Skip to contentSkip to conversation

Pull request for branch for-2011.02/toolchain-improvements

GIT PULL New Thomas Petazzoni 3 participants 35 messages Dec 5, 2010 – Dec 12, 2010 Kbuild

lore
Share & tools
lore
#0 lore raw

Hello,

Here is a set of patches that mainly improves the toolchain configuration and support, with two majors directions :

* Turn BR2_LARGEFILE, BR2_INET_RPC, BR2_INET_RPC and similar options into hidden options, and let the toolchain backends present them as they wish. For example for RPC, xthe Buildroot backend says "Do you want RPC ?" while the external toolchain backend doesn't show the option when glibc is used, or present an option "Does your toolchain has RPC ?" when uclibc is used.

So really those options should now be called BR2_TOOLCHAIN_HAS_LARGEFILE, BR2_TOOLCHAIN_HAS_INET_RPC, etc, but we kept the old names in this patch series because changing them impacts all packages. This will be done later.

A first commit handles largefile, rpc, ipv6, locale and program invocation options, and some later commits handle the C++ option and thread options.

* Extend external toolchain with "profiles", i.e well-known external toolchains. For those toolchains, Buildroot can automatically download and install them and will automatically select the appropriate configuration options (C library used, RPC, IPv6, locale, etc.). So an user can know tell Buildroot to use a CodeSourcery toolchain, and Buildroot will do all the job. Such toolchains are extracted in $(O)/external-toolchain, but my intent is to move it to $(O)/toolchain once Gustavoz's toolchain cleanup work is done.

This work keeps the ability of doing some more manual things with external toolchains:

- For an external toolchain profile, you can tell Buildroot *not* to download it, and in that case, you must provide Buildroot with the location of the external toolchain.

- You keep the ability to specify a custom profile, where you will have to tell Buildroot everything: location of the toolchain, toolchain prefix, C library used, etc.

In my opinion, those two directions really make the external toolchains easier and simpler to use. One topic left is the gdb options, but I decided not to change it, in order to not conflict too much with Gustavoz's work.

The rest of the changes are more minor modifications, related to these toolchain improvements.

Regards,

Thomas

The following changes since commit 7e9c3a387820154fd1355f23c2669072c0c3a5f7:
Peter Korsgaard (1):
docs/news.html: add 2010.11 announce link

are available in the git repository at:

git://git.busybox.net/~tpetazzoni/git/buildroot for-2011.02/toolchain-improvements

Thomas Petazzoni (15):
toolchain: Improve C library option selection
toolchain: remove ending semi-colon in helpers
toolchain: add support for external toolchain profiles and download
toolchain: remove toolchain-specific stripping
toolchain: rework C++ options
toolchain: move Stack Protection Support option
ltp-testsuite: bump version and use autotargets
toolchain: rework thread options
toolchain: check that the thread option selection is correct
gcc: disable libgomp when thread support is disabled
uclibc: add patch for uClibc snapshot to fix build issue
toolchain: BR2_NEEDS_GETTEXT must be y when uClibc is used
toolchain: expose thread options in the Crosstool-NG backend
toolchain: more verbose in crosstool-ng backend
package: pass TARGET_MAKE_ENV at install/clean/uninstall steps

 Makefile                                           |    3 +-
 package/Makefile.autotools.in                      |   10 +-
 package/Makefile.in                                |   12 +-
 package/dmalloc/dmalloc.mk                         |    6 +-
 package/ltp-testsuite/Config.in                    |   15 +-
 ...tp-testsuite-20101031-disable-controllers.patch |   25 ++
 .../ltp-testsuite-disable-ipv6-tests.patch         |  123 ----------
 .../ltp-testsuite-enable-openposix-for-nptl.patch  |   33 ---
 .../ltp-testsuite-generate-needs-bash.patch        |    8 -
 .../ltp-testsuite-sched-getaffinity.patch          |   11 -
 .../ltp-testsuite-uclibc-syscalls.patch            |   11 -
 package/ltp-testsuite/ltp-testsuite.mk             |   68 +------
 package/lzma/Config.in                             |    4 +-
 package/multimedia/ffmpeg/ffmpeg.mk                |    6 +-
 package/openvpn/openvpn.mk                         |    6 +-
 package/rpm/Config.in                              |    4 +-
 toolchain/Config.in                                |    5 +-
 toolchain/Makefile.in                              |    6 -
 toolchain/gcc/gcc-uclibc-4.x.mk                    |    8 +-
 toolchain/helpers.mk                               |   16 +-
 toolchain/toolchain-buildroot/Config.in.2          |  101 ++++++++
 toolchain/toolchain-common.in                      |  121 ++---------
 toolchain/toolchain-crosstool-ng/Config.in         |   95 ++++++++-
 .../toolchain-crosstool-ng/crosstool-ng.config     |    8 +-
 toolchain/toolchain-crosstool-ng/crosstool-ng.mk   |   25 ++-
 toolchain/toolchain-external/Config.in             |  240 +++++++++++++++++++-
 toolchain/toolchain-external/Config.in.2           |   17 --
 toolchain/toolchain-external/ext-tool.mk           |  212 ++++++++++++------
 ...tconf-only-use-specs-names-when-available.patch |  131 +++++++++++
 toolchain/uClibc/uclibc.mk                         |    2 +-
 30 files changed, 827 insertions(+), 505 deletions(-)
 create mode 100644 package/ltp-testsuite/ltp-testsuite-20101031-disable-controllers.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-disable-ipv6-tests.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-enable-openposix-for-nptl.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-generate-needs-bash.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-sched-getaffinity.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-uclibc-syscalls.patch
 delete mode 100644 toolchain/toolchain-external/Config.in.2
 create mode 100644 toolchain/uClibc/uClibc.snapshot.getconf-only-use-specs-names-when-available.patch

Thanks,

Signature · 1 line
Thomas Petazzoni
#1 lore raw

PATCH 1/15 toolchain: Improve C library option selection

Turn BR2_LARGEFILE, BR2_INET_IPV6, BR2_INET_RPC, BR2_USE_WCHAR, BR2_ENABLE_LOCALE and BR2_PROGRAM_INVOCATION into hidden options.

Then, for Buildroot toolchains, external toolchains and Crosstool-NG toolchains, provide visible options that selects the hidden options.

This allows :

* To show a different label and help text in the case of Buildroot toolchain (do you want to enable feature X ?) and in the case of external toolchain (is feature X available in your toolchain ?)

* To not show any option when a glibc external toolchain is selected (since glibc is assumed to support all of largefile, IPv6, RPC, WCHAR, locale and program invocation) and have them all selected in that case.

There is some amount of duplication between Buildroot toolchain config options and Crosstool-NG toolchain config options, because kconfig doesn't allow to source the same Config.in file twice (even if under mutually exclusive conditions). This duplication is more readable that the hack that consists in splitting files in multiple pieces.

---
 toolchain/toolchain-buildroot/Config.in.2  |   50 +++++++++++++++++++++++
 toolchain/toolchain-common.in              |   57 +++++----------------------
 toolchain/toolchain-crosstool-ng/Config.in |   53 +++++++++++++++++++++++++
 toolchain/toolchain-external/Config.in     |   59 ++++++++++++++++++++++++++++
 4 files changed, 172 insertions(+), 47 deletions(-)
toolchain/toolchain-buildroot/Config.in.2 +50 -0
index 2a17796..d052b84 100644
@@ -1,6 +1,56 @@
 # Buildroot backend specific options
 
 if BR2_TOOLCHAIN_BUILDROOT
+
+comment "Toolchain Options"
+
+config BR2_TOOLCHAIN_BUILDROOT_LARGEFILE
+	bool "Enable large file (files > 2 GB) support?"
+	select BR2_LARGEFILE
+	help
+	  Enable this option if you want your toolchain to support
+	  files bigger than 2 GB.
+
+config BR2_TOOLCHAIN_BUILDROOT_INET_IPV6
+	bool "Enable IPv6"
+	select BR2_INET_IPV6
+	help
+	  Enable this option if you want your toolchain to support
+	  IPv6.
+
+config BR2_TOOLCHAIN_BUILDROOT_INET_RPC
+	bool "Enable RPC"
+	select BR2_INET_RPC
+	help
+	  Enable this option if you want your toolchain to support
+	  RPC (needed for NFS, for example).
+
+config BR2_TOOLCHAIN_BUILDROOT_WCHAR
+	bool "Enable WCHAR support"
+	select BR2_USE_WCHAR
+	help
+	  Enable this option if you want your toolchain to support
+	  wide characters (i.e characters longer than 8 bits, needed
+	  for locale support).
+
+config BR2_TOOLCHAIN_BUILDROOT_LOCALE
+	bool "Enable toolchain locale/i18n support"
+	select BR2_TOOLCHAIN_BUILDROOT_WCHAR
+	select BR2_ENABLE_LOCALE
+	help
+	  Enable this option if you want your toolchain to support
+	  localization and internationalization.
+
+config BR2_TOOLCHAIN_BUILDROOT_PROGRAM_INVOCATION
+	bool "Enable 'program invocation name'"
+	select BR2_PROGRAM_INVOCATION
+	help
+	  Enable this option if you want your toolchain to support the
+	  GNU-specific program_invocation_name and
+	  program_invocation_short_name strings.  Some GNU packages
+	  (like tar and coreutils) utilize these for extra useful
+	  output, but in general are not required.
+
 source "toolchain/elf2flt/Config.in"
 source "toolchain/mklibs/Config.in"
 source "toolchain/sstrip/Config.in"
toolchain/toolchain-common.in +10 -47

82 lines not on this page · raw message

toolchain/toolchain-crosstool-ng/Config.in +53 -0

59 lines not on this page · raw message

toolchain/toolchain-external/Config.in +59 -0

68 lines not on this page · raw message

Signature · 1 line
1.7.0.4
#2 lore raw

PATCH 1/15 toolchain: Improve C library option selection

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> Turn BR2_LARGEFILE, BR2_INET_IPV6, BR2_INET_RPC, BR2_USE_WCHAR, Thomas> BR2_ENABLE_LOCALE and BR2_PROGRAM_INVOCATION into hidden options.

Thomas> Then, for Buildroot toolchains, external toolchains and Crosstool-NG Thomas> toolchains, provide visible options that selects the hidden options.

Thomas> This allows :

Thomas> * To show a different label and help text in the case of Buildroot
Thomas> toolchain (do you want to enable feature X ?) and in the case of
Thomas> external toolchain (is feature X available in your toolchain ?)

Thomas> * To not show any option when a glibc external toolchain is selected

 Thomas>    (since glibc is assumed to support all of largefile, IPv6, RPC,
 Thomas>    WCHAR, locale and program invocation) and have them all selected in
 Thomas>    that case.

Thomas> There is some amount of duplication between Buildroot toolchain config Thomas> options and Crosstool-NG toolchain config options, because kconfig Thomas> doesn't allow to source the same Config.in file twice (even if under Thomas> mutually exclusive conditions). This duplication is more readable that Thomas> the hack that consists in splitting files in multiple pieces.

This breaks the toolchain setup of people's existing .config's, but OK - Perhaps add a note about it in the commit message though.

Thomas> +++ b/toolchain/toolchain-buildroot/Config.in.2 Thomas> +config BR2_TOOLCHAIN_BUILDROOT_LOCALE
Thomas> + bool "Enable toolchain locale/i18n support"
Thomas> + select BR2_TOOLCHAIN_BUILDROOT_WCHAR
Thomas> + select BR2_ENABLE_LOCALE
Thomas> + help
Thomas> + Enable this option if you want your toolchain to support
Thomas> + localization and internationalization.
Thomas> +
Thomas> +config BR2_TOOLCHAIN_BUILDROOT_PROGRAM_INVOCATION
Thomas> + bool "Enable 'program invocation name'"

The other places we have 'Enable ... SUPPORT'

Thomas> +++ b/toolchain/toolchain-external/Config.in Thomas> @@ -10,9 +10,68 @@ config BR2_TOOLCHAIN_EXTERNAL_UCLIBC

Thomas> config BR2_TOOLCHAIN_EXTERNAL_GLIBC
Thomas> bool "glibc"
Thomas> + select BR2_LARGEFILE
Thomas> + select BR2_INET_IPV6
Thomas> + select BR2_INET_RPC
Thomas> + select BR2_USE_WCHAR
Thomas> + select BR2_ENABLE_LOCALE
Thomas> + select BR2_PROGRAM_INVOCATION

Thomas> endchoice

Thomas> +if BR2_TOOLCHAIN_EXTERNAL_UCLIBC
Thomas> +
Thomas> +config BR2_TOOLCHAIN_EXTERNAL_LARGEFILE
Thomas> + bool "Toolchain has large file support ?"

We normally don't do spaces before question mark (several places in this file).

Signature · 1 line
Bye, Peter Korsgaard
#3 lore raw

PATCH 1/15 toolchain: Improve C library option selection

Quoted 2 lines · On Sat, 11 Dec 2010 23:14:36 +0100 Peter Korsgaard <[email protected]> wrote:

This breaks the toolchain setup of people's existing .config's, but OK - Perhaps add a note about it in the commit message though.

Done.

Quoted 1 line

The other places we have 'Enable ... SUPPORT'

Ok, done. For many other places as well.

Quoted 2 lines

We normally don't do spaces before question mark (several places in this file).

Ok, done.

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#4 lore raw

PATCH 2/15 toolchain: remove ending semi-colon in helpers

In reply to Thomas Petazzoni

Some helpers had their final line ending with a semi-colon, some did not. For consistency, remove the final semi-colon from all helpers, it's the responsability of the caller to add the final semi-colon as needed.

