov_local_to_r64(emit->as, emit->stack_start + emit->stack_size - 1, reg); emit_pre_raw(emit, -1); break; case NEED_TO_PUSH_R64: emit_pre_raw(emit, 0); - if (emit->last_r64 != r64) { - asm_x64_mov_r64_to_r64(emit->as, emit->last_r64, r64); + *vtype = emit->last_vtype; + if (emit->last_reg != reg) { + asm_x64_mov_r64_to_r64(emit->as, emit->last_reg, reg); } break; case NEED_TO_PUSH_I64: emit_pre_raw(emit, 0); - asm_x64_mov_i64_to_r64_optimised(emit->as, emit->last_i64, r64); + *vtype = emit->last_vtype; + asm_x64_mov_i64_to_r64_optimised(emit->as, emit->last_imm, reg); break; } emit->need_to_push = NEED_TO_PUSH_NOTHING; } -static void emit_pre_pop_r64_r64(emit_t *emit, int r64a, int r64b) { - emit_pre_pop_r64(emit, r64a); +static void emit_pre_pop_reg_reg(emit_t *emit, vtype_kind_t *vtypea, int r64a, vtype_kind_t *vtypeb, int r64b) { + emit_pre_pop_reg(emit, vtypea, r64a); + *vtypeb = emit->stack_vtype[emit->stack_size - 1]; asm_x64_mov_local_to_r64(emit->as, emit->stack_start + emit->stack_size - 1, r64b); adjust_stack(emit, -1); } -static void emit_pre_pop_r64_r64_r64(emit_t *emit, int r64a, int r64b, int r64c) { - emit_pre_pop_r64(emit, r64a); +static void emit_pre_pop_reg_reg_reg(emit_t *emit, vtype_kind_t *vtypea, int r64a, vtype_kind_t *vtypeb, int r64b, vtype_kind_t *vtypec, int r64c) { + emit_pre_pop_reg(emit, vtypea, r64a); + *vtypeb = emit->stack_vtype[emit->stack_size - 1]; + *vtypec = emit->stack_vtype[emit->stack_size - 2]; asm_x64_mov_local_to_r64(emit->as, emit->stack_start + emit->stack_size - 1, r64b); asm_x64_mov_local_to_r64(emit->as, emit->stack_start + emit->stack_size - 2, r64c); adjust_stack(emit, -2); @@ -205,31 +270,42 @@ static void emit_pre_pop_r64_r64_r64(emit_t *emit, int r64a, int r64b, int r64c) static void emit_post(emit_t *emit) { } -static void emit_post_push_r64(emit_t *emit, int r64) { +static void emit_post_push_reg(emit_t *emit, vtype_kind_t vtype, int reg) { emit->need_to_push = NEED_TO_PUSH_R64; - emit->last_r64 = r64; + emit->last_vtype = vtype; + emit->last_reg = reg; } -static void emit_post_push_i64(emit_t *emit, int64_t i64) { +static void emit_post_push_imm(emit_t *emit, vtype_kind_t vtype, int64_t imm) { emit->need_to_push = NEED_TO_PUSH_I64; - emit->last_i64 = i64; + emit->last_vtype = vtype; + emit->last_imm = imm; } -static void emit_post_push_r64_r64(emit_t *emit, int r64a, int r64b) { +static void emit_post_push_reg_reg(emit_t *emit, vtype_kind_t vtypea, int r64a, vtype_kind_t vtypeb, int r64b) { + emit->stack_vtype[emit->stack_size] = vtypea; asm_x64_mov_r64_to_local(emit->as, r64a, emit->stack_start + emit->stack_size); emit->need_to_push = NEED_TO_PUSH_R64; - emit->last_r64 = r64b; + emit->last_vtype = vtypeb; + emit->last_reg = r64b; adjust_stack(emit, 1); } -static void emit_post_push_r64_r64_r64(emit_t *emit, int r64a, int r64b, int r64c) { +static void emit_post_push_reg_reg_reg(emit_t *emit, vtype_kind_t vtypea, int r64a, vtype_kind_t vtypeb, int r64b, vtype_kind_t vtypec, int r64c) { + emit->stack_vtype[emit->stack_size] = vtypea; + emit->stack_vtype[emit->stack_size + 1] = vtypeb; + emit->stack_vtype[emit->stack_size + 2] = vtypec; asm_x64_mov_r64_to_local(emit->as, r64a, emit->stack_start + emit->stack_size); asm_x64_mov_r64_to_local(emit->as, r64b, emit->stack_start + emit->stack_size + 1); asm_x64_mov_r64_to_local(emit->as, r64c, emit->stack_start + emit->stack_size + 2); adjust_stack(emit, 3); } -static void emit_post_push_r64_r64_r64_r64(emit_t *emit, int r64a, int r64b, int r64c, int r64d) { +static void emit_post_push_reg_reg_reg_reg(emit_t *emit, vtype_kind_t vtypea, int r64a, vtype_kind_t vtypeb, int r64b, vtype_kind_t vtypec, int