Toolchain updates
cover Toolchain updates
Hello,
Here is a set of updates/improvements to the internal toolchain backend (with some impacts on the external toolchain one). Mainly it:
* Allows to build the toolchain with only two passes of gcc build as opposed to three. Of course, this is only possible in the appropriate situations. This patch had already been posted previously to the list, and was rebased on top of current master.
* Fixes Fortran and Objective-C support, moves the C++ option close to the one for Fortran and Objective-C.
* Adds support for glibc, in addition to eglibc. Note that as suggested by Arnout, I'm proposing a single package to handle both libraries.
* Refactors the Stack Smashing Protection support.
* Refactors the Mudflap support. Note however that I didn't manage to get a Busybox system with mudflap enabled to boot. I am personaly not that much interested in mudflap support, so unless someone steps up to look at the issue, I think I'll propose to remove Mudflap support entirely.
* Makes a small number of related cleanups.
Thomas Petazzoni (12):
package/Makefile.in: fix incorrect comment
eglibc: libstdc++ is not part of the C library
toolchain-internal: skip gcc-intermediate when possible
gcc: fix Fortran support
gcc: fix Objective-C support
gcc: move C++ support option next to Fortran/Objective-C
toolchain: add support for glibc
gcc/gcc-final: use the common HOST_GCC_CONFIGURE_SYMLINK macro
gcc: force symlink creation
toolchain: refactor Stack Smashing Protection support
toolchain: move elf2flt option inclusion to
toolchain-buildroot/Config.in
toolchain: improve mudflap support
Config.in | 37 +++++++++++++++
package/Makefile.in | 17 +++++--
package/gcc/Config.in.host | 53 ++++++++++++++++++---
package/gcc/gcc-final/gcc-final.mk | 31 ++++++++----
package/gcc/gcc-initial/gcc-initial.mk | 8 ++++
package/gcc/gcc.mk | 8 +++-
package/{eglibc/eglibc.mk => glibc/glibc.mk} | 70 ++++++++++++++++------------
package/psmisc/psmisc.mk | 2 +-
package/uclibc/Config.in | 11 +++++
package/uclibc/uclibc.mk | 5 +-
toolchain/Config.in | 11 ++++-
toolchain/helpers.mk | 15 +++++-
toolchain/toolchain-buildroot/Config.in | 25 +++++++++-
toolchain/toolchain-buildroot/Config.in.2 | 31 ------------
toolchain/toolchain-common.in | 6 +++
toolchain/toolchain-external/Config.in | 18 +++++++
toolchain/toolchain-external/ext-tool.mk | 11 +++++
17 files changed, 270 insertions(+), 89 deletions(-)
rename package/{eglibc/eglibc.mk => glibc/glibc.mk} (58%)
delete mode 100644 toolchain/toolchain-buildroot/Config.in.2Signature · 1 line
1.8.1.2
PATCH 1/12 package/Makefile.in: fix incorrect comment
The TARGET_{CC,CXX,LD,...} variables no longer contain any --sysroot option, since we're now using a toolchain wrapper for external toolchains.
--- package/Makefile.in | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
package/Makefile.in +1 -2
@@ -143,8 +143,7 @@ endif # Quotes are needed for spaces et al in path components. TARGET_PATH="$(HOST_DIR)/bin:$(HOST_DIR)/usr/bin:$(HOST_DIR)/usr/sbin/:$(PATH)" -# Define TARGET_xx variables for all common binutils/gcc tools by -# including the --sysroot option where necessary. +# Define TARGET_xx variables for all common binutils/gcc TARGET_AR = $(TARGET_CROSS)ar TARGET_AS = $(TARGET_CROSS)as TARGET_CC = $(TARGET_CROSS)gcc
Signature · 1 line
1.8.1.2
PATCH 2/12 eglibc: libstdc++ is not part of the C library
In reply to Thomas PetazzoniFor some reason, the eglibc.mk file was trying to install libstdc++.so from eglibc to the target. But the C++ standard library is provided by GCC, not by the C library.
--- package/eglibc/eglibc.mk | 7 ------- 1 file changed, 7 deletions(-)
package/eglibc/eglibc.mk +0 -7
@@ -89,17 +89,10 @@ ifeq ($(BR2_PACKAGE_GDB_SERVER),y) EGLIBC_LIBS_LIB += libthread_db.so endif -ifeq ($(BR2_INSTALL_LIBSTDCPP),y) -EGLIBC_LIBS_USR_LIB += libstdc++.so -endif - define EGLIBC_INSTALL_TARGET_CMDS for libs in $(EGLIBC_LIBS_LIB); do \ $(call copy_toolchain_lib_root,$(STAGING_DIR)/,,lib,$$libs,/lib) ; \ done - for libs in $(EGLIBC_LIBS_USR_LIB); do \ - $(call copy_toolchain_lib_root,$(STAGING_DIR)/,,lib,$$libs,/usr/lib) ; \ - done endef $(eval $(autotools-package))
Signature · 1 line
1.8.1.2
PATCH 3/12 toolchain-internal: skip gcc-intermediate when possible
In reply to Thomas PetazzoniWhen NPTL support was introduced, gcc required a three stages build process. Since gcc 4.7, this is no longer necessary, and it is possible to get back to a two stages build process. This patch takes advantage of this, by doing a two stages build process when possible.
We introduce a few hidden kconfig options:
* BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD, which is set by the gcc Config.in logic to indicate that the compiler might need a three stages build. Currently, all versions prior to 4.7.x are selecting this kconfig option.
* BR2_TOOLCHAIN_LIBC_NEEDS_THREE_STAGE_BUILD, which indicates whether the C library might need a three stages build. This is the case for eglibc, and uClibc when NPTL is enabled.
* BR2_TOOLCHAIN_NEEDS_THREE_STAGE_BUILD finally is enabled when both of the previous options are enabled. It indicates that a three stages build is actually needed.
In addition to those options, the uClibc/gcc build logic is changed to take use only a two stages build process when possible.