---
 toolchain/helpers.mk |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)
toolchain/helpers.mk +4 -4
index 05d43e7..43c6db3 100644
@@ -201,7 +201,7 @@ check_uclibc = \
 	$(call check_uclibc_feature,__UCLIBC_HAS_RPC__,BR2_INET_RPC,$${UCLIBC_CONFIG_FILE},RPC support) ;\
 	$(call check_uclibc_feature,__UCLIBC_HAS_LOCALE__,BR2_ENABLE_LOCALE,$${UCLIBC_CONFIG_FILE},Locale support) ;\
 	$(call check_uclibc_feature,__UCLIBC_HAS_WCHAR__,BR2_USE_WCHAR,$${UCLIBC_CONFIG_FILE},Wide char support) ;\
-	$(call check_uclibc_feature,__UCLIBC_HAS_PROGRAM_INVOCATION_NAME__,BR2_PROGRAM_INVOCATION,$${UCLIBC_CONFIG_FILE},Program invocation support) ;\
+	$(call check_uclibc_feature,__UCLIBC_HAS_PROGRAM_INVOCATION_NAME__,BR2_PROGRAM_INVOCATION,$${UCLIBC_CONFIG_FILE},Program invocation support)
 
 #
 # Check that the Buildroot configuration of the ABI matches the
@@ -221,7 +221,7 @@ check_arm_abi = \
 	if [ x$(BR2_ARM_EABI) = x"y" -a $${EXT_TOOLCHAIN_ABI} = "oabi" ] ; then \
 		echo "Incorrect ABI setting" ; \
 		exit 1 ; \
-	fi ; \
+	fi
 
 #
 # Check that the external toolchain supports C++
@@ -231,7 +231,7 @@ check_cplusplus = \
 	if test $$? -ne 0 ; then \
 		echo "BR2_INSTALL_LIBSTDCPP is selected but C++ support not available in external toolchain" ; \
 		exit 1 ; \
-	fi ; \
+	fi
 
 #
 # Check that the cross-compiler given in the configuration exists
@@ -241,4 +241,4 @@ check_cross_compiler_exists = \
 	if test $$? -ne 0 ; then \
 		echo "Cannot execute cross-compiler '$(TARGET_CC)'" ; \
 		exit 1 ; \
-	fi ; \
+	fi
Signature · 1 line
1.7.0.4
#5 lore raw

PATCH 3/15 toolchain: add support for external toolchain profiles and download

In reply to Thomas Petazzoni

Instead of letting the user define all the details of his external toolchain, we define a set of profiles for well-known external toolchains (CodeSourcery ones only at the moment, can easily be extended with other toolchains).

Once a profile has been choosen, the user is offered the choice of either letting Buildroot download and install the external toolchain, or (as before) to tell Buildroot where the toolchain is installed on the system.

We of course provide a "custom profile", through which the user can configure Buildroot to use a custom external toolchain for which no profile is available.

---
 Makefile                                 |    3 +-
 package/Makefile.in                      |   10 +-
 toolchain/Config.in                      |    5 +-
 toolchain/toolchain-external/Config.in   |  169 ++++++++++++++++++++++--
 toolchain/toolchain-external/Config.in.2 |   17 ---
 toolchain/toolchain-external/ext-tool.mk |  210 +++++++++++++++++++++---------
 6 files changed, 316 insertions(+), 98 deletions(-)
 delete mode 100644 toolchain/toolchain-external/Config.in.2
Makefile +2 -1
index 596470f..c62c269 100644
@@ -272,6 +272,7 @@ STAMP_DIR:=$(BASE_DIR)/stamps
 BINARIES_DIR:=$(BASE_DIR)/images
 TARGET_DIR:=$(BASE_DIR)/target
 TOOLCHAIN_DIR=$(BASE_DIR)/toolchain
+TOOLCHAIN_EXTERNAL_DIR=$(BASE_DIR)/external-toolchain
 TARGET_SKELETON=$(TOPDIR)/fs/skeleton
 
 BR2_DEPENDS_DIR=$(BUILD_DIR)/buildroot-config
@@ -586,7 +587,7 @@ endif
 
 clean:
 	rm -rf $(STAGING_DIR) $(TARGET_DIR) $(BINARIES_DIR) $(HOST_DIR) \
-		$(STAMP_DIR) $(BUILD_DIR) $(TOOLCHAIN_DIR)
+		$(STAMP_DIR) $(BUILD_DIR) $(TOOLCHAIN_DIR) $(TOOLCHAIN_EXTERNAL_DIR)
 
 distclean: clean
 ifeq ($(DL_DIR),$(TOPDIR)/dl)
package/Makefile.in +5 -5
index 2ca8c31..d8b85fa 100644
@@ -114,13 +114,13 @@ REAL_GNU_TARGET_NAME=$(ARCH)-unknown-linux-$(LIBC)$(ABI)
 ifeq ($(BR2_TOOLCHAIN_BUILDROOT),y)
 TARGET_CROSS=$(STAGING_DIR)/usr/bin/$(REAL_GNU_TARGET_NAME)-
 else ifeq ($(BR2_TOOLCHAIN_EXTERNAL),y)
-TOOLCHAIN_EXTERNAL_PREFIX:=$(call qstrip,$(BR2_TOOLCHAIN_EXTERNAL_PREFIX))
-TOOLCHAIN_EXTERNAL_PATH:=$(call qstrip,$(BR2_TOOLCHAIN_EXTERNAL_PATH))
-ifneq ($(TOOLCHAIN_EXTERNAL_PATH),)
-TARGET_CROSS=$(TOOLCHAIN_EXTERNAL_PATH)/bin/$(TOOLCHAIN_EXTERNAL_PREFIX)-
+TOOLCHAIN_EXTERNAL_PREFIX=$(call qstrip,$(BR2_TOOLCHAIN_EXTERNAL_PREFIX))
+ifeq ($(BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD),y)
+TOOLCHAIN_EXTERNAL_LOCATION=$(TOOLCHAIN_EXTERNAL_DIR)
 else
-TARGET_CROSS=$(TOOLCHAIN_EXTERNAL_PREFIX)-
+TOOLCHAIN_EXTERNAL_LOCATION=$(call qstrip,$(BR2_TOOLCHAIN_EXTERNAL_PATH))
 endif
+TARGET_CROSS=$(TOOLCHAIN_EXTERNAL_LOCATION)/bin/$(TOOLCHAIN_EXTERNAL_PREFIX)-
 else ifeq ($(BR2_TOOLCHAIN_CTNG),y)
 TARGET_CROSS=$(TOOLCHAIN_DIR)/bin/$(REAL_GNU_TARGET_NAME)-
 endif
toolchain/Config.in +4 -1
index d476cbf..8baa0c1 100644
@@ -13,6 +13,10 @@ config BR2_TOOLCHAIN_BUILDROOT
 
 config BR2_TOOLCHAIN_EXTERNAL
 	bool "External toolchain"
+	help
+	  Select if you want to use an existing cross-compiling
+	  toolchain. Buildroot can either download automatically a
+	  toolchain, or use an already installed toolchain.
 
 config BR2_TOOLCHAIN_CTNG
 	bool "Crosstool-NG toolchain"
@@ -29,6 +33,5 @@ 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"
-source "toolchain/toolchain-external/Config.in.2"
 
 endmenu
toolchain/toolchain-external/Config.in +159 -10

195 lines not on this page · raw message

toolchain/toolchain-external/Config.in.2 deleted +0 -17
deleted file mode 100644
index 489558c..0000000
@@ -1,17 +0,0 @@
-#
-
-if BR2_TOOLCHAIN_EXTERNAL
-config BR2_TOOLCHAIN_EXTERNAL_PATH
-	string "External toolchain path"
-	default "/path/to/staging_dir/usr"
-	help
-	  Path to where the external toolchain is installed.
-
-config BR2_TOOLCHAIN_EXTERNAL_PREFIX
-	string "External toolchain prefix"
-	default "$(ARCH)-linux"
-	help
-	  This the the external toolchain prefix. For example:
-	    armeb-unknown-linux-gnu, mipsel-unknown-linux-gnu, etc.
-
-endif
toolchain/toolchain-external/ext-tool.mk +146 -64

232 lines not on this page · raw message

Signature · 1 line
1.7.0.4
#6 lore raw

PATCH 3/15 toolchain: add support for external toolchain profiles and download

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Hi,

Thomas> Instead of letting the user define all the details of his external Thomas> toolchain, we define a set of profiles for well-known external Thomas> toolchains (CodeSourcery ones only at the moment, can easily be Thomas> extended with other toolchains).

Thomas> Once a profile has been choosen, the user is offered the choice of Thomas> either letting Buildroot download and install the external toolchain, Thomas> or (as before) to tell Buildroot where the toolchain is installed on Thomas> the system.

Thomas> We of course provide a "custom profile", through which the user can Thomas> configure Buildroot to use a custom external toolchain for which no Thomas> profile is available.

Thomas> +++ b/toolchain/toolchain-external/Config.in

 Thomas>  config BR2_TOOLCHAIN_EXTERNAL_STRIP
 Thomas>  	bool
 Thomas>  	default y

Thomas> + depends on BR2_TOOLCHAIN_EXTERNAL

This is all in an if BR2_TOOLCHAIN_EXTERNAL conditional. Did you mean BR2_TOOLCHAIN_EXTERNAL_CUSTOM instead?

 Thomas>  	prompt "Strip shared libraries"
 Thomas>  	help
 Thomas>  	  Strip shared libraries copied from the external toolchain.

Thomas> -endif
Thomas> +
Thomas> +endif # BR2_TOOLCHAIN_EXTERNAL
Thomas> +

Signature · 1 line
Bye, Peter Korsgaard
#7 lore raw

PATCH 3/15 toolchain: add support for external toolchain profiles and download

Quoted 7 lines · On Sat, 11 Dec 2010 23:26:38 +0100 Peter Korsgaard <[email protected]> wrote:
 Thomas>  config BR2_TOOLCHAIN_EXTERNAL_STRIP
 Thomas>  	bool
 Thomas>  	default y

Thomas> + depends on BR2_TOOLCHAIN_EXTERNAL

This is all in an if BR2_TOOLCHAIN_EXTERNAL conditional. Did you mean BR2_TOOLCHAIN_EXTERNAL_CUSTOM instead?

This option is removed by the next patch anyway, but I fixed this patch (just removed this added line).

Regards,

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#8 lore raw

PATCH 4/15 toolchain: remove toolchain-specific stripping

In reply to Thomas Petazzoni

We already handle the stripping of libraries in $(TARGET_DIR) at the global level, so there's no need to have toolchain-specific option and code for this.

---
 toolchain/helpers.mk                             |    9 ---------
 toolchain/toolchain-crosstool-ng/Config.in       |    6 ------
 toolchain/toolchain-crosstool-ng/crosstool-ng.mk |    4 ++--
 toolchain/toolchain-external/Config.in           |    8 --------
 toolchain/toolchain-external/ext-tool.mk         |    4 ++--
 5 files changed, 4 insertions(+), 27 deletions(-)
toolchain/helpers.mk +0 -9
index 43c6db3..24f379d 100644
@@ -15,13 +15,11 @@
 # $1: arch specific sysroot directory
 # $2: library name
 # $3: destination directory of the libary, relative to $(TARGET_DIR)
-# $4: strip (y|n), default is to strip
 #
 copy_toolchain_lib_root = \
 	ARCH_SYSROOT_DIR="$(strip $1)"; \
 	LIB="$(strip $2)"; \
 	DESTDIR="$(strip $3)" ; \
-	STRIP="$(strip $4)"; \
  \
 	LIBS=`(cd $${ARCH_SYSROOT_DIR}; \
 		find -L . -path "./lib/$${LIB}.*"     -o \
@@ -39,13 +37,6 @@ copy_toolchain_lib_root = \
 				cp -d $${FULLPATH} $(TARGET_DIR)/$${DESTDIR}/; \
 			elif test -f $${FULLPATH}; then \
 				$(INSTALL) -D -m0755 $${FULLPATH} $(TARGET_DIR)/$${DESTDIR}/$${LIB}; \
-				case "$${STRIP}" in \
-				(0 | n | no) \
-;; \
-				(*) \
-					$(TARGET_CROSS)strip "$(TARGET_DIR)/$${DESTDIR}/$${LIB}"; \
-;; \
-				esac; \
 			else \
 				exit -1; \
 			fi; \
toolchain/toolchain-crosstool-ng/Config.in +0 -6
index db3506e..5ba1071 100644
@@ -98,10 +98,4 @@ config BR2_TOOLCHAIN_CTNG_uClibc_PROGRAM_INVOCATION
 
