This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
[GOLD] PowerPC tweak relnum tests
- From: Alan Modra <amodra at gmail dot com>
- To: binutils at sourceware dot org
- Date: Fri, 28 Jun 2019 11:46:26 +0930
- Subject: [GOLD] PowerPC tweak relnum tests
There is a call of relocate() to perform a single relocation. In that
case the "relnum" parameter is -1U and of course it isn't appropriate
to consider any of the PowerPC code sequence optimisations triggered
by a following relocation.
* powerpc.cc (Target_powerpc::Relocate::relocate): Don't look
at next/previous reloc when relnum is -1.
diff --git a/gold/powerpc.cc b/gold/powerpc.cc
index 43010b8c05..6f0be0975a 100644
--- a/gold/powerpc.cc
+++ b/gold/powerpc.cc
@@ -8986,7 +8986,7 @@ Target_powerpc<size, big_endian>::Relocate::relocate(
size_t reloc_count = shdr.get_sh_size() / reloc_size;
if (size == 64
&& ent->r2save_
- && relnum + 1 < reloc_count)
+ && relnum < reloc_count - 1)
{
Reltype next_rela(preloc + reloc_size);
if (elfcpp::elf_r_type<size>(next_rela.get_r_info())
@@ -9757,7 +9757,7 @@ Target_powerpc<size, big_endian>::Relocate::relocate(
// addi 2,2,.TOC.@l
// if .TOC. is in range. */
if (value + address - 4 + 0x80008000 <= 0xffffffff
- && relnum != 0
+ && relnum + 1 > 1
&& preloc != NULL
&& target->abiversion() >= 2
&& !parameters->options().output_is_position_independent()
--
Alan Modra
Australia Development Lab, IBM