--- package/eglibc/eglibc.mk | 5 +++-- package/gcc/Config.in.host | 11 +++++++++++ package/gcc/gcc-initial/gcc-initial.mk | 8 ++++++++ package/uclibc/Config.in | 1 + package/uclibc/uclibc.mk | 5 +++-- toolchain/Config.in | 9 +++++++++ 6 files changed, 35 insertions(+), 4 deletions(-)
package/eglibc/eglibc.mk +3 -2
@@ -14,8 +14,9 @@ EGLIBC_LICENSE_FILES = libc/COPYING libc/COPYING.LIB libc/LICENSES # cross-compiler and the kernel headers EGLIBC_DEPENDENCIES = host-gcc-initial linux-headers host-gawk -# Before eglibc is built, we must have the second stage cross-compiler -eglibc-build: host-gcc-intermediate +# Before eglibc is built, we must have the second stage +# cross-compiler, for some gcc versions +eglibc-build: $(if $(BR2_TOOLCHAIN_NEEDS_THREE_STAGE_BUILD),host-gcc-intermediate) EGLIBC_SUBDIR = build
package/gcc/Config.in.host +11 -0
@@ -3,6 +3,12 @@ comment "GCC Options" config BR2_GCC_NEEDS_MPC bool +# Until gcc 4.7, a three stage build process was needed when using +# NPTL. This hidden option tells whether gcc is a version that +# requires this three stage build process. +config BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD + bool + choice prompt "GCC compiler Version" default BR2_GCC_VERSION_4_4_X if BR2_sparc_sparchfleon || BR2_sparc_sparchfleonv8 || BR2_sparc_sparcsfleon || BR2_sparc_sparcsfleonv8
@@ -16,11 +22,13 @@ choice config BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 depends on BR2_avr32 bool "gcc 4.2.2-avr32-2.1.5" + select BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD config BR2_GCC_VERSION_4_3_X depends on !BR2_arc && !BR2_avr32 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_cortex_a5 && !BR2_cortex_a7 && !BR2_cortex_a8 && !BR2_cortex_a9 && !BR2_cortex_a15 && !BR2_x86_atom && !BR2_powerpc_e300c2 && !BR2_powerpc_e300c3 && !BR2_powerpc_e500mc && !BR2_powerpc_464 && !BR2_powerpc_464fp && !BR2_powerpc_476 && !BR2_powerpc_476fp && !BR2_fa526 && !BR2_pj4 depends on !BR2_ARM_EABIHF bool "gcc 4.3.x" + select BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD config BR2_GCC_VERSION_4_4_X depends on !BR2_arc && !BR2_avr32 && !BR2_cortex_a5 && !BR2_cortex_a7 && !BR2_cortex_a15 && !BR2_x86_atom && !BR2_powerpc_476 && !BR2_powerpc_476fp && !BR2_fa526 && !BR2_pj4
@@ -29,6 +37,7 @@ choice depends on !BR2_ARM_EABIHF # VFPv4 support appeared in gcc 4.5 depends on !BR2_ARM_FPU_VFPV4 && !BR2_ARM_FPU_VFPV4D16 + select BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD config BR2_GCC_VERSION_4_5_X depends on !BR2_arc && !BR2_avr32 && !BR2_cortex_a7 && !BR2_cortex_a15 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_fa526 && !BR2_pj4
@@ -36,11 +45,13 @@ choice # ARM EABIhf support appeared in gcc 4.6 depends on !BR2_ARM_EABIHF bool "gcc 4.5.x" + select BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD config BR2_GCC_VERSION_4_6_X depends on !BR2_arc && !BR2_avr32 && !BR2_bfin && !BR2_cortex_a7 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_pj4 select BR2_GCC_NEEDS_MPC bool "gcc 4.6.x" + select BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD config BR2_GCC_VERSION_4_7_X depends on !BR2_arc && !BR2_avr32 && !BR2_bfin && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_pj4
package/gcc/gcc-initial/gcc-initial.mk +8 -0
@@ -29,6 +29,7 @@ HOST_GCC_INITIAL_CONF_OPT = \ --enable-languages=c \ --disable-shared \ --without-headers \ + --disable-threads \ --with-newlib \ --disable-largefile \ --disable-nls \
@@ -37,4 +38,11 @@ HOST_GCC_INITIAL_CONF_OPT = \ HOST_GCC_INITIAL_MAKE_OPT = all-gcc HOST_GCC_INITIAL_INSTALL_OPT = install-gcc +ifeq ($(BR2_TOOLCHAIN_NEEDS_THREE_STAGE_BUILD),) +ifeq ($(BR2_GCC_SUPPORTS_FINEGRAINEDMTUNE),y) +HOST_GCC_INITIAL_MAKE_OPT += all-target-libgcc +HOST_GCC_INITIAL_INSTALL_OPT += install-target-libgcc +endif +endif + $(eval $(host-autotools-package))
package/uclibc/Config.in +1 -0
@@ -115,6 +115,7 @@ choice config BR2_PTHREADS_NATIVE bool "Native POSIX Threading (NPTL)" select BR2_TOOLCHAIN_HAS_THREADS + select BR2_TOOLCHAIN_LIBC_NEEDS_THREE_STAGE_BUILD depends on !BR2_arc depends on !BR2_avr32 depends on !BR2_bfin
package/uclibc/uclibc.mk +3 -2
@@ -22,8 +22,9 @@ UCLIBC_INSTALL_STAGING = YES # cross-compiler and the kernel headers UCLIBC_DEPENDENCIES = host-gcc-initial linux-headers -# Before uClibc is built, we must have the second stage cross-compiler -uclibc-build: host-gcc-intermediate +# Before uClibc is built, we must have the second stage +# cross-compiler, for some gcc versions, and when NPTL is used. +uclibc-build: $(if $(BR2_TOOLCHAIN_NEEDS_THREE_STAGE_BUILD),host-gcc-intermediate) # specifying UCLIBC_CONFIG_FILE on the command-line overrides the .config # setting.
toolchain/Config.in +9 -0
@@ -1,5 +1,13 @@ menu "Toolchain" +config BR2_TOOLCHAIN_NEEDS_THREE_STAGE_BUILD + bool + default y if BR2_TOOLCHAIN_LIBC_NEEDS_THREE_STAGE_BUILD && \ + BR2_GCC_VERSION_NEEDS_THREE_STAGE_BUILD + +config BR2_TOOLCHAIN_LIBC_NEEDS_THREE_STAGE_BUILD + bool + # Should be selected for glibc or eglibc config BR2_TOOLCHAIN_USES_GLIBC bool
@@ -10,6 +18,7 @@ config BR2_TOOLCHAIN_USES_GLIBC select BR2_TOOLCHAIN_HAS_THREADS select BR2_TOOLCHAIN_HAS_THREADS_DEBUG select BR2_TOOLCHAIN_HAS_SHADOW_PASSWORDS + select BR2_TOOLCHAIN_LIBC_NEEDS_THREE_STAGE_BUILD config BR2_TOOLCHAIN_USES_UCLIBC bool
Signature · 1 line
1.8.1.2
PATCH 4/12 gcc: fix Fortran support
In reply to Thomas PetazzoniThis commit fixes the Fortran support in the gcc package by making sure that the runtime Fortran library is properly installed to the target.
