[PATCH v2] replace sprintf with strcpy to avoid GCC warning [BZ#28439]

[Martin Sebor]

On 10/10/21 2:28 AM, Florian Weimer wrote:
> * Martin Sebor:
> 
>> On 10/9/21 3:15 PM, Florian Weimer wrote:
>>> * Martin Sebor:
>>>
>>>> diff --git a/resolv/res_query.c b/resolv/res_query.c
>>>> index 75b0e5f2f7..adc8a13f75 100644
>>>> --- a/resolv/res_query.c
>>>> +++ b/resolv/res_query.c
>>>> @@ -589,10 +589,9 @@ __res_context_querydomain (struct resolv_context *ctx,
>>>>    	struct __res_state *statp = ctx->resp;
>>>>    	char nbuf[MAXDNAME];
>>>>    	const char *longname = nbuf;
>>>> -	size_t n, d;
>>>>    
>>>>    	if (domain == NULL) {
>>>> -		n = strlen(name);
>>>> +		size_t n = strlen(name);
>>>>    
>>>>    		/* Decrement N prior to checking it against MAXDNAME
>>>>    		   so that we detect a wrap to SIZE_MAX and return
>>>> @@ -603,15 +602,13 @@ __res_context_querydomain (struct resolv_context *ctx,
>>>>    			return (-1);
>>>>    		}
>>>>    		longname = name;
>>>> -	} else {
>>>> -		n = strlen(name);
>>>> -		d = strlen(domain);
>>>> -		if (n + d + 1 >= MAXDNAME) {
>>>> -			RES_SET_H_ERRNO(statp, NO_RECOVERY);
>>>> -			return (-1);
>>>> -		}
>>>> -		sprintf(nbuf, "%s.%s", name, domain);
>>>>    	}
>>>> +	else if (__snprintf (nbuf, sizeof nbuf, "%s.%s", name, domain)
>>>> +		 >= sizeof nbuf)
>>>> +	  {
>>>> +	    RES_SET_H_ERRNO(statp, NO_RECOVERY);
>>>> +	    return -1;
>>>> +	  }
>>>>    	return __res_context_query (ctx, longname, class, type, answer,
>>>>    				    anslen, answerp, answerp2, nanswerp2,
>>>>    				    resplen2, answerp2_malloced);
>>>
>>> Maybe add a comment about EOVERFLOW?  I think it still works because
>>> the -1 from snprintf turns into SIZE_MAX.
>>
>> snprintf returns "the number of bytes that would have been written
>> if sizeof buf had been sufficiently large" no?  Or is __snprintf
>> different?
> 
> The return type is int, not size_t, and there are two input arguments.
> So there is potential for overflow.

Ah, I see what you meant by EOVERFLOW now.   Yes, the conversion
to size_t would have handled the case of any error but I agree
that calling out the overflow might have been helpful.

Martin