https://sourceware.org/bugzilla/show_bug.cgi?id=25803
--- Comment #18 from gagan singh sidhu (gagz, broly, w/e u want) <broly at mac
dot com> ---
looking at the binutils and searching for the keyword, is it possible 'gas' is
the problem?:
GagansMacPro:mkbu Gagan$ grep -rn gnu_indirect_function .
Binary file ./gas/as-new matches
Binary file ./gas/config/obj-elf.o matches
./gold/config.log:907:| __asm__(".type invoke, %gnu_indirect_function");
looking at the output for gas/config/obj-elf.c:
else if (strcmp (type_name, "gnu_indirect_function") == 0
|| strcmp (type_name, "10") == 0
|| strcmp (type_name, "STT_GNU_IFUNC") == 0)
{
struct elf_backend_data *bed;
bed = (struct elf_backend_data *) get_elf_backend_data (stdoutput);
if (bed->elf_osabi == ELFOSABI_NONE)
bed->elf_osabi = ELFOSABI_GNU;
else if (bed->elf_osabi != ELFOSABI_GNU
&& bed->elf_osabi != ELFOSABI_FREEBSD)
as_bad (_("symbol type \"%s\" is supported only by GNU "
"and FreeBSD targets"), type_name);
elf_tdata (stdoutput)->has_gnu_osabi |= elf_gnu_osabi_ifunc;
type = BSF_FUNCTION | BSF_GNU_INDIRECT_FUNCTION;
}
it seems using the 'none' causes bed->elf_osabi = ELFOSABI_GNU, whereas it
should be falling into the else if condition? sorry i don't know this code
well. i know the tuple is a 'none-linux-gnu' though and i'm curious if that
causes it to be assigned ELFOSABI_NONE first, and then ELFOSABI_GNU, or if it
is assigned ELFOSABI_GNU initially.
--
You are receiving this mail because:
You are on the CC list for the bug.