It also takes this opportunity to slightly reword the Config.in option help text, to no longer mention gcc < 4.2 since they are no longer supported in Buildroot, and to have a more meaningful explanation.
Finally, it fixes the gcc-final.mk code to use the correct BR2_INSTALL_FORTRAN option rather than the non-existing BR2_GCC_CROSS_FORTRAN option, to decide whether Fortran support should be enabled or not in the cross-compiler.
--- package/gcc/Config.in.host | 8 ++++---- package/gcc/gcc-final/gcc-final.mk | 6 +++++- 2 files changed, 9 insertions(+), 5 deletions(-)
package/gcc/Config.in.host +4 -4
@@ -119,13 +119,13 @@ config BR2_INSTALL_OBJC Build/install Objective-C compiler and runtime? config BR2_INSTALL_FORTRAN - bool "Build/install Fortran compiler and runtime?" + bool "Enable Fortran support" depends on !BR2_avr32 select BR2_PACKAGE_LIBMPFR help - Build/install Fortran compiler and runtime? - Note that it is highly recommended NOT to use gfortran - from gcc older than 4.2.0 + Enable the support for the Fortran language in the + cross-compiler, and install the Fortran runtime to the + target. config BR2_GCC_ENABLE_TLS bool "Enable compiler tls support" if BR2_TOOLCHAIN_BUILDROOT_UCLIBC
package/gcc/gcc-final/gcc-final.mk +5 -1
@@ -34,7 +34,7 @@ HOST_GCC_FINAL_PRE_CONFIGURE_HOOKS += HOST_GCC_FINAL_CONFIGURE_SYMLINK # Languages supported by the cross-compiler GCC_FINAL_CROSS_LANGUAGES-y = c GCC_FINAL_CROSS_LANGUAGES-$(BR2_INSTALL_LIBSTDCPP) += c++ -GCC_FINAL_CROSS_LANGUAGES-$(BR2_GCC_CROSS_FORTRAN) += fortran +GCC_FINAL_CROSS_LANGUAGES-$(BR2_INSTALL_FORTRAN) += fortran GCC_FINAL_CROSS_LANGUAGES-$(BR2_GCC_CROSS_OBJC) += objc GCC_FINAL_CROSS_LANGUAGES = $(subst $(space),$(comma),$(GCC_FINAL_CROSS_LANGUAGES-y))
@@ -127,6 +127,10 @@ ifeq ($(BR2_GCC_ENABLE_OPENMP),y) HOST_GCC_FINAL_USR_LIBS += libgomp endif +ifeq ($(BR2_INSTALL_FORTRAN),y) +HOST_GCC_FINAL_USR_LIBS += libgfortran +endif + ifneq ($(HOST_GCC_FINAL_USR_LIBS),) define HOST_GCC_FINAL_INSTALL_USR_LIBS mkdir -p $(TARGET_DIR)/usr/lib
Signature · 1 line
1.8.1.2
PATCH 5/12 gcc: fix Objective-C support
In reply to Thomas PetazzoniThis commit fixes the Objective-C support in the gcc package by making sure that the runtime Objective-C library is properly installed to the target.
It also takes this opportunity to slightly reword the Config.in option help text to have a more meaningful explanation.
Finally, it fixes the gcc-final.mk code to use the correct BR2_INSTALL_OBJC option rather than the non-existing BR2_GCC_CROSS_OBJC option, to decide whether Objective-C support should be enabled or not in the cross-compiler.
--- package/gcc/Config.in.host | 6 ++++-- package/gcc/gcc-final/gcc-final.mk | 6 +++++- 2 files changed, 9 insertions(+), 3 deletions(-)
package/gcc/Config.in.host +4 -2
@@ -113,10 +113,12 @@ config BR2_EXTRA_GCC_CONFIG_OPTIONS initial, gcc intermediate and gcc final passes. config BR2_INSTALL_OBJC - bool "Build/install Objective-C compiler and runtime?" + bool "Enable Objective-C support" depends on !BR2_avr32 help - Build/install Objective-C compiler and runtime? + Enable the support for the Objective-C language in the + cross-compiler, and install the Objective-C runtime to the + target. config BR2_INSTALL_FORTRAN bool "Enable Fortran support"
package/gcc/gcc-final/gcc-final.mk +5 -1
@@ -35,7 +35,7 @@ HOST_GCC_FINAL_PRE_CONFIGURE_HOOKS += HOST_GCC_FINAL_CONFIGURE_SYMLINK GCC_FINAL_CROSS_LANGUAGES-y = c GCC_FINAL_CROSS_LANGUAGES-$(BR2_INSTALL_LIBSTDCPP) += c++ GCC_FINAL_CROSS_LANGUAGES-$(BR2_INSTALL_FORTRAN) += fortran -GCC_FINAL_CROSS_LANGUAGES-$(BR2_GCC_CROSS_OBJC) += objc +GCC_FINAL_CROSS_LANGUAGES-$(BR2_INSTALL_OBJC) += objc GCC_FINAL_CROSS_LANGUAGES = $(subst $(space),$(comma),$(GCC_FINAL_CROSS_LANGUAGES-y)) HOST_GCC_FINAL_CONF_OPT = \
@@ -131,6 +131,10 @@ ifeq ($(BR2_INSTALL_FORTRAN),y) HOST_GCC_FINAL_USR_LIBS += libgfortran endif +ifeq ($(BR2_INSTALL_OBJC),y) +HOST_GCC_FINAL_USR_LIBS += libobjc +endif + ifneq ($(HOST_GCC_FINAL_USR_LIBS),) define HOST_GCC_FINAL_INSTALL_USR_LIBS mkdir -p $(TARGET_DIR)/usr/lib
Signature · 1 line
1.8.1.2
PATCH 6/12 gcc: move C++ support option next to Fortran/Objective-C
In reply to Thomas PetazzoniThe option to enable C++ support was still located in toolchain/toolchain-buildroot/Config.in.2, with misc other toolchain options. It seems more logical to have this option with the other options to select the languages supported by the cross-compiler, so we move it next to the Fortran/Objective-C options in package/gcc/Config.in.host.