 endif # BR2_TOOLCHAIN_CTNG_uClibc
 
-config BR2_TOOLCHAIN_CTNG_STRIP_LIBS
-	bool "Strip libs copied to target"
-	default !BR2_STRIP_none
-	help
-	  Strip shared libraries copied from the toolchain.
-
 endif # BR2_TOOLCHAIN_CTNG
toolchain/toolchain-crosstool-ng/crosstool-ng.mk +2 -2
index 31797c4..1c5e864 100644
@@ -65,10 +65,10 @@ $(STAMP_DIR)/ct-ng-toolchain-installed: $(STAMP_DIR)/ct-ng-toolchain-built
 	    echo "CTNG_SYSROOT='$${CTNG_SYSROOT}'";                         \
 	    echo "Copy external toolchain libraries to target...";          \
 	    for libs in $(CTNG_LIBS_LIB); do                                \
-	        $(call copy_toolchain_lib_root,$${CTNG_SYSROOT},$$libs,/lib,$(BR2_TOOLCHAIN_CTNG_STRIP_LIBS)); \
+	        $(call copy_toolchain_lib_root,$${CTNG_SYSROOT},$$libs,/lib); \
 	    done;                                                           \
 	    for libs in $(CTNG_LIBS_USR_LIB); do                            \
-	        $(call copy_toolchain_lib_root,$${CTNG_SYSROOT},$$libs,/usr/lib,$(BR2_TOOLCHAIN_CTNG_STRIP_LIBS)); \
+	        $(call copy_toolchain_lib_root,$${CTNG_SYSROOT},$$libs,/usr/lib); \
 	    done;                                                           \
 	    echo "Copy external toolchain sysroot to staging...";           \
 	    $(call copy_toolchain_sysroot,$${CTNG_SYSROOT},$${CTNG_SYSROOT},)
toolchain/toolchain-external/Config.in +0 -8
index 60866e1..fe4fc17 100644
@@ -218,13 +218,5 @@ endif # BR2_TOOLCHAIN_EXTERNAL_CUSTOM_UCLIBC
 
 endif # BR2_TOOLCHAIN_EXTERNAL_CUSTOM
 
-config BR2_TOOLCHAIN_EXTERNAL_STRIP
-	bool
-	default y
-	depends on BR2_TOOLCHAIN_EXTERNAL
-	prompt "Strip shared libraries"
-	help
-	  Strip shared libraries copied from the external toolchain.
-
 endif # BR2_TOOLCHAIN_EXTERNAL
 
toolchain/toolchain-external/ext-tool.mk +2 -2
index 900c06d..6f6ec29 100644
@@ -184,10 +184,10 @@ $(STAMP_DIR)/ext-toolchain-installed: $(TOOLCHAIN_EXTERNAL_DEPENDENCIES)
 	mkdir -p $(TARGET_DIR)/lib ; \
 	echo "Copy external toolchain libraries to target..." ; \
 	for libs in $(LIB_EXTERNAL_LIBS); do \
-	 	$(call copy_toolchain_lib_root,$${ARCH_SYSROOT_DIR},$$libs,/lib,$(BR2_TOOLCHAIN_EXTERNAL_STRIP)); \
+	 	$(call copy_toolchain_lib_root,$${ARCH_SYSROOT_DIR},$$libs,/lib); \
 	done ; \
 	for libs in $(USR_LIB_EXTERNAL_LIBS); do \
-	 	$(call copy_toolchain_lib_root,$${ARCH_SYSROOT_DIR},$$libs,/usr/lib,$(BR2_TOOLCHAIN_EXTERNAL_STRIP)); \
+	 	$(call copy_toolchain_lib_root,$${ARCH_SYSROOT_DIR},$$libs,/usr/lib); \
 	done ; \
 	echo "Copy external toolchain sysroot to staging..." ; \
 	$(call copy_toolchain_sysroot,$${SYSROOT_DIR},$${ARCH_SYSROOT_DIR},$${ARCH_SUBDIR}) ; \
Signature · 1 line
1.7.0.4
#9 lore raw

PATCH 5/15 toolchain: rework C++ options

In reply to Thomas Petazzoni

Instead of having BR2_GCC_CROSS_CXX and BR2_INSTALL_LIBSTDCPP, with BR2_GCC_CROSS_CXX not being visible (and therefore being useless), let's just keep BR2_INSTALL_LIBSTDCPP to enable C++ in the toolchain and install C++ libraries on the target.

We also take that opportunity to make BR2_INSTALL_LIBSTDCPP an hidden option, which is selected by an option in Buildroot toolchain support or an option in External toolchain support, just as we did for other toolchain features.

Some work definitely remains to be done :

- The name BR2_INSTALL_LIBSTDCPP is ugly, but we keep it for the moment in order to avoid changing all packages.

- We could have two independent options: one to tell whether the toolchain has C++ support (on which packages would depend on) and another to tell whether the C++ library should be installed (that packages would select). This would allow to have a C++-free target filesystem even if the toolchain used actually supports C++.

- We should clarify the other language-related options (Fortran, Java, Objective-C, etc.).

---
 package/Makefile.in                        |    2 +-
 package/lzma/Config.in                     |    4 ++--
 package/rpm/Config.in                      |    4 ++--
 toolchain/gcc/gcc-uclibc-4.x.mk            |    2 +-
 toolchain/toolchain-buildroot/Config.in.2  |   16 ++++++++++++++++
 toolchain/toolchain-common.in              |   25 +++----------------------
 toolchain/toolchain-crosstool-ng/Config.in |    8 ++++++++
 toolchain/toolchain-external/Config.in     |    9 +++++++++
 8 files changed, 42 insertions(+), 28 deletions(-)
package/Makefile.in +1 -1
index d8b85fa..b56b706 100644
@@ -299,7 +299,7 @@ ifneq ($(BR2_INET_IPV6),y)
 DISABLE_IPV6= --disable-ipv6
 endif
 
-ifneq ($(BR2_GCC_CROSS_CXX),y)
+ifneq ($(BR2_INSTALL_LIBSTDCPP),y)
 TARGET_CONFIGURE_OPTS+=CXX=false
 endif
 
package/lzma/Config.in +2 -2
index 4904721..2b94dde 100644
@@ -1,6 +1,6 @@
 config BR2_PACKAGE_LZMA
 	bool "lzma"
-	depends on BR2_GCC_CROSS_CXX
+	depends on BR2_INSTALL_LIBSTDCPP
 	depends on BR2_DEPRECATED
 	help
 	  Lempel Ziv compression method (LZMA) is a compression
@@ -9,4 +9,4 @@ config BR2_PACKAGE_LZMA
 	  http://tukaani.org/lzma/
 
 comment "lzma requires a toolchain with C++ support"
-        depends on !BR2_GCC_CROSS_CXX && BR2_DEPRECATED
+        depends on !BR2_INSTALL_LIBSTDCPP && BR2_DEPRECATED
package/rpm/Config.in +2 -2
index 448606c..8d4c6a3 100644
@@ -25,11 +25,11 @@ config BR2_PACKAGE_RPM_BZIP2_PAYLOADS
 config BR2_PACKAGE_RPM_XZ_PAYLOADS
 	bool "support for xz payloads"
 	depends on BR2_PACKAGE_RPM
-	depends on BR2_GCC_CROSS_CXX
+	depends on BR2_INSTALL_LIBSTDCPP
 	help
 	  Support for xz payloads in RPM.
 
 comment "xz payload support requires a toolchain with c++ support"
-	depends on !BR2_GCC_CROSS_CXX
+	depends on !BR2_INSTALL_LIBSTDCPP
 
 
toolchain/gcc/gcc-uclibc-4.x.mk +1 -1
index af085f7..eec6523 100644
@@ -67,7 +67,7 @@ GCC_STAGING_PREREQ+=$(STAGING_DIR)/usr/lib/libc.a
 GCC_TARGET_LANGUAGES:=c
 
 GCC_CROSS_LANGUAGES:=c
-ifeq ($(BR2_GCC_CROSS_CXX),y)
+ifeq ($(BR2_INSTALL_LIBSTDCPP),y)
 GCC_CROSS_LANGUAGES:=$(GCC_CROSS_LANGUAGES),c++
 endif
 ifeq ($(BR2_GCC_CROSS_FORTRAN),y)
toolchain/toolchain-buildroot/Config.in.2 +16 -0
index d052b84..dd459fc 100644
@@ -51,6 +51,22 @@ config BR2_TOOLCHAIN_BUILDROOT_PROGRAM_INVOCATION
 	  (like tar and coreutils) utilize these for extra useful
 	  output, but in general are not required.
 
+config BR2_TOOLCHAIN_BUILDROOT_C_PLUS_PLUS
+	bool "Enable C++ support"
+	select BR2_INSTALL_LIBSTDCPP
+	depends on !(! BR2_GCC_SUPPORTS_FINEGRAINEDMTUNE && \
+		       BR2_TOOLCHAIN_BUILDROOT_LOCALE    && \
+		       BR2_UCLIBC_VERSION_0_9_31)
+	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 0.9.31 with locale enabled with gcc 4.2"
+	depends on !BR2_GCC_SUPPORTS_FINEGRAINEDMTUNE && \
+		    BR2_TOOLCHAIN_BUILDROOT_LOCALE    && \
+		    BR2_UCLIBC_VERSION_0_9_31
+
 source "toolchain/elf2flt/Config.in"
 source "toolchain/mklibs/Config.in"
 source "toolchain/sstrip/Config.in"
toolchain/toolchain-common.in +3 -22
index ea47466..2c54acd 100644
@@ -23,6 +23,9 @@ config BR2_ENABLE_LOCALE
 config BR2_PROGRAM_INVOCATION
 	bool
 
+config BR2_INSTALL_LIBSTDCPP
+	bool
+
 config BR2_ENABLE_LOCALE_PURGE
 	bool "Purge unwanted locales"
 	help
@@ -108,28 +111,6 @@ choice
 		depends on BR2_UCLIBC_VERSION_SNAPSHOT
 endchoice
 
-config BR2_GCC_CROSS_CXX
-	bool
-	help
-	  If you are building your own toolchain and want to build 
-	  a C++ cross-compiler this needs to be enabled.
-	  If you have an external binary toolchain that has a C++ compiler
-	  and you want to use it then you need to enable this option.
-
-config BR2_INSTALL_LIBSTDCPP
-	bool "Build/install c++ compiler and libstdc++?"
-	select BR2_GCC_CROSS_CXX
-	depends on !(! BR2_GCC_SUPPORTS_FINEGRAINEDMTUNE && BR2_ENABLE_LOCALE && BR2_UCLIBC_VERSION_0_9_31)
-	help
-	  If you are building your own toolchain and want to build and install
-	  the C++ compiler and library then you need to enable this option.
-	  If you have an external toolchain that has been built with C++ 
-	  support and you want to use the compiler / library then you need 
-	  to select this option.
-
-comment "C++ support broken in uClibc 0.9.31 with locale enabled with gcc 4.2"
-	depends on !BR2_GCC_SUPPORTS_FINEGRAINEDMTUNE && BR2_ENABLE_LOCALE && BR2_UCLIBC_VERSION_0_9_31
-
 config BR2_TARGET_OPTIMIZATION
 	string "Target Optimizations"
 	default "-pipe"
toolchain/toolchain-crosstool-ng/Config.in +8 -0
index 5ba1071..b345297 100644
@@ -98,4 +98,12 @@ config BR2_TOOLCHAIN_CTNG_uClibc_PROGRAM_INVOCATION
 
 endif # BR2_TOOLCHAIN_CTNG_uClibc
 
+config BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS
+	bool "Enable C++ support"
+	select BR2_INSTALL_LIBSTDCPP
+	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.
+
 endif # BR2_TOOLCHAIN_CTNG
toolchain/toolchain-external/Config.in +9 -0
index fe4fc17..1366f07 100644
@@ -143,6 +143,7 @@ config BR2_TOOLCHAIN_EXTERNAL_GLIBC
 	select BR2_USE_WCHAR
 	select BR2_ENABLE_LOCALE
 	select BR2_PROGRAM_INVOCATION
+	select BR2_INSTALL_LIBSTDCPP
 
 config BR2_TOOLCHAIN_EXTERNAL_UCLIBC
 	bool
@@ -214,6 +215,14 @@ config BR2_TOOLCHAIN_EXTERNAL_PROGRAM_INVOCATION
 	  invocation support. If you don't know, leave the default
 	  value, Buildroot will tell you if it's correct or not.
 
+config BR2_TOOLCHAIN_EXTERNAL_C_PLUS_PLUS
+	bool "Toolchain has C++ support ?"
+	select BR2_INSTALL_LIBSTDCPP
+	help
+	  Enable this option if you want your external toolchain has
+	  support for C++. 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
 
 endif # BR2_TOOLCHAIN_EXTERNAL_CUSTOM
Signature · 1 line
1.7.0.4
#10 lore raw

PATCH 5/15 toolchain: rework C++ options

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Hi,

Thomas> Instead of having BR2_GCC_CROSS_CXX and BR2_INSTALL_LIBSTDCPP, with Thomas> BR2_GCC_CROSS_CXX not being visible (and therefore being useless), Thomas> let's just keep BR2_INSTALL_LIBSTDCPP to enable C++ in the toolchain Thomas> and install C++ libraries on the target.

Thomas> We also take that opportunity to make BR2_INSTALL_LIBSTDCPP an hidden Thomas> option, which is selected by an option in Buildroot toolchain support Thomas> or an option in External toolchain support, just as we did for other Thomas> toolchain features.

Thomas> Some work definitely remains to be done :

Thomas> - The name BR2_INSTALL_LIBSTDCPP is ugly, but we keep it for the
Thomas> moment in order to avoid changing all packages.

Thomas> - We could have two independent options: one to tell whether the

 Thomas>    toolchain has C++ support (on which packages would depend on) and
 Thomas>    another to tell whether the C++ library should be installed (that
 Thomas>    packages would select). This would allow to have a C++-free target
 Thomas>    filesystem even if the toolchain used actually supports C++.

Is that really important? I would expect people who really care about such stuff to build a custom toolchain for their project.

If there's interest, perhaps it would make more sense to generalize this library-cleanup step for all libs and do something like the mklibs stuff.

