PR 21224, PowerPC64 abort due to dynamic relocs on hidden undefweak

PR 21224, PowerPC64 abort due to dynamic relocs on hidden undefweak

[Alan Modra]

binutils-2.28 was out for less than 3 hours before this PR was
raised.  I wonder if that is some sort of record?

ppc64_elf_relocate_section lacked a check which meant that it emitted
dynamic relocs against a hidden undefweak symbol for which no dynamic
relocs had been allocated.

	PR 21224
	PR 20519
	* elf64-ppc.c (ppc64_elf_relocate_section): Add missing
	dyn_relocs check.

diff --git a/bfd/ChangeLog b/bfd/ChangeLog
index 92682d5..9560ad2 100644
--- a/bfd/ChangeLog
+++ b/bfd/ChangeLog
@@ -1,3 +1,10 @@
+2017-03-07  Alan Modra  <amodra@gmail.com>
+
+	PR 21224
+	PR 20519
+	* elf64-ppc.c (ppc64_elf_relocate_section): Add missing
+	dyn_relocs check.
+
 2017-03-05  Alan Modra  <amodra@gmail.com>
 
 	* elf-bfd.h (struct eh_cie_fde): Add u.cie.per_encoding_aligned8.
diff --git a/bfd/elf64-ppc.c b/bfd/elf64-ppc.c
index 84cb214..d0673c9 100644
--- a/bfd/elf64-ppc.c
+++ b/bfd/elf64-ppc.c
@@ -14804,8 +14804,10 @@ ppc64_elf_relocate_section (bfd *output_bfd,
 	    break;
 
 	  if (bfd_link_pic (info)
-	      ? ((h != NULL && pc_dynrelocs (h))
-		 || must_be_dyn_reloc (info, r_type))
+	      ? ((h == NULL
+		  || h->dyn_relocs != NULL)
+		 && ((h != NULL && pc_dynrelocs (h))
+		     || must_be_dyn_reloc (info, r_type)))
 	      : (h != NULL
 		 ? h->dyn_relocs != NULL
 		 : ELF_ST_TYPE (sym->st_info) == STT_GNU_IFUNC))

-- 
Alan Modra
Australia Development Lab, IBM