So are there any plans to enforce my proposed policy in GCC, or the
upcoming C standard?
On Wed, Nov 26, 2014 at 4:56 AM, xinglp <xinglp@xxxxxxxxx> wrote:
>> I don't think you need a warning in this particular case, perhaps a
>> better way would be for the compiler to enforce a rule that forbids
>> access of function pointers of inner functions that referencing local
>> variables of its parent function.
> Yes, this is better. For now, when this happend, there's only a
> segfault and it's diffcult to dig out.
>>
>> For example, both:
>>
>> void *foo(int x)
>> {
>> int a;
>> void foo2(int i)
>> {
>> // I want warning for the below code.
>> if (i == a){}
>> x += i;
>> a += i;
>> }
>> foo2(a);
>>
>> return foo2;
>> }
>>
>> and:
>>
>> void *bar;
>>
>> void foo(int x)
>> {
>> int a;
>> void foo2(int i)
>> {
>> // I want warning for the below code.
>> if (i == a){}
>> x += i;
>> a += i;
>> }
>> foo2(a);
>>
>> bar = foo2;
>> }
>>
>>
>> should return an error.
