harden compiler parsing, recovery, and deep-expression handling

This commit is contained in:
2026-06-12 00:28:45 +02:00
parent a5ceb727c1
commit 99ba907f59
13 changed files with 1159 additions and 503 deletions
+157 -127
View File
@@ -2,6 +2,7 @@ package lower
import "../hir"
import "../ir"
import "../source"
import "../types"
import "core:fmt"
import "core:mem"
@@ -11,6 +12,7 @@ State :: struct {
instructions: [dynamic]ir.Instruction,
local_values: []int,
local_slots: []int,
expr_stack: [dynamic]Lower_Expr_Frame,
allocator: mem.Allocator,
}
@@ -35,142 +37,25 @@ sentinel :: proc(value_type: types.Type) -> i64 {
}
}
lower_expr :: proc(state: ^State, expr_id: int) -> int {
if expr_id < 0 || expr_id >= len(state.hir_module.exprs) {
trap := append_instruction(state, ir.Instruction{
op=.Trap,
type=types.VOID,
target=-1,
a=-1,
b=-1,
diagnostic=-1,
})
_ = trap
return append_instruction(state, ir.Instruction{
op=.Const,
type=types.I64,
integer=sentinel(types.I64),
target=-1,
a=-1,
b=-1,
diagnostic=-1,
})
}
expr := state.hir_module.exprs[expr_id]
switch expr.kind {
case .Invalid:
append_instruction(state, ir.Instruction{
op=.Trap,
span=expr.span,
type=types.VOID,
target=-1,
a=-1,
b=-1,
diagnostic=expr.diagnostic,
})
fallback := expr.type
if !types.is_concrete_integer(fallback) {
fallback = types.I64
}
return append_instruction(state, ir.Instruction{
op=.Const,
span=expr.span,
type=fallback,
integer=sentinel(fallback),
target=-1,
a=-1,
b=-1,
diagnostic=-1,
})
case .Integer:
return append_instruction(state, ir.Instruction{
op=.Const,
span=expr.span,
type=expr.type,
integer=expr.integer,
target=-1,
a=-1,
b=-1,
diagnostic=-1,
})
case .Local:
if expr.target >= 0 && expr.target < len(state.local_slots) && state.local_slots[expr.target] >= 0 {
return append_instruction(state, ir.Instruction{
op=.Load,
span=expr.span,
type=expr.type,
target=-1,
a=state.local_slots[expr.target],
b=-1,
diagnostic=-1,
})
}
if expr.target >= 0 && expr.target < len(state.local_values) {
return state.local_values[expr.target]
}
case .Global:
return append_instruction(state, ir.Instruction{
op=.Load_Global,
span=expr.span,
type=expr.type,
target=expr.target,
a=-1,
b=-1,
diagnostic=-1,
})
case .Widen:
value := lower_expr(state, expr.left)
return append_instruction(state, ir.Instruction{
op=.Widen,
span=expr.span,
type=expr.type,
target=-1,
a=value,
b=-1,
diagnostic=-1,
})
case .Add:
left := lower_expr(state, expr.left)
right := lower_expr(state, expr.right)
return append_instruction(state, ir.Instruction{
op=.Add_Checked,
span=expr.span,
type=expr.type,
target=-1,
a=left,
b=right,
diagnostic=-1,
})
case .Call:
args := make([]int, len(expr.args), state.allocator)
for arg, index in expr.args {
args[index] = lower_expr(state, arg)
}
return append_instruction(state, ir.Instruction{
op=.Call,
span=expr.span,
type=expr.type,
target=expr.target,
a=-1,
b=-1,
args=args,
diagnostic=-1,
})
}
append_recovery_value :: proc(state: ^State, span: source.Span, value_type: types.Type, diagnostic := -1) -> int {
append_instruction(state, ir.Instruction{
op=.Trap,
span=expr.span,
span=span,
type=types.VOID,
target=-1,
a=-1,
b=-1,
diagnostic=expr.diagnostic,
diagnostic=diagnostic,
})
fallback := value_type
if !types.is_concrete_integer(fallback) {
fallback = types.I64
}
return append_instruction(state, ir.Instruction{
op=.Const,
span=expr.span,
type=types.I64,
integer=sentinel(types.I64),
span=span,
type=fallback,
integer=sentinel(fallback),
target=-1,
a=-1,
b=-1,
@@ -178,6 +63,133 @@ lower_expr :: proc(state: ^State, expr_id: int) -> int {
})
}
Lower_Expr_Frame :: struct {
expr: int,
stage: u8,
left: int,
arg_index: int,
args: []int,
}
lower_expr :: proc(state: ^State, expr_id: int) -> int {
stack := state.expr_stack
clear_dynamic_array(&stack)
defer {
for frame in stack {
delete(frame.args, state.allocator)
}
clear_dynamic_array(&stack)
state.expr_stack = stack
}
append(&stack, Lower_Expr_Frame{expr=expr_id})
last := -1
for len(stack) > 0 {
frame_index := len(stack)-1
frame := stack[frame_index]
if frame.expr < 0 || frame.expr >= len(state.hir_module.exprs) {
last = append_recovery_value(state, source.Span{}, types.I64)
_ = pop(&stack)
continue
}
expr := state.hir_module.exprs[frame.expr]
if frame.stage == 0 {
switch expr.kind {
case .Invalid:
last = append_recovery_value(state, expr.span, expr.type, expr.diagnostic)
_ = pop(&stack)
case .Integer:
last = append_instruction(state, ir.Instruction{
op=.Const, span=expr.span, type=expr.type, integer=expr.integer,
target=-1, a=-1, b=-1, diagnostic=-1,
})
_ = pop(&stack)
case .Local:
last = -1
