+193 -1

src/p.c

View file Reviewed

··· 1619 1619 } 1620 1620 } 1621 1621 1622 1622 - static void semantic_validate(parser *p, const tobi_ir *root) { 1622 1622 + typedef struct method_state { 1623 1623 + parser *p; 1624 1624 + tobi_ir *method; 1625 1625 + uint16_t locals; 1626 1626 + uint16_t local_reads; 1627 1627 + uint16_t local_writes; 1628 1628 + uint16_t arg_reads; 1629 1629 + uint16_t arg_writes; 1630 1630 + unsigned warnings; 1631 1631 + } method_state; 1632 1632 + 1633 1633 + static int ref_index(const tobi_ir *n, const char *prefix, unsigned max, unsigned *idx) { 1634 1634 + size_t plen = strlen(prefix); 1635 1635 + if (!n || n->kind != TOBI_IR_REF || !n->name || strncmp(n->name, prefix, plen) != 0) { 1636 1636 + return 0; 1637 1637 + } 1638 1638 + char c = n->name[plen]; 1639 1639 + if (n->name[plen + 1u] != '\0' || c < '0' || c > (char)('0' + max)) { 1640 1640 + return 0; 1641 1641 + } 1642 1642 + *idx = (unsigned)(c - '0'); 1643 1643 + return 1; 1644 1644 + } 1645 1645 + 1646 1646 + static void method_mark_ref(method_state *st, tobi_ir *n, int write_target) { 1647 1647 + unsigned idx = 0; 1648 1648 + if (ref_index(n, "Local", 7, &idx)) { 1649 1649 + if (write_target) { 1650 1650 + st->locals |= (uint16_t)(1u << idx); 1651 1651 + st->local_writes |= (uint16_t)(1u << idx); 1652 1652 + tobi_ir_set_str(n, "method_symbol=local write=initialised"); 1653 1653 + } else if ((st->locals & (uint16_t)(1u << idx)) == 0) { 1654 1654 + st->local_reads |= (uint16_t)(1u << idx); 1655 1655 + st->warnings++; 1656 1656 + sem_diag(st->p, n->off, "Local%u may be read before being initialised in method %s", 1657 1657 + idx, st->method->path ? st->method->path : st->method->name); 1658 1658 + tobi_ir_set_str(n, "method_symbol=local read=uninitialised"); 1659 1659 + } else { 1660 1660 + st->local_reads |= (uint16_t)(1u << idx); 1661 1661 + tobi_ir_set_str(n, "method_symbol=local read=initialised"); 1662 1662 + } 1663 1663 + } else if (ref_index(n, "Arg", 6, &idx)) { 1664 1664 + if (write_target) { 1665 1665 + st->arg_writes |= (uint16_t)(1u << idx); 1666 1666 + } else { 1667 1667 + st->arg_reads |= (uint16_t)(1u << idx); 1668 1668 + } 1669 1669 + if (idx >= st->method->method_args) { 1670 1670 + st->warnings++; 1671 1671 + sem_diag(st->p, n->off, "Arg%u %s exceeds declared argument count %u in method %s", 1672 1672 + idx, write_target ? "write" : "read", 1673 1673 + st->method->method_args, st->method->path ? st->method->path : st->method->name); 1674 1674 + tobi_ir_set_str(n, write_target ? "method_symbol=arg write=out_of_range" : 1675 1675 + "method_symbol=arg read=out_of_range"); 1676 1676 + } else if (write_target) { 1677 1677 + tobi_ir_set_str(n, "method_symbol=arg write=declared"); 1678 1678 + } else { 1679 1679 + tobi_ir_set_str(n, "method_symbol=arg read=declared"); 1680 1680 + } 1681 1681 + } 1682 1682 + } 1683 1683 + 1684 1684 + static void method_symbols_node(method_state *st, tobi_ir *n); 1685 1685 + 1686 1686 + static void method_symbols_block(method_state *st, tobi_ir *b) { 1687 1687 + if (!b) { 1688 1688 + return; 1689 1689 + } 1690 1690 + for (size_t i = 0; i < b->child_len; i++) { 1691 1691 + method_symbols_node(st, b->child[i]); 1692 1692 + } 1693 1693 + } 1694 1694 + 1695 1695 + static void method_symbols_node(method_state *st, tobi_ir *n) { 1696 1696 + if (!n) { 1697 1697 + return; 1698 1698 + } 1699 1699 + if (n->kind == TOBI_IR_STORE) { 1700 1700 + if (n->child_len > 0) { 1701 1701 + method_symbols_node(st, n->child[0]); 1702 1702 + } 1703 1703 + if (n->child_len > 1) { 1704 1704 + unsigned idx = 0; 1705 1705 + if (ref_index(n->child[1], "Local", 7, &idx) || ref_index(n->child[1], "Arg", 6, &idx)) { 1706 1706 + method_mark_ref(st, n->child[1], 1); 1707 1707 + } else { 1708 1708 + method_symbols_node(st, n->child[1]); 1709 1709 + } 1710 1710 + } 1711 1711 + return; 1712 1712 + } 1713 1713 + if (n->kind == TOBI_IR_REF) { 1714 1714 + method_mark_ref(st, n, 0); 1715 1715 + return; 1716 1716 + } 1717 1717 + if (n->kind == TOBI_IR_EXPR && n->name) { 1718 1718 + size_t target_start = n->child_len; 1719 1719 + if (strcmp(n->name, "div") == 0 && n->child_len >= 4) { 1720 1720 + target_start = 2; 1721 1721 + } else if ((strcmp(n->name, "add") == 