r64c, vtype_kind_t vtyped, int r64d) { + emit->stack_vtype[emit->stack_size] = vtypea; + emit->stack_vtype[emit->stack_size + 1] = vtypeb; + emit->stack_vtype[emit->stack_size + 2] = vtypec; + emit->stack_vtype[emit->stack_size + 3] = vtyped; asm_x64_mov_r64_to_local(emit->as, r64a, emit->stack_start + emit->stack_size); asm_x64_mov_r64_to_local(emit->as, r64b, emit->stack_start + emit->stack_size + 1); asm_x64_mov_r64_to_local(emit->as, r64c, emit->stack_start + emit->stack_size + 2); @@ -237,13 +313,18 @@ static void emit_post_push_r64_r64_r64_r64(emit_t *emit, int r64a, int r64b, int adjust_stack(emit, 4); } -static void emit_get_stack_pointer_to_r64_for_pop(emit_t *emit, int r64, int n_pop) { - asm_x64_mov_local_addr_to_r64(emit->as, emit->stack_start + emit->stack_size - 1, r64); +// vtype of all n_pop objects is VTYPE_PYOBJ +static void emit_get_stack_pointer_to_reg_for_pop(emit_t *emit, int reg, int n_pop) { + asm_x64_mov_local_addr_to_r64(emit->as, emit->stack_start + emit->stack_size - 1, reg); adjust_stack(emit, -n_pop); } -static void emit_get_stack_pointer_to_r64_for_push(emit_t *emit, int r64, int n_push) { - asm_x64_mov_local_addr_to_r64(emit->as, emit->stack_start + emit->stack_size + n_push - 1, r64); +// vtype of all n_push objects is VTYPE_PYOBJ +static void emit_get_stack_pointer_to_reg_for_push(emit_t *emit, int reg, int n_push) { + for (int i = 0; i < n_push; i++) { + emit->stack_vtype[emit->stack_size + i] = VTYPE_PYOBJ; + } + asm_x64_mov_local_addr_to_r64(emit->as, emit->stack_start + emit->stack_size + n_push - 1, reg); adjust_stack(emit, n_push); } @@ -251,20 +332,30 @@ static void emit_call(emit_t *emit, void *fun) { asm_x64_call_ind(emit->as, fun, REG_RAX); } -static void emit_call_with_i64_arg(emit_t *emit, void *fun, int64_t arg_val, int arg_r64) { - asm_x64_mov_i64_to_r64_optimised(emit->as, arg_val, arg_r64); +static void emit_call_with_imm_arg(emit_t *emit, void *fun, int64_t arg_val, int arg_reg) { + asm_x64_mov_i64_to_r64_optimised(emit->as, arg_val, arg_reg); asm_x64_call_ind(emit->as, fun, REG_RAX); } static void emit_x64_load_id(emit_t *emit, qstr qstr) { - emit_common_load_id(emit, &emit_x64_method_table, emit->scope, qstr); + // check for built-ins + if (strcmp(qstr_str(qstr), "v_int") == 0) { + emit_pre(emit); + //emit_post_push_blank(emit, VTYPE_BUILTIN_V_INT); + + // not a built-in, so do usual thing + } else { + emit_common_load_id(emit, &emit_x64_method_table, emit->scope, qstr); + } } static void emit_x64_store_id(emit_t *emit, qstr qstr) { + // TODO check for built-ins and disallow emit_common_store_id(emit, &emit_x64_method_table, emit->scope, qstr); } static void emit_x64_delete_id(emit_t *emit, qstr qstr) { + // TODO check for built-ins and disallow emit_common_delete_id(emit, &emit_x64_method_table, emit->scope, qstr); } @@ -273,184 +364,285 @@ static void emit_x64_label_assign(emit_t *emit, int l) { } static void emit_x64_import_name(emit_t *emit, qstr qstr) { + // not supported assert(0); } + static void emit_x64_import_from(emit_t *emit, qstr qstr) { + // not supported assert(0); } + static void emit_x64_import_star(emit_t *emit) { + // not supported assert(0); } static void emit_x64_load_const_tok(emit_t *emit, py_token_kind_t tok) { emit_pre(emit); - py_obj_t o; - switch (tok) { - case PY_TOKEN_KW_NONE: o = py_const_none; break; - case PY_TOKEN_KW_FALSE: o = py_const_false; break; - case PY_TOKEN_KW_TRUE: o = py_const_true; break; - default: assert(0); // shouldn't happen + int vtype; + machine_uint_t val; + if (emit->do_viper_types) { + switch (tok) { + case PY_TOKEN_KW_NONE: vtype = VTYPE_PTR_NONE; val = 0; break; + case PY_TOKEN_KW_FALSE: vtype = VTYPE_BOOL; val = 0; break; + case PY_TOKEN_KW_TRUE: vtype = VTYPE_BOOL; val = 1; break; + default: assert(0); vtype = 0; val = 0; // shouldn't happen + } + } else { + vtype = VTYPE_PYOBJ; + switch (tok) { + case PY_TOKEN_KW_NONE: val = (machine_uint_t)py_const_none; break; + case PY_TOKEN_KW_FALSE: val = (machine_uint_t)py_const_false; break; + case PY_TOKEN_KW_TRUE: val = (machine_uint_t)py_const_true; break; + default: assert(0); vtype = 0; val = 0; // shouldn't happen + } } - emit_post_push_i64(emit, (uint64_t)o); + emit_post_push_imm(emit, vtype, val); } static void emit_x64_load_const_small_int(emit_t *emit, int arg) { emit_pre(emit); - if (emit->do_native_types) { - emit_post_push_i64(emit, arg); + if (emit->do_viper_types) { + emit_post_push_imm(emit, VTYPE_INT, arg); } else { - emit_post_push_i64(emit, (arg << 1) | 1); + emit_post_push_imm(emit, VTYPE_PYOBJ, (arg << 1) | 1); } } static void emit_x64_load_const_int(emit_t *emit, qstr qstr) { + // not implemented + // load integer, check fits in 32 bits assert(0); } + static void emit_x64_load_const_dec(emit_t *emit, qstr qstr) { + // not supported (although, could support floats in future) assert(0); } + static void emit_x64_load_const_id(emit_t *emit, qstr qstr) { + // not supported? assert(0); } static void emit_x64_load_const_str(emit_t *emit, qstr qstr, bool bytes) { emit_pre(emit); - emit_call_with_i64_arg(emit, rt_load_const_str, qstr, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); + if (emit->do_viper_types) { + // not implemented properly + // load a pointer to the asciiz string? + assert(0); + emit_post_push_imm(emit, VTYPE_PTR, (machine_uint_t)qstr_str(qstr)); + } else { + emit_call_with_imm_arg(emit, rt_load_const_str, qstr, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); + } } static void emit_x64_load_const_verbatim_start(emit_t *emit) { + // not supported/needed assert(0); } + static void emit_x64_load_const_verbatim_int(emit_t *emit, int val) { + // not supported/needed assert(0); } + static void emit_x64_load_const_verbatim_str(emit_t *emit, const char *str) { + // not supported/needed assert(0); } + static void emit_x64_load_const_verbatim_strn(emit_t *emit, const char *str, int len) { + // not supported/needed assert(0); } + static void emit_x64_load_const_verbatim_quoted_str(emit_t *emit, qstr qstr, bool bytes) { + // not supported/needed assert(0); } + static void emit_x64_load_const_verbatim_end(emit_t *emit) { + // not supported/needed assert(0); } static void emit_x64_load_fast(emit_t *emit, qstr qstr, int local_num) { + vtype_kind_t vtype = emit->local_vtype[local_num]; + if (vtype == VTYPE_UNBOUND) { + printf("ViperTypeError: local %s used before type known\n", qstr_str(qstr)); + } if (local_num == 0) { emit_pre(emit); - emit_post_push_r64(emit, REG_LOCAL_1); + emit_post_push_reg(emit, vtype, REG_LOCAL_1); } else { emit_pre(emit); asm_x64_mov_local_to_r64(emit->as, local_num - 1, REG_RAX); - emit_post_push_r64(emit, REG_RAX); + emit_post_push_reg(emit, vtype, REG_RAX); } } static void emit_x64_load_name(emit_t *emit, qstr qstr) { emit_pre(emit); - emit_call_with_i64_arg(emit, rt_load_name, qstr, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); + emit_call_with_imm_arg(emit, rt_load_name, qstr, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); } static void emit_x64_load_global(emit_t *emit, qstr qstr) { emit_pre(emit); - emit_call_with_i64_arg(emit, rt_load_global, qstr, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); + emit_call_with_imm_arg(emit, rt_load_global, qstr, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); } static void emit_x64_load_deref(emit_t *emit, qstr