Andrzej Hajda <[email protected]> wrote:
> Semantic patch finds comparisons of types:
> unsigned < 0
> unsigned >= 0
> The former is always false, the latter is always true.
> Such comparisons are useless, so theoretically they could be
> safely removed, but their presence quite often indicates bugs.
Or someone has left them in because they don't matter and there's the
possibility that the type being tested might be or become signed under some
circumstances. If the comparison is useless, I'd expect the compiler to just
discard it - for such cases your patch is pointless.
If I have, for example:
unsigned x;
if (x == 0 || x > 27)
give_a_range_error();
I will write this as:
unsigned x;
if (x <= 0 || x > 27)
give_a_range_error();
because it that gives a way to handle x being changed to signed at some point
in the future for no cost. In which case, your changing the <= to an ==
"because the < part of the case is useless" is arguably wrong.
David
On 09/21/2015 03:42 PM, David Howells wrote:
> Andrzej Hajda <[email protected]> wrote:
>
>> Semantic patch finds comparisons of types:
>> unsigned < 0
>> unsigned >= 0
>> The former is always false, the latter is always true.
>> Such comparisons are useless, so theoretically they could be
>> safely removed, but their presence quite often indicates bugs.
>
> Or someone has left them in because they don't matter and there's the
> possibility that the type being tested might be or become signed under some
> circumstances. If the comparison is useless, I'd expect the compiler to just
> discard it - for such cases your patch is pointless.
>
> If I have, for example:
>
> unsigned x;
>
> if (x == 0 || x > 27)
> give_a_range_error();
>
> I will write this as:
>
> unsigned x;
>
> if (x <= 0 || x > 27)
> give_a_range_error();
>
> because it that gives a way to handle x being changed to signed at some point
> in the future for no cost. In which case, your changing the <= to an ==
> "because the < part of the case is useless" is arguably wrong.
This is why I have not checked for such cases - I have skipped checks of type
unsigned <= 0
exactly for the reasons above.
However I have left two other checks as they seems to me more suspicious - they
are always true or false. But as Dmitry and Andrew pointed out Linus have quite
strong opinion against removing range checks in such cases as he finds it
clearer. I think it applies to patches 29-36. I am not sure about patches 26-28,37.
Regards
Andrzej
>
> David
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
On 09/22/2015 11:13 AM, Andrzej Hajda wrote:
> On 09/21/2015 03:42 PM, David Howells wrote:
>> Andrzej Hajda <[email protected]> wrote:
>>
>>> Semantic patch finds comparisons of types:
>>> unsigned < 0
>>> unsigned >= 0
>>> The former is always false, the latter is always true.
>>> Such comparisons are useless, so theoretically they could be
>>> safely removed, but their presence quite often indicates bugs.
>>
>> Or someone has left them in because they don't matter and there's the
>> possibility that the type being tested might be or become signed under some
>> circumstances. If the comparison is useless, I'd expect the compiler to just
>> discard it - for such cases your patch is pointless.
>>
>> If I have, for example:
>>
>> unsigned x;
>>
>> if (x == 0 || x > 27)
>> give_a_range_error();
>>
>> I will write this as:
>>
>> unsigned x;
>>
>> if (x <= 0 || x > 27)
>> give_a_range_error();
>>
>> because it that gives a way to handle x being changed to signed at some point
>> in the future for no cost. In which case, your changing the <= to an ==
>> "because the < part of the case is useless" is arguably wrong.
>
> This is why I have not checked for such cases - I have skipped checks of type
> unsigned <= 0
> exactly for the reasons above.
>
> However I have left two other checks as they seems to me more suspicious - they
> are always true or false. But as Dmitry and Andrew pointed out Linus have quite
> strong opinion against removing range checks in such cases as he finds it
> clearer. I think it applies to patches 29-36. I am not sure about patches 26-28,37.
Dropped 30/38 and 31/38 from LED tree then.
--
Best Regards,
Jacek Anaszewski