--- package/gcc/Config.in.host | 13 +++++++++++++ toolchain/toolchain-buildroot/Config.in.2 | 13 ------------- 2 files changed, 13 insertions(+), 13 deletions(-)
package/gcc/Config.in.host +13 -0
@@ -112,6 +112,19 @@ config BR2_EXTRA_GCC_CONFIG_OPTIONS include. Those options are applied for all of the gcc initial, gcc intermediate and gcc final passes. +config BR2_TOOLCHAIN_BUILDROOT_CXX + bool "Enable C++ support" + select BR2_INSTALL_LIBSTDCPP + depends on !(BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 && \ + BR2_TOOLCHAIN_BUILDROOT_LOCALE) + help + Enable this option if you want your toolchain to support the + C++ language and you want C++ libraries to be installed on + your target system. + +comment "C++ support broken in uClibc with locale enabled with gcc 4.2" + depends on BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 && BR2_TOOLCHAIN_BUILDROOT_LOCALE + config BR2_INSTALL_OBJC bool "Enable Objective-C support" depends on !BR2_avr32
toolchain/toolchain-buildroot/Config.in.2 +0 -13
@@ -4,19 +4,6 @@ if BR2_TOOLCHAIN_BUILDROOT comment "Toolchain Options" -config BR2_TOOLCHAIN_BUILDROOT_CXX - bool "Enable C++ support" - select BR2_INSTALL_LIBSTDCPP - depends on !(BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 && \ - BR2_TOOLCHAIN_BUILDROOT_LOCALE) - help - Enable this option if you want your toolchain to support the - C++ language and you want C++ libraries to be installed on - your target system. - -comment "C++ support broken in uClibc with locale enabled with gcc 4.2" - depends on BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 && BR2_TOOLCHAIN_BUILDROOT_LOCALE - config BR2_TOOLCHAIN_BUILDROOT_USE_SSP bool "Enable stack protection support" help
Signature · 1 line
1.8.1.2
PATCH 7/12 toolchain: add support for glibc
In reply to Thomas PetazzoniThe support for eglibc 2.17 was added to the internal toolchain backend for 2013.08. This commit now adds glibc 2.18 support to the internal toolchain backend.
Since the building procedure is very similar to the one of eglibc, we have renamed the 'eglibc' package to 'glibc', and made it capable of handling either glibc or eglibc.
---
package/gcc/Config.in.host | 2 +-
package/gcc/gcc-final/gcc-final.mk | 4 +-
package/{eglibc/eglibc.mk => glibc/glibc.mk} | 62 +++++++++++++++++-----------
toolchain/toolchain-buildroot/Config.in | 24 ++++++++++-
4 files changed, 64 insertions(+), 28 deletions(-)
rename package/{eglibc/eglibc.mk => glibc/glibc.mk} (59%)package/gcc/Config.in.host +1 -1
@@ -145,7 +145,7 @@ config BR2_INSTALL_FORTRAN config BR2_GCC_ENABLE_TLS bool "Enable compiler tls support" if BR2_TOOLCHAIN_BUILDROOT_UCLIBC default y - depends on BR2_PTHREADS_NATIVE || BR2_TOOLCHAIN_BUILDROOT_EGLIBC + depends on BR2_PTHREADS_NATIVE || BR2_TOOLCHAIN_BUILDROOT_EGLIBC || BR2_TOOLCHAIN_BUILDROOT_GLIBC help Enable the compiler to generate code for accessing thread local storage variables
package/gcc/gcc-final/gcc-final.mk +2 -2
@@ -91,11 +91,11 @@ endef HOST_GCC_FINAL_POST_INSTALL_HOOKS += HOST_GCC_FINAL_CREATE_SIMPLE_SYMLINKS -# In gcc 4.7.x, the ARM EABIhf library loader path for eglibc was not +# In gcc 4.7.x, the ARM EABIhf library loader path for (e)glibc was not # correct, so we create a symbolic link to make things work # properly. eglibc installs the library loader as ld-linux-armhf.so.3, # but gcc creates binaries that reference ld-linux.so.3. -ifeq ($(BR2_arm)$(BR2_ARM_EABIHF)$(BR2_GCC_VERSION_4_7_X)$(BR2_TOOLCHAIN_BUILDROOT_EGLIBC),yyyy) +ifeq ($(BR2_arm)$(BR2_ARM_EABIHF)$(BR2_GCC_VERSION_4_7_X)$(BR2_TOOLCHAIN_USES_GLIBC),yyyy) define HOST_GCC_FINAL_LD_LINUX_LINK ln -sf ld-linux-armhf.so.3 $(TARGET_DIR)/lib/ld-linux.so.3 ln -sf ld-linux-armhf.so.3 $(STAGING_DIR)/lib/ld-linux.so.3
package/glibc/glibc.mk renamed from package/eglibc/eglibc.mk +38 -24
103 lines not on this page · raw message
toolchain/toolchain-buildroot/Config.in +23 -1
@@ -34,12 +34,34 @@ config BR2_TOOLCHAIN_BUILDROOT_EGLIBC http://eglibc.org +config BR2_TOOLCHAIN_BUILDROOT_GLIBC + bool "glibc (experimental)" + depends on BR2_arm || BR2_armeb || BR2_aarch64 || \ + BR2_i386 || BR2_mips || BR2_mipsel || \ + BR2_mips64 || BR2_mips64el || BR2_powerpc || \ + BR2_sh || BR2_sh64 || BR2_sparc || \ + BR2_x86_64 + select BR2_TOOLCHAIN_USES_GLIBC + # our glibc.mk enables RPC support + select BR2_TOOLCHAIN_HAS_NATIVE_RPC + help + This option selects glibc as the C library for the + cross-compilation toolchain. + + The glibc support in the Buildroot toolchain backend is + very recent, and therefore marked experimental. + + http://www.gnu.org/software/libc/ + endchoice config BR2_TOOLCHAIN_BUILDROOT_LIBC string default "uclibc" if BR2_TOOLCHAIN_BUILDROOT_UCLIBC - default "eglibc" if BR2_TOOLCHAIN_BUILDROOT_EGLIBC + # Both glibc and eglibc are handled by the package called + # 'glibc' + default "glibc" if BR2_TOOLCHAIN_BUILDROOT_EGLIBC + default "glibc" if BR2_TOOLCHAIN_BUILDROOT_GLIBC source "package/uclibc/Config.in"
Signature · 1 line
1.8.1.2
PATCH 8/12 gcc/gcc-final: use the common HOST_GCC_CONFIGURE_SYMLINK macro
In reply to Thomas PetazzoniEach stage of the gcc build needs to make a small dance before the configuration step to create a build sub-directory and a symbolic link to the configure script. The common gcc.mk had a HOST_GCC_CONFIGURE_SYMLINK already used by the gcc-initial and gcc-intermediate steps, but the gcc-final step wasn't using it.