qstr) { + // not implemented + // in principle could support this quite easily (ldr r0, [r0, #0]) and then get closed over variables! assert(0); } + static void emit_x64_load_closure(emit_t *emit, qstr qstr) { + // not implemented assert(0); } static void emit_x64_load_attr(emit_t *emit, qstr qstr) { - emit_pre_pop_r64(emit, REG_ARG_1); // arg1 = base - emit_call_with_i64_arg(emit, rt_load_attr, qstr, REG_ARG_2); // arg2 = attribute name - emit_post_push_r64(emit, REG_RET); + // depends on type of subject: + // - integer, function, pointer to integers: error + // - pointer to structure: get member, quite easy + // - Python object: call rt_load_attr, and needs to be typed to convert result + vtype_kind_t vtype_base; + emit_pre_pop_reg(emit, &vtype_base, REG_ARG_1); // arg1 = base + assert(vtype_base == VTYPE_PYOBJ); + emit_call_with_imm_arg(emit, rt_load_attr, qstr, REG_ARG_2); // arg2 = attribute name + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); } static void emit_x64_load_method(emit_t *emit, qstr qstr) { - emit_pre_pop_r64(emit, REG_ARG_1); // arg1 = base - emit_get_stack_pointer_to_r64_for_push(emit, REG_ARG_3, 2); // arg3 = dest ptr - emit_call_with_i64_arg(emit, rt_load_method, qstr, REG_ARG_2); // arg2 = method name + vtype_kind_t vtype_base; + emit_pre_pop_reg(emit, &vtype_base, REG_ARG_1); // arg1 = base + assert(vtype_base == VTYPE_PYOBJ); + emit_get_stack_pointer_to_reg_for_push(emit, REG_ARG_3, 2); // arg3 = dest ptr + emit_call_with_imm_arg(emit, rt_load_method, qstr, REG_ARG_2); // arg2 = method name } static void emit_x64_load_build_class(emit_t *emit) { + // not supported assert(0); -} // basically load __build_class__ from builtins +} static void emit_x64_store_fast(emit_t *emit, qstr qstr, int local_num) { + vtype_kind_t vtype; if (local_num == 0) { - emit_pre_pop_r64(emit, REG_LOCAL_1); + emit_pre_pop_reg(emit, &vtype, REG_LOCAL_1); emit_post(emit); } else { - emit_pre_pop_r64(emit, REG_RAX); + emit_pre_pop_reg(emit, &vtype, REG_RAX); asm_x64_mov_r64_to_local(emit->as, REG_RAX, local_num - 1); emit_post(emit); } + + // check types + if (emit->local_vtype[local_num] == VTYPE_UNBOUND) { + // first time this local is assigned, so give it a type of the object stored in it + emit->local_vtype[local_num] = vtype; + } else if (emit->local_vtype[local_num] != vtype) { + // type of local is not the same as object stored in it + printf("ViperTypeError: type mismatch, local %s has type %d but source object has type %d\n", qstr_str(qstr), emit->local_vtype[local_num], vtype); + } } static void emit_x64_store_name(emit_t *emit, qstr qstr) { - emit_pre_pop_r64(emit, REG_ARG_2); - emit_call_with_i64_arg(emit, rt_store_name, qstr, REG_ARG_1); // arg1 = name + // rt_store_name, but needs conversion of object (maybe have rt_viper_store_name(obj, type)) + vtype_kind_t vtype; + emit_pre_pop_reg(emit, &vtype, REG_ARG_2); + assert(vtype == VTYPE_PYOBJ); + emit_call_with_imm_arg(emit, rt_store_name, qstr, REG_ARG_1); // arg1 = name emit_post(emit); } static void emit_x64_store_global(emit_t *emit, qstr qstr) { + // not implemented assert(0); } static void emit_x64_store_deref(emit_t *emit, qstr qstr) { + // not implemented assert(0); } + static void emit_x64_store_attr(emit_t *emit, qstr qstr) { + // not implemented assert(0); } + static void emit_x64_store_locals(emit_t *emit) { + // not supported assert(0); } static void emit_x64_store_subscr(emit_t *emit) { - emit_pre_pop_r64_r64_r64(emit, REG_ARG_2, REG_ARG_1, REG_ARG_3); // index, base, value to store + // depends on type of subject: + // - integer, function, pointer to structure: error + // - pointer to integers: store as per array + // - Python