fix potential leak found by coverity

[chipitsine]

*** CID 446774:  Resource leaks  (RESOURCE_LEAK)
/DriverManager/SQLDriverConnect.c: 469 in __get_pair() 463
464         __get_attr( cp, &keyword, &value );
465         if ( keyword )
466         {
467             con_p = malloc( sizeof( *con_p ));
468             if ( !con_p )
>>>     CID 446774:  Resource leaks  (RESOURCE_LEAK)
>>>     Variable "value" going out of scope leaks the storage it points to.
469                 return NULL;
470             con_p -> keyword = keyword;
471             con_p -> attribute = value;
472             return con_p;
473         }
474         else

*** CID 446773:  Resource leaks  (RESOURCE_LEAK)
/DriverManager/SQLDriverConnect.c: 469 in __get_pair() 463
464         __get_attr( cp, &keyword, &value );
465         if ( keyword )
466         {
467             con_p = malloc( sizeof( *con_p ));
468             if ( !con_p )
>>>     CID 446773:  Resource leaks  (RESOURCE_LEAK)
>>>     Variable "keyword" going out of scope leaks the storage it points to.
469                 return NULL;
470             con_p -> keyword = keyword;
471             con_p -> attribute = value;
472             return con_p;
473         }
474         else

[chipitsine]

from my view there's bigger problem when local variables are returned out of a function.
I wonder why coverity does not worry about it, I plan to have look at it later

[lurcher]

On 27/10/2024 18:06, Ilya Shipitsin wrote: from my view there's bigger problem when local variables are returned out of a function. I wonder why coverity does not worry about it, I plan to have look at it later

It would worry me as well if that was what was happening. I think I trust coverity less and less as time goes on.

[chipitsine]

the following is suspicious

but that code is there for ~15 years. If there were a problem, it must appear somehow. What makes me think my assumption (about returning local variables|) is wrong

[chipitsine]

@lurcher , can we merge it since it addresses leaks ?

[lurcher]

I don't understand where the leak is. the local variables you mention, are pointers, they are set to point to allocated memory in the call to __get_attr, then the pointed to, allocated memory is returned in the returned allocated structure. Just where does the problem occur?

[chipitsine]

if NULL is returned, so con_p is leaked

[lurcher]

Ahh, ok, I can see that. Will fix that.