Re: [PATCH] stdlib: tests: don't double-define _FORTIFY_SOURCE

[Sam James <sam@gentoo.org>]

[-- Attachment #1: Type: text/plain, Size: 1499 bytes --]



> On 10 Aug 2022, at 15:29, Siddhesh Poyarekar <siddhesh@gotplt.org> wrote:
> 
> On 2022-08-08 18:30, Sam James wrote:
>> This is a fair point, although I now see we've actually got libc_cv_predef_fortify_source
>> which sets CPPUNDEFS for exactly this sort of problem anyway.
>> I don't see the warning with gcc-11 + -Werror + F_S=3 on a test program. I can build some
>> older GCCs as I should probably keep them around anyway though.
> 
> Interesting, because you should have seen the warning about _FORTIFY_SOURCE=3 not being supported; I'm surprised that it doesn't fail due to that warning.

... and I see it now. I don't know what I did the other day. Oops.

> 
>>> Alternatively, some magic here to determine the maximum fortification level wouldn't hurt, but I won't gate your patch on that :)  I can work on that bit.
>> I started looking at that and I'm not sure there's a point. includes/features.h downgrades us appropriately. I think we can unconditionally
>> set F_S=3 if I'm right about GCC not caring, as all the logic is on the glibc side, right?
> 
> It's the warning I'm thinking about avoiding.
> 
>> We can always split this into two if you want:
>> 1. The original commit (I can convert it to use libc_cv_predef_fortify_source's result) & backport it to 2.36
> 
> Sounds good to me.
> 
>> 2. Another to crank to =3 and don't backport it in case I'm missing something.
> 

Okay, sounds like a plan, cheers. I'll get on it.

> Thanks,
> Sid


[-- Attachment #2: Message signed with OpenPGP --]
[-- Type: application/pgp-signature, Size: 358 bytes --]