Thomas> +++ b/toolchain/toolchain-crosstool-ng/Config.in
Thomas> @@ -98,4 +98,12 @@ config BR2_TOOLCHAIN_CTNG_uClibc_PROGRAM_INVOCATION

Thomas> endif # BR2_TOOLCHAIN_CTNG_uClibc

Thomas> +config BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS
Thomas> + bool "Enable C++ support"
Thomas> + select BR2_INSTALL_LIBSTDCPP
Thomas> + help

Thomas> +	  Enable this option if you want your toolchain to support the
Thomas> +	  C++ language and you want C++ libraries to be installed on
Thomas> +	  your target system.

Thomas> +

Here we add an option to build a ctng toolchain (glibc/uclibc) with/without C++ support.

Thomas> +++ b/toolchain/toolchain-external/Config.in Thomas> @@ -143,6 +143,7 @@ config BR2_TOOLCHAIN_EXTERNAL_GLIBC

 Thomas>  	select BR2_USE_WCHAR
 Thomas>  	select BR2_ENABLE_LOCALE
 Thomas>  	select BR2_PROGRAM_INVOCATION

Thomas> + select BR2_INSTALL_LIBSTDCPP

.. But here we say that all external glibc toolchains have C++ support. That doesn't seem to be always true.

Thomas> config BR2_TOOLCHAIN_EXTERNAL_UCLIBC
Thomas> bool
Thomas> @@ -214,6 +215,14 @@ config BR2_TOOLCHAIN_EXTERNAL_PROGRAM_INVOCATION

 Thomas>  	  invocation support. If you don't know, leave the default
 Thomas>  	  value, Buildroot will tell you if it's correct or not.

Thomas> +config BR2_TOOLCHAIN_EXTERNAL_C_PLUS_PLUS
Thomas> + bool "Toolchain has C++ support ?"

No space before '?'

Thomas> + select BR2_INSTALL_LIBSTDCPP
Thomas> + help
Thomas> + Enable this option if you want your external toolchain has

'Enable this option if your external toolchain has'

Thomas> + support for C++. If you don't know, leave the default value,
Thomas> + Buildroot will tell you if it's correct or not.

Signature · 1 line
Bye, Peter Korsgaard
#11 lore raw

PATCH 5/15 toolchain: rework C++ options

Quoted 9 lines · On Sat, 11 Dec 2010 23:38:38 +0100 Peter Korsgaard <[email protected]> wrote:

Thomas> - We could have two independent options: one to tell Thomas> whether the toolchain has C++ support (on which packages Thomas> would depend on) and another to tell whether the C++ library Thomas> should be installed (that packages would select). This would Thomas> allow to have a C++-free target filesystem even if the Thomas> toolchain used actually supports C++.

Is that really important? I would expect people who really care about such stuff to build a custom toolchain for their project.

No strong opinion on this, so I just deleted this part from the commit log.

Quoted 3 lines

If there's interest, perhaps it would make more sense to generalize this library-cleanup step for all libs and do something like the mklibs stuff.

Yeah, would probably be a better idea.

Quoted 5 lines

Here we add an option to build a ctng toolchain (glibc/uclibc) with/without C++ support.

.. But here we say that all external glibc toolchains have C++ support. That doesn't seem to be always true.

Indeed. Generally, most glibc toolchains will probably have C++ support, but that's not necessarly the case. So I move the option outside the CUSTOM_UCLIBC thing, so that it is visible for any CUSTOM external toolchain.

Quoted 4 lines

Thomas> +config BR2_TOOLCHAIN_EXTERNAL_C_PLUS_PLUS
Thomas> + bool "Toolchain has C++ support ?"

No space before '?'

Fixed.

Quoted 1 line

'Enable this option if your external toolchain has'

Fixed.

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#12 lore raw

PATCH 5/15 toolchain: rework C++ options

In reply to Thomas Petazzoni

Thomas, All,

Quoted 4 lines · On Sunday 05 December 2010 10:25:32 Thomas Petazzoni wrote:

Instead of having BR2_GCC_CROSS_CXX and BR2_INSTALL_LIBSTDCPP, with BR2_GCC_CROSS_CXX not being visible (and therefore being useless), let's just keep BR2_INSTALL_LIBSTDCPP to enable C++ in the toolchain and install C++ libraries on the target.

[--SNIP--]

toolchain/toolchain-crosstool-ng/Config.in config BR2_TOOLCHAIN_CTNG_uClibc_PROGRAM_INVOCATION · line 101

Quoted 9 lines
toolchain/toolchain-crosstool-ng/Config.in +1 -0
index 5ba1071..b345297 100644
@@ -98,4 +98,12 @@ config BR2_TOOLCHAIN_CTNG_uClibc_PROGRAM_INVOCATION
 
 endif # BR2_TOOLCHAIN_CTNG_uClibc
 
+config BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS

C_PLUS_PLUS --> CXX

'CXX' is a common, and well understood, variable name for 'C++'. Besides, it is much shorter to write. :-)

toolchain/toolchain-crosstool-ng/Config.in config BR2_TOOLCHAIN_CTNG_uClibc_PROGRAM_INVOCATION · line 107

Quoted 6 lines
toolchain/toolchain-crosstool-ng/Config.in +6 -0
+	bool "Enable C++ support"
+	select BR2_INSTALL_LIBSTDCPP
+	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.

I think this option should be used in the .mk (yes, I know, I should have run a real git send-email). Could you add this hunk:

toolchain/toolchain-crosstool-ng/crosstool-ng.mk +1 -1
@@ -207,7 +207,7 @@ CTNG_FIX_DOT_CONFIG_SED += s:^(CT_CC_PKGVERSION)="(.*)":\1="crosstool-NG $(CTNG_
 ifneq ($(call qstrip,$(BR2_PACKAGE_GDB_SERVER))$(call qstrip,$(BR2_PACKAGE_GDB_HOST)),)
 CTNG_FIX_DOT_CONFIG_SED += s:^(CT_DEBUG_gdb)=.*:\# \1 is not set:;
 endif
-ifneq ($(call qstrip,$(BR2_INSTALL_LIBSTDCPP)),)
+ifeq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS)),y)
 CTNG_FIX_DOT_CONFIG_SED += s:^\# (CT_CC_LANG_CXX) is not set:\1=y:;
 else
 CTNG_FIX_DOT_CONFIG_SED += s:^(CT_CC_LANG_CXX)=.*:\# \1 is not set:;

Regards,
Yann E. MORIN.

Signature · 6 lines
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'
#13 lore raw

PATCH 5/15 toolchain: rework C++ options

Quoted 6 lines · On Sun, 12 Dec 2010 11:25:24 +0100 "Yann E. MORIN"…
Quoted 1 line

+config BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS

C_PLUS_PLUS --> CXX

'CXX' is a common, and well understood, variable name for 'C++'. Besides, it is much shorter to write. :-)

Done.

Quoted 2 lines

I think this option should be used in the .mk (yes, I know, I should have run a real git send-email). Could you add this hunk:

Done. As the change was small, I just squashed it into the existing 5/15 patch.

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
#14 lore raw

PATCH 6/15 toolchain: move Stack Protection Support option

In reply to Thomas Petazzoni

The BR2_USE_SSP option is only used inside the uClibc build, so only meaningful for Buildroot internal toolchains. Therefore, the option is moved to the right location so that it isn't visible when working with external toolchains.

---
 toolchain/toolchain-buildroot/Config.in.2 |    9 +++++++++
 toolchain/toolchain-common.in             |    9 ---------
 toolchain/uClibc/uclibc.mk                |    2 +-
 3 files changed, 10 insertions(+), 10 deletions(-)
toolchain/toolchain-buildroot/Config.in.2 +9 -0
index dd459fc..9cf54d4 100644
@@ -67,6 +67,15 @@ comment "C++ support broken in uClibc 0.9.31 with locale enabled with gcc 4.2"
 		    BR2_TOOLCHAIN_BUILDROOT_LOCALE    && \
 		    BR2_UCLIBC_VERSION_0_9_31
 
+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 "toolchain/elf2flt/Config.in"
 source "toolchain/mklibs/Config.in"
 source "toolchain/sstrip/Config.in"
toolchain/toolchain-common.in +0 -9
index 2c54acd..5502ab4 100644
@@ -76,15 +76,6 @@ config BR2_SOFT_FLOAT
 	  functions, then everything will need to be compiled with soft
 	  floating point support (-msoft-float).
 
-config BR2_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.
-
 choice
 	prompt "Thread library implementation"
 	default BR2_PTHREADS_OLD
toolchain/uClibc/uclibc.mk +1 -1
index 1ed40cc..e05e16c 100644
@@ -271,7 +271,7 @@ else
 		-e 's,.*UCLIBC_HAS_FPU.*,UCLIBC_HAS_FPU=y\nHAS_FPU=y\nUCLIBC_HAS_FLOATS=y\n,g' \
 		$(UCLIBC_DIR)/.oldconfig
 endif
-ifeq ($(BR2_USE_SSP),y)
+ifeq ($(BR2_TOOLCHAIN_BUILDROOT_USE_SSP),y)
 	$(SED) 's,^.*UCLIBC_HAS_SSP[^_].*,UCLIBC_HAS_SSP=y,g' $(UCLIBC_DIR)/.oldconfig
 else
 	$(SED) 's,^.*UCLIBC_HAS_SSP[^_].*,UCLIBC_HAS_SSP=n,g' $(UCLIBC_DIR)/.oldconfig
Signature · 1 line
1.7.0.4
#15 lore raw

PATCH 7/15 ltp-testsuite: bump version and use autotargets

In reply to Thomas Petazzoni

Reworking ltp-testsuite is needed in order to get rid of some thread-specific options that will be cleaned-up in the next commit.

---
 package/ltp-testsuite/Config.in                    |   15 ++-
 ...tp-testsuite-20101031-disable-controllers.patch |   25 ++++
 .../ltp-testsuite-disable-ipv6-tests.patch         |  123 --------------------
 .../ltp-testsuite-enable-openposix-for-nptl.patch  |   33 -----
 .../ltp-testsuite-generate-needs-bash.patch        |    8 --
 .../ltp-testsuite-sched-getaffinity.patch          |   11 --
 .../ltp-testsuite-uclibc-syscalls.patch            |   11 --
 package/ltp-testsuite/ltp-testsuite.mk             |   68 +----------
 8 files changed, 42 insertions(+), 252 deletions(-)
 create mode 100644 package/ltp-testsuite/ltp-testsuite-20101031-disable-controllers.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-disable-ipv6-tests.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-enable-openposix-for-nptl.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-generate-needs-bash.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-sched-getaffinity.patch
 delete mode 100644 package/ltp-testsuite/ltp-testsuite-uclibc-syscalls.patch
package/ltp-testsuite/Config.in +13 -2
index efe0db7..e059af3 100644
@@ -1,7 +1,18 @@
-config BR2_PACKAGE_LTP-TESTSUITE
+config BR2_PACKAGE_LTP_TESTSUITE
 	bool "ltp-testsuite"
-	depends on BROKEN
+	depends on !BR2_PTHREADS_NONE
+	# Several tests of LTP use legacy functions that are usually
+	# not available in uClibc.
+	depends on BR2_TOOLCHAIN_EXTERNAL_GLIBC || \
+		   BR2_TOOLCHAIN_CTNG_glibc	|| \
+		   BR2_TOOLCHAIN_CTNG_eglibc
 	help
 	  The Linux Test Project provides a huge testsuite for Linux.
 
 	  http://ltp.sourceforge.net/
+
+comment "ltp-testsuite requires a glibc toolchain with thread support"
+	depends on (!BR2_TOOLCHAIN_EXTERNAL_GLIBC && \
+		    !BR2_TOOLCHAIN_CTNG_glibc	  && \
+		    !BR2_TOOLCHAIN_CTNG_eglibc)   || \
+                    BR2_PTHREADS_NONE
package/ltp-testsuite/ltp-testsuite-20101031-disable-controllers.patch new file +25 -0
new file mode 100644
index 0000000..57d6038
@@ -0,0 +1,25 @@
+Disable controllers testcases
+
+The cpuset controllers testcases do not build due to bug
+https://sourceforge.net/tracker/?func=detail&aid=3126942&group_id=3382&atid=103382. Disabling
+just the cpuset controllers do not seem to be easily possible, and
+those controller features are rarely used on embedded systems anyway.
+
+Signed-off-by: Thomas Petazzoni <[email protected]>
+---
+ testcases/kernel/Makefile |    2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+Index: ltp-testsuite-20101031/testcases/kernel/Makefile
+===================================================================
+--- ltp-testsuite-20101031.orig/testcases/kernel/Makefile
++++ ltp-testsuite-20101031/testcases/kernel/Makefile
+@@ -28,7 +28,7 @@
+ #
+ # KEEP THIS LIST ALPHABETIZED PLEASE!
+ ifneq ($(UCLINUX),1)
+-SUBDIRS		:= containers connectors controllers fs hotplug io ipc mem \
++SUBDIRS		:= containers connectors fs hotplug io ipc mem \
+ 		   numa performance_counters power_management pty sched \
+ 		   security syscalls timers tracing
+ else
package/ltp-testsuite/ltp-testsuite-disable-ipv6-tests.patch deleted +0 -123

