The variable TARGET_LDFLAGS. isnt set but MODULE_LDFLAGS is instead.
if test "x$grub_cv_prog_ld_build_id_none" = xyes; then
MODULE_LDFLAGS="$MODULE_LDFLAGS -Wl,--build-id=none"
I could change to use TARGET_LDFLAGS if it wont brake other
architectures.
We were using PPC_BUILD_ID_FLAG because it was intended to be only to
PowerPC at first. I can tell if it is needed for others architectures
and if it wont brake others as well.
On Tue, 2008-12-02 at 13:34 -0500, Pavel Roskin wrote:
> On Tue, 2008-12-02 at 19:14 +0200, Vesa Jääskeläinen wrote:
>
> > No no. I mean just KERNEL_LDFLAGS like comparison to MODULE_LDFLAGS. No
> > elf or nothing like that. Is there any problem if one provides this
> > setting even on x86 if linker recognizes it? (what is being tested
> > here). PPC_BUILD_ID_FLAG just sounds too specific.
> >
> > Isn't EFI also using ELF? I assume this would be beneficial also there.
>
> I agree, we should not be be multiplying hacks. There is already code
> adding -Wl,--build-id=none to TARGET_LDFLAGS. It's a macro
> grub_PROG_LD_BUILD_ID_NONE in aclocal.m4. What's wrong with it? Does
> it fail to add -Wl,--build-id=none on PowerPC or is TARGET_LDFLAGS not
> used to link kernel.elf?
>
> What are the symptoms caused by not using -Wl,--build-id=none?
>
--
Best Regards,
Manoel Abranches <[EMAIL PROTECTED]>
IBM Linux Technology Center Brazil
_______________________________________________
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
