Re: [PATCH] net: gro: use IS_ERR before PTR_ERR

[Guo Zhengkui <guozhengkui@vivo.com>]

On 2021/12/7 22:41, Alexander Lobakin wrote:
> From: Guo Zhengkui <guozhengkui@vivo.com>
> Date: Tue,  7 Dec 2021 15:31:09 +0800
> 
> Hi, thanks for your patch.
> 
>> fix following cocci warning:
>> ./net/core/gro.c:493:5-12: ERROR: PTR_ERR applied after initialization to constant on line 441
>>
>> Signed-off-by: Guo Zhengkui <guozhengkui@vivo.com>
>> ---
>>   net/core/gro.c | 8 +++++---
>>   1 file changed, 5 insertions(+), 3 deletions(-)
>>
>> diff --git a/net/core/gro.c b/net/core/gro.c
>> index 8ec8b44596da..ee08f7b23793 100644
>> --- a/net/core/gro.c
>> +++ b/net/core/gro.c
>> @@ -490,9 +490,11 @@ static enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff
>>   	if (&ptype->list == head)
>>   		goto normal;
>>   
>> -	if (PTR_ERR(pp) == -EINPROGRESS) {
>> -		ret = GRO_CONSUMED;
>> -		goto ok;
>> +	if (IS_ERR(pp)) {
>> +		if (PTR_ERR(pp) == -EINPROGRESS) {
>> +			ret = GRO_CONSUMED;
>> +			goto ok;
>> +		}
>>   	}
> 
> `if (PTR_ERR(ptr) == -ERRNO)` itself is correct without a check for
> IS_ERR(). The former basically is a more precise test comparing to
> the latter.

Yes, even without `IS_ERR`, it runs well.

At least, `IS_ERR` before `PTR_ERR` is a good habit. :)

Zhengkui

> Not sure if compilers can get it well, but in ideal case the first
> will be omitted from the object code at all, and so do we.
> 
> In case I'm wrong and this is a correct fix, it at least shouldn't
> increase the indentation by one, these two conditions can be placed
> into one `if` statement.
> 
> NAK.
> 
>>   
>>   	same_flow = NAPI_GRO_CB(skb)->same_flow;
>> -- 
>> 2.20.1
> 
> Al
>