This commit fixes this inconsistency, and therefore removes the HOST_GCC_FINAL_CONFIGURE_SYMLINK macro.
--- package/gcc/gcc-final/gcc-final.mk | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-)
package/gcc/gcc-final/gcc-final.mk +1 -6
@@ -24,12 +24,7 @@ HOST_GCC_FINAL_POST_PATCH_HOOKS += HOST_GCC_APPLY_PATCHES # subdirectory in the gcc sources, and build from there. HOST_GCC_FINAL_SUBDIR = build -define HOST_GCC_FINAL_CONFIGURE_SYMLINK - mkdir -p $(@D)/build - ln -s ../configure $(@D)/build/configure -endef - -HOST_GCC_FINAL_PRE_CONFIGURE_HOOKS += HOST_GCC_FINAL_CONFIGURE_SYMLINK +HOST_GCC_FINAL_PRE_CONFIGURE_HOOKS += HOST_GCC_CONFIGURE_SYMLINK # Languages supported by the cross-compiler GCC_FINAL_CROSS_LANGUAGES-y = c
Signature · 1 line
1.8.1.2
PATCH 9/12 gcc: force symlink creation
In reply to Thomas PetazzoniThe HOST_GCC_CONFIGURE_SYMLINK creates a symbolic link $(@D)/build/configure -> $(@D)/configure for each build step of gcc. However, it was only using 'ln -s' and not 'ln -sf', which was causing problems when doing 'make host-gcc-final-reconfigure' for example, because the configure commands (including this macro) are being re-executed, but the symbolic link already exists.
Changing this symbolic link creation to 'ln -sf' fixes this problem.
--- package/gcc/gcc.mk | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
package/gcc/gcc.mk +1 -1
@@ -76,7 +76,7 @@ endef define HOST_GCC_CONFIGURE_SYMLINK mkdir -p $(@D)/build - ln -s ../configure $(@D)/build/configure + ln -sf ../configure $(@D)/build/configure endef #
Signature · 1 line
1.8.1.2
PATCH 10/12 toolchain: refactor Stack Smashing Protection support
In reply to Thomas PetazzoniThis commit refactors how Stack Smashing Protection support is handled in Buildroot:
*) It turns the BR2_TOOLCHAIN_BUILDROOT_USE_SSP option into an option
that only enables the SSP support in uClibc, when using the internal toolchain backend.
*) It adds an hidden BR2_TOOLCHAIN_HAS_SSP option that gets enabled
when the toolchain has SSP support. Here we have the usual dance:
glibc/eglibc in internal/external backend always select this option, in the case of uClibc/internal, it gets selected when BR2_TOOLCHAIN_BUILDROOT_USE_SSP is enabled, in the case of uClibc/external, there is a new configuration option that the user must select (or not) depending on whether the toolchain has SSP support.
*) It adds a new options BR2_ENABLE_SSP in the "Build options" menu,
to enable the usage of SSP support, by adding -fstack-protector-all to the CFLAGS.
--- Config.in | 17 +++++++++++++++++ package/Makefile.in | 2 +- package/psmisc/psmisc.mk | 2 +- package/uclibc/Config.in | 10 ++++++++++ toolchain/Config.in | 1 + toolchain/helpers.mk | 3 ++- toolchain/toolchain-buildroot/Config.in.2 | 9 --------- toolchain/toolchain-common.in | 3 +++ toolchain/toolchain-external/Config.in | 9 +++++++++ 9 files changed, 44 insertions(+), 12 deletions(-)
Config.in +17 -0
@@ -398,6 +398,23 @@ config BR2_OPTIMIZE_S endchoice +config BR2_ENABLE_SSP + bool "build code with Stack Smashing Protection" + depends on BR2_TOOLCHAIN_HAS_SSP + help + Enable stack smashing protection support using GCCs + -fstack-protector-all option. + + See http://www.linuxfromscratch.org/hints/downloads/files/ssp.txt + for details. + + Note that this requires the toolchain to have SSP + support. This is always the case for glibc and eglibc + toolchain, but is optional in uClibc toolchains. + +comment "enabling Stack Smashing Protection requires support in the toolchain" + depends on !BR2_TOOLCHAIN_HAS_SSP + config BR2_PREFER_STATIC_LIB bool "prefer static libraries" help
package/Makefile.in +1 -1
@@ -129,7 +129,7 @@ TARGET_CFLAGS += -msep-data TARGET_CXXFLAGS += -msep-data endif -ifeq ($(BR2_TOOLCHAIN_BUILDROOT_USE_SSP),y) +ifeq ($(BR2_ENABLE_SSP),y) TARGET_CFLAGS += -fstack-protector-all TARGET_CXXFLAGS += -fstack-protector-all endif
package/psmisc/psmisc.mk +1 -1
@@ -10,7 +10,7 @@ PSMISC_LICENSE = GPLv2 PSMISC_LICENSE_FILES = COPYING PSMISC_DEPENDENCIES = ncurses $(if $(BR2_NEEDS_GETTEXT_IF_LOCALE),gettext) -ifneq ($(BR2_TOOLCHAIN_BUILDROOT_USE_SSP),y) +ifneq ($(BR2_ENABLE_SSP),y) # Don't force -fstack-protector PSMISC_CONF_OPT = --disable-harden-flags endif
package/uclibc/Config.in +10 -0
@@ -130,6 +130,16 @@ config BR2_PTHREAD_DEBUG help Build the thread library with debugging enabled. +config BR2_TOOLCHAIN_BUILDROOT_USE_SSP + bool "Enable stack protection support" + select BR2_TOOLCHAIN_HAS_SSP + help + Enable stack smashing protection support using GCCs + -fstack-protector-all option in uClibc. + + See http://www.linuxfromscratch.org/hints/downloads/files/ssp.txt + for details. + config BR2_UCLIBC_INSTALL_UTILS bool "Compile and install uClibc utilities" depends on !BR2_bfin
toolchain/Config.in +1 -0
@@ -18,6 +18,7 @@ config BR2_TOOLCHAIN_USES_GLIBC select BR2_TOOLCHAIN_HAS_THREADS select BR2_TOOLCHAIN_HAS_THREADS_DEBUG select BR2_TOOLCHAIN_HAS_SHADOW_PASSWORDS + select BR2_TOOLCHAIN_HAS_SSP select BR2_TOOLCHAIN_LIBC_NEEDS_THREE_STAGE_BUILD config BR2_TOOLCHAIN_USES_UCLIBC
toolchain/helpers.mk +2 -1
