On 03/10/2016 04:45 PM, Remy Horton wrote:
>
> On 10/03/2016 13:42, Panu Matilainen wrote:
>> On 02/25/2016 08:48 PM, Aaron Conole wrote:
>>> Silence a compiler warning that this variable may be used uninitialized.
>>>
>>> Signed-off-by: Aaron Conole <aconole at redhat.com>
> [..]
>>
>> The patch looks ok as such, but then again warning looks like a false
>> positive to me: assignment and dereferencing depend on the same value of
>> eop, which cannot change between the two.
>
> In two minds about this. It is a logical impossibility, but these days
> optimising compilers are getting very aggressive. For instance GCC has a
> delightfully-named -fdelete-null-pointer-checks option, which caused
> security holes..
Indeed, that's why silencing a false positive (assuming it actually is
one) by throwing some more NULL-checks for the allegedly impossible
makes me a bit nervous. Besides compiler optimizations going crazy, I've
seen such extra NULL-checks turn into actual bugs when surroundings
subtly change.
- Panu -