object: call runtime with converted object or type info + vtype_kind_t vtype_index, vtype_base, vtype_value; + emit_pre_pop_reg_reg_reg(emit, &vtype_index, REG_ARG_2, &vtype_base, REG_ARG_1, &vtype_value, REG_ARG_3); // index, base, value to store + assert(vtype_index == VTYPE_PYOBJ); + assert(vtype_base == VTYPE_PYOBJ); + assert(vtype_value == VTYPE_PYOBJ); emit_call(emit, rt_store_subscr); } static void emit_x64_delete_fast(emit_t *emit, qstr qstr, int local_num) { + // not implemented + // could support for Python types, just set to None (so GC can reclaim it) assert(0); } + static void emit_x64_delete_name(emit_t *emit, qstr qstr) { + // not implemented + // use rt_delete_name assert(0); } + static void emit_x64_delete_global(emit_t *emit, qstr qstr) { + // not implemented + // use rt_delete_global assert(0); } + static void emit_x64_delete_deref(emit_t *emit, qstr qstr) { + // not supported assert(0); } + static void emit_x64_delete_attr(emit_t *emit, qstr qstr) { + // not supported assert(0); } + static void emit_x64_delete_subscr(emit_t *emit) { + // not supported assert(0); } static void emit_x64_dup_top(emit_t *emit) { - emit_pre_pop_r64(emit, REG_RAX); - emit_post_push_r64_r64(emit, REG_RAX, REG_RAX); + vtype_kind_t vtype; + emit_pre_pop_reg(emit, &vtype, REG_RAX); + emit_post_push_reg_reg(emit, vtype, REG_RAX, vtype, REG_RAX); } static void emit_x64_dup_top_two(emit_t *emit) { - emit_pre_pop_r64_r64(emit, REG_RAX, REG_RDI); - emit_post_push_r64_r64_r64_r64(emit, REG_RDI, REG_RAX, REG_RDI, REG_RAX); + vtype_kind_t vtype1, vtype2; + emit_pre_pop_reg_reg(emit, &vtype1, REG_RAX, &vtype2, REG_RDI); + emit_post_push_reg_reg_reg_reg(emit, vtype2, REG_RDI, vtype1, REG_RAX, vtype2, REG_RDI, vtype1, REG_RAX); } static void emit_x64_pop_top(emit_t *emit) { - emit_pre_pop_r64(emit, REG_RAX); + vtype_kind_t vtype; + emit_pre_pop_reg(emit, &vtype, REG_RAX); emit_post(emit); } @@ -459,8 +651,9 @@ static void emit_x64_rot_two(emit_t *emit) { } static void emit_x64_rot_three(emit_t *emit) { - emit_pre_pop_r64_r64_r64(emit, REG_RAX, REG_RDI, REG_RSI); - emit_post_push_r64_r64_r64(emit, REG_RAX, REG_RSI, REG_RDI); + vtype_kind_t vtype_rax, vtype_rdi, vtype_rsi; + emit_pre_pop_reg_reg_reg(emit, &vtype_rax, REG_RAX, &vtype_rdi, REG_RDI, &vtype_rsi, REG_RSI); + emit_post_push_reg_reg_reg(emit, vtype_rax, REG_RAX, vtype_rsi, REG_RSI, vtype_rdi, REG_RDI); } static void emit_x64_jump(emit_t *emit, int label) { @@ -470,17 +663,21 @@ static void emit_x64_jump(emit_t *emit, int label) { } static void emit_x64_pop_jump_if_false(emit_t *emit, int label) { - if (emit->do_native_types) { - emit_pre_pop_r64(emit, REG_RET); - asm_x64_test_r8_with_r8(emit->as, REG_RET, REG_RET); + vtype_kind_t vtype = peek_vtype(emit); + if (vtype == VTYPE_BOOL) { + emit_pre_pop_reg(emit, &vtype, REG_RAX); + asm_x64_test_r8_with_r8(emit->as, REG_RAX, REG_RAX); asm_x64_jcc_label(emit->as, JCC_JZ, label); emit_post(emit); - } else { - emit_pre_pop_r64(emit, REG_ARG_1); + } else if (vtype == VTYPE_PYOBJ) { + emit_pre_pop_reg(emit, &vtype, REG_ARG_1); emit_call(emit, rt_is_true); asm_x64_test_r8_with_r8(emit->as, REG_RET, REG_RET); asm_x64_jcc_label(emit->as, JCC_JZ, label); emit_post(emit); + } else { + printf("ViperTypeError: expecting a bool or pyobj, got %d\n", vtype); + assert(0); } } @@ -506,6 +703,7 @@ static void emit_x64_continue_loop(emit_t *emit, int label) { assert(0); } static void emit_x64_setup_with(emit_t *emit, int label) { + // not supported, or could be with runtime call assert(0); } static void emit_x64_with_cleanup(emit_t *emit) { @@ -521,6 +719,8 @@ static void emit_x64_end_finally(emit_t *emit) { assert(0); } static