@@ -269,7 +269,8 @@ check_uclibc = \ $(call check_uclibc_feature,__UCLIBC_HAS_LOCALE__,BR2_ENABLE_LOCALE,$${UCLIBC_CONFIG_FILE},Locale support) ;\ $(call check_uclibc_feature,__UCLIBC_HAS_WCHAR__,BR2_USE_WCHAR,$${UCLIBC_CONFIG_FILE},Wide char support) ;\ $(call check_uclibc_feature,__UCLIBC_HAS_THREADS__,BR2_TOOLCHAIN_HAS_THREADS,$${UCLIBC_CONFIG_FILE},Thread support) ;\ - $(call check_uclibc_feature,__PTHREADS_DEBUG_SUPPORT__,BR2_TOOLCHAIN_HAS_THREADS_DEBUG,$${UCLIBC_CONFIG_FILE},Thread debugging support) + $(call check_uclibc_feature,__PTHREADS_DEBUG_SUPPORT__,BR2_TOOLCHAIN_HAS_THREADS_DEBUG,$${UCLIBC_CONFIG_FILE},Thread debugging support) ;\ + $(call check_uclibc_feature,__UCLIBC_HAS_SSP__,BR2_TOOLCHAIN_HAS_SSP,$${UCLIBC_CONFIG_FILE},Stack Smashing Protection support) # # Check that the Buildroot configuration of the ABI matches the
toolchain/toolchain-buildroot/Config.in.2 +0 -9
@@ -4,15 +4,6 @@ if BR2_TOOLCHAIN_BUILDROOT comment "Toolchain Options" -config BR2_TOOLCHAIN_BUILDROOT_USE_SSP - bool "Enable stack protection support" - help - Enable stack smashing protection support using GCCs - -fstack-protector-all option. - - See http://www.linuxfromscratch.org/hints/downloads/files/ssp.txt - for details. - source "package/elf2flt/Config.in.host" endif
toolchain/toolchain-common.in +3 -0
@@ -32,6 +32,9 @@ config BR2_TOOLCHAIN_HAS_THREADS_DEBUG config BR2_TOOLCHAIN_HAS_SHADOW_PASSWORDS bool +config BR2_TOOLCHAIN_HAS_SSP + bool + config BR2_ENABLE_LOCALE_PURGE bool "Purge unwanted locales" help
toolchain/toolchain-external/Config.in +9 -0
@@ -921,6 +921,15 @@ config BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS_DEBUG debugging support. If you don't know, leave the default value, Buildroot will tell you if it's correct or not. +config BR2_TOOLCHAIN_EXTERNAL_HAS_SSP + bool "Toolchain has SSP support?" + select BR2_TOOLCHAIN_HAS_SSP + help + Selection this option if your external toolchain has Stack + Smashing Protection support enabled. If you don't know, + leave the default value, Buildroot will tell you if it's + correct or not. + endif # BR2_TOOLCHAIN_EXTERNAL_CUSTOM_UCLIBC config BR2_TOOLCHAIN_EXTERNAL_INET_RPC
Signature · 1 line
1.8.1.2
PATCH 10/12 toolchain: refactor Stack Smashing Protection support
On 08/28/13 08:35, Thomas Petazzoni wrote:
[snip]
package/psmisc/psmisc.mk PSMISC_LICENSE = GPLv2 · line 16
Quoted 13 lines
package/psmisc/psmisc.mk +1
-1@@ -10,7 +10,7 @@ PSMISC_LICENSE = GPLv2 PSMISC_LICENSE_FILES = COPYING PSMISC_DEPENDENCIES = ncurses $(if $(BR2_NEEDS_GETTEXT_IF_LOCALE),gettext) -ifneq ($(BR2_TOOLCHAIN_BUILDROOT_USE_SSP),y) +ifneq ($(BR2_ENABLE_SSP),y) # Don't force -fstack-protector PSMISC_CONF_OPT = --disable-harden-flags endif
If psmisc normally uses -fstack-protector, then I wouldn't disable that
just because the user hasn't chosen to enable the option globally.
In other words, I'd put
ifneq ($(BR2_TOOLCHAIN_HAS_SSP),y)
Regards,
Arnout
Signature · 6 lines
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
PATCH 10/12 toolchain: refactor Stack Smashing Protection support
Dear Arnout Vandecappelle,
Quoted 19 lines · Arnout Vandecappelle
On 08/28/13 08:35, Thomas Petazzoni wrote:
[snip]package/psmisc/psmisc.mk PSMISC_LICENSE = GPLv2 · line 16
Quoted 13 lines
package/psmisc/psmisc.mk +1
-1@@ -10,7 +10,7 @@ PSMISC_LICENSE = GPLv2 PSMISC_LICENSE_FILES = COPYING PSMISC_DEPENDENCIES = ncurses $(if $(BR2_NEEDS_GETTEXT_IF_LOCALE),gettext) -ifneq ($(BR2_TOOLCHAIN_BUILDROOT_USE_SSP),y) +ifneq ($(BR2_ENABLE_SSP),y) # Don't force -fstack-protector PSMISC_CONF_OPT = --disable-harden-flags endifIf psmisc normally uses -fstack-protector, then I wouldn't disable that
just because the user hasn't chosen to enable the option globally.
Quoted 3 lines
In other words, I'd put
ifneq ($(BR2_TOOLCHAIN_HAS_SSP),y)
Sure, why not.
Thomas
Signature · 4 lines
Thomas Petazzoni, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
PATCH 11/12 toolchain: move elf2flt option inclusion to toolchain-buildroot/Config.in
In reply to Thomas PetazzoniThe only remaining thing in toolchain-buildroot/Config.in.2 is the inclusion of the elf2flt option. It doesn't really make sense to have a separate Config.in file for that, so let's move this to toolchain-buildroot/Config.in.
--- toolchain/Config.in | 1 - toolchain/toolchain-buildroot/Config.in | 1 + toolchain/toolchain-buildroot/Config.in.2 | 9 --------- 3 files changed, 1 insertion(+), 10 deletions(-) delete mode 100644 toolchain/toolchain-buildroot/Config.in.2
toolchain/Config.in +0 -1
@@ -62,6 +62,5 @@ source "toolchain/toolchain-buildroot/Config.in" source "toolchain/toolchain-external/Config.in" source "toolchain/toolchain-crosstool-ng/Config.in" source "toolchain/toolchain-common.in" -source "toolchain/toolchain-buildroot/Config.in.2" endmenu
toolchain/toolchain-buildroot/Config.in +1 -0
@@ -67,4 +67,5 @@ source "package/uclibc/Config.in" source "package/binutils/Config.in.host" source "package/gcc/Config.in.host" +source "package/elf2flt/Config.in.host" endif
toolchain/toolchain-buildroot/Config.in.2 deleted +0 -9
@@ -1,9 +0,0 @@ -# Buildroot backend specific options - -if BR2_TOOLCHAIN_BUILDROOT - -comment "Toolchain Options" - -source "package/elf2flt/Config.in.host" - -endif
Signature · 1 line
1.8.1.2
PATCH 12/12 toolchain: improve mudflap support
In reply to Thomas PetazzoniThe mudflap library is only useful if one uses the -fmudflap gcc option, to do more checks on pointers/arrays, which is typically not the case.