123 lines not on this page · raw message

package/ltp-testsuite/ltp-testsuite-enable-openposix-for-nptl.patch deleted +0 -33
deleted file mode 100644
index ebb888c..0000000
@@ -1,33 +0,0 @@
-diff -ur ltp-full-20060615/testcases/Makefile ltp-full-20060615-patched/testcases/Makefile
---- ltp-full-20060615/testcases/Makefile	2006-01-31 08:48:38.000000000 -0600
-+++ ltp-full-20060615-patched/testcases/Makefile	2006-07-04 22:20:40.298850250 -0500
-@@ -1,4 +1,4 @@
--SUBDIRS = `ls */Makefile | sed "s/Makefile//g" | grep -v open | grep -v pounder | grep -v DOTS`
-+SUBDIRS = `ls */Makefile | sed "s/Makefile//g" | grep -v open_hpi | grep -v pounder | grep -v DOTS`
- UCLINUX_SUBDIRS = kernel
- 
- all:
-diff -ur ltp-full-20060615/testcases/open_posix_testsuite/LDFLAGS ltp-full-20060615-patched/testcases/open_posix_testsuite/LDFLAGS
---- ltp-full-20060615/testcases/open_posix_testsuite/LDFLAGS	2005-06-03 11:29:49.000000000 -0500
-+++ ltp-full-20060615-patched/testcases/open_posix_testsuite/LDFLAGS	2006-07-04 22:21:34.322226500 -0500
-@@ -14,3 +14,5 @@
- # allow for the NPTL-specific compilation (used in some test cases)
- # Note: this sometimes require the package nptl-devel.*.rpm to be installed.
- #-I /usr/include/nptl -L /usr/lib/nptl -D_XOPEN_SOURCE=600 -lpthread -lrt -lm
-+
-+-D_XOPEN_SOURCE=600 -lpthread -lrt -lm -D_GNU_SOURCE
-diff -ur ltp-full-20060615/testcases/open_posix_testsuite/Makefile ltp-full-20060615-patched/testcases/open_posix_testsuite/Makefile
---- ltp-full-20060615/testcases/open_posix_testsuite/Makefile	2006-05-21 18:38:52.000000000 -0500
-+++ ltp-full-20060615-patched/testcases/open_posix_testsuite/Makefile	2006-07-04 22:22:16.660872500 -0500
-@@ -45,7 +45,10 @@
- TIMEOUT = $(top_builddir)/t0 $(TIMEOUT_VAL)
- 
- 
--all: build-tests run-tests 
-+all: build-tests $(top_builddir)/t0
-+
-+install:
-+	@echo " "
- 
- build-tests: $(BUILD_TESTS:.c=.test)
- run-tests: $(RUN_TESTS:.test=.run-test)
package/ltp-testsuite/ltp-testsuite-generate-needs-bash.patch deleted +0 -8
deleted file mode 100644
index 1f28956..0000000
@@ -1,8 +0,0 @@
---- ltp-full-20061121/testcases/network/generate.sh.orig	2006-12-07 22:21:56.000000000 -0700
-+++ ltp-full-20061121/testcases/network/generate.sh	2006-12-07 22:22:05.000000000 -0700
-@@ -1,4 +1,4 @@
--#!/bin/sh
-+#!/bin/bash
- #
- #   Copyright (c) International Business Machines  Corp., 2001
- #
package/ltp-testsuite/ltp-testsuite-sched-getaffinity.patch deleted +0 -11
deleted file mode 100644
index 9e4a9d8..0000000
@@ -1,11 +0,0 @@
---- ltp-full-20090630/testcases/kernel/syscalls/sched_getaffinity/Makefile	2009-06-15 19:38:04.000000000 +0100
-+++ ltp-full-20090630.mod/testcases/kernel/syscalls/sched_getaffinity/Makefile	2009-07-02 15:02:33.000000000 +0100
-@@ -16,7 +16,7 @@
- #  Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- #
- 
--CFLAGS += -I../../../../include -Wall
-+CFLAGS += -I../../../../include -Wall -D_GNU_SOURCE
- LDLIBS += -L../../../../lib -lltp
- 
- SRCS    = $(wildcard *.c)
package/ltp-testsuite/ltp-testsuite-uclibc-syscalls.patch deleted +0 -11
deleted file mode 100644
index 866a56c..0000000
@@ -1,11 +0,0 @@
---- ltp-full-20090630/testcases/kernel/syscalls/Makefile	2009-02-26 11:37:28.000000000 +0000
-+++ ltp-full-20090630.mod/testcases/kernel/syscalls/Makefile	2009-07-02 14:57:30.000000000 +0100
-@@ -25,7 +25,7 @@
- #
- # Commented this out since there are directories here we don't want built by default
- #
--SUBDIR = `ls */Makefile | sed "s/Makefile//g"` 
-+SUBDIR = `ls */Makefile | sed "s/Makefile//g" | grep -vE "^clock_nanosleep|quotactl|sched_getaffinity|sigreturn|unshare"` 
- UCLINUX_SUBDIR = `ls */Makefile | sed "s/Makefile//g" | grep -vE "^fork|epoll|capget|capset|chmod|chown|llseek|nftw|clone|profil|getcontext|remap_file_pages"`
- 
- all:
package/ltp-testsuite/ltp-testsuite.mk +4 -64

73 lines not on this page · raw message

Signature · 1 line
1.7.0.4
#16 lore raw

PATCH 7/15 ltp-testsuite: bump version and use autotargets

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> Reworking ltp-testsuite is needed in order to get rid of some Thomas> thread-specific options that will be cleaned-up in the next commit.

Thomas> +++ b/package/ltp-testsuite/Config.in
Thomas> @@ -1,7 +1,18 @@
Thomas> -config BR2_PACKAGE_LTP-TESTSUITE
Thomas> +config BR2_PACKAGE_LTP_TESTSUITE
Thomas> bool "ltp-testsuite"
Thomas> - depends on BROKEN
Thomas> + depends on !BR2_PTHREADS_NONE
Thomas> + # Several tests of LTP use legacy functions that are usually
Thomas> + # not available in uClibc.
Thomas> + depends on BR2_TOOLCHAIN_EXTERNAL_GLIBC || \
Thomas> + BR2_TOOLCHAIN_CTNG_glibc || \
Thomas> + BR2_TOOLCHAIN_CTNG_eglibc
Thomas> help
Thomas> The Linux Test Project provides a huge testsuite for Linux.

Thomas> http://ltp.sourceforge.net/
Thomas> +
Thomas> +comment "ltp-testsuite requires a glibc toolchain with thread support"
Thomas> + depends on (!BR2_TOOLCHAIN_EXTERNAL_GLIBC && \
Thomas> + !BR2_TOOLCHAIN_CTNG_glibc && \
Thomas> + !BR2_TOOLCHAIN_CTNG_eglibc) || \
Thomas> + BR2_PTHREADS_NONE

Does it really not work on uClibc? This is afaik the first package we have that is glibc only.

Signature · 1 line
Bye, Peter Korsgaard
#17 lore raw

PATCH 7/15 ltp-testsuite: bump version and use autotargets

Quoted 2 lines · On Sat, 11 Dec 2010 23:42:03 +0100 Peter Korsgaard <[email protected]> wrote:

Does it really not work on uClibc? This is afaik the first package we have that is glibc only.

ltp-testsuite uses sigset(), which is obsolete. uClibc can implement it, but not with our default configuration. I thought ltp-testsuite was not of sufficiently general usage to decide to make the change of our default uClibc configuration. The option is UCLIBC_HAS_OBSOLETE_BSD_SIGNAL.

BTW, I have the same problem with the mrouted package, which needs UCLIBC_SUPPORT_AI_ADDRCONFIG enabled in uClibc to compile. For this package, I just added a comment in the Config.in help text mentionning this. Should I do the same with ltp-testsuite ?

I'm not sure how to handle those types of packages, that can work with uClibc, but not with our default uClibc configuration, and who are probably not of enough general use to require a change of our default uClibc configuration.

Thoughts ?

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#18 lore raw

PATCH 7/15 ltp-testsuite: bump version and use autotargets

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> On Sat, 11 Dec 2010 23:42:03 +0100

Quoted 2 lines · Thomas> Peter Korsgaard <[email protected]> wrote:
Quoted 2 lines

Does it really not work on uClibc? This is afaik the first package we have that is glibc only.

Thomas> ltp-testsuite uses sigset(), which is obsolete. uClibc can Thomas> implement it, but not with our default configuration. I thought Thomas> ltp-testsuite was not of sufficiently general usage to decide Thomas> to make the change of our default uClibc configuration. The Thomas> option is UCLIBC_HAS_OBSOLETE_BSD_SIGNAL.

Ok.

Thomas> BTW, I have the same problem with the mrouted package, which Thomas> needs UCLIBC_SUPPORT_AI_ADDRCONFIG enabled in uClibc to Thomas> compile. For this package, I just added a comment in the Thomas> Config.in help text mentionning this. Should I do the same with Thomas> ltp-testsuite ?

Yeah, I would say this is good enough.

Thomas> I'm not sure how to handle those types of packages, that can Thomas> work with uClibc, but not with our default uClibc Thomas> configuration, and who are probably not of enough general use Thomas> to require a change of our default uClibc configuration.

We could ofcourse add toolchain knobs for all of these, but as you say they are probably not worth the effort / extra complexity. I think just leaving them like this is good enough.

Signature · 1 line
Bye, Peter Korsgaard
#19 lore raw

PATCH 7/15 ltp-testsuite: bump version and use autotargets

Quoted 8 lines · On Sun, 12 Dec 2010 23:21:24 +0100 Peter Korsgaard <[email protected]> wrote:

Thomas> ltp-testsuite uses sigset(), which is obsolete. uClibc can Thomas> implement it, but not with our default configuration. I Thomas> thought ltp-testsuite was not of sufficiently general usage Thomas> to decide to make the change of our default uClibc Thomas> configuration. The option is UCLIBC_HAS_OBSOLETE_BSD_SIGNAL.

Ok.

Quoted 7 lines

Thomas> BTW, I have the same problem with the mrouted package, which Thomas> needs UCLIBC_SUPPORT_AI_ADDRCONFIG enabled in uClibc to Thomas> compile. For this package, I just added a comment in the Thomas> Config.in help text mentionning this. Should I do the same Thomas> with ltp-testsuite ?

Yeah, I would say this is good enough.

So you mean that just like for mrouted, I should allow the selection of ltp-testsuite with uClibc, even though it would not build with our default uClibc configuration ?

Quoted 3 lines

We could ofcourse add toolchain knobs for all of these, but as you say they are probably not worth the effort / extra complexity. I think just leaving them like this is good enough.

Obviously, I don't think it's reasonable to add dozens of toolchain configuration knobs.

Regards,

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#20 lore raw

PATCH 7/15 ltp-testsuite: bump version and use autotargets

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Hi,

Thomas> So you mean that just like for mrouted, I should allow the Thomas> selection of ltp-testsuite with uClibc, even though it would Thomas> not build with our default uClibc configuration ?

Yes.

Quoted 3 lines
Quoted 3 lines

We could ofcourse add toolchain knobs for all of these, but as you say they are probably not worth the effort / extra complexity. I think just leaving them like this is good enough.

Thomas> Obviously, I don't think it's reasonable to add dozens of toolchain Thomas> configuration knobs.

Me neither.

Signature · 1 line
Bye, Peter Korsgaard
#21 lore raw

PATCH 8/15 toolchain: rework thread options

In reply to Thomas Petazzoni

The selection of linuxthreads, linuxthreads old or NPTL doesn't make a lot of sense for external toolchains. So, instead, we :

* Introduce an hidden BR2_TOOLCHAIN_HAS_THREADS option, which must be selected by toolchain specific options when thread support is available. Package needing to test thread support should use this option.

* Move the none/linuxthreads/linuxthreads old/NPTL selection to Buildroot internal toolchain configuration.

* Add an option in external toolchain to tell if thread support is available or not in the external toolchain.

ffmpeg, dmalloc and openvpn are fixed to use the new BR2_TOOLCHAIN_HAS_THREADS option. For openvpn, --enable-threads=posix is no longer used, as the configure script doesn't even understand this option.