0 || strcmp(n->name, "concat") == 0 || 1722 1722 + strcmp(n->name, "sub") == 0 || strcmp(n->name, "mul") == 0 || 1723 1723 + strcmp(n->name, "shl") == 0 || strcmp(n->name, "shr") == 0 || 1724 1724 + strcmp(n->name, "and") == 0 || strcmp(n->name, "nand") == 0 || 1725 1725 + strcmp(n->name, "or") == 0 || strcmp(n->name, "nor") == 0 || 1726 1726 + strcmp(n->name, "xor") == 0 || strcmp(n->name, "not") == 0 || 1727 1727 + strcmp(n->name, "find_set_left_bit") == 0 || 1728 1728 + strcmp(n->name, "find_set_right_bit") == 0 || 1729 1729 + strcmp(n->name, "concat_res") == 0 || strcmp(n->name, "mod") == 0 || 1730 1730 + strcmp(n->name, "index") == 0 || strcmp(n->name, "to_buffer") == 0 || 1731 1731 + strcmp(n->name, "to_decimal_string") == 0 || 1732 1732 + strcmp(n->name, "to_hex_string") == 0 || strcmp(n->name, "to_integer") == 0 || 1733 1733 + strcmp(n->name, "to_string") == 0 || strcmp(n->name, "copy_object") == 0 || 1734 1734 + strcmp(n->name, "mid") == 0) && n->child_len >= 3) { 1735 1735 + target_start = n->child_len - 1u; 1736 1736 + } 1737 1737 + for (size_t i = 0; i < target_start; i++) { 1738 1738 + method_symbols_node(st, n->child[i]); 1739 1739 + } 1740 1740 + for (size_t i = target_start; i < n->child_len; i++) { 1741 1741 + unsigned idx = 0; 1742 1742 + if (ref_index(n->child[i], "Local", 7, &idx) || ref_index(n->child[i], "Arg", 6, &idx)) { 1743 1743 + method_mark_ref(st, n->child[i], 1); 1744 1744 + } else { 1745 1745 + method_symbols_node(st, n->child[i]); 1746 1746 + } 1747 1747 + } 1748 1748 + return; 1749 1749 + } 1750 1750 + if (n->kind == TOBI_IR_IF) { 1751 1751 + if (n->child_len > 0) { 1752 1752 + method_symbols_node(st, n->child[0]); 1753 1753 + } 1754 1754 + uint16_t base = st->locals; 1755 1755 + uint16_t then_locals = base; 1756 1756 + uint16_t else_locals = base; 1757 1757 + if (n->child_len > 1) { 1758 1758 + st->locals = base; 1759 1759 + method_symbols_block(st, n->child[1]); 1760 1760 + then_locals = st->locals; 1761 1761 + } 1762 1762 + if (n->child_len > 2) { 1763 1763 + st->locals = base; 1764 1764 + method_symbols_block(st, n->child[2]); 1765 1765 + else_locals = st->locals; 1766 1766 + st->locals = then_locals & else_locals; 1767 1767 + } else { 1768 1768 + st->locals = base; 1769 1769 + } 1770 1770 + return; 1771 1771 + } 1772 1772 + if (n->kind == TOBI_IR_WHILE) { 1773 1773 + if (n->child_len > 0) { 1774 1774 + method_symbols_node(st, n->child[0]); 1775 1775 + } 1776 1776 + uint16_t base = st->locals; 1777 1777 + if (n->child_len > 1) { 1778 1778 + st->locals = base; 1779 1779 + method_symbols_block(st, n->child[1]); 1780 1780 + } 1781 1781 + st->locals = base; 1782 1782 + return; 1783 1783 + } 1784 1784 + for (size_t i = 0; i < n->child_len; i++) { 1785 1785 + method_symbols_node(st, n->child[i]); 1786 1786 + } 1787 1787 + } 1788 1788 + 1789 1789 + static void semantic_methods(parser *p, tobi_ir *n) { 1790 1790 + if (!n) { 1791 1791 + return; 1792 1792 + } 1793 1793 + if (n->kind == TOBI_IR_METHOD) { 1794 1794 + method_state st; 1795 1795 + memset(&st, 0, sizeof(st)); 1796 1796 + st.p = p; 1797 1797 + st.method = n; 1798 1798 + if (n->child_len > 0) { 1799 1799 + method_symbols_block(&st, n->child[0]); 1800 1800 + } 1801 1801 + char summary[160]; 1802 1802 + (void)snprintf(summary, sizeof(summary), 1803 1803 + "method_symbols args=%u local_reads=0x%x local_writes=0x%x final_locals=0x%x arg_reads=0x%x arg_writes=0x%x warnings=%u", 1804 1804 + n->method_args, st.local_reads, st.local_writes, st.locals, 1805 1805 + st.arg_reads, st.arg_writes, st.warnings); 1806 1806 + tobi_ir_set_target(n, summary); 1807 1807 + } 1808 1808 + for (size_t i = 0; i < n->child_len; i++) { 1809 1809 + semantic_methods(p, n->child[i]); 1810 1810 + } 1811 1811 + } 1812 1812 + 1813 1813 + static void semantic_validate(parser *p, tobi_ir *root) { 1623 1814 for (size_t i = 0; i < p->ns->len; i++) { 1624 1815 const tobi_ns_ent *e = &p->ns->items[i]; 1625 1816 if (e->external && e->kind == TOBI_NS_METHOD) { ··· 1634 1825 } 1635 1826 } 1636 1827 semantic_walk(p, root, NULL); 1828 1828 + semantic_methods(p, root); 1637 1829 } 1638 1830 1639 1831 int tobi_parse(const uint8_t *data, size_t len, int strict, tobi_parse_result *out) {