This commit:
* Adds an option to enable/disable mudflap support at the gcc level. By default, it is disabled, which saves a little bit of build time.
* Adds a way for the external toolchain backend to tell whether mudflap is supported or not by the external toolchain.
* Adds a global BR2_TOOLCHAIN_HAS_MUDFLAP hidden option, that indicates whether the toolchain (internal or external) supports mudflap.
* Adds a global BR2_ENABLE_MUDFLAP in "Build options" that allows the user to build all packages with mudflap support. It depends on BR2_TOOLCHAIN_HAS_MUDFLAP.
WARNING WARNING: this currently doesn't result into a working system, even with just Busybox. Busybox init crashes with "/sbin/init: symbol lookup error: /lib/libmudflapth.so.0: undefined symbol:
main". According to
http://gcc.gnu.org/ml/gcc-help/2008-03/msg00165.html it is caused by -Wl,--gc-sections (which is used by Busybox), but even after removing it, it still doesn't work. I don't personaly have much interest in mudflap support, so as I couldn't get it to work easily, my suggestion would be to drop support for it entirely, unless someone steps up to fix this issue.
--- Config.in | 20 ++++++++++++++++++++ package/Makefile.in | 12 ++++++++++++ package/gcc/Config.in.host | 13 +++++++++++++ package/gcc/gcc-final/gcc-final.mk | 8 ++++++++ package/gcc/gcc.mk | 6 ++++++ toolchain/helpers.mk | 12 ++++++++++++ toolchain/toolchain-common.in | 3 +++ toolchain/toolchain-external/Config.in | 9 +++++++++ toolchain/toolchain-external/ext-tool.mk | 11 +++++++++++ 9 files changed, 94 insertions(+)
Config.in +20 -0
@@ -415,6 +415,26 @@ config BR2_ENABLE_SSP comment "enabling Stack Smashing Protection requires support in the toolchain" depends on !BR2_TOOLCHAIN_HAS_SSP +config BR2_ENABLE_MUDFLAP + bool "build code with Mudflap pointer debugging" + depends on BR2_TOOLCHAIN_HAS_MUDFLAP + help + Instrument all risky pointer/array dereferencing + operations, some standard library string/heap functions, + and some other associated constructs with range/validity + tests. Modules so instrumented should be immune to buffer + overflows, invalid heap use, and some other classes of + C/C++ programming errors. + + See http://gcc.gnu.org/wiki/Mudflap_Pointer_Debugging for + more details. + + Note that this requires the toolchain to have Mudflap + support. + +comment "enabling Mudflap pointer debugging requires support in the toolchain" + depends on !BR2_TOOLCHAIN_HAS_MUDFLAP + config BR2_PREFER_STATIC_LIB bool "prefer static libraries" help
package/Makefile.in +12 -0
@@ -134,6 +134,18 @@ TARGET_CFLAGS += -fstack-protector-all TARGET_CXXFLAGS += -fstack-protector-all endif +ifeq ($(BR2_ENABLE_MUDFLAP),y) +ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y) +TARGET_CFLAGS += -fmudflapth +TARGET_CXXFLAGS += fmudflapth +TARGET_LDFLAGS += -fmudflapth -ldl -lpthread -lmudflapth +else +TARGET_CFLAGS += -fmudflap +TARGET_CXXFLAGS += -fmudflap +TARGET_LDFLAGS += -fmudflap -ldl -lmudflap +endif +endif + ifeq ($(BR2_TOOLCHAIN_BUILDROOT)$(BR2_TOOLCHAIN_CTNG),y) TARGET_CROSS=$(HOST_DIR)/usr/bin/$(GNU_TARGET_NAME)- else
package/gcc/Config.in.host +13 -0
@@ -155,3 +155,16 @@ config BR2_GCC_ENABLE_OPENMP depends on !BR2_PTHREADS_NONE && !BR2_avr32 && !BR2_arc help Enable OpenMP support for the compiler + +config BR2_GCC_ENABLE_LIBMUDFLAP + bool "Enable libmudflap support" + select BR2_TOOLCHAIN_HAS_MUDFLAP + help + libmudflap is a gcc library used for the mudflap pointer + debugging functionality. It is only needed if you intend to + use the -fmudflap gcc flag. + + See http://gcc.gnu.org/wiki/Mudflap_Pointer_Debugging and + the help of the gcc -fmudflap option for more details. + + If you're unsure, leave this option disabled.
