Is PRE architecture dependent? aarch64 vs x86_64

[David Brown]

On 18/07/2025 18:55, Segher Boessenkool wrote:
> Hi!
> 
> On Fri, Jul 18, 2025 at 06:39:41PM +0200, David Brown wrote:
> 
>> Are you able to give an example of the C code for which the optimisation
>> above applies, and values for which the result is affected?  (When thinking
>> about overflows, I always like to use 16-bit int because the numbers are
>> smaller and easier to work with.)
> 
> 16 bits?  Why so big!  :-)

Well, we could test it on the AVR target with "-mint8" in effect, but 
some people might worry that 8-bit ints are non-conforming...


> 
>> Conversion to signed integer types is implementation-defined behaviour in
>> the C standards, not undefined behaviour.  That means the compiler must pick
>> a specific tactic which is documented (in section 4.5 of the gcc manual) and
>> applied consistently.  It is not undefined behaviour - code that relies on
>> two's complement conversion of unsigned types to signed types is not
>> incorrect code, merely non-portable code.  (In practice, of course, it is
>> portable, as all real-world compilers use the same tactic on two's
>> complement targets.)
> 
> Leaving it as UB is a correct implementation.  If that is documented it
> is IB then, but what is the difference here :-)
> 

No, UB is /not/ a correct implementation of signed integer conversions. 
It has to be defined and documented, but the implementation can choose 
the details.