---
 package/dmalloc/dmalloc.mk                |    6 +++---
 package/ltp-testsuite/Config.in           |    4 ++--
 package/multimedia/ffmpeg/ffmpeg.mk       |    6 +++---
 package/openvpn/openvpn.mk                |    6 +++---
 toolchain/Makefile.in                     |    6 ------
 toolchain/gcc/gcc-uclibc-4.x.mk           |    6 ++++++
 toolchain/toolchain-buildroot/Config.in.2 |   26 ++++++++++++++++++++++++++
 toolchain/toolchain-common.in             |   29 +++--------------------------
 toolchain/toolchain-external/Config.in    |    9 +++++++++
 toolchain/toolchain-external/ext-tool.mk  |    2 +-
 10 files changed, 56 insertions(+), 44 deletions(-)
package/dmalloc/dmalloc.mk +3 -3
index 60ad763..dca0488 100644
@@ -16,10 +16,10 @@ else
 DMALLOC_CONF_OPT+=--disable-cxx
 endif
 
-ifeq ($(BR2_PTHREADS_NONE),y)
-DMALLOC_CONF_OPT+=--disable-threads
-else
+ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y)
 DMALLOC_CONF_OPT+=--enable-threads
+else
+DMALLOC_CONF_OPT+=--disable-threads
 endif
 
 define DMALLOC_POST_PATCH
package/ltp-testsuite/Config.in +2 -2
index e059af3..5024cc3 100644
@@ -1,6 +1,6 @@
 config BR2_PACKAGE_LTP_TESTSUITE
 	bool "ltp-testsuite"
-	depends on !BR2_PTHREADS_NONE
+	depends on BR2_TOOLCHAIN_HAS_THREADS
 	# Several tests of LTP use legacy functions that are usually
 	# not available in uClibc.
 	depends on BR2_TOOLCHAIN_EXTERNAL_GLIBC || \
@@ -15,4 +15,4 @@ comment "ltp-testsuite requires a glibc toolchain with thread support"
 	depends on (!BR2_TOOLCHAIN_EXTERNAL_GLIBC && \
 		    !BR2_TOOLCHAIN_CTNG_glibc	  && \
 		    !BR2_TOOLCHAIN_CTNG_eglibc)   || \
-                    BR2_PTHREADS_NONE
+                    !BR2_TOOLCHAIN_HAS_THREADS
package/multimedia/ffmpeg/ffmpeg.mk +3 -3
index 959efa3..f0ec930 100644
@@ -111,10 +111,10 @@ else
 FFMPEG_CONF_OPT += --disable-outdevs
 endif
 
-ifeq ($(BR2_PTHREADS_NONE),y)
-FFMPEG_CONF_OPT += --disable-pthreads
-else
+ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y)
 FFMPEG_CONF_OPT += --enable-pthreads
+else
+FFMPEG_CONF_OPT += --disable-pthreads
 endif
 
 ifeq ($(BR2_PACKAGE_ZLIB),y)
package/openvpn/openvpn.mk +3 -3
index 7d8fb0c..ba91690 100644
@@ -8,10 +8,10 @@ OPENVPN_VERSION = 2.1.3
 OPENVPN_SITE = http://openvpn.net/release
 OPENVPN_CONF_OPT = --enable-small
 
-ifeq ($(BR2_PTHREADS_NATIVE),y)
-	OPENVPN_CONF_OPT += --enable-threads=posix
-else
+ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y)
 	OPENVPN_CONF_OPT += --enable-pthread
+else
+	OPENVPN_CONF_OPT += --disable-pthread
 endif
 
 ifeq ($(BR2_PACKAGE_OPENVPN_LZO),y)
toolchain/Makefile.in +0 -6
index 371509f..2cb6e95 100644
@@ -1,9 +1,3 @@
-ifeq ($(BR2_PTHREADS_NONE),y)
-THREADS:=--disable-threads
-else
-THREADS:=--enable-threads
-endif
-
 BR2_CONFIGURE_DEVEL_SYSROOT=--with-sysroot=$(TOOLCHAIN_DIR)/uClibc_dev/
 BR2_CONFIGURE_STAGING_SYSROOT=--with-sysroot=$(STAGING_DIR)
 BR2_CONFIGURE_BUILD_TOOLS=--with-build-time-tools=$(STAGING_DIR)/usr/$(REAL_GNU_TARGET_NAME)/bin
toolchain/gcc/gcc-uclibc-4.x.mk +6 -0
index eec6523..5f69906 100644
@@ -138,6 +138,12 @@ else
 GCC_TLS:=--disable-tls
 endif
 
+ifeq ($(BR2_PTHREADS_NONE),y)
+THREADS:=--disable-threads
+else
+THREADS:=--enable-threads
+endif
+
 ifeq ($(BR2_GCC_SUPPORTS_FINEGRAINEDMTUNE),y)
 GCC_DECIMAL_FLOAT:=--disable-decimal-float
 endif
toolchain/toolchain-buildroot/Config.in.2 +26 -0
index 9cf54d4..39f9b4d 100644
@@ -76,6 +76,32 @@ config BR2_TOOLCHAIN_BUILDROOT_USE_SSP
 	  See http://www.linuxfromscratch.org/hints/downloads/files/ssp.txt
 	  for details.
 
+choice
+	prompt "Thread library implementation"
+	default BR2_PTHREADS_OLD
+	help
+	  If you are building your own toolchain then select the type
+	  of libpthreads you want to use. Not all thread variants work
+	  with all versions of uClibc, the "linuxthreads (stable/old)"
+	  may be a working fallback if you need threading at all.
+
+	config BR2_PTHREADS_NONE
+		bool "none"
+
+	config BR2_PTHREADS
+		bool "linuxthreads"
+	        select BR2_TOOLCHAIN_HAS_THREADS
+
+	config BR2_PTHREADS_OLD
+		bool "linuxthreads (stable/old)"
+	        select BR2_TOOLCHAIN_HAS_THREADS
+
+	config BR2_PTHREADS_NATIVE
+		bool "Native POSIX Threading (NPTL)"
+	        select BR2_TOOLCHAIN_HAS_THREADS
+		depends on BR2_UCLIBC_VERSION_SNAPSHOT
+endchoice
+
 source "toolchain/elf2flt/Config.in"
 source "toolchain/mklibs/Config.in"
 source "toolchain/sstrip/Config.in"
toolchain/toolchain-common.in +3 -26
index 5502ab4..34d70a0 100644
@@ -26,6 +26,9 @@ config BR2_PROGRAM_INVOCATION
 config BR2_INSTALL_LIBSTDCPP
 	bool
 
+config BR2_TOOLCHAIN_HAS_THREADS
+	bool
+
 config BR2_ENABLE_LOCALE_PURGE
 	bool "Purge unwanted locales"
 	help
@@ -76,32 +79,6 @@ config BR2_SOFT_FLOAT
 	  functions, then everything will need to be compiled with soft
 	  floating point support (-msoft-float).
 
-choice
-	prompt "Thread library implementation"
-	default BR2_PTHREADS_OLD
-	help
-	  If you are building your own toolchain then select the type of 
-	  libpthreads you want to use.
-	  Not all thread variants work with all versions of uClibc,
-	  the "linuxthreads (stable/old)" may be a working fallback
-	  if you need threading at all.
-	  If you have an external binary toolchain then select the type 
-	  of libpthreads it was built with.
-
-	config BR2_PTHREADS_NONE
-		bool "none"
-
-	config BR2_PTHREADS
-		bool "linuxthreads"
-
-	config BR2_PTHREADS_OLD
-		bool "linuxthreads (stable/old)"
-
-	config BR2_PTHREADS_NATIVE
-		bool "Native POSIX Threading (NPTL)"
-		depends on BR2_UCLIBC_VERSION_SNAPSHOT
-endchoice
-
 config BR2_TARGET_OPTIMIZATION
 	string "Target Optimizations"
 	default "-pipe"
toolchain/toolchain-external/Config.in +9 -0
index 1366f07..ebfd3a0 100644
@@ -144,6 +144,7 @@ config BR2_TOOLCHAIN_EXTERNAL_GLIBC
 	select BR2_ENABLE_LOCALE
 	select BR2_PROGRAM_INVOCATION
 	select BR2_INSTALL_LIBSTDCPP
+	select BR2_TOOLCHAIN_HAS_THREADS
 
 config BR2_TOOLCHAIN_EXTERNAL_UCLIBC
 	bool
@@ -223,6 +224,14 @@ config BR2_TOOLCHAIN_EXTERNAL_C_PLUS_PLUS
 	  support for C++. If you don't know, leave the default value,
 	  Buildroot will tell you if it's correct or not.
 
+config BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS
+	bool "Toolchain has threads support ?"
+	select BR2_TOOLCHAIN_HAS_THREADS
+	help
+	  Enable this option if you want your external toolchain has
+	  support for C++. 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
 
 endif # BR2_TOOLCHAIN_EXTERNAL_CUSTOM
toolchain/toolchain-external/ext-tool.mk +1 -1
index 6f6ec29..e17b0da 100644
@@ -53,7 +53,7 @@ ifeq ($(BR2_INSTALL_LIBSTDCPP),y)
 USR_LIB_EXTERNAL_LIBS+=libstdc++.so
 endif
 
-ifneq ($(BR2_PTHREADS_NONE),y)
+ifeq ($(BR2_TOOLCHAIN_HAS_THREADS),y)
 LIB_EXTERNAL_LIBS+=libpthread.so
 ifeq ($(BR2_PACKAGE_GDB_SERVER),y)
 LIB_EXTERNAL_LIBS+=libthread_db.so
Signature · 1 line
1.7.0.4
#22 lore raw

PATCH 8/15 toolchain: rework thread options

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> The selection of linuxthreads, linuxthreads old or NPTL doesn't make a Thomas> lot of sense for external toolchains. So, instead, we :

Thomas> * Introduce an hidden BR2_TOOLCHAIN_HAS_THREADS option, which must be

 Thomas>    selected by toolchain specific options when thread support is
 Thomas>    available. Package needing to test thread support should use this
 Thomas>    option.

Thomas> * Move the none/linuxthreads/linuxthreads old/NPTL selection to
Thomas> Buildroot internal toolchain configuration.

Thomas> * Add an option in external toolchain to tell if thread support is
Thomas> available or not in the external toolchain.

Thomas> ffmpeg, dmalloc and openvpn are fixed to use the new Thomas> BR2_TOOLCHAIN_HAS_THREADS option. For openvpn, --enable-threads=posix Thomas> is no longer used, as the configure script doesn't even understand Thomas> this option.

Thomas> +++ b/toolchain/toolchain-buildroot/Config.in.2 Thomas> @@ -76,6 +76,32 @@ config BR2_TOOLCHAIN_BUILDROOT_USE_SSP

 Thomas>  	  See http://www.linuxfromscratch.org/hints/downloads/files/ssp.txt
 Thomas>  	  for details.

Thomas> +choice
Thomas> + prompt "Thread library implementation"
Thomas> + default BR2_PTHREADS_OLD
Thomas> + help
Thomas> + If you are building your own toolchain then select the type

This is now only shown if we're building a toolchain, so this part of the help text could go.

 Thomas> +	  of libpthreads you want to use. Not all thread variants work
 Thomas> +	  with all versions of uClibc, the "linuxthreads (stable/old)"
 Thomas> +	  may be a working fallback if you need threading at all.

Thomas> +++ b/toolchain/toolchain-external/Config.in Thomas> @@ -144,6 +144,7 @@ config BR2_TOOLCHAIN_EXTERNAL_GLIBC

 Thomas>  	select BR2_ENABLE_LOCALE
 Thomas>  	select BR2_PROGRAM_INVOCATION
 Thomas>  	select BR2_INSTALL_LIBSTDCPP

Thomas> + select BR2_TOOLCHAIN_HAS_THREADS

Thomas> config BR2_TOOLCHAIN_EXTERNAL_UCLIBC
Thomas> bool
Thomas> @@ -223,6 +224,14 @@ config BR2_TOOLCHAIN_EXTERNAL_C_PLUS_PLUS

 Thomas>  	  support for C++. If you don't know, leave the default value,
 Thomas>  	  Buildroot will tell you if it's correct or not.

Thomas> +config BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS
Thomas> + bool "Toolchain has threads support ?"

No space before '?'

Thomas> + select BR2_TOOLCHAIN_HAS_THREADS
Thomas> + help

 Thomas> +	  Enable this option if you want your external toolchain has
 Thomas> +	  support for C++. If you don't know, leave the default value,
 Thomas> +	  Buildroot will tell you if it's correct or not.

Cut'n'paste? ;) It shouldn't be about C++ and reworded.

Signature · 1 line
Bye, Peter Korsgaard
#23 lore raw

PATCH 8/15 toolchain: rework thread options

Quoted 9 lines · On Sat, 11 Dec 2010 23:45:26 +0100 Peter Korsgaard <[email protected]> wrote:

Thomas> +choice
Thomas> + prompt "Thread library implementation"
Thomas> + default BR2_PTHREADS_OLD
Thomas> + help
Thomas> + If you are building your own toolchain then
Thomas> select the type

This is now only shown if we're building a toolchain, so this part of the help text could go.

Fixed.

Quoted 4 lines

Thomas> +config BR2_TOOLCHAIN_EXTERNAL_HAS_THREADS
Thomas> + bool "Toolchain has threads support ?"

No space before '?'

Ditto.

Quoted 9 lines

Thomas> + select BR2_TOOLCHAIN_HAS_THREADS
Thomas> + help
Thomas> + Enable this option if you want your external
Thomas> toolchain has
Thomas> + support for C++. If you don't know, leave the
Thomas> default value,
Thomas> + Buildroot will tell you if it's correct or not.

Cut'n'paste? ;) It shouldn't be about C++ and reworded.

Indeed, bad cut and paste. Fixed.

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#24 lore raw

PATCH 9/15 toolchain: check that the thread option selection is correct

In reply to Thomas Petazzoni

When an external toolchain is used, the user is responsible for telling Buildroot what the configuration of the toolchain is. In order to make sure that the selection is correct, we already perform checks against the configuration of the uClibc in the selected external toolchain. This commit adds the check for thread support.

---
 toolchain/helpers.mk |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
toolchain/helpers.mk +2 -1
index 24f379d..ee902fa 100644
@@ -192,7 +192,8 @@ check_uclibc = \
 	$(call check_uclibc_feature,__UCLIBC_HAS_RPC__,BR2_INET_RPC,$${UCLIBC_CONFIG_FILE},RPC support) ;\
 	$(call check_uclibc_feature,__UCLIBC_HAS_LOCALE__,BR2_ENABLE_LOCALE,$${UCLIBC_CONFIG_FILE},Locale support) ;\
 	$(call check_uclibc_feature,__UCLIBC_HAS_WCHAR__,BR2_USE_WCHAR,$${UCLIBC_CONFIG_FILE},Wide char support) ;\
-	$(call check_uclibc_feature,__UCLIBC_HAS_PROGRAM_INVOCATION_NAME__,BR2_PROGRAM_INVOCATION,$${UCLIBC_CONFIG_FILE},Program invocation support)
+	$(call check_uclibc_feature,__UCLIBC_HAS_PROGRAM_INVOCATION_NAME__,BR2_PROGRAM_INVOCATION,$${UCLIBC_CONFIG_FILE},Program invocation support) ;\
+	$(call check_uclibc_feature,__UCLIBC_HAS_THREADS__,BR2_TOOLCHAIN_HAS_THREADS,$${UCLIBC_CONFIG_FILE},Thread support)
 
 #
 # Check that the Buildroot configuration of the ABI matches the
Signature · 1 line
1.7.0.4
#25 lore raw

PATCH 10/15 gcc: disable libgomp when thread support is disabled

In reply to Thomas Petazzoni

libgomp needs thread support on the target to build properly.

