[RFC PATCH] powerpc/Makefile: Add option to build with gold

[RFC PATCH] powerpc/Makefile: Add option to build with gold

[Michael Ellerman]

This adds a Kconfig option which can be turned on to enable use of the
gold linker. It depends on DEBUG_KERNEL to hopefully hide it from users
who don't really care about it, and also !COMPILE_TEST so that it will
be turned off for allyes/modconfig builds.

In the wrapper script we currently always use ${CROSS}ld, so instead
make that the default if ${LD} isn't already set. When we're called as
part of a normal build ${LD} will be set, but if the wrapper is called
separately we will fall back to ${CROSS}ld as before.

With this enabled I have successfully built a ppc64[le]_defconfig using
a toolchain which only has ld.gold.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
 arch/powerpc/Kconfig.debug | 7 +++++++
 arch/powerpc/Makefile      | 4 ++++
 arch/powerpc/boot/wrapper  | 4 +++-
 3 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/Kconfig.debug b/arch/powerpc/Kconfig.debug
index 63292f64b25a..c63ec2bb08ae 100644
--- a/arch/powerpc/Kconfig.debug
+++ b/arch/powerpc/Kconfig.debug
@@ -19,6 +19,13 @@ config PPC_WERROR
 	depends on !PPC_DISABLE_WERROR
 	default y
 
+config PPC_USE_GOLD
+	bool "Build with the gold linker"
+	depends on DEBUG_KERNEL && !COMPILE_TEST
+	default n
+	  Build the kernel with the ld.gold linker, rather than ld.bfd.
+	  If you don't know what this means say N.
+
 config PRINT_STACK_DEPTH
 	int "Stack depth to print" if DEBUG_KERNEL
 	default 64
diff --git a/arch/powerpc/Makefile b/arch/powerpc/Makefile
index 1934707bf321..005c63fd8be1 100644
--- a/arch/powerpc/Makefile
+++ b/arch/powerpc/Makefile
@@ -65,6 +65,10 @@ endif
 
 UTS_MACHINE := $(OLDARCH)
 
+ifdef CONFIG_PPC_USE_GOLD
+LD := $(LD).gold
+endif
+
 ifeq ($(CONFIG_CPU_LITTLE_ENDIAN),y)
 override LD	+= -EL
 override CROSS32AS += -mlittle-endian
diff --git a/arch/powerpc/boot/wrapper b/arch/powerpc/boot/wrapper
index 6681ec3625c9..00f13d18fd6f 100755
--- a/arch/powerpc/boot/wrapper
+++ b/arch/powerpc/boot/wrapper
@@ -140,6 +140,8 @@ while [ "$#" -gt 0 ]; do
     shift
 done
 
+LD=${LD:-${CROSS}ld}
+
 if [ -n "$dts" ]; then
     if [ ! -r "$dts" -a -r "$object/dts/$dts" ]; then
 	dts="$object/dts/$dts"
@@ -417,7 +419,7 @@ if [ "$platform" != "miboot" ]; then
     if [ -n "$link_address" ] ; then
         text_start="-Ttext $link_address"
     fi
-    ${CROSS}ld -m $format -T $lds $text_start $pie -o "$ofile" \
+    ${LD} -m $format -T $lds $text_start $pie -o "$ofile" \
 	$platformo $tmp $object/wrapper.a
     rm $tmp
 fi
-- 
2.7.4

Re: [RFC PATCH] powerpc/Makefile: Add option to build with gold

[Nicholas Piggin]

On Fri, 12 Aug 2016 21:30:21 +1000
Michael Ellerman <mpe@ellerman.id.au> wrote:

> This adds a Kconfig option which can be turned on to enable use of the
> gold linker. It depends on DEBUG_KERNEL to hopefully hide it from users
> who don't really care about it, and also !COMPILE_TEST so that it will
> be turned off for allyes/modconfig builds.
> 
> In the wrapper script we currently always use ${CROSS}ld, so instead
> make that the default if ${LD} isn't already set. When we're called as
> part of a normal build ${LD} will be set, but if the wrapper is called
> separately we will fall back to ${CROSS}ld as before.
> 
> With this enabled I have successfully built a ppc64[le]_defconfig using
> a toolchain which only has ld.gold.

Nice. Out of curiosity, I ran some testing.

On allyesconfig with sfr's thin archives patch and --gc-sections,
ld.gold is about 10% faster than ld.bfd. About 11 seconds vs 12. A lot
of that is in utf8 conversions from "fnmatch" it's using to match
section names though. Setting LANG=C brings it down to 8 seconds and
helps ld.bfd a tiny bit but not much. So that's a pretty good win.
Threading still has some crashes.

ld.gold creates a larger executable though. It seems to be creating a
lot more branch stubs (19000 vs 6500 for ld.bfd). 

I should keep in mind we link the kernel about 4 times every build
before complain too much about linker performance to the binutils guys.
I wonder if we couldn't objcopy the kallsyms onto the end of the image
to avoid that relinking.