US-CERT Vulnerability Note VU#162289

[Joel Sherrill]

Robert Dewar wrote:
> Paul Koning wrote:
>
>   
>> That said, it certainly is helpful if the compiler can detect some
>> undefined actions and warn about them.  But that doesn't create a duty
>> to warn about all of them.
>>     
>
> If it were reasonable to require a compiler to generate a warning
> for a particular case, the standard would have made it an error.
> The whole point in allowing undefined behavior is that in certain
> cases, it is too onerous for a compiler to be required to detect
> the undefined behavior, so it is not required to do so.
>   
I recall something in the Ada LRM that a conforming
Ada program did not depend on undefined or implementation
dependent behavior.  The example I remember being
used to explain it to me was a program depending upon
the precise order of tasks executing.  That can obviously
vary based upon interrupts, CPU speed, time slice
quantum and a number of tasking implementation decisions.

When you talk undefined, the program is questionable
at best.  I like the Ada LRM because it tries to be very
clean about this in a way that a programmer can understand
and try to do the right thing.

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel.sherrill@OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985