+18

test/t_p.c

View file Reviewed

··· 286 286 T_CHECK(r.diag.len >= 1); 287 287 T_STR(r.diag.items[0].msg, "alias cycle"); 288 288 tobi_parse_result_free(&r); 289 289 + 290 290 + uint8_t symbols[] = { 291 291 + 0x14,0x0a,'S','Y','M','0',0x01, 292 292 + 0xa4,0x60,0xa4,0x69 293 293 + }; 294 294 + T_CHECK(tobi_parse(symbols, sizeof(symbols), 0, &r)); 295 295 + T_CHECK(r.diag.len >= 2); 296 296 + T_STR(r.diag.items[0].msg, "Local0 may be read before"); 297 297 + T_STR(r.diag.items[1].msg, "Arg1 read exceeds declared argument count"); 298 298 + T_STR(r.root->child[0]->target, "method_symbols"); 299 299 + T_STR(r.root->child[0]->target, "local_reads=0x1"); 300 300 + T_STR(r.root->child[0]->target, "arg_reads=0x2"); 301 301 + T_STR(r.root->child[0]->target, "warnings=2"); 302 302 + tobi_ir *sym_ret = r.root->child[0]->child[0]->child[0]; 303 303 + T_CHECK(sym_ret->kind == TOBI_IR_RETURN); 304 304 + T_STR(sym_ret->child[0]->str, "uninitialised"); 305 305 + T_STR(r.root->child[0]->child[0]->child[1]->child[0]->str, "out_of_range"); 306 306 + tobi_parse_result_free(&r); 289 307 return 0; 290 308 }