Re: behavior-preserving optimization in C, was compiler bugs

"Christopher Glaeser" <cdg@nullstone.com> writes:




> Suppose you are responsible for maintaining a C compiler and you
> receive a defect report with a program that works fine without
> optimization but generates a divide by zero exception with
> optimization. You analyze the program and determine that the divide
> by zero is in a function that references an uninitialized local array
> element. Without optimization, the array element has a value of 1 due
> to the more-or-less random state of the stack on entry to the
> function. With optimization, changes to register and stack
> assignments cause changes to the state of the stack on entry to the
> function, and now the uninitialized local array element has the value
> 0, which causes the divide by zero exception.
>
> How would you respond to this defect report?


I would tell the programmer what the problem is and, if time allows,
add an analysis that detects potential uses of uninitialised variables
or array elements and issue warnings if such occur. I would also make
sure the policy for bug reports will only accept programs where the
compiler does not issue warnings.


But for a language like C, it is a futile job. There are so many
unspecified behaviours that you can't realistically preserve all or
issue warnings where you don't.


And, to a degree, preserving unspecified behaviour at optimisation is
not doing the programmer any favours: If the same compiler does not
preserve this behaviour across optimisation levels, it is very
unlikely that it will be preserved if you change compiler (or even
target architecture).


Torben

Return to the comp.compilers page.
Search the comp.compilers archives again.