On Sun, Aug 16, 2020 at 8:02 AM Arvind Sankar <nivedita@alum.mit.edu> wrote:
>
> On Sun, Aug 16, 2020 at 07:22:35AM +0200, Sedat Dilek wrote:
> > On Sun, Aug 16, 2020 at 2:19 AM 'Fangrui Song' via Clang Built Linux
> > <clang-built-linux@googlegroups.com> wrote:
> > >
> > > Adding a definition without a declaration for stpcpy looks good.
> > > Clang LTO will work.
> > >
> > > (If the kernel does not want to provide these routines,
> > > is http://git.kernel.org/linus/6edfba1b33c701108717f4e036320fc39abe1912
> > > probably wrong? (why remove -ffreestanding from the main Makefile) )
> > >
> >
> > We had some many issues in arch/x86 where *FLAGS were removed or used
> > differently and had to re-add them :-(.
> >
> > So if -ffreestanding is used in arch/x86 and was! used in top-level
> > Makefile - it makes sense to re-add it back?
> > ( I cannot speak for archs other than x86. )
> >
> > - Sedat -
>
> -ffreestanding disables _all_ builtins and libcall optimizations, which
> is probably not desirable. If we added it back, we'd need to also go
> back to #define various string functions to the __builtin versions.
>
> Though I don't understand the original issue, with -ffreestanding,
> sprintf shouldn't have been turned into strcpy in the first place.
>
> 32-bit still has -ffreestanding -- I wonder if that's actually necessary
> any more?
>
> Why does -fno-builtin-stpcpy not work with clang LTO? Isn't that a
> compiler bug?

I just confirmed that adding -fno-builtin-stpcpy to KBUILD_CFLAGS does
work with LTO as well.

Sami