shishantbiswas/c3c
1
1
Fork 0
mirror of https://github.com/c3lang/c3c.git synced 2026-02-27 12:01:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix bugs in "trap-on-wrap" #1434.

Browse Source
This commit is contained in:
Christoffer Lerno
2024-09-14 16:19:40 +02:00
parent 3ab201ce10
commit f51230e793
2 changed files with 22 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
releasenotes.md
View File

@@ -25,6 +25,7 @@
- Fix bug where inline index access to array in a struct would crash the compiler.
- Asserts are now correctly included and traced in when running tests.
- Use atexit to fix finalizers on Windows #1361.
- Fix bugs in "trap-on-wrap" #1434.
### Stdlib changes
- Additional init functions for hashmap.

34
src/compiler/llvm_codegen_expr.c
View File

@@ -330,7 +330,7 @@ LLVMValueRef llvm_emit_const_padding(GenContext *c, AlignSize size)
static inline LLVMValueRef llvm_emit_add_int(GenContext *c, Type *type, LLVMValueRef left, LLVMValueRef right, SourceSpan loc)
{
if (compiler.build.feature.trap_on_wrap)
if (compiler.build.feature.trap_on_wrap && type->type_kind != TYPE_VECTOR)
{
LLVMTypeRef type_to_use = llvm_get_type(c, type->canonical);
LLVMValueRef args[2] = { left, right };
@@ -566,11 +566,11 @@ void llvm_emit_convert_value_from_coerced(GenContext *c, BEValue *result, LLVMTy
static inline LLVMValueRef llvm_emit_sub_int(GenContext *c, Type *type, LLVMValueRef left, LLVMValueRef right, SourceSpan loc)
{
if (compiler.build.feature.trap_on_wrap)
if (compiler.build.feature.trap_on_wrap && type->type_kind != TYPE_VECTOR)
{
LLVMTypeRef type_to_use = llvm_get_type(c, type);
LLVMValueRef args[2] = { left, right };
assert(type->canonical == type);
assert(type_lowering(type) == type);
LLVMValueRef add_res;
if (type_is_unsigned(type))
{
@@ -2369,8 +2369,8 @@ static inline LLVMValueRef llvm_emit_inc_dec_value(GenContext *c, SourceSpan spa
: LLVMBuildNUWSub(c->builder, original->value, diff_value, "subnuw");
}
return diff > 0
? llvm_emit_add_int(c, type, original->value, diff_value, span)
: llvm_emit_sub_int(c, type, original->value, diff_value, span);
? llvm_emit_add_int(c, original->type, original->value, diff_value, span)
: llvm_emit_sub_int(c, original->type, original->value, diff_value, span);
}
case TYPE_VECTOR:
{
@@ -2394,8 +2394,8 @@ static inline LLVMValueRef llvm_emit_inc_dec_value(GenContext *c, SourceSpan spa
if (is_integer)
{
return diff > 0
? llvm_emit_add_int(c, type, original->value, val, span)
: llvm_emit_sub_int(c, type, original->value, val, span);
? llvm_emit_add_int(c, original->type, original->value, val, span)
: llvm_emit_sub_int(c, original->type, original->value, val, span);
}
else
{
@@ -2958,7 +2958,7 @@ static void llvm_emit_slice_values(GenContext *c, Expr *slice, BEValue *parent_r
// Walk from end if it is a slice from the back.
if (start_from_end)
{
start_index.value = llvm_emit_sub_int(c, start_type, len.value, start_index.value, slice->span);
start_index.value = llvm_emit_sub_int(c, start_index.type, len.value, start_index.value, slice->span);
}
// Check that index does not extend beyond the length.
@@ -3006,12 +3006,12 @@ static void llvm_emit_slice_values(GenContext *c, Expr *slice, BEValue *parent_r
if (end_from_end)
{
assert(range.range_type == RANGE_DYNAMIC);
end_index.value = llvm_emit_sub_int(c, end_type, len.value, end_index.value, slice->span);
end_index.value = llvm_emit_sub_int(c, end_index.type, len.value, end_index.value, slice->span);
llvm_value_rvalue(c, &end_index);
}
if (is_len_range)
{
end_index.value = llvm_emit_add_int(c, end_type, start_index.value, end_index.value, slice->span);
end_index.value = llvm_emit_add_int(c, end_index.type, start_index.value, end_index.value, slice->span);
}
// This will trap any bad negative index, so we're fine.
@@ -3024,7 +3024,7 @@ static void llvm_emit_slice_values(GenContext *c, Expr *slice, BEValue *parent_r
BEValue excess;
llvm_emit_int_comp(c, &excess, &len, &end_index, BINARYOP_LT);
BEValue actual_end_index = end_index;
actual_end_index.value = llvm_emit_sub_int(c, end_type, end_index.value, llvm_const_int(c, type_isz, 1), slice->span);
actual_end_index.value = llvm_emit_sub_int(c, end_index.type, end_index.value, llvm_const_int(c, type_isz, 1), slice->span);
llvm_emit_panic_if_true(c, &excess, "End index out of bounds", slice->span, "End index out of bounds (end index of %d exceeds size of %d)", &actual_end_index, &len);
}
}
@@ -3246,12 +3246,20 @@ static void llvm_emit_slice_assign(GenContext *c, BEValue *be_value, Expr *expr)
// Create the new offset
LLVMValueRef next_offset = llvm_emit_add_int(c, start.type, offset, llvm_const_int(c, start.type, 1), expr->span);
LLVMBasicBlockRef assign_block_end = llvm_get_current_block_if_in_use(c);
// And jump back
llvm_emit_br(c, cond_block);
// Finally set up our phi
llvm_set_phi(offset, start.value, start_block, next_offset, assign_block);
if (!assign_block_end)
{
offset = start.value;
}
else
{
llvm_set_phi(offset, start.value, start_block, next_offset, assign_block_end);
}
// And emit the exit block.
llvm_emit_block(c, exit_block);
@@ -3604,7 +3612,7 @@ static void llvm_emit_struct_comparison(GenContext *c, BEValue *result, BEValue
static inline LLVMValueRef llvm_emit_mult_int(GenContext *c, Type *type, LLVMValueRef left, LLVMValueRef right, SourceSpan loc)
{
if (compiler.build.feature.trap_on_wrap)
if (compiler.build.feature.trap_on_wrap && type->type_kind != TYPE_VECTOR)
{
LLVMTypeRef type_to_use = llvm_get_type(c, type);
LLVMValueRef args[2] = { left, right };