This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: An unreviewed patch
On Fri, May 19, 2006 at 02:59:43PM -0700, Roland McGrath wrote:
> > This is also impacting powerpc64 when we compile with gcc-4.2. We where
> > planning to submit a patch based on Makefile changes to assert -fno-inline,
> > but the referenced patch looks like a better (more general) solution.
>
> It would be better to have a solution that does not add the useless extra
> function just to make this one conditional function call.
But that is precisely what generates the problematic constant pool
(testing the weak, external function). Many architectures will have
no way to do this given the constraints of _init. I can't even picture
a way to do it on PPC.
It's never been inlined before; previous versions of GCC either didn't
automatically inline it, or had that feature explicitly disabled via
-fno-unit-at-a-time.
--
Daniel Jacobowitz
CodeSourcery