lclint-interest message 70

From evs Tue Apr 16 15:33:49 1996
Date: Tue, 16 Apr 96 15:20:10 -0400
From: evs (David Evans)
To: dea@gandalf.us.cis.philips.com
Cc: lclint-interest@larch.lcs.mit.edu
In-Reply-To: Debra Aubrey's message of Tue, 16 Apr 1996 12:52:15 -0400 <199604161652.MAA02723@gandalf.knx.dvs.philips.com>
Subject: Questions about LCLint 2.0


> Here is one code snippet:
> 
>    ptr = (procid *) buffer;
>    if (ptr->_state != PS_DEAD)
>    {
>      return (SUCCESS);
>    }
>
> Here is the LCLint message:
>
>    cp_childcheck.c:125,7: Field ptr->_state used before definition

Is buffer->_state defined before the assignment to ptr?  (Apparently,
lclint doesn't think it is, but I can't tell without seeing the code
before this.)

> Along the same lines is this snippet:
>
>    switch (whichProcess)
>    {
>      case CP_CONFIG:
>        procPtr = &CP_Config;
>        break;
>
>      case CP_DISPLAY:
>        procPtr = &CP_Display;
>        break;
>
>      case CP_EBS:
>        procPtr = &CP_Ebs;
>        break;
>        
>      default:
>        break;
>    }
>
>    if (procPtr->procID == CP_DEAD)
>    {
>      return ((error_code)FAILURE);
>    }
>
> And here is the LCLint message:
> 
>    cp_childcheck.c:111,7: Variable procPtr used before definition
>    An rvalue is used that may not be initialized to a value on some execution
>    path. Use -usedef to suppress message.
>
> Again, procPtr has been initialized as a pointer in the switch statement.
> I don't understand why LCLint is complaining that it has not been
> initialized?

Ah, but its not.  If the default path through the switch is taken, then
procPtr is not initialized!  Recall, that lclint will report a
use-before-definition error if there is any path to the use on wich the
variable is not defined.  

--- Dave