void emit_x64_get_iter(emit_t *emit) { + // perhaps the difficult one, as we want to rewrite for loops using native code + // in cases where we iterate over a Python object, can we use normal runtime calls? assert(0); } // tos = getiter(tos) static void emit_x64_for_iter(emit_t *emit, int label) { @@ -529,41 +729,90 @@ static void emit_x64_for_iter(emit_t *emit, int label) { static void emit_x64_for_iter_end(emit_t *emit) { assert(0); } + +static void emit_x64_pop_block(emit_t *emit) { + emit_pre(emit); + emit_post(emit); +} + static void emit_x64_pop_except(emit_t *emit) { assert(0); } static void emit_x64_unary_op(emit_t *emit, rt_unary_op_t op) { - emit_pre_pop_r64(emit, REG_ARG_2); - emit_call_with_i64_arg(emit, rt_unary_op, op, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); + vtype_kind_t vtype; + emit_pre_pop_reg(emit, &vtype, REG_ARG_2); + assert(vtype == VTYPE_PYOBJ); + emit_call_with_imm_arg(emit, rt_unary_op, op, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); +} + +static void emit_x64_binary_op(emit_t *emit, rt_binary_op_t op) { + vtype_kind_t vtype_lhs, vtype_rhs; + emit_pre_pop_reg_reg(emit, &vtype_rhs, REG_ARG_3, &vtype_lhs, REG_ARG_2); + if (vtype_lhs == VTYPE_INT && vtype_rhs == VTYPE_INT) { + assert(op == RT_BINARY_OP_ADD); + asm_x64_add_r64_to_r64(emit->as, REG_ARG_3, REG_ARG_2); + emit_post_push_reg(emit, VTYPE_INT, REG_ARG_2); + } else if (vtype_lhs == VTYPE_PYOBJ && vtype_rhs == VTYPE_PYOBJ) { + emit_call_with_imm_arg(emit, rt_binary_op, op, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); + } else { + printf("ViperTypeError: can't do binary op between types %d and %d\n", vtype_lhs, vtype_rhs); + assert(0); + } +} + +static void emit_x64_compare_op(emit_t *emit, rt_compare_op_t op) { + vtype_kind_t vtype_lhs, vtype_rhs; + emit_pre_pop_reg_reg(emit, &vtype_rhs, REG_ARG_3, &vtype_lhs, REG_ARG_2); + if (vtype_lhs == VTYPE_INT && vtype_rhs == VTYPE_INT) { + assert(op == RT_COMPARE_OP_LESS); + asm_x64_xor_r64_to_r64(emit->as, REG_RET, REG_RET); + asm_x64_cmp_r64_with_r64(emit->as, REG_ARG_3, REG_ARG_2); + asm_x64_setcc_r8(emit->as, JCC_JL, REG_RET); + emit_post_push_reg(emit, VTYPE_BOOL, REG_RET); + } else if (vtype_lhs == VTYPE_PYOBJ && vtype_rhs == VTYPE_PYOBJ) { + emit_call_with_imm_arg(emit, rt_compare_op, op, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); + } else { + printf("ViperTypeError: can't do comparison between types %d and %d\n", vtype_lhs, vtype_rhs); + assert(0); + } } static void emit_x64_build_tuple(emit_t *emit, int n_args) { + // call runtime, with types of args + // if wrapped in byte_array, or something, allocates memory and fills it assert(0); } static void emit_x64_build_list(emit_t *emit, int n_args) { emit_pre(emit); - emit_get_stack_pointer_to_r64_for_pop(emit, REG_ARG_2, n_args); // pointer to items in reverse order - emit_call_with_i64_arg(emit, rt_build_list, n_args, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); // new list + emit_get_stack_pointer_to_reg_for_pop(emit, REG_ARG_2, n_args); // pointer to items in reverse order + emit_call_with_imm_arg(emit, rt_build_list, n_args, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); // new list } static void emit_x64_list_append(emit_t *emit, int list_index) { + // only used in list comprehension, so call runtime assert(0); } static void emit_x64_build_map(emit_t *emit, int n_args) { emit_pre(emit); - emit_call_with_i64_arg(emit, rt_build_map, n_args, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); // new map + emit_call_with_imm_arg(emit, rt_build_map, n_args, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); // new map } static void emit_x64_store_map(emit_t *emit) { - emit_pre_pop_r64_r64_r64(emit, REG_ARG_2, REG_ARG_3, REG_ARG_1); // key, value, map + vtype_kind_t vtype_key, vtype_value, vtype_map; + emit_pre_pop_reg_reg_reg(emit, &vtype_key, REG_ARG_2, &vtype_value, REG_ARG_3, &vtype_map, REG_ARG_1); // key, value, map + assert(vtype_key == VTYPE_PYOBJ); + assert(vtype_value == VTYPE_PYOBJ); + assert(vtype_map == VTYPE_PYOBJ); emit_call(emit, rt_store_map); - emit_post_push_r64(emit, REG_RET); // map + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); // map } static void emit_x64_map_add(emit_t *emit, int map_index) { @@ -572,9 +821,9 @@ static void emit_x64_map_add(emit_t *emit, int map_index) { static void emit_x64_build_set(emit_t *emit, int n_args) { emit_pre(emit); - emit_get_stack_pointer_to_r64_for_pop(emit, REG_ARG_2, n_args); // pointer to items in reverse order - emit_call_with_i64_arg(emit, rt_build_set, n_args, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); // new set + emit_get_stack_pointer_to_reg_for_pop(emit, REG_ARG_2, n_args); // pointer to items in reverse order + emit_call_with_imm_arg(emit, rt_build_set, n_args, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); // new set } static void emit_x64_set_add(emit_t *emit, int set_index) { @@ -584,6 +833,7 @@ static void emit_x64_build_slice(emit_t *emit, int n_args) { assert(0); } static void emit_x64_unpack_sequence(emit_t *emit, int n_args) { + // call runtime, needs type decl assert(0); } static void emit_x64_unpack_ex(emit_t *emit, int n_left, int n_right) { @@ -591,10 +841,11 @@ static void emit_x64_unpack_ex(emit_t *emit, int n_left, int n_right) { } static void emit_x64_make_function(emit_t *emit, scope_t *scope, int n_dict_params, int n_default_params) { + // call runtime, with type info for args, or don't support dict/default params, or only support Python objects for them assert(n_default_params == 0 && n_dict_params == 0); emit_pre(emit); - emit_call_with_i64_arg(emit, rt_make_function_from_id, scope->unique_code_id, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); + emit_call_with_imm_arg(emit, rt_make_function_from_id, scope->unique_code_id, REG_ARG_1); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); } static void emit_x64_make_closure(emit_t *emit, scope_t *scope, int n_dict_params, int n_default_params) { @@ -602,88 +853,83 @@ static void emit_x64_make_closure(emit_t *emit, scope_t *scope, int n_dict_param } static void emit_x64_call_function(emit_t *emit, int n_positional, int n_keyword, bool have_star_arg, bool have_dbl_star_arg) { + // call special viper runtime routine with type info for args, and wanted type info for return assert(n_keyword == 0 && !have_star_arg && !have_dbl_star_arg); if (n_positional == 0) { - emit_pre_pop_r64(emit, REG_ARG_1); // the function + vtype_kind_t vtype_fun; + emit_pre_pop_reg(emit, &vtype_fun, REG_ARG_1); // the function + assert(vtype_fun == VTYPE_PYOBJ); emit_call(emit, rt_call_function_0); } else if (n_positional == 1) { - emit_pre_pop_r64_r64(emit, REG_ARG_2, REG_ARG_1); // the single argument, the function + vtype_kind_t vtype_fun, vtype_arg1; + emit_pre_pop_reg_reg(emit, &vtype_arg1, REG_ARG_2, &vtype_fun, REG_ARG_1); // the single argument, the function + assert(vtype_fun == VTYPE_PYOBJ); + assert(vtype_arg1 == VTYPE_PYOBJ); emit_call(emit, rt_call_function_1); } else if (n_positional == 2) { - emit_pre_pop_r64_r64_r64(emit, REG_ARG_3, REG_ARG_2, REG_ARG_1); // the second argument, the first argument, the function + vtype_kind_t vtype_fun, vtype_arg1, vtype_arg2; + emit_pre_pop_reg_reg_reg(emit, &vtype_arg2, REG_ARG_3, &vtype_arg1, REG_ARG_2, &vtype_fun, REG_ARG_1); // the second argument, the