From: Alan Modra <amodra@gmail.com>
Date: Thu, 16 Apr 2020 03:15:30 +0000 (+0930)
Subject: PowerPC64 GOT reloc reserving PLT entry for ifunc
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=2165dc8d90ef9618a1db9bc596f433b02b7cc54a;p=binutils-gdb.git

PowerPC64 GOT reloc reserving PLT entry for ifunc

I can't see any reason why ELFv2 should create a PLT entry for ifuncs
referenced by GOT relocs as long as the GOT entry remains.  The GOT
entry ought to be resolved by ld.so to the value returned by the ifunc
resolver, or if there is global entry stub created for some other
reason, by the linker to the stub address.

	* elf64-ppc.c (ppc64_elf_check_relocs): Don't create plt entries
	for GOT relocs against ifuncs.
---

diff --git a/bfd/elf64-ppc.c b/bfd/elf64-ppc.c
index 25a3e468a73..63de3aba594 100644
--- a/bfd/elf64-ppc.c
+++ b/bfd/elf64-ppc.c
@@ -4774,14 +4774,6 @@ ppc64_elf_check_relocs (bfd *abfd, struct bfd_link_info *info,
 	    if (!update_local_sym_info (abfd, symtab_hdr, r_symndx,
 					rel->r_addend, tls_type))
 	      return FALSE;
-
-	  /* We may also need a plt entry if the symbol turns out to be
-	     an ifunc.  */
-	  if (h != NULL && !bfd_link_pic (info) && abiversion (abfd) != 1)
-	    {
-	      if (!update_plt_info (abfd, &h->plt.plist, rel->r_addend))
-		return FALSE;
-	    }
 	  break;
 
 	case R_PPC64_PLT16_HA: