Re: [PATCH v2] newlib: fix build with <gcc-5 versions

>> It's not just about old GCC, it's about any C compiler that doesn't have
>> that builtin.
> 
> Well, I guess, GTG then.


Let's see if I understand the situation:

You are committing a replacement implementation of __builtin_mul_overflow() for older GCC versions and for any other compiler which does not have it.

The only significant extra feature about that function is the detection of integer overflow.

The implementation lives in libc/include/sys/cdefs.h , so it is accessible not just by some special malloc code which should never overflow because targets wouldn't have that much memory or whatever.

The replacement implementation is known to be broken and therefore poses a risk on anybody relying on the original, documented behaviour.

There are no mitigation measures. There is not even a comment next to the replacement implementation that states it is broken.

And you guys are fine with that.

Is that correct?

Regards,
   rdiez