first argument, the function + assert(vtype_fun == VTYPE_PYOBJ); + assert(vtype_arg1 == VTYPE_PYOBJ); + assert(vtype_arg2 == VTYPE_PYOBJ); emit_call(emit, rt_call_function_2); } else { assert(0); } - emit_post_push_r64(emit, REG_RET); + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); } static void emit_x64_call_method(emit_t *emit, int n_positional, int n_keyword, bool have_star_arg, bool have_dbl_star_arg) { assert(n_keyword == 0 && !have_star_arg && !have_dbl_star_arg); if (n_positional == 0) { - emit_pre_pop_r64_r64(emit, REG_ARG_2, REG_ARG_1); // the self object (or NULL), the method + vtype_kind_t vtype_meth, vtype_self; + emit_pre_pop_reg_reg(emit, &vtype_self, REG_ARG_2, &vtype_meth, REG_ARG_1); // the self object (or NULL), the method + assert(vtype_meth == VTYPE_PYOBJ); + assert(vtype_self == VTYPE_PYOBJ); emit_call(emit, rt_call_method_1); } else if (n_positional == 1) { - emit_pre_pop_r64_r64_r64(emit, REG_ARG_3, REG_ARG_2, REG_ARG_1); // the first argument, the self object (or NULL), the method + vtype_kind_t vtype_meth, vtype_self, vtype_arg1; + emit_pre_pop_reg_reg_reg(emit, &vtype_arg1, REG_ARG_3, &vtype_self, REG_ARG_2, &vtype_meth, REG_ARG_1); // the first argument, the self object (or NULL), the method + assert(vtype_meth == VTYPE_PYOBJ); + assert(vtype_self == VTYPE_PYOBJ); + assert(vtype_arg1 == VTYPE_PYOBJ); emit_call(emit, rt_call_method_2); } else { assert(0); } - emit_post_push_r64(emit, REG_RET); -} - -static void emit_x64_pop_block(emit_t *emit) { - emit_pre(emit); - emit_post(emit); -} - -static void emit_x64_binary_op(emit_t *emit, rt_binary_op_t op) { - if (emit->do_native_types) { - assert(op == RT_BINARY_OP_ADD); - emit_pre_pop_r64_r64(emit, REG_ARG_2, REG_RET); - asm_x64_add_r64_to_r64(emit->as, REG_ARG_2, REG_RET); - emit_post_push_r64(emit, REG_RET); - } else { - emit_pre_pop_r64_r64(emit, REG_ARG_3, REG_ARG_2); - emit_call_with_i64_arg(emit, rt_binary_op, op, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); - } + emit_post_push_reg(emit, VTYPE_PYOBJ, REG_RET); } -static void emit_x64_compare_op(emit_t *emit, rt_compare_op_t op) { - if (emit->do_native_types) { - assert(op == RT_COMPARE_OP_LESS); - emit_pre_pop_r64_r64(emit, REG_ARG_3, REG_ARG_2); - asm_x64_xor_r64_to_r64(emit->as, REG_RET, REG_RET); - asm_x64_cmp_r64_with_r64(emit->as, REG_ARG_3, REG_ARG_2); - asm_x64_setcc_r8(emit->as, JCC_JL, REG_RET); - emit_post_push_r64(emit, REG_RET); +static void emit_x64_return_value(emit_t *emit) { + // easy. since we don't know who we return to, just return the raw value. + // runtime needs then to know our type signature, but I think that's possible. + vtype_kind_t vtype; + emit_pre_pop_reg(emit, &vtype, REG_RAX); + if (emit->do_viper_types) { + assert(vtype == VTYPE_PTR_NONE); } else { - emit_pre_pop_r64_r64(emit, REG_ARG_3, REG_ARG_2); - emit_call_with_i64_arg(emit, rt_compare_op, op, REG_ARG_1); - emit_post_push_r64(emit, REG_RET); + assert(vtype == VTYPE_PYOBJ); } -} - -static void emit_x64_return_value(emit_t *emit) { - emit_pre_pop_r64(emit, REG_RAX); emit->last_emit_was_return_value = true; //asm_x64_call_ind(emit->as, 0, REG_RAX); to seg fault for debugging with gdb asm_x64_exit(emit->as); } static void emit_x64_raise_varargs(emit_t *emit, int n_args) { + // call runtime assert(0); } static void emit_x64_yield_value(emit_t *emit) { + // not supported (for now) assert(0); } static void emit_x64_yield_from(emit_t *emit) { + // not supported (for now) assert(0); } const emit_method_table_t emit_x64_method_table = { - emit_x64_set_native_types, + emit_x64_set_viper_types, emit_x64_start_pass, emit_x64_end_pass, emit_x64_last_emit_was_return_value, |