Re: [PATCH 1/2] explicit_bzero-tests: pacify GCC

[Paul Eggert]

On 7/18/21 5:23 PM, Bruno Haible wrote:
> Such compiler optimizations would need to be backed by the standards.
> Are there initiatives to "outlaw" references to uninitialized storage
> in recent C or C++ standards?

No initiatives are needed, at least for C. Using uninitialized storage 
is undefined behavior in the current C standard and this has been true 
ever since C was standardized. I imagine C++ is similar.


> I hope code such as
>
>    int x; /* uninitialized */
>    if (!((x & 1) == 0 || (x & 1) == 1))
>      abort ();
>
> will never crash. x & 1 can only be 0 or 1. Tertium not datur.

The C standard doesn't guarantee that code will never crash. For 
example, the standard allows an implementation that uses two's 
complement but where INT_MIN == -INT_MAX and where the bit pattern 
0x80000000 is a trap value (i.e., your program aborts if it reads an int 
whose machine value is 0x80000000).


> GCC does not _know_ that the array
> is uninitialized. It's only a "maybe uninitialized".

That's what GCC's diagnostic says, yes. But in cases like these GCC 
actually "knows" that variables are uninitialized and it sometimes 
optimizes based on this knowledge. For example, for:

  _Bool f (void) {  char *p; return !p; }

gcc -O2 (GCC 11.1.1 20210531 (Red Hat 11.1.1-3)) "knows" that P is 
uninitialized and generates code equivalent to that of:

  _Bool f (void) { return 1; }

That is, GCC optimizes away the access to p's value, which GCC can do 
because the behavior is undefined.


>   If GCC ever
> infers that it is "certainly uninitialized", we could defeat that
> through a use of 'volatile', such as

Yes, some use of volatile should do the trick for GCC (which is what my 
patch did). However, one would still have problems with a debugging 
implementation, e.g., if GCC ever supports an -fsanitize=uninitialized 
option that catches use of uninitialized storage.

This is all low priority of course.