---
 toolchain/gcc/gcc-uclibc-4.x.mk |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
toolchain/gcc/gcc-uclibc-4.x.mk +1 -1
index 5f69906..cd63992 100644
@@ -139,7 +139,7 @@ GCC_TLS:=--disable-tls
 endif
 
 ifeq ($(BR2_PTHREADS_NONE),y)
-THREADS:=--disable-threads
+THREADS:=--disable-threads --disable-libgomp
 else
 THREADS:=--enable-threads
 endif
Signature · 1 line
1.7.0.4
#26 lore raw

PATCH 11/15 uclibc: add patch for uClibc snapshot to fix build issue

In reply to Thomas Petazzoni

On older glibc (like 2.7), the _SC_V7_* symbols are not defined, making the build of current uClibc fail on Debian Lenny systems and other systems using a fairly old glibc. So we make sure to only use the specs names that are defined by the host C library.

---
 ...tconf-only-use-specs-names-when-available.patch |  131 ++++++++++++++++++++
 1 files changed, 131 insertions(+), 0 deletions(-)
 create mode 100644 toolchain/uClibc/uClibc.snapshot.getconf-only-use-specs-names-when-available.patch
toolchain/uClibc/uClibc.snapshot.getconf-only-use-specs-names-when-available.patch new file +131 -0
new file mode 100644
index 0000000..1086ca7
@@ -0,0 +1,131 @@
+From 1b36ac390724a83455e349fdf6487dd513aeac05 Mon Sep 17 00:00:00 2001
+From: Thomas Petazzoni <[email protected]>
+Date: Sat, 4 Dec 2010 09:58:38 +0100
+Subject: [PATCH] getconf: only use specs names when available
+
+On older glibc (like 2.7), the _SC_V7_* symbols are not defined,
+making the build of current uClibc fail on Debian Lenny systems and
+other systems using a fairly old glibc. So we make sure to only use
+the specs names that are defined by the host C library.
+
+Signed-off-by: Thomas Petazzoni <[email protected]>
+---
+ utils/getconf.c |   72 ++++++++++++++++++++++++++++++++++++------------------
+ 1 files changed, 48 insertions(+), 24 deletions(-)
+
+diff --git a/utils/getconf.c b/utils/getconf.c
+index 81566df..812783e 100644
+--- a/utils/getconf.c
++++ b/utils/getconf.c
+@@ -1014,18 +1014,42 @@ static const struct conf vars[] =
+ 
+ static const struct { const char *name; int num; } specs[] =
+   {
++#ifdef _SC_XBS5_ILP32_OFF32
+     { "XBS5_ILP32_OFF32", _SC_XBS5_ILP32_OFF32 },
++#endif
++#ifdef _SC_XBS5_ILP32_OFFBIG
+     { "XBS5_ILP32_OFFBIG", _SC_XBS5_ILP32_OFFBIG },
++#endif
++#ifdef _SC_XBS5_LP64_OFF64
+     { "XBS5_LP64_OFF64", _SC_XBS5_LP64_OFF64 },
++#endif
++#ifdef _SC_XBS5_LPBIG_OFFBIG
+     { "XBS5_LPBIG_OFFBIG", _SC_XBS5_LPBIG_OFFBIG },
++#endif
++#ifdef _SC_V6_ILP32_OFF32
+     { "POSIX_V6_ILP32_OFF32", _SC_V6_ILP32_OFF32 },
++#endif
++#ifdef _SC_V6_ILP32_OFFBIG
+     { "POSIX_V6_ILP32_OFFBIG", _SC_V6_ILP32_OFFBIG },
++#endif
++#ifdef _SC_V6_LP64_OFF64
+     { "POSIX_V6_LP64_OFF64", _SC_V6_LP64_OFF64 },
++#endif
++#ifdef _SC_V6_LPBIG_OFFBIG
+     { "POSIX_V6_LPBIG_OFFBIG", _SC_V6_LPBIG_OFFBIG },
++#endif
++#ifdef _SC_V7_ILP32_OFF32
+     { "POSIX_V7_ILP32_OFF32", _SC_V7_ILP32_OFF32 },
++#endif
++#ifdef _SC_V7_ILP32_OFFBIG
+     { "POSIX_V7_ILP32_OFFBIG", _SC_V7_ILP32_OFFBIG },
++#endif
++#ifdef _SC_V7_LP64_OFF64
+     { "POSIX_V7_LP64_OFF64", _SC_V7_LP64_OFF64 },
++#endif
++#ifdef _SC_V7_LPBIG_OFFBIG
+     { "POSIX_V7_LPBIG_OFFBIG", _SC_V7_LPBIG_OFFBIG },
++#endif
+   };
+ static const int nspecs = sizeof (specs) / sizeof (specs[0]);
+ 
+@@ -1177,41 +1201,41 @@ environment SPEC.\n\n");
+ 
+       switch (specs[i].num)
+ 	{
+-#ifndef _XBS5_ILP32_OFF32
+-	  case _SC_XBS5_ILP32_OFF32:
++#if !defined(_XBS5_ILP32_OFF32) && defined(_SC_XBS5_ILP32_OFF32)
++	case _SC_XBS5_ILP32_OFF32:
+ #endif
+-#ifndef _XBS5_ILP32_OFFBIG
+-	  case _SC_XBS5_ILP32_OFFBIG:
++#if !defined(_XBS5_ILP32_OFFBIG) && defined(_SC_XBS5_ILP32_OFFBIG)
++	case _SC_XBS5_ILP32_OFFBIG:
+ #endif
+-#ifndef _XBS5_LP64_OFF64
+-	  case _SC_XBS5_LP64_OFF64:
++#if !defined(_XBS5_LP64_OFF64) && defined(_SC_XBS5_LP64_OFF64)
++	case _SC_XBS5_LP64_OFF64:
+ #endif
+-#ifndef _XBS5_LPBIG_OFFBIG
+-	  case _SC_XBS5_LPBIG_OFFBIG:
++#if !defined(_XBS5_LPBIG_OFFBIG) && defined(_SC_XBS5_LPBIG_OFFBIG)
++	case _SC_XBS5_LPBIG_OFFBIG:
+ #endif
+-#ifndef _POSIX_V6_ILP32_OFF32
+-	  case _SC_V6_ILP32_OFF32:
++#if !defined(_POSIX_V6_ILP32_OFF32) && defined(_SC_V6_ILP32_OFF32)
++	case _SC_V6_ILP32_OFF32:
+ #endif
+-#ifndef _POSIX_V6_ILP32_OFFBIG
+-	  case _SC_V6_ILP32_OFFBIG:
++#if !defined(_POSIX_V6_ILP32_OFFBIG) && defined(_SC_V6_ILP32_OFFBIG)
++	case _SC_V6_ILP32_OFFBIG:
+ #endif
+-#ifndef _POSIX_V6_LP64_OFF64
+-	  case _SC_V6_LP64_OFF64:
++#if !defined(_POSIX_V6_LP64_OFF64) && defined(_SC_V6_LP64_OFF64)
++	case _SC_V6_LP64_OFF64:
+ #endif
+-#ifndef _POSIX_V6_LPBIG_OFFBIG
+-	  case _SC_V6_LPBIG_OFFBIG:
++#if !defined(_POSIX_V6_LPBIG_OFFBIG) && defined(_SC_V6_LPBIG_OFFBIG)
++	case _SC_V6_LPBIG_OFFBIG:
+ #endif
+-#ifndef _POSIX_V7_ILP32_OFF32
+-	  case _SC_V7_ILP32_OFF32:
++#if !defined(_POSIX_V7_ILP32_OFF32) && defined(_SC_V7_ILP32_OFF32)
++	case _SC_V7_ILP32_OFF32:
+ #endif
+-#ifndef _POSIX_V7_ILP32_OFFBIG
+-	  case _SC_V7_ILP32_OFFBIG:
++#if !defined(_POSIX_V7_ILP32_OFFBIG) && defined(_SC_V7_ILP32_OFFBIG)
++	case _SC_V7_ILP32_OFFBIG:
+ #endif
+-#ifndef _POSIX_V7_LP64_OFF64
+-	  case _SC_V7_LP64_OFF64:
++#if !defined(_POSIX_V7_LP64_OFF64) && defined(_SC_V7_LP64_OFF64)
++	case _SC_V7_LP64_OFF64:
+ #endif
+-#ifndef _POSIX_V7_LPBIG_OFFBIG
+-	  case _SC_V7_LPBIG_OFFBIG:
++#if !defined(_POSIX_V7_LPBIG_OFFBIG) && defined(_SC_V7_LPBIG_OFFBIG)
++	case _SC_V7_LPBIG_OFFBIG:
+ #endif
+ 	    {
+ 	      const char *args[argc + 3];
+-- 
+1.7.0.4
+
Signature · 1 line
1.7.0.4
#27 lore raw

PATCH 12/15 toolchain: BR2_NEEDS_GETTEXT must be y when uClibc is used

In reply to Thomas Petazzoni

BR2_NEEDS_GETTEXT was already set to y for Buildroot internal toolchain and external toolchains based on uClibc, but wasn't set for toolchains based on uClibC built by the Crosstool-NG backend.

---
 toolchain/toolchain-common.in |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
toolchain/toolchain-common.in +1 -0
index 34d70a0..bdc6e76 100644
@@ -60,6 +60,7 @@ config BR2_NEEDS_GETTEXT
 	bool
 	default y if BR2_TOOLCHAIN_BUILDROOT
 	default y if BR2_TOOLCHAIN_EXTERNAL_UCLIBC
+	default y if BR2_TOOLCHAIN_CTNG_uClibc
 
 config BR2_NEEDS_GETTEXT_IF_LOCALE
 	bool
Signature · 1 line
1.7.0.4
#28 lore raw

PATCH 13/15 toolchain: expose thread options in the Crosstool-NG backend

In reply to Thomas Petazzoni
---
 toolchain/toolchain-crosstool-ng/Config.in       |   32 ++++++++++++++++++++++
 toolchain/toolchain-crosstool-ng/crosstool-ng.mk |   21 ++++++++++++++
 2 files changed, 53 insertions(+), 0 deletions(-)
toolchain/toolchain-crosstool-ng/Config.in +32 -0
index b345297..9d8cb24 100644
@@ -106,4 +106,36 @@ config BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS
 	  C++ language and you want C++ libraries to be installed on
 	  your target system.
 
+choice
+	prompt "Thread library implementation"
+	default BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD \
+		if BR2_TOOLCHAIN_CTNG_uClibc
+	default BR2_TOOLCHAIN_CTNG_THREADS_NPTL		\
+		if BR2_TOOLCHAIN_CTNG_eglibc ||		\
+		   BR2_TOOLCHAIN_CTNG_glibc
+	help
+	  Enable thread support and select thread implementation. Not
+	  all thread variants work with all versions of uClibc, the
+	  "linuxthreads (stable/old)" may be a working fallback if you
+	  need threading at all.
+
+	config BR2_TOOLCHAIN_CTNG_THREADS_NONE
+		bool "none"
+
+	config BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS
+		bool "linuxthreads"
+		select BR2_TOOLCHAIN_HAS_THREADS
+
+	config BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD
+		bool "linuxthreads (stable/old)"
+		select BR2_TOOLCHAIN_HAS_THREADS
+		depends on BR2_TOOLCHAIN_CTNG_uClibc
+
+	config BR2_TOOLCHAIN_CTNG_THREADS_NPTL
+		bool "Native POSIX Threading (NPTL)"
+		select BR2_TOOLCHAIN_HAS_THREADS
+		depends on BR2_TOOLCHAIN_CTNG_eglibc || \
+			   BR2_TOOLCHAIN_CTNG_glibc
+endchoice
+
 endif # BR2_TOOLCHAIN_CTNG
toolchain/toolchain-crosstool-ng/crosstool-ng.mk +21 -0
index 1c5e864..b7f88ee 100644
@@ -213,6 +213,27 @@ else
 CTNG_FIX_DOT_CONFIG_SED += s:^(CT_CC_LANG_CXX)=.*:\# \1 is not set:;
 endif
 
+# Thread implementation selection
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NONE).*:\# \2 is not set:;
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_LINUXTHREADS).*:\# \2 is not set:;
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NPTL).*:\# \2 is not set:;
+ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS))$(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_LINUXTHREADS).*:\2=y:;
+ ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_uClibc)),)
+  ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_NEW).*:\# \2 is not set:;
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_OLD).*:\2=y:;
+  else
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_OLD).*:\# \2 is not settted:;
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_NEW).*:\2=y:;
+  endif
+ endif
+else ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_NPTL)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NPTL).*:\2=y:;
+else ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_NONE)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NONE).*:\2=y:;
+endif
+
 #--------------
 # And the specials for paths
 CTNG_FIX_DOT_CONFIG_PATHS_SED += s:^(CT_PREFIX_DIR)=.*:\1="$(TOOLCHAIN_DIR)":;
Signature · 1 line
1.7.0.4
#29 lore raw

PATCH 13/15 toolchain: expose thread options in the Crosstool-NG backend

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> Signed-off-by: Thomas Petazzoni <[email protected]> Thomas> ---

 Thomas>  toolchain/toolchain-crosstool-ng/Config.in       |   32 ++++++++++++++++++++++
 Thomas>  toolchain/toolchain-crosstool-ng/crosstool-ng.mk |   21 ++++++++++++++

Thomas> 2 files changed, 53 insertions(+), 0 deletions(-)

Thomas> diff --git a/toolchain/toolchain-crosstool-ng/Config.in b/toolchain/toolchain-crosstool-ng/Config.in Thomas> index b345297..9d8cb24 100644
Thomas> --- a/toolchain/toolchain-crosstool-ng/Config.in
Thomas> +++ b/toolchain/toolchain-crosstool-ng/Config.in
Thomas> @@ -106,4 +106,36 @@ config BR2_TOOLCHAIN_CTNG_C_PLUS_PLUS

 Thomas>  	  C++ language and you want C++ libraries to be installed on
 Thomas>  	  your target system.