package/gcc/gcc-final/gcc-final.mk +8 -0
@@ -130,6 +130,14 @@ ifeq ($(BR2_INSTALL_OBJC),y) HOST_GCC_FINAL_USR_LIBS += libobjc endif +ifeq ($(BR2_GCC_ENABLE_LIBMUDFLAP),y) +ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y) +HOST_GCC_FINAL_USR_LIBS += libmudflapth +else +HOST_GCC_FINAL_USR_LIBS += libmudflap +endif +endif + ifneq ($(HOST_GCC_FINAL_USR_LIBS),) define HOST_GCC_FINAL_INSTALL_USR_LIBS mkdir -p $(TARGET_DIR)/usr/lib
package/gcc/gcc.mk +6 -0
@@ -119,6 +119,12 @@ else HOST_GCC_COMMON_CONF_OPT += --disable-tls endif +ifeq ($(BR2_GCC_ENABLE_LIBMUDFLAP),y) +HOST_GCC_COMMON_CONF_OPT += --enable-libmudflap +else +HOST_GCC_COMMON_CONF_OPT += --disable-libmudflap +endif + ifeq ($(BR2_PTHREADS_NONE),y) HOST_GCC_COMMON_CONF_OPT += \ --disable-threads \
toolchain/helpers.mk +12 -0
@@ -315,6 +315,18 @@ check_cplusplus = \ fi # +# Check that the external toolchain has mudflap support +# +# $1: cross-gcc path +# +check_mudflap = \ + __CROSS_CC=$(strip $1) ; \ + if ! test -f `$${__CROSS_CC} -print-file-name=libmudflap.a` ; then \ + echo "Mudflap support is selected but is not available in external toolchain" ; \ + exit 1 ; \ + fi + +# # Check that the cross-compiler given in the configuration exists # # $1: cross-gcc path
toolchain/toolchain-common.in +3 -0
@@ -35,6 +35,9 @@ config BR2_TOOLCHAIN_HAS_SHADOW_PASSWORDS config BR2_TOOLCHAIN_HAS_SSP bool +config BR2_TOOLCHAIN_HAS_MUDFLAP + bool + config BR2_ENABLE_LOCALE_PURGE bool "Purge unwanted locales" help
toolchain/toolchain-external/Config.in +9 -0
@@ -13,6 +13,7 @@ config BR2_TOOLCHAIN_EXTERNAL_LINARO_2013_06 select BR2_TOOLCHAIN_HAS_NATIVE_RPC select BR2_INSTALL_LIBSTDCPP select BR2_HOSTARCH_NEEDS_IA32_LIBS + select BR2_TOOLCHAIN_HAS_MUDFLAP help Linaro toolchain for the ARM architecture. It uses Linaro GCC 2013.06 (based on gcc 4.8), Linaro GDB 2013.06 (based on
@@ -949,6 +950,14 @@ config BR2_TOOLCHAIN_EXTERNAL_CXX support. If you don't know, leave the default value, Buildroot will tell you if it's correct or not. +config BR2_TOOLCHAIN_EXTERNAL_MUDFLAP + bool "Toolchain has mudflap support?" + select BR2_TOOLCHAIN_HAS_MUDFLAP + help + Select this option if your external toolchain has mudflap + support. If you don't know, leave the default value, + Buildroot will tell you if it's correct or not. + config BR2_TOOLCHAIN_EXTRA_EXTERNAL_LIBS string "Extra toolchain libraries to be copied to target" help
toolchain/toolchain-external/ext-tool.mk +11 -0
@@ -55,6 +55,14 @@ ifeq ($(BR2_TOOLCHAIN_EXTERNAL_GLIBC),y) LIB_EXTERNAL_LIBS+=libnss_files.so libnss_dns.so endif +ifeq ($(BR2_ENABLE_MUDFLAP),y) +ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y) +LIB_EXTERNAL_LIBS+=libmudflapth.so +else +LIB_EXTERNAL_LIBS+=libmudflap.so +endif +endif + ifeq ($(BR2_INSTALL_LIBSTDCPP),y) USR_LIB_EXTERNAL_LIBS+=libstdc++.so endif
@@ -376,6 +384,9 @@ $(STAMP_DIR)/ext-toolchain-checked: $(TOOLCHAIN_EXTERNAL_DEPENDENCIES) if test "$(BR2_INSTALL_LIBSTDCPP)" = "y" ; then \ $(call check_cplusplus,$(TOOLCHAIN_EXTERNAL_CXX)) ; \ fi ; \ + if test "$(BR2_TOOLCHAIN_EXTERNAL_MUDFLAP)" = "y" ; then \ + $(call check_mudflap,$(TOOLCHAIN_EXTERNAL_CC)) ; \ + fi ; \ if test "$(BR2_TOOLCHAIN_EXTERNAL_UCLIBC)" = "y" ; then \ $(call check_uclibc,$${SYSROOT_DIR}) ; \ else \
Signature · 1 line
1.8.1.2
PATCH 12/12 toolchain: improve mudflap support
Quoted 21 lines · Thomas Petazzoni
The mudflap library is only useful if one uses the -fmudflap gcc option, to do more checks on pointers/arrays, which is typically not the case.
This commit:
* Adds an option to enable/disable mudflap support at the gcc level. By default, it is disabled, which saves a little bit of build time.
* Adds a way for the external toolchain backend to tell whether mudflap is supported or not by the external toolchain.
* Adds a global
BR2_TOOLCHAIN_HAS_MUDFLAPhidden option, that indicates whether the toolchain (internal or external) supports mudflap.* Adds a global
BR2_ENABLE_MUDFLAPin "Build options" that allows the user to build all packages with mudflap support. It depends onBR2_TOOLCHAIN_HAS_MUDFLAP.
Quoted 10 lines
WARNING WARNING: this currently doesn't result into a working system, even with just Busybox. Busybox init crashes with "
/sbin/init: symbol lookup error:/lib/libmudflapth.so.0: undefined symbol:
main". According to
http://gcc.gnu.org/ml/gcc-help/2008-03/msg00165.html it is caused by -Wl,--gc-sections (which is used by Busybox), but even after removing it, it still doesn't work. I don't personaly have much interest in mudflap support, so as I couldn't get it to work easily, my suggestion would be to drop support for it entirely, unless someone steps up to fix this issue.
I would drop the option to enable -fmudflap globally, but keep the
option to build mudflap support in the internal toolchain. As you wrote in another mail, mudflap is a debugging tool and not really a hardening tool like SSP.
Regards,
Arnout
Signature · 6 lines
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
PATCH 12/12 toolchain: improve mudflap support
Dear Arnout Vandecappelle,
Quoted 11 lines · Arnout Vandecappelle
Quoted 10 lines
WARNING WARNING: this currently doesn't result into a working system, even with just Busybox. Busybox init crashes with "
/sbin/init: symbol lookup error:/lib/libmudflapth.so.0: undefined symbol:
main". According to
http://gcc.gnu.org/ml/gcc-help/2008-03/msg00165.html it is caused by -Wl,--gc-sections (which is used by Busybox), but even after removing it, it still doesn't work. I don't personaly have much interest in mudflap support, so as I couldn't get it to work easily, my suggestion would be to drop support for it entirely, unless someone steps up to fix this issue.
Quoted 4 lines
I would drop the option to enable -fmudflap globally, but keep the
option to build mudflap support in the internal toolchain. As you wrote in another mail, mudflap is a debugging tool and not really a hardening tool like SSP.
I also agree with this. Gustavo, are you fine with this?
Thanks,
Thomas
Signature · 4 lines
Thomas Petazzoni, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
PATCH 12/12 toolchain: improve mudflap support
Quoted 6 lines · Thomas Petazzoni
Quoted 4 lines
I would drop the option to enable -fmudflap globally, but keep the
option to build mudflap support in the internal toolchain. As you wrote in another mail, mudflap is a debugging tool and not really a hardening tool like SSP.I also agree with this. Gustavo, are you fine with this?
+1 we really don't want to make a Kconfig option out of each possible compiler flag and, well, busybox doesn't behave with it which many targets will use.
Regards.