Piotr Stefaniak <postgres@piotr-stefaniak.me> writes:
> On 06/22/2015 08:55 PM, Tom Lane wrote:
>> If I recall that code correctly, the assumption was that if the third
>> argument is zero then memcmp() must not fetch any bytes (not should not,
>> but MUST not) and therefore it doesn't matter if we pass a NULL.
> It's not about fetching any bytes, it's about passing an invalid pointer
> (a null pointer in this case) which gives a compiler the opportunity to
> apply an optimization. For example, glibc has memcpy marked as
> __nonnull(1,2).
If they do, that's a glibc bug, per the above observation.
regards, tom lane