From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnout Vandecappelle Date: Thu, 05 Mar 2015 22:41:28 +0100 Subject: [Buildroot] [PATCH 1/1] toolchain: add link-time-optimization support In-Reply-To: <1425479715-5716-1-git-send-email-syntheticpp@gmx.net> References: <1425479715-5716-1-git-send-email-syntheticpp@gmx.net> Message-ID: <54F8CD88.1080006@mind.be> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Hi Peter, On 04/03/15 15:35, Peter K?mmel wrote: I would start the commit log with mentioning that LTO is not used by default. > GCC's link-time-optimization could be enabled by the flag '-flto'. > With LTO enabled, ar and runlib must be called with an argument which runlib -> ranlib > triggers the usage of a LTO plugin. > GCC provides wrappers for ar and ranlib which implicitly pass the LTO > arguments to ar/ranlib. This way existing Makefiles don't need to be > changed for LTO support when these wrappers around ar and ranlib are used. > Also the LTO and plugin support must be enabled in the host's binutils. This part should mention explicitly that it applies to the external toolchains. Example of an (IMHO) improved commit message: >>> Add a new option BR2_GCC_ENABLE_LTO that builds gcc and binutils with LTO support. Individual packages still have to enable LTO explicitly by passing '-flto' to GCC, which passes it on to the linker. This option does not add that flag globally. Some packages detect if the compiler supports LTO and enable the flag if it does. To support LTO, ar and ranlib must be called with an argument which triggers the usage of the LTO plugin. Since GCC doesn't call these tools itself, it instead provides wrappers for ar and ranlib that pass the LTO arguments. This way existing Makefiles don't need to be changed for LTO support. However, these wrappers are called -gcc-ar which matches the pattern to link to the buildroot wrapper in the external toolchain logic. So the external toolchain logic is updated to provide the correct symlink. <<< > > Signed-off-by: Peter K?mmel Can you add a patch changelog in the future? Look at other patches on the list to see how it is done. Also, put the version number of the patch in the subject. You can easily do this using 'git format-patch -v5' (or for older git versions 'git format-patch --subject-prefix="PATCH v5"') > --- > package/binutils/binutils.mk | 4 ++++ > package/gcc/Config.in.host | 7 +++++++ > package/gcc/gcc.mk | 4 ++++ > toolchain/toolchain-external/toolchain-external.mk | 8 ++++++++ > 4 files changed, 23 insertions(+) > > diff --git a/package/binutils/binutils.mk b/package/binutils/binutils.mk > index 9e99253..daa8c45 100644 > --- a/package/binutils/binutils.mk > +++ b/package/binutils/binutils.mk > @@ -96,5 +96,9 @@ BINUTILS_PRE_PATCH_HOOKS += BINUTILS_XTENSA_PRE_PATCH > HOST_BINUTILS_PRE_PATCH_HOOKS += BINUTILS_XTENSA_PRE_PATCH > endif > > +ifeq ($(BR2_GCC_ENABLE_LTO),y) > +HOST_BINUTILS_CONF_OPTS += --enable-plugins --enable-lto > +endif > + > $(eval $(autotools-package)) > $(eval $(host-autotools-package)) > diff --git a/package/gcc/Config.in.host b/package/gcc/Config.in.host > index dd61f51..3c83579 100644 > --- a/package/gcc/Config.in.host > +++ b/package/gcc/Config.in.host > @@ -116,6 +116,13 @@ config BR2_GCC_ENABLE_TLS > Enable the compiler to generate code for accessing > thread local storage variables > > +config BR2_GCC_ENABLE_LTO > + bool "Enable compiler link-time-optimization support" > + depends on !BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 > + help > + Since version 4.5 GCC supports lto. Build GCC with lto support enabled. This line should be wrapped. And perhaps an improved help text would be: Since version 4.5, GCC supports link-time optimization (LTO). Select this option to turn on LTO support in GCC. > + Needed when -flto should be used. > + > config BR2_GCC_ENABLE_OPENMP > bool "Enable compiler OpenMP support" > depends on !BR2_PTHREADS_NONE && !BR2_arc && !BR2_microblaze > diff --git a/package/gcc/gcc.mk b/package/gcc/gcc.mk > index ffac15c..b5d2ddb 100644 > --- a/package/gcc/gcc.mk > +++ b/package/gcc/gcc.mk > @@ -132,6 +132,10 @@ else > HOST_GCC_COMMON_CONF_OPTS += --disable-tls > endif > > +ifeq ($(BR2_GCC_ENABLE_LTO),y) > +HOST_GCC_COMMON_CONF_OPTS += --enable-plugins --enable-lto > +endif > + > ifeq ($(BR2_GCC_ENABLE_LIBMUDFLAP),y) > HOST_GCC_COMMON_CONF_OPTS += --enable-libmudflap > else > diff --git a/toolchain/toolchain-external/toolchain-external.mk b/toolchain/toolchain-external/toolchain-external.mk > index ab73f9a..d7a8499 100644 > --- a/toolchain/toolchain-external/toolchain-external.mk > +++ b/toolchain/toolchain-external/toolchain-external.mk > @@ -656,12 +656,20 @@ endif > # Build toolchain wrapper for preprocessor, C and C++ compiler and setup > # symlinks for everything else. Skip gdb symlink when we are building our > # own gdb to prevent two gdb's in output/host/usr/bin. > +# When the link-time-optimazation flag '-flto' is used, then the compiler > +# and binutils have to support lto. ar/ranlib needs to be called with the > +# lto plugin. The wrappers *-gcc-ar and *-gcc-ranlib provided by GCC could > +# be used as drop-ins for ar/runlib when Makefiles are used which do not > +# pass the lto arguments. > define TOOLCHAIN_EXTERNAL_INSTALL_WRAPPER > $(Q)$(call MESSAGE,"Building ext-toolchain wrapper") > mkdir -p $(HOST_DIR)/usr/bin; cd $(HOST_DIR)/usr/bin; \ > for i in $(TOOLCHAIN_EXTERNAL_CROSS)*; do \ > base=$${i##*/}; \ > case "$$base" in \ > + *-ar|*-ranlib|*-nm) \ > + ln -sf $$(echo $$i | sed 's%^$(HOST_DIR)%../..%') .; \ > + ;; \ This part could actually be a separate patch, since it fixes a problem that is exists even if we don't support LTO in buildroot. However, since this patch is already in good shape, (just some text changes), it's not really necessary for me to still split it. > *cc|*cc-*|*++|*++-*|*cpp) \ > ln -sf ext-toolchain-wrapper $$base; \ > ;; \ > -- Arnout Vandecappelle arnout at mind be Senior Embedded Software Architect +32-16-286500 Essensium/Mind http://www.mind.be G.Geenslaan 9, 3001 Leuven, Belgium BE 872 984 063 RPR Leuven LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle GPG fingerprint: 7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F