sh4: fix toolchain creation
PATCH v3 1/1 sh4: fix toolchain creation
The Linux kernel does force compile with -m4-nofpu, which is only available when building a multilib toolchain.
The interesting part here is, that buildroot use --disable-multilib for gcc configure, but enables --with-multilib-list=m4,m4-nofpu in the default configuration for Qemu targeting r2d emulation. This results in a toolchain, which can be used for the kernel and for userland without creating a multilib toolchain with different kinds of libgcc version. In the multilib case there would be subdirectories created (!m4 and m4-nofpu). As buildroot uses a short version of toolchain creation, a multilib enabled gcc build fails when creating libgcc.
So the best solution is to just keep multilib disabled, but always add --with-multilib-list when sh4/sh4eb/sh4a/sh4aeb is choosen.
Tested with sh4/sh4a toolchain build and qemu defconfig with gcc 4.8.x/4.9.x (with and without C++ enabled), uClibc and glibc.
Disable sh4a/sh4aeb for uClibc, as it does not implemented, yet.
---
Changes v3:
- add local variable for complete directory as suggested by Thomas Pettazoni
- disable sh4a/sh4aeb for uClibc as suggested by Thomas Pettazoni
Changes v2:
- fix build when C++ is enabled, reported by Thomas Pettazoni
- tested with sh4a/sh4aeb
--- configs/qemu_sh4_r2d_defconfig | 4 ---- package/gcc/gcc-final/gcc-final.mk | 22 +++++++++++++++++----- toolchain/toolchain-buildroot/Config.in | 4 ++-- 3 files changed, 19 insertions(+), 11 deletions(-)
configs/qemu_sh4_r2d_defconfig +0 -4
@@ -10,10 +10,6 @@ BR2_TARGET_GENERIC_GETTY_PORT="ttySC1" BR2_TARGET_ROOTFS_EXT2=y # BR2_TARGET_ROOTFS_TAR is not set -# The kernel wants to use the -m4-nofpu option to make sure that it -# doesn't use floating point operations. -BR2_EXTRA_GCC_CONFIG_OPTIONS="--with-multilib-list=m4,m4-nofpu" - # Lock to 3.19 headers to avoid breaking with newer kernels BR2_KERNEL_HEADERS_VERSION=y BR2_DEFAULT_KERNEL_VERSION="3.19"
package/gcc/gcc-final/gcc-final.mk +17 -5
@@ -52,6 +52,18 @@ HOST_GCC_FINAL_CONF_OPTS = \ --enable-poison-system-directories \ --with-build-time-tools=$(HOST_DIR)/usr/$(GNU_TARGET_NAME)/bin +HOST_GCC_FINAL_GCC_LIB_DIR = $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib* +# The kernel wants to use the -m4-nofpu option to make sure that it +# doesn't use floating point operations. +ifeq ($(BR2_sh4)$(BR2_sh4eb),y) +HOST_GCC_FINAL_CONF_OPTS += "--with-multilib-list=m4,m4-nofpu" +HOST_GCC_FINAL_GCC_LIB_DIR = $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib/!m4* +endif +ifeq ($(BR2_sh4a)$(BR2_sh4aeb),y) +HOST_GCC_FINAL_CONF_OPTS += "--with-multilib-list=m4a,m4a-nofpu" +HOST_GCC_FINAL_GCC_LIB_DIR = $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib/!m4* +endif + # Disable shared libs like libstdc++ if we do static since it confuses linking ifeq ($(BR2_STATIC_LIBS),y) HOST_GCC_FINAL_CONF_OPTS += --disable-shared
@@ -110,9 +122,9 @@ endif # Cannot use the HOST_GCC_FINAL_USR_LIBS mechanism below, because we want # libgcc_s to be installed in /lib and not /usr/lib. define HOST_GCC_FINAL_INSTALL_LIBGCC - -cp -dpf $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib*/libgcc_s* \ + -cp -dpf $(HOST_GCC_FINAL_GCC_LIB_DIR)/libgcc_s* \ $(STAGING_DIR)/lib/ - -cp -dpf $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib*/libgcc_s* \ + -cp -dpf $(HOST_GCC_FINAL_GCC_LIB_DIR)/libgcc_s* \ $(TARGET_DIR)/lib/ endef
@@ -140,7 +152,7 @@ endif ifneq ($(HOST_GCC_FINAL_USR_LIBS),) define HOST_GCC_FINAL_INSTALL_STATIC_LIBS for i in $(HOST_GCC_FINAL_USR_LIBS) ; do \ - cp -dpf $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib*/$${i}.a \ + cp -dpf $(HOST_GCC_FINAL_GCC_LIB_DIR)/$${i}.a \ $(STAGING_DIR)/usr/lib/ ; \ done endef
@@ -148,9 +160,9 @@ endef ifeq ($(BR2_STATIC_LIBS),) define HOST_GCC_FINAL_INSTALL_SHARED_LIBS for i in $(HOST_GCC_FINAL_USR_LIBS) ; do \ - cp -dpf $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib*/$${i}.so* \ + cp -dpf $(HOST_GCC_FINAL_GCC_LIB_DIR)/$${i}.so* \ $(STAGING_DIR)/usr/lib/ ; \ - cp -dpf $(HOST_DIR)/usr/$(GNU_TARGET_NAME)/lib*/$${i}.so* \ + cp -dpf $(HOST_GCC_FINAL_GCC_LIB_DIR)/$${i}.so* \ $(TARGET_DIR)/usr/lib/ ; \ done endef
toolchain/toolchain-buildroot/Config.in +2 -2
@@ -32,8 +32,8 @@ config BR2_TOOLCHAIN_BUILDROOT_UCLIBC depends on BR2_arcle || BR2_arceb || BR2_arm || BR2_armeb || \ BR2_bfin || BR2_i386 || BR2_m68k || \ BR2_mips || BR2_mipsel || BR2_mips64 || BR2_mips64el || \ - BR2_powerpc || BR2_sh || BR2_sparc || BR2_xtensa || \ - BR2_x86_64 + BR2_powerpc || BR2_sh2a || BR2_sh4 || BR2_sh4eb || \ + BR2_sparc || BR2_xtensa || BR2_x86_64 help This option selects uClibc as the C library for the cross-compilation toolchain.
Signature · 1 line
1.7.10.4
Thomas Petazzoni
AppliedPATCH v3 1/1 sh4: fix toolchain creation
Dear Waldemar Brodkorb,
Quoted 18 lines · Waldemar Brodkorb
The Linux kernel does force compile with -m4-nofpu, which is only available when building a multilib toolchain.
The interesting part here is, that buildroot use --disable-multilib for gcc configure, but enables --with-multilib-list=m4,m4-nofpu in the default configuration for Qemu targeting r2d emulation. This results in a toolchain, which can be used for the kernel and for userland without creating a multilib toolchain with different kinds of libgcc version. In the multilib case there would be subdirectories created (!m4 and m4-nofpu). As buildroot uses a short version of toolchain creation, a multilib enabled gcc build fails when creating libgcc.So the best solution is to just keep multilib disabled, but always add --with-multilib-list when sh4/sh4eb/sh4a/sh4aeb is choosen.
Tested with sh4/sh4a toolchain build and qemu defconfig with gcc 4.8.x/4.9.x (with and without C++ enabled), uClibc and glibc.
Quoted 7 lines
Disable sh4a/sh4aeb for uClibc, as it does not implemented, yet.
Signed-off-by Waldemar Brodkorb <[email protected]>---
Changes v3:
- add local variable for complete directory as suggested by Thomas Pettazoni
- disable sh4a/sh4aeb for uClibc as suggested by Thomas Pettazoni
Applied, thanks a lot!
Thomas
Signature · 3 lines
Thomas Petazzoni, CTO, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com