[PATCH] build: Use -frandom-seed=0

[Chris Johns]

On 29/2/2024 6:36 pm, Sebastian Huber wrote:
> On 29.02.24 00:29, Chris Johns wrote:
>> On 28/2/2024 6:24 pm, Sebastian Huber wrote:
>>> On 28.02.24 06:34, Chris Johns wrote:
>>>> The manual says:
>>>>
>>>>     The string can either be a number (decimal, octal or hex) or
>>>>     an arbitrary string (in which case it’s converted to a number by
>>>>     computing CRC32).
>>>>
>>>>     The string should be different for every file you compile.
>>>>
>>>> I take this to mean the option `-frandom-seed=0` uses `0` as a number
>>>> however it
>>>> is the same for every file and the manual clearly says it must be different?
>>> Using -frandom-seed=0 seems to be quite common on the internet. The random seed
>>> is rarely used in GCC. The only use case in RTEMS I found is related to the gcov
>>> code coverage instrumentation.
>> There are lots of things that are common on the internet I ignore 😉
>>
>> Is this a bug in the documentation?
> 
> From my point of view this random seed is a gray area in the compiler. The cases
> in which it is used should not matter for the RTEMS build (except for the code
> coverage):
> 
> https://gcc.gnu.org/pipermail/gcc-help/2024-February/143324.html
> 
> I will try to add it only to the coverage flags.

I would prefer knowing which path is right. If the documentation for gcc says
not to do something then I am not sure it is good for us to add it anywhere. For
example it is working now in your testing but a future release of gcc changes
and subtle issues appear our testing does not pick up. If the gcc maintainers
say it is OK then I am fine but they need to update their documentation. I know
nothing about this options so it is difficult for me to say yes to the change
when it is in conflict to the published documentation. Others may have a
different view.

Chris