[PATCH] inet: Increase htontest coverage to run-time evaluation, library
Florian Weimer
fweimer@redhat.com
Mon Jun 17 13:27:17 GMT 2024
* Florian Weimer:
> * Andreas Schwab:
>
>> On Jun 14 2024, Florian Weimer wrote:
>>
>>> +/* Use a volatile function pointer as a compiler barrier, to test the
>>> + library implementation, not the compiler intrinsic. */
>>> +uint32_t (*volatile htonl_ptr) (uint32_t) = htonl;
>>> +uint32_t (*volatile ntohl_ptr) (uint32_t) = ntohl;
>>> +uint16_t (*volatile htons_ptr) (uint16_t) = htons;
>>> +uint16_t (*volatile ntohs_ptr) (uint16_t) = ntohs;
>>
>> Perhaps -fno-builtin is already sufficient?
>
> Hmm. There is no compiler builtin yet, which means that (htonl)
> etc. are currently sufficient for ensuring that the implementation is
> called. I'd still like to make this change in case one gets added.
Let's drop this patch.
Thanks,
Florian
More information about the Libc-alpha
mailing list