Thomas> +choice
Thomas> + prompt "Thread library implementation"
Thomas> + default BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD \
Thomas> + if BR2_TOOLCHAIN_CTNG_uClibc
Thomas> + default BR2_TOOLCHAIN_CTNG_THREADS_NPTL \
Thomas> + if BR2_TOOLCHAIN_CTNG_eglibc || \
Thomas> + BR2_TOOLCHAIN_CTNG_glibc
Thomas> + help

 Thomas> +	  Enable thread support and select thread implementation. Not
 Thomas> +	  all thread variants work with all versions of uClibc, the
 Thomas> +	  "linuxthreads (stable/old)" may be a working fallback if you
 Thomas> +	  need threading at all.

Thomas> +
Thomas> + config BR2_TOOLCHAIN_CTNG_THREADS_NONE
Thomas> + bool "none"

So you can build a glibc/eglibc ctng toolchain without threading support? Then we shouldn't unconditionally enable thread support for external glibc toolchains (in patch 8).

Signature · 1 line
Bye, Peter Korsgaard
#30 lore raw

PATCH 13/15 toolchain: expose thread options in the Crosstool-NG backend

In reply to Thomas Petazzoni

Thomas, All,

On Sunday 05 December 2010 10:25:40 Thomas Petazzoni wrote:
[--SNIP--]

Quoted 10 lines
toolchain/toolchain-crosstool-ng/crosstool-ng.mk +2 -0
index 1c5e864..b7f88ee 100644
@@ -213,6 +213,27 @@ else
 CTNG_FIX_DOT_CONFIG_SED += s:^(CT_CC_LANG_CXX)=.*:\# \1 is not set:;
 endif
 
+# Thread implementation selection
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NONE).*:\# \2 is not set:;

toolchain/toolchain-crosstool-ng/crosstool-ng.mk else · line 227

Quoted 10 lines
toolchain/toolchain-crosstool-ng/crosstool-ng.mk +10 -0
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_LINUXTHREADS).*:\# \2 is not set:;
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NPTL).*:\# \2 is not set:;
+ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS))$(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_LINUXTHREADS).*:\2=y:;
+ ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_uClibc)),)
+  ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_PTHREADS_OLD)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_NEW).*:\# \2 is not set:;
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_OLD).*:\2=y:;
+  else
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_OLD).*:\# \2 is not settted:;

'settted' ?

toolchain/toolchain-crosstool-ng/crosstool-ng.mk else · line 239

Quoted 12 lines
toolchain/toolchain-crosstool-ng/crosstool-ng.mk +9 -0
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_NEW).*:\2=y:;
+  endif
+ endif
+else ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_NPTL)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NPTL).*:\2=y:;
+else ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_NONE)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NONE).*:\2=y:;
+endif
+
 #--------------
 # And the specials for paths
 CTNG_FIX_DOT_CONFIG_PATHS_SED += s:^(CT_PREFIX_DIR)=.*:\1="$(TOOLCHAIN_DIR)":;

I am not so sure what to say about this one...

There is already a place where we check the C library, on line 223 (upstream) and I would say that should go there.

I'll look at it tonight...

Regards,
Yann E. MORIN.

Signature · 6 lines
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'
#31 lore raw

PATCH 13/15 toolchain: expose thread options in the Crosstool-NG backend

Quoted 3 lines · On Sun, 12 Dec 2010 11:37:22 +0100 "Yann E. MORIN"…
Quoted 1 line

s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_OLD).*:\# \2 is not settted:;

'settted' ?

Leftover from a crappy debugging session. Thanks for spotting this, I fixed it.

Quoted 17 lines
Quoted 13 lines

+CTNG_FIX_DOT_CONFIG_SED +=
s:^(|\# )(CT_LIBC_UCLIBC_LNXTHRD_NEW).*:\2=y:;

toolchain/toolchain-crosstool-ng/crosstool-ng.mk +8 -0
+  endif
+ endif
+else ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_NPTL)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NPTL).*:\2=y:;
+else ifneq ($(call qstrip,$(BR2_TOOLCHAIN_CTNG_THREADS_NONE)),)
+CTNG_FIX_DOT_CONFIG_SED += s:^(|\# )(CT_THREADS_NONE).*:\2=y:;
+endif
+
 #--------------
 # And the specials for paths
 CTNG_FIX_DOT_CONFIG_PATHS_SED +=
Quoted 1 line

s:^(CT_PREFIX_DIR)=.*:\1="$(TOOLCHAIN_DIR)":;

I am not so sure what to say about this one...

Quoted 2 lines

There is already a place where we check the C library, on line 223 (upstream) and I would say that should go there.

I am not sure how it would fit in the existing checks. Can we fix this as a separate patchset ?

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
#32 lore raw

PATCH 14/15 toolchain: more verbose in crosstool-ng backend

In reply to Thomas Petazzoni

By default, Crosstool-NG hides all the details of the toolchain build process. However, this is inconsistent with the typical Buildroot build, and the rotating progress thing makes log file resulting from redirecting the Buildroot output a bit unusable. Therefore, switch the Crosstool-NG configuration to a more verbose default.

---
 .../toolchain-crosstool-ng/crosstool-ng.config     |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)
toolchain/toolchain-crosstool-ng/crosstool-ng.config +4 -4
index d5ee244..506f78a 100644
@@ -72,12 +72,12 @@ CT_CONFIG_SHELL_CUSTOM_PATH="/bin/sh"
 # CT_LOG_ERROR is not set
 # CT_LOG_WARN is not set
 # CT_LOG_INFO is not set
-CT_LOG_EXTRA=y
-# CT_LOG_ALL is not set
+# CT_LOG_EXTRA is not set
+CT_LOG_ALL=y
 # CT_LOG_DEBUG is not set
-CT_LOG_LEVEL_MAX="EXTRA"
+CT_LOG_LEVEL_MAX="ALL"
 # CT_LOG_SEE_TOOLS_WARN is not set
-CT_LOG_PROGRESS_BAR=y
+# CT_LOG_PROGRESS_BAR is not set
 CT_LOG_TO_FILE=y
 CT_LOG_FILE_COMPRESS=y
 
Signature · 1 line
1.7.0.4
#33 lore raw

PATCH 15/15 package: pass TARGET_MAKE_ENV at install/clean/uninstall steps

In reply to Thomas Petazzoni

TARGET_MAKE_ENV contains the PATH to the host tools built in $(HOST_DIR), and not having this directory in the PATH while installing packages may cause troubles.

The specific example that trigerred the problem is the following one. libglib2 ./configure tries to find a awk version, either gawk or mawk. On my limited testing system, only mawk is available, so normally, libglib2 detects mawk and uses it, and it works perfectly.

However, when host-gawk has been compiled before libglib2 (which happens for example if you have used the Crosstool-NG backend), gawk is well detected in $(HOST_DIR)/usr/bin at ./configure time ($(HOST_DIR)/usr/bin is in the PATH at ./configure time, thanks to TARGET_MAKE_ENV). But at install time, when libglib2 uses gawk, it doesn't find it anymore because it is not in the PATH.

This also makes things coherent with host installation, in which $(HOST_MAKE_ENV) was already used.

---
 package/Makefile.autotools.in |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)
package/Makefile.autotools.in +5 -5
index 7d04e44..de0a355 100644
@@ -210,7 +210,7 @@ endif
 #
 ifndef $(2)_INSTALL_STAGING_CMDS
 define $(2)_INSTALL_STAGING_CMDS
-	$$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_INSTALL_STAGING_OPT) -C $$($$(PKG)_SRCDIR)
+	$(TARGET_MAKE_ENV) $$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_INSTALL_STAGING_OPT) -C $$($$(PKG)_SRCDIR)
 	for i in $$$$(find $(STAGING_DIR)/usr/lib* -name "*.la"); do \
 		cp -f $$$$i $$$$i~; \
 		$$(SED) "s:\(['= ]\)/usr:\\1$(STAGING_DIR)/usr:g" $$$$i; \
@@ -224,7 +224,7 @@ endif
 #
 ifndef $(2)_INSTALL_TARGET_CMDS
 define $(2)_INSTALL_TARGET_CMDS
-	$$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_INSTALL_TARGET_OPT) -C $$($$(PKG)_SRCDIR)
+	$(TARGET_MAKE_ENV) $$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_INSTALL_TARGET_OPT) -C $$($$(PKG)_SRCDIR)
 endef
 endif
 
@@ -234,7 +234,7 @@ endif
 #
 ifndef $(2)_CLEAN_CMDS
 define $(2)_CLEAN_CMDS
-	-$$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE)  $$($$(PKG)_CLEAN_OPT) -C $$($$(PKG)_SRCDIR)
+	-$(TARGET_MAKE_ENV) $$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE)  $$($$(PKG)_CLEAN_OPT) -C $$($$(PKG)_SRCDIR)
 endef
 endif
 
@@ -244,7 +244,7 @@ endif
 #
 ifndef $(2)_UNINSTALL_STAGING_CMDS
 define $(2)_UNINSTALL_STAGING_CMDS
-	$$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_UNINSTALL_STAGING_OPT) -C $$($$(PKG)_SRCDIR)
+	$(TARGET_MAKE_ENV) $$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_UNINSTALL_STAGING_OPT) -C $$($$(PKG)_SRCDIR)
 endef
 endif
 
@@ -254,7 +254,7 @@ endif
 #
 ifndef $(2)_UNINSTALL_TARGET_CMDS
 define $(2)_UNINSTALL_TARGET_CMDS
-	$$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_UNINSTALL_TARGET_OPT) -C $$($$(PKG)_SRCDIR)
+	$(TARGET_MAKE_ENV) $$($$(PKG)_MAKE_ENV) $$($$(PKG)_MAKE) $$($$(PKG)_UNINSTALL_TARGET_OPT) -C $$($$(PKG)_SRCDIR)
 endef
 endif
 
Signature · 1 line
1.7.0.4
#34 lore raw
In reply to Thomas Petazzoni
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> Hello,
Thomas> Here is a set of patches that mainly improves the toolchain Thomas> configuration and support, with two majors directions :

Thomas> * Turn BR2_LARGEFILE, BR2_INET_RPC, BR2_INET_RPC and similar options

 Thomas>    into hidden options, and let the toolchain backends present them as
 Thomas>    they wish. For example for RPC, xthe Buildroot backend says "Do you
 Thomas>    want RPC ?" while the external toolchain backend doesn't show the
 Thomas>    option when glibc is used, or present an option "Does your
 Thomas>    toolchain has RPC ?" when uclibc is used.
 Thomas>    So really those options should now be called
 Thomas>    BR2_TOOLCHAIN_HAS_LARGEFILE, BR2_TOOLCHAIN_HAS_INET_RPC, etc, but
 Thomas>    we kept the old names in this patch series because changing them
 Thomas>    impacts all packages. This will be done later.
 Thomas>    A first commit handles largefile, rpc, ipv6, locale and program
 Thomas>    invocation options, and some later commits handle the C++ option
 Thomas>    and thread options.

Thomas> * Extend external toolchain with "profiles", i.e well-known external

 Thomas>    toolchains. For those toolchains, Buildroot can automatically
 Thomas>    download and install them and will automatically select the
 Thomas>    appropriate configuration options (C library used, RPC, IPv6,
 Thomas>    locale, etc.). So an user can know tell Buildroot to use a
 Thomas>    CodeSourcery toolchain, and Buildroot will do all the job. Such
 Thomas>    toolchains are extracted in $(O)/external-toolchain, but my intent
 Thomas>    is to move it to $(O)/toolchain once Gustavoz's toolchain cleanup
 Thomas>    work is done.

Thomas> This work keeps the ability of doing some more manual things with
Thomas> external toolchains:

Thomas> - For an external toolchain profile, you can tell Buildroot *not*
Thomas> to download it, and in that case, you must provide Buildroot
Thomas> with the location of the external toolchain.

Thomas> - You keep the ability to specify a custom profile, where you
Thomas> will have to tell Buildroot everything: location of the
Thomas> toolchain, toolchain prefix, C library used, etc.

Thomas> In my opinion, those two directions really make the external Thomas> toolchains easier and simpler to use. One topic left is the gdb Thomas> options, but I decided not to change it, in order to not conflict too Thomas> much with Gustavoz's work.

Thanks, looks good. Care to fix the few issues I pointed out and send a new pull request? Then I'll pull it.

Signature · 1 line
Bye, Peter Korsgaard