[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gnupg-users
Subject:    Re: Strange message seen on FreeBSD 14.0 amd64
From:       Jacob Bachmeyer via Gnupg-users <gnupg-users () gnupg ! org>
Date:       2023-07-12 1:30:41
Message-ID: 64AE0241.6020607 () gmail ! com
[Download RAW message or body]

Dennis Clarke via Gnupg-users wrote:
>
> Dear GnuPG type folks :
>
>     I don't know what this means. Can we just compile with a decent C
>  compiler such as the LLVM/Clang in FreeBSD ?
>
>
> [...]
>
>         Libgcrypt v1.10.2 has been configured as follows:
>
> [...]
>
>    Please not that your compiler does not support the GCC style
>    aligned attribute. Using this software may evoke bus errors.
>
> hydra$
>
>
> So what does that mean ? I *must* use GCC to compile this source ?

It means that the sources use a GNU extension that configure has 
detected that Clang does not properly implement.  The specific example 
cited ("aligned") should be non-critical for you, since you are running 
on AMD64 and that architecture does not actually require proper 
alignment.  The resultant executables should work in your case, but at 
reduced performance (unaligned accesses are permitted on x86-64, but are 
slower than aligned accesses) unless SSE (which *does* have hard 
alignment requirements) is used.  Since I note that you are disabling 
the use of assembler modules, SSE will probably *not* be used in your 
executable.

In short, try it---if it works for you, great!  If GPG crashes with 
SIGBUS, try rebuilding it with GCC before reporting a bug in GPG.  If it 
works when built with GCC, you have found a bug (a missing feature that 
Clang claims to have) in Clang.  Clang typically defines __GNUC__, thus 
claiming to support GNU extensions, so this is a bug in Clang if your 
Clang-compiled GPG does not work.


-- Jacob


_______________________________________________
Gnupg-users mailing list
Gnupg-users@gnupg.org
https://lists.gnupg.org/mailman/listinfo/gnupg-users
[prev in list] [next in list] [prev in thread] [next in thread]