if expr.target >= 0 && expr.target < len(state.local_slots) && state.local_slots[expr.target] >= 0 {
last = append_instruction(state, ir.Instruction{
op=.Load, span=expr.span, type=expr.type, target=-1,
a=state.local_slots[expr.target], b=-1, diagnostic=-1,
})
} else if expr.target >= 0 && expr.target < len(state.local_values) &&
state.local_values[expr.target] >= 0 {
last = state.local_values[expr.target]
}
if last < 0 {
last = append_recovery_value(state, expr.span, expr.type, expr.diagnostic)
}
_ = pop(&stack)
case .Global:
if expr.target < 0 || expr.target >= len(state.hir_module.globals) {
last = append_recovery_value(state, expr.span, expr.type, expr.diagnostic)
} else {
last = append_instruction(state, ir.Instruction{
op=.Load_Global, span=expr.span, type=expr.type, target=expr.target,
a=-1, b=-1, diagnostic=-1,
})
}
_ = pop(&stack)
case .Widen:
stack[frame_index].stage = 1
append(&stack, Lower_Expr_Frame{expr=expr.left})
case .Add:
stack[frame_index].stage = 2
append(&stack, Lower_Expr_Frame{expr=expr.left})
case .Call:
if expr.target < 0 || expr.target >= len(state.hir_module.functions) {
last = append_recovery_value(state, expr.span, expr.type, expr.diagnostic)
_ = pop(&stack)
continue
}
stack[frame_index].args = make([]int, len(expr.args), state.allocator)
stack[frame_index].stage = 4
if len(expr.args) > 0 {
append(&stack, Lower_Expr_Frame{expr=expr.args[0]})
}
}
continue
}
if frame.stage == 1 {
last = append_instruction(state, ir.Instruction{
op=.Widen, span=expr.span, type=expr.type, target=-1,
a=last, b=-1, diagnostic=-1,
})
_ = pop(&stack)
continue
}
if frame.stage == 2 {
stack[frame_index].left = last
stack[frame_index].stage = 3
append(&stack, Lower_Expr_Frame{expr=expr.right})
continue
}
if frame.stage == 3 {
last = append_instruction(state, ir.Instruction{
op=.Add_Checked, span=expr.span, type=expr.type, target=-1,
a=frame.left, b=last, diagnostic=-1,
})
_ = pop(&stack)
continue
}
if frame.stage == 4 {
if frame.arg_index < len(expr.args) {
stack[frame_index].args[frame.arg_index] = last
stack[frame_index].arg_index += 1
if frame.arg_index+1 < len(expr.args) {
append(&stack, Lower_Expr_Frame{expr=expr.args[frame.arg_index+1]})
continue
}
}
last = append_instruction(state, ir.Instruction{
op=.Call, span=expr.span, type=expr.type, target=expr.target,
a=-1, b=-1, args=stack[frame_index].args, diagnostic=-1,
})
stack[frame_index].args = nil
_ = pop(&stack)
}
}
return last
}
lower_body :: proc(hir_module: ^hir.Module, function: hir.Function, allocator: mem.Allocator) -> []ir.Instruction {
state := State{
hir_module=hir_module,
@@ -186,9 +198,11 @@ lower_body :: proc(hir_module: ^hir.Module, function: hir.Function, allocator: m
local_slots=make([]int, len(function.locals), allocator),
}
state.instructions.allocator = allocator
state.expr_stack.allocator = allocator
defer {
delete(state.local_values, allocator)
delete(state.local_slots, allocator)
delete(state.expr_stack)
}
for _, index in state.local_values {
state.local_values[index] = -1
@@ -211,6 +225,13 @@ lower_body :: proc(hir_module: ^hir.Module, function: hir.Function, allocator: m
switch statement.kind {
case .Declaration:
value := lower_expr(&state, statement.expr)
if statement.local < 0 || statement.local >= len(function.locals) {
append_instruction(&state, ir.Instruction{
op=.Trap, span=statement.span, type=types.VOID,
target=-1, a=-1, b=-1, diagnostic=statement.diagnostic,
})
continue
}
local := function.locals[statement.local]
if local.mutable {
slot := append_instruction(&state, ir.Instruction{
@@ -241,6 +262,13 @@ lower_body :: proc(hir_module: ^hir.Module, function: hir.Function, allocator: m
if statement.local >= 0 && statement.local < len(state.local_slots) {
slot = state.local_slots[statement.local]
}
if slot < 0 || statement.local < 0 || statement.local >= len(function.locals) {
append_instruction(&state, ir.Instruction{
op=.Trap, span=statement.span, type=types.VOID,
target=-1, a=-1, b=-1, diagnostic=statement.diagnostic,
})
continue
}
append_instruction(&state, ir.Instruction{
op=.Store,
span=statement.span,
@@ -311,6 +339,8 @@ lower_body :: proc(hir_module: ^hir.Module, function: hir.Function, allocator: m
lower_global_initializer :: proc(hir_module: ^hir.Module, global: hir.Global, allocator: mem.Allocator) -> []ir.Instruction {
state := State{hir_module=hir_module, allocator=allocator}
state.instructions.allocator = allocator
state.expr_stack.allocator = allocator
defer delete(state.expr_stack)
value := lower_expr(&state, global.expr)
append_instruction(&state, ir